diff options
Diffstat (limited to 'drivers/block')
39 files changed, 3035 insertions, 10788 deletions
diff --git a/drivers/block/DAC960.c b/drivers/block/DAC960.c index 245a879b036e..255591ab3716 100644 --- a/drivers/block/DAC960.c +++ b/drivers/block/DAC960.c @@ -1678,9 +1678,12 @@ static bool DAC960_V1_ReadControllerConfiguration(DAC960_Controller_T Enquiry2->FirmwareID.FirmwareType = '0'; Enquiry2->FirmwareID.TurnID = 0; } - sprintf(Controller->FirmwareVersion, "%d.%02d-%c-%02d", - Enquiry2->FirmwareID.MajorVersion, Enquiry2->FirmwareID.MinorVersion, - Enquiry2->FirmwareID.FirmwareType, Enquiry2->FirmwareID.TurnID); + snprintf(Controller->FirmwareVersion, sizeof(Controller->FirmwareVersion), + "%d.%02d-%c-%02d", + Enquiry2->FirmwareID.MajorVersion, + Enquiry2->FirmwareID.MinorVersion, + Enquiry2->FirmwareID.FirmwareType, + Enquiry2->FirmwareID.TurnID); if (!((Controller->FirmwareVersion[0] == '5' && strcmp(Controller->FirmwareVersion, "5.06") >= 0) || (Controller->FirmwareVersion[0] == '4' && @@ -6588,7 +6591,8 @@ static void DAC960_CreateProcEntries(DAC960_Controller_T *Controller) &dac960_proc_fops); } - sprintf(Controller->ControllerName, "c%d", Controller->ControllerNumber); + snprintf(Controller->ControllerName, sizeof(Controller->ControllerName), + "c%d", Controller->ControllerNumber); ControllerProcEntry = proc_mkdir(Controller->ControllerName, DAC960_ProcDirectoryEntry); proc_create_data("initial_status", 0, ControllerProcEntry, &dac960_initial_status_proc_fops, Controller); diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig index 8ddc98279c8f..4a438b8abe27 100644 --- a/drivers/block/Kconfig +++ b/drivers/block/Kconfig @@ -17,6 +17,7 @@ if BLK_DEV config BLK_DEV_NULL_BLK tristate "Null test block driver" + depends on CONFIGFS_FS config BLK_DEV_FD tristate "Normal floppy disk support" @@ -111,33 +112,6 @@ source "drivers/block/mtip32xx/Kconfig" source "drivers/block/zram/Kconfig" -config BLK_CPQ_CISS_DA - tristate "Compaq Smart Array 5xxx support" - depends on PCI - select CHECK_SIGNATURE - select BLK_SCSI_REQUEST - help - This is the driver for Compaq Smart Array 5xxx controllers. - Everyone using these boards should say Y here. - See <file:Documentation/blockdev/cciss.txt> for the current list of - boards supported by this driver, and for further information - on the use of this driver. - -config CISS_SCSI_TAPE - bool "SCSI tape drive support for Smart Array 5xxx" - depends on BLK_CPQ_CISS_DA && PROC_FS - depends on SCSI=y || SCSI=BLK_CPQ_CISS_DA - help - When enabled (Y), this option allows SCSI tape drives and SCSI medium - changers (tape robots) to be accessed via a Compaq 5xxx array - controller. (See <file:Documentation/blockdev/cciss.txt> for more details.) - - "SCSI support" and "SCSI tape support" must also be enabled for this - option to work. - - When this option is disabled (N), the SCSI portion of the driver - is not compiled. - config BLK_DEV_DAC960 tristate "Mylex DAC960/DAC1100 PCI RAID Controller support" depends on PCI @@ -470,7 +444,7 @@ config VIRTIO_BLK depends on VIRTIO ---help--- This is the virtual block driver for virtio. It can be used with - lguest or QEMU based VMMs (like KVM or Xen). Say Y or M. + QEMU based VMMs (like KVM or Xen). Say Y or M. config VIRTIO_BLK_SCSI bool "SCSI passthrough request for the Virtio block driver" diff --git a/drivers/block/Makefile b/drivers/block/Makefile index ec8c36897b75..1f456d86a190 100644 --- a/drivers/block/Makefile +++ b/drivers/block/Makefile @@ -15,7 +15,6 @@ obj-$(CONFIG_ATARI_FLOPPY) += ataflop.o obj-$(CONFIG_AMIGA_Z2RAM) += z2ram.o obj-$(CONFIG_BLK_DEV_RAM) += brd.o obj-$(CONFIG_BLK_DEV_LOOP) += loop.o -obj-$(CONFIG_BLK_CPQ_CISS_DA) += cciss.o obj-$(CONFIG_BLK_DEV_DAC960) += DAC960.o obj-$(CONFIG_XILINX_SYSACE) += xsysace.o obj-$(CONFIG_CDROM_PKTCDVD) += pktcdvd.o diff --git a/drivers/block/brd.c b/drivers/block/brd.c index 104b71c0490d..bbd0d186cfc0 100644 --- a/drivers/block/brd.c +++ b/drivers/block/brd.c @@ -294,14 +294,13 @@ out: static blk_qc_t brd_make_request(struct request_queue *q, struct bio *bio) { - struct block_device *bdev = bio->bi_bdev; - struct brd_device *brd = bdev->bd_disk->private_data; + struct brd_device *brd = bio->bi_disk->private_data; struct bio_vec bvec; sector_t sector; struct bvec_iter iter; sector = bio->bi_iter.bi_sector; - if (bio_end_sector(bio) > get_capacity(bdev->bd_disk)) + if (bio_end_sector(bio) > get_capacity(bio->bi_disk)) goto io_error; bio_for_each_segment(bvec, bio, iter) { @@ -326,7 +325,11 @@ static int brd_rw_page(struct block_device *bdev, sector_t sector, struct page *page, bool is_write) { struct brd_device *brd = bdev->bd_disk->private_data; - int err = brd_do_bvec(brd, page, PAGE_SIZE, 0, is_write, sector); + int err; + + if (PageTransHuge(page)) + return -ENOTSUPP; + err = brd_do_bvec(brd, page, PAGE_SIZE, 0, is_write, sector); page_endio(page, is_write, err); return err; } diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c deleted file mode 100644 index 678af946be30..000000000000 --- a/drivers/block/cciss.c +++ /dev/null @@ -1,5415 +0,0 @@ -/* - * Disk Array driver for HP Smart Array controllers. - * (C) Copyright 2000, 2007 Hewlett-Packard Development Company, L.P. - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; version 2 of the License. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA - * 02111-1307, USA. - * - * Questions/Comments/Bugfixes to iss_storagedev@hp.com - * - */ - -#include <linux/module.h> -#include <linux/interrupt.h> -#include <linux/types.h> -#include <linux/pci.h> -#include <linux/pci-aspm.h> -#include <linux/kernel.h> -#include <linux/slab.h> -#include <linux/delay.h> -#include <linux/major.h> -#include <linux/fs.h> -#include <linux/bio.h> -#include <linux/blkpg.h> -#include <linux/timer.h> -#include <linux/proc_fs.h> -#include <linux/seq_file.h> -#include <linux/init.h> -#include <linux/jiffies.h> -#include <linux/hdreg.h> -#include <linux/spinlock.h> -#include <linux/compat.h> -#include <linux/mutex.h> -#include <linux/bitmap.h> -#include <linux/io.h> -#include <linux/uaccess.h> - -#include <linux/dma-mapping.h> -#include <linux/blkdev.h> -#include <linux/genhd.h> -#include <linux/completion.h> -#include <scsi/scsi.h> -#include <scsi/sg.h> -#include <scsi/scsi_ioctl.h> -#include <scsi/scsi_request.h> -#include <linux/cdrom.h> -#include <linux/scatterlist.h> -#include <linux/kthread.h> - -#define CCISS_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin)) -#define DRIVER_NAME "HP CISS Driver (v 3.6.26)" -#define DRIVER_VERSION CCISS_DRIVER_VERSION(3, 6, 26) - -/* Embedded module documentation macros - see modules.h */ -MODULE_AUTHOR("Hewlett-Packard Company"); -MODULE_DESCRIPTION("Driver for HP Smart Array Controllers"); -MODULE_SUPPORTED_DEVICE("HP Smart Array Controllers"); -MODULE_VERSION("3.6.26"); -MODULE_LICENSE("GPL"); -static int cciss_tape_cmds = 6; -module_param(cciss_tape_cmds, int, 0644); -MODULE_PARM_DESC(cciss_tape_cmds, - "number of commands to allocate for tape devices (default: 6)"); -static int cciss_simple_mode; -module_param(cciss_simple_mode, int, S_IRUGO|S_IWUSR); -MODULE_PARM_DESC(cciss_simple_mode, - "Use 'simple mode' rather than 'performant mode'"); - -static int cciss_allow_hpsa; -module_param(cciss_allow_hpsa, int, S_IRUGO|S_IWUSR); -MODULE_PARM_DESC(cciss_allow_hpsa, - "Prevent cciss driver from accessing hardware known to be " - " supported by the hpsa driver"); - -static DEFINE_MUTEX(cciss_mutex); -static struct proc_dir_entry *proc_cciss; - -#include "cciss_cmd.h" -#include "cciss.h" -#include <linux/cciss_ioctl.h> - -/* define the PCI info for the cards we can control */ -static const struct pci_device_id cciss_pci_device_id[] = { - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISS, 0x0E11, 0x4070}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, 0x0E11, 0x4080}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, 0x0E11, 0x4082}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, 0x0E11, 0x4083}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x4091}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409A}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409B}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409C}, - {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409D}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSA, 0x103C, 0x3225}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSC, 0x103C, 0x3223}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSC, 0x103C, 0x3234}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSC, 0x103C, 0x3235}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSD, 0x103C, 0x3211}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSD, 0x103C, 0x3212}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSD, 0x103C, 0x3213}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSD, 0x103C, 0x3214}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSD, 0x103C, 0x3215}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSC, 0x103C, 0x3237}, - {PCI_VENDOR_ID_HP, PCI_DEVICE_ID_HP_CISSC, 0x103C, 0x323D}, - {0,} -}; - -MODULE_DEVICE_TABLE(pci, cciss_pci_device_id); - -/* board_id = Subsystem Device ID & Vendor ID - * product = Marketing Name for the board - * access = Address of the struct of function pointers - */ -static struct board_type products[] = { - {0x40700E11, "Smart Array 5300", &SA5_access}, - {0x40800E11, "Smart Array 5i", &SA5B_access}, - {0x40820E11, "Smart Array 532", &SA5B_access}, - {0x40830E11, "Smart Array 5312", &SA5B_access}, - {0x409A0E11, "Smart Array 641", &SA5_access}, - {0x409B0E11, "Smart Array 642", &SA5_access}, - {0x409C0E11, "Smart Array 6400", &SA5_access}, - {0x409D0E11, "Smart Array 6400 EM", &SA5_access}, - {0x40910E11, "Smart Array 6i", &SA5_access}, - {0x3225103C, "Smart Array P600", &SA5_access}, - {0x3223103C, "Smart Array P800", &SA5_access}, - {0x3234103C, "Smart Array P400", &SA5_access}, - {0x3235103C, "Smart Array P400i", &SA5_access}, - {0x3211103C, "Smart Array E200i", &SA5_access}, - {0x3212103C, "Smart Array E200", &SA5_access}, - {0x3213103C, "Smart Array E200i", &SA5_access}, - {0x3214103C, "Smart Array E200i", &SA5_access}, - {0x3215103C, "Smart Array E200i", &SA5_access}, - {0x3237103C, "Smart Array E500", &SA5_access}, - {0x323D103C, "Smart Array P700m", &SA5_access}, -}; - -/* How long to wait (in milliseconds) for board to go into simple mode */ -#define MAX_CONFIG_WAIT 30000 -#define MAX_IOCTL_CONFIG_WAIT 1000 - -/*define how many times we will try a command because of bus resets */ -#define MAX_CMD_RETRIES 3 - -#define MAX_CTLR 32 - -/* Originally cciss driver only supports 8 major numbers */ -#define MAX_CTLR_ORIG 8 - -static ctlr_info_t *hba[MAX_CTLR]; - -static struct task_struct *cciss_scan_thread; -static DEFINE_MUTEX(scan_mutex); -static LIST_HEAD(scan_q); - -static void do_cciss_request(struct request_queue *q); -static irqreturn_t do_cciss_intx(int irq, void *dev_id); -static irqreturn_t do_cciss_msix_intr(int irq, void *dev_id); -static int cciss_open(struct block_device *bdev, fmode_t mode); -static int cciss_unlocked_open(struct block_device *bdev, fmode_t mode); -static void cciss_release(struct gendisk *disk, fmode_t mode); -static int cciss_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long arg); -static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo); - -static int cciss_revalidate(struct gendisk *disk); -static int rebuild_lun_table(ctlr_info_t *h, int first_time, int via_ioctl); -static int deregister_disk(ctlr_info_t *h, int drv_index, - int clear_all, int via_ioctl); - -static void cciss_read_capacity(ctlr_info_t *h, int logvol, - sector_t *total_size, unsigned int *block_size); -static void cciss_read_capacity_16(ctlr_info_t *h, int logvol, - sector_t *total_size, unsigned int *block_size); -static void cciss_geometry_inquiry(ctlr_info_t *h, int logvol, - sector_t total_size, - unsigned int block_size, InquiryData_struct *inq_buff, - drive_info_struct *drv); -static void cciss_interrupt_mode(ctlr_info_t *); -static int cciss_enter_simple_mode(struct ctlr_info *h); -static void start_io(ctlr_info_t *h); -static int sendcmd_withirq(ctlr_info_t *h, __u8 cmd, void *buff, size_t size, - __u8 page_code, unsigned char scsi3addr[], - int cmd_type); -static int sendcmd_withirq_core(ctlr_info_t *h, CommandList_struct *c, - int attempt_retry); -static int process_sendcmd_error(ctlr_info_t *h, CommandList_struct *c); - -static int add_to_scan_list(struct ctlr_info *h); -static int scan_thread(void *data); -static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c); -static void cciss_hba_release(struct device *dev); -static void cciss_device_release(struct device *dev); -static void cciss_free_gendisk(ctlr_info_t *h, int drv_index); -static void cciss_free_drive_info(ctlr_info_t *h, int drv_index); -static inline u32 next_command(ctlr_info_t *h); -static int cciss_find_cfg_addrs(struct pci_dev *pdev, void __iomem *vaddr, - u32 *cfg_base_addr, u64 *cfg_base_addr_index, - u64 *cfg_offset); -static int cciss_pci_find_memory_BAR(struct pci_dev *pdev, - unsigned long *memory_bar); -static inline u32 cciss_tag_discard_error_bits(ctlr_info_t *h, u32 tag); -static int write_driver_ver_to_cfgtable(CfgTable_struct __iomem *cfgtable); - -/* performant mode helper functions */ -static void calc_bucket_map(int *bucket, int num_buckets, int nsgs, - int *bucket_map); -static void cciss_put_controller_into_performant_mode(ctlr_info_t *h); - -#ifdef CONFIG_PROC_FS -static void cciss_procinit(ctlr_info_t *h); -#else -static void cciss_procinit(ctlr_info_t *h) -{ -} -#endif /* CONFIG_PROC_FS */ - -#ifdef CONFIG_COMPAT -static int cciss_compat_ioctl(struct block_device *, fmode_t, - unsigned, unsigned long); -#endif - -static const struct block_device_operations cciss_fops = { - .owner = THIS_MODULE, - .open = cciss_unlocked_open, - .release = cciss_release, - .ioctl = cciss_ioctl, - .getgeo = cciss_getgeo, -#ifdef CONFIG_COMPAT - .compat_ioctl = cciss_compat_ioctl, -#endif - .revalidate_disk = cciss_revalidate, -}; - -/* set_performant_mode: Modify the tag for cciss performant - * set bit 0 for pull model, bits 3-1 for block fetch - * register number - */ -static void set_performant_mode(ctlr_info_t *h, CommandList_struct *c) -{ - if (likely(h->transMethod & CFGTBL_Trans_Performant)) - c->busaddr |= 1 | (h->blockFetchTable[c->Header.SGList] << 1); -} - -/* - * Enqueuing and dequeuing functions for cmdlists. - */ -static inline void addQ(struct list_head *list, CommandList_struct *c) -{ - list_add_tail(&c->list, list); -} - -static inline void removeQ(CommandList_struct *c) -{ - /* - * After kexec/dump some commands might still - * be in flight, which the firmware will try - * to complete. Resetting the firmware doesn't work - * with old fw revisions, so we have to mark - * them off as 'stale' to prevent the driver from - * falling over. - */ - if (WARN_ON(list_empty(&c->list))) { - c->cmd_type = CMD_MSG_STALE; - return; - } - - list_del_init(&c->list); -} - -static void enqueue_cmd_and_start_io(ctlr_info_t *h, - CommandList_struct *c) -{ - unsigned long flags; - set_performant_mode(h, c); - spin_lock_irqsave(&h->lock, flags); - addQ(&h->reqQ, c); - h->Qdepth++; - if (h->Qdepth > h->maxQsinceinit) - h->maxQsinceinit = h->Qdepth; - start_io(h); - spin_unlock_irqrestore(&h->lock, flags); -} - -static void cciss_free_sg_chain_blocks(SGDescriptor_struct **cmd_sg_list, - int nr_cmds) -{ - int i; - - if (!cmd_sg_list) - return; - for (i = 0; i < nr_cmds; i++) { - kfree(cmd_sg_list[i]); - cmd_sg_list[i] = NULL; - } - kfree(cmd_sg_list); -} - -static SGDescriptor_struct **cciss_allocate_sg_chain_blocks( - ctlr_info_t *h, int chainsize, int nr_cmds) -{ - int j; - SGDescriptor_struct **cmd_sg_list; - - if (chainsize <= 0) - return NULL; - - cmd_sg_list = kmalloc(sizeof(*cmd_sg_list) * nr_cmds, GFP_KERNEL); - if (!cmd_sg_list) - return NULL; - - /* Build up chain blocks for each command */ - for (j = 0; j < nr_cmds; j++) { - /* Need a block of chainsized s/g elements. */ - cmd_sg_list[j] = kmalloc((chainsize * - sizeof(*cmd_sg_list[j])), GFP_KERNEL); - if (!cmd_sg_list[j]) { - dev_err(&h->pdev->dev, "Cannot get memory " - "for s/g chains.\n"); - goto clean; - } - } - return cmd_sg_list; -clean: - cciss_free_sg_chain_blocks(cmd_sg_list, nr_cmds); - return NULL; -} - -static void cciss_unmap_sg_chain_block(ctlr_info_t *h, CommandList_struct *c) -{ - SGDescriptor_struct *chain_sg; - u64bit temp64; - - if (c->Header.SGTotal <= h->max_cmd_sgentries) - return; - - chain_sg = &c->SG[h->max_cmd_sgentries - 1]; - temp64.val32.lower = chain_sg->Addr.lower; - temp64.val32.upper = chain_sg->Addr.upper; - pci_unmap_single(h->pdev, temp64.val, chain_sg->Len, PCI_DMA_TODEVICE); -} - -static int cciss_map_sg_chain_block(ctlr_info_t *h, CommandList_struct *c, - SGDescriptor_struct *chain_block, int len) -{ - SGDescriptor_struct *chain_sg; - u64bit temp64; - - chain_sg = &c->SG[h->max_cmd_sgentries - 1]; - chain_sg->Ext = CCISS_SG_CHAIN; - chain_sg->Len = len; - temp64.val = pci_map_single(h->pdev, chain_block, len, - PCI_DMA_TODEVICE); - if (dma_mapping_error(&h->pdev->dev, temp64.val)) { - dev_warn(&h->pdev->dev, - "%s: error mapping chain block for DMA\n", - __func__); - return -1; - } - chain_sg->Addr.lower = temp64.val32.lower; - chain_sg->Addr.upper = temp64.val32.upper; - - return 0; -} - -#include "cciss_scsi.c" /* For SCSI tape support */ - -static const char *raid_label[] = { "0", "4", "1(1+0)", "5", "5+1", "ADG", - "UNKNOWN" -}; -#define RAID_UNKNOWN (ARRAY_SIZE(raid_label)-1) - -#ifdef CONFIG_PROC_FS - -/* - * Report information about this controller. - */ -#define ENG_GIG 1000000000 -#define ENG_GIG_FACTOR (ENG_GIG/512) -#define ENGAGE_SCSI "engage scsi" - -static void cciss_seq_show_header(struct seq_file *seq) -{ - ctlr_info_t *h = seq->private; - - seq_printf(seq, "%s: HP %s Controller\n" - "Board ID: 0x%08lx\n" - "Firmware Version: %c%c%c%c\n" - "IRQ: %d\n" - "Logical drives: %d\n" - "Current Q depth: %d\n" - "Current # commands on controller: %d\n" - "Max Q depth since init: %d\n" - "Max # commands on controller since init: %d\n" - "Max SG entries since init: %d\n", - h->devname, - h->product_name, - (unsigned long)h->board_id, - h->firm_ver[0], h->firm_ver[1], h->firm_ver[2], - h->firm_ver[3], (unsigned int)h->intr[h->intr_mode], - h->num_luns, - h->Qdepth, h->commands_outstanding, - h->maxQsinceinit, h->max_outstanding, h->maxSG); - -#ifdef CONFIG_CISS_SCSI_TAPE - cciss_seq_tape_report(seq, h); -#endif /* CONFIG_CISS_SCSI_TAPE */ -} - -static void *cciss_seq_start(struct seq_file *seq, loff_t *pos) -{ - ctlr_info_t *h = seq->private; - unsigned long flags; - - /* prevent displaying bogus info during configuration - * or deconfiguration of a logical volume - */ - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) { - spin_unlock_irqrestore(&h->lock, flags); - return ERR_PTR(-EBUSY); - } - h->busy_configuring = 1; - spin_unlock_irqrestore(&h->lock, flags); - - if (*pos == 0) - cciss_seq_show_header(seq); - - return pos; -} - -static int cciss_seq_show(struct seq_file *seq, void *v) -{ - sector_t vol_sz, vol_sz_frac; - ctlr_info_t *h = seq->private; - unsigned ctlr = h->ctlr; - loff_t *pos = v; - drive_info_struct *drv = h->drv[*pos]; - - if (*pos > h->highest_lun) - return 0; - - if (drv == NULL) /* it's possible for h->drv[] to have holes. */ - return 0; - - if (drv->heads == 0) - return 0; - - vol_sz = drv->nr_blocks; - vol_sz_frac = sector_div(vol_sz, ENG_GIG_FACTOR); - vol_sz_frac *= 100; - sector_div(vol_sz_frac, ENG_GIG_FACTOR); - - if (drv->raid_level < 0 || drv->raid_level > RAID_UNKNOWN) - drv->raid_level = RAID_UNKNOWN; - seq_printf(seq, "cciss/c%dd%d:" - "\t%4u.%02uGB\tRAID %s\n", - ctlr, (int) *pos, (int)vol_sz, (int)vol_sz_frac, - raid_label[drv->raid_level]); - return 0; -} - -static void *cciss_seq_next(struct seq_file *seq, void *v, loff_t *pos) -{ - ctlr_info_t *h = seq->private; - - if (*pos > h->highest_lun) - return NULL; - *pos += 1; - - return pos; -} - -static void cciss_seq_stop(struct seq_file *seq, void *v) -{ - ctlr_info_t *h = seq->private; - - /* Only reset h->busy_configuring if we succeeded in setting - * it during cciss_seq_start. */ - if (v == ERR_PTR(-EBUSY)) - return; - - h->busy_configuring = 0; -} - -static const struct seq_operations cciss_seq_ops = { - .start = cciss_seq_start, - .show = cciss_seq_show, - .next = cciss_seq_next, - .stop = cciss_seq_stop, -}; - -static int cciss_seq_open(struct inode *inode, struct file *file) -{ - int ret = seq_open(file, &cciss_seq_ops); - struct seq_file *seq = file->private_data; - - if (!ret) - seq->private = PDE_DATA(inode); - - return ret; -} - -static ssize_t -cciss_proc_write(struct file *file, const char __user *buf, - size_t length, loff_t *ppos) -{ - int err; - char *buffer; - -#ifndef CONFIG_CISS_SCSI_TAPE - return -EINVAL; -#endif - - if (!buf || length > PAGE_SIZE - 1) - return -EINVAL; - - buffer = memdup_user_nul(buf, length); - if (IS_ERR(buffer)) - return PTR_ERR(buffer); - -#ifdef CONFIG_CISS_SCSI_TAPE - if (strncmp(ENGAGE_SCSI, buffer, sizeof ENGAGE_SCSI - 1) == 0) { - struct seq_file *seq = file->private_data; - ctlr_info_t *h = seq->private; - - err = cciss_engage_scsi(h); - if (err == 0) - err = length; - } else -#endif /* CONFIG_CISS_SCSI_TAPE */ - err = -EINVAL; - /* might be nice to have "disengage" too, but it's not - safely possible. (only 1 module use count, lock issues.) */ - - kfree(buffer); - return err; -} - -static const struct file_operations cciss_proc_fops = { - .owner = THIS_MODULE, - .open = cciss_seq_open, - .read = seq_read, - .llseek = seq_lseek, - .release = seq_release, - .write = cciss_proc_write, -}; - -static void cciss_procinit(ctlr_info_t *h) -{ - struct proc_dir_entry *pde; - - if (proc_cciss == NULL) - proc_cciss = proc_mkdir("driver/cciss", NULL); - if (!proc_cciss) - return; - pde = proc_create_data(h->devname, S_IWUSR | S_IRUSR | S_IRGRP | - S_IROTH, proc_cciss, - &cciss_proc_fops, h); -} -#endif /* CONFIG_PROC_FS */ - -#define MAX_PRODUCT_NAME_LEN 19 - -#define to_hba(n) container_of(n, struct ctlr_info, dev) -#define to_drv(n) container_of(n, drive_info_struct, dev) - -/* List of controllers which cannot be hard reset on kexec with reset_devices */ -static u32 unresettable_controller[] = { - 0x3223103C, /* Smart Array P800 */ - 0x3234103C, /* Smart Array P400 */ - 0x3235103C, /* Smart Array P400i */ - 0x3211103C, /* Smart Array E200i */ - 0x3212103C, /* Smart Array E200 */ - 0x3213103C, /* Smart Array E200i */ - 0x3214103C, /* Smart Array E200i */ - 0x3215103C, /* Smart Array E200i */ - 0x3237103C, /* Smart Array E500 */ - 0x323D103C, /* Smart Array P700m */ - 0x40800E11, /* Smart Array 5i */ - 0x409C0E11, /* Smart Array 6400 */ - 0x409D0E11, /* Smart Array 6400 EM */ - 0x40700E11, /* Smart Array 5300 */ - 0x40820E11, /* Smart Array 532 */ - 0x40830E11, /* Smart Array 5312 */ - 0x409A0E11, /* Smart Array 641 */ - 0x409B0E11, /* Smart Array 642 */ - 0x40910E11, /* Smart Array 6i */ -}; - -/* List of controllers which cannot even be soft reset */ -static u32 soft_unresettable_controller[] = { - 0x40800E11, /* Smart Array 5i */ - 0x40700E11, /* Smart Array 5300 */ - 0x40820E11, /* Smart Array 532 */ - 0x40830E11, /* Smart Array 5312 */ - 0x409A0E11, /* Smart Array 641 */ - 0x409B0E11, /* Smart Array 642 */ - 0x40910E11, /* Smart Array 6i */ - /* Exclude 640x boards. These are two pci devices in one slot - * which share a battery backed cache module. One controls the - * cache, the other accesses the cache through the one that controls - * it. If we reset the one controlling the cache, the other will - * likely not be happy. Just forbid resetting this conjoined mess. - */ - 0x409C0E11, /* Smart Array 6400 */ - 0x409D0E11, /* Smart Array 6400 EM */ -}; - -static int ctlr_is_hard_resettable(u32 board_id) -{ - int i; - - for (i = 0; i < ARRAY_SIZE(unresettable_controller); i++) - if (unresettable_controller[i] == board_id) - return 0; - return 1; -} - -static int ctlr_is_soft_resettable(u32 board_id) -{ - int i; - - for (i = 0; i < ARRAY_SIZE(soft_unresettable_controller); i++) - if (soft_unresettable_controller[i] == board_id) - return 0; - return 1; -} - -static int ctlr_is_resettable(u32 board_id) -{ - return ctlr_is_hard_resettable(board_id) || - ctlr_is_soft_resettable(board_id); -} - -static ssize_t host_show_resettable(struct device *dev, - struct device_attribute *attr, - char *buf) -{ - struct ctlr_info *h = to_hba(dev); - - return snprintf(buf, 20, "%d\n", ctlr_is_resettable(h->board_id)); -} -static DEVICE_ATTR(resettable, S_IRUGO, host_show_resettable, NULL); - -static ssize_t host_store_rescan(struct device *dev, - struct device_attribute *attr, - const char *buf, size_t count) -{ - struct ctlr_info *h = to_hba(dev); - - add_to_scan_list(h); - wake_up_process(cciss_scan_thread); - wait_for_completion_interruptible(&h->scan_wait); - - return count; -} -static DEVICE_ATTR(rescan, S_IWUSR, NULL, host_store_rescan); - -static ssize_t host_show_transport_mode(struct device *dev, - struct device_attribute *attr, - char *buf) -{ - struct ctlr_info *h = to_hba(dev); - - return snprintf(buf, 20, "%s\n", - h->transMethod & CFGTBL_Trans_Performant ? - "performant" : "simple"); -} -static DEVICE_ATTR(transport_mode, S_IRUGO, host_show_transport_mode, NULL); - -static ssize_t dev_show_unique_id(struct device *dev, - struct device_attribute *attr, - char *buf) -{ - drive_info_struct *drv = to_drv(dev); - struct ctlr_info *h = to_hba(drv->dev.parent); - __u8 sn[16]; - unsigned long flags; - int ret = 0; - - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) - ret = -EBUSY; - else - memcpy(sn, drv->serial_no, sizeof(sn)); - spin_unlock_irqrestore(&h->lock, flags); - - if (ret) - return ret; - else - return snprintf(buf, 16 * 2 + 2, - "%02X%02X%02X%02X%02X%02X%02X%02X" - "%02X%02X%02X%02X%02X%02X%02X%02X\n", - sn[0], sn[1], sn[2], sn[3], - sn[4], sn[5], sn[6], sn[7], - sn[8], sn[9], sn[10], sn[11], - sn[12], sn[13], sn[14], sn[15]); -} -static DEVICE_ATTR(unique_id, S_IRUGO, dev_show_unique_id, NULL); - -static ssize_t dev_show_vendor(struct device *dev, - struct device_attribute *attr, - char *buf) -{ - drive_info_struct *drv = to_drv(dev); - struct ctlr_info *h = to_hba(drv->dev.parent); - char vendor[VENDOR_LEN + 1]; - unsigned long flags; - int ret = 0; - - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) - ret = -EBUSY; - else - memcpy(vendor, drv->vendor, VENDOR_LEN + 1); - spin_unlock_irqrestore(&h->lock, flags); - - if (ret) - return ret; - else - return snprintf(buf, sizeof(vendor) + 1, "%s\n", drv->vendor); -} -static DEVICE_ATTR(vendor, S_IRUGO, dev_show_vendor, NULL); - -static ssize_t dev_show_model(struct device *dev, - struct device_attribute *attr, - char *buf) -{ - drive_info_struct *drv = to_drv(dev); - struct ctlr_info *h = to_hba(drv->dev.parent); - char model[MODEL_LEN + 1]; - unsigned long flags; - int ret = 0; - - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) - ret = -EBUSY; - else - memcpy(model, drv->model, MODEL_LEN + 1); - spin_unlock_irqrestore(&h->lock, flags); - - if (ret) - return ret; - else - return snprintf(buf, sizeof(model) + 1, "%s\n", drv->model); -} -static DEVICE_ATTR(model, S_IRUGO, dev_show_model, NULL); - -static ssize_t dev_show_rev(struct device *dev, - struct device_attribute *attr, - char *buf) -{ - drive_info_struct *drv = to_drv(dev); - struct ctlr_info *h = to_hba(drv->dev.parent); - char rev[REV_LEN + 1]; - unsigned long flags; - int ret = 0; - - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) - ret = -EBUSY; - else - memcpy(rev, drv->rev, REV_LEN + 1); - spin_unlock_irqrestore(&h->lock, flags); - - if (ret) - return ret; - else - return snprintf(buf, sizeof(rev) + 1, "%s\n", drv->rev); -} -static DEVICE_ATTR(rev, S_IRUGO, dev_show_rev, NULL); - -static ssize_t cciss_show_lunid(struct device *dev, - struct device_attribute *attr, char *buf) -{ - drive_info_struct *drv = to_drv(dev); - struct ctlr_info *h = to_hba(drv->dev.parent); - unsigned long flags; - unsigned char lunid[8]; - - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) { - spin_unlock_irqrestore(&h->lock, flags); - return -EBUSY; - } - if (!drv->heads) { - spin_unlock_irqrestore(&h->lock, flags); - return -ENOTTY; - } - memcpy(lunid, drv->LunID, sizeof(lunid)); - spin_unlock_irqrestore(&h->lock, flags); - return snprintf(buf, 20, "0x%02x%02x%02x%02x%02x%02x%02x%02x\n", - lunid[0], lunid[1], lunid[2], lunid[3], - lunid[4], lunid[5], lunid[6], lunid[7]); -} -static DEVICE_ATTR(lunid, S_IRUGO, cciss_show_lunid, NULL); - -static ssize_t cciss_show_raid_level(struct device *dev, - struct device_attribute *attr, char *buf) -{ - drive_info_struct *drv = to_drv(dev); - struct ctlr_info *h = to_hba(drv->dev.parent); - int raid; - unsigned long flags; - - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) { - spin_unlock_irqrestore(&h->lock, flags); - return -EBUSY; - } - raid = drv->raid_level; - spin_unlock_irqrestore(&h->lock, flags); - if (raid < 0 || raid > RAID_UNKNOWN) - raid = RAID_UNKNOWN; - - return snprintf(buf, strlen(raid_label[raid]) + 7, "RAID %s\n", - raid_label[raid]); -} -static DEVICE_ATTR(raid_level, S_IRUGO, cciss_show_raid_level, NULL); - -static ssize_t cciss_show_usage_count(struct device *dev, - struct device_attribute *attr, char *buf) -{ - drive_info_struct *drv = to_drv(dev); - struct ctlr_info *h = to_hba(drv->dev.parent); - unsigned long flags; - int count; - - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) { - spin_unlock_irqrestore(&h->lock, flags); - return -EBUSY; - } - count = drv->usage_count; - spin_unlock_irqrestore(&h->lock, flags); - return snprintf(buf, 20, "%d\n", count); -} -static DEVICE_ATTR(usage_count, S_IRUGO, cciss_show_usage_count, NULL); - -static struct attribute *cciss_host_attrs[] = { - &dev_attr_rescan.attr, - &dev_attr_resettable.attr, - &dev_attr_transport_mode.attr, - NULL -}; - -static struct attribute_group cciss_host_attr_group = { - .attrs = cciss_host_attrs, -}; - -static const struct attribute_group *cciss_host_attr_groups[] = { - &cciss_host_attr_group, - NULL -}; - -static struct device_type cciss_host_type = { - .name = "cciss_host", - .groups = cciss_host_attr_groups, - .release = cciss_hba_release, -}; - -static struct attribute *cciss_dev_attrs[] = { - &dev_attr_unique_id.attr, - &dev_attr_model.attr, - &dev_attr_vendor.attr, - &dev_attr_rev.attr, - &dev_attr_lunid.attr, - &dev_attr_raid_level.attr, - &dev_attr_usage_count.attr, - NULL -}; - -static struct attribute_group cciss_dev_attr_group = { - .attrs = cciss_dev_attrs, -}; - -static const struct attribute_group *cciss_dev_attr_groups[] = { - &cciss_dev_attr_group, - NULL -}; - -static struct device_type cciss_dev_type = { - .name = "cciss_device", - .groups = cciss_dev_attr_groups, - .release = cciss_device_release, -}; - -static struct bus_type cciss_bus_type = { - .name = "cciss", -}; - -/* - * cciss_hba_release is called when the reference count - * of h->dev goes to zero. - */ -static void cciss_hba_release(struct device *dev) -{ - /* - * nothing to do, but need this to avoid a warning - * about not having a release handler from lib/kref.c. - */ -} - -/* - * Initialize sysfs entry for each controller. This sets up and registers - * the 'cciss#' directory for each individual controller under - * /sys/bus/pci/devices/<dev>/. - */ -static int cciss_create_hba_sysfs_entry(struct ctlr_info *h) -{ - device_initialize(&h->dev); - h->dev.type = &cciss_host_type; - h->dev.bus = &cciss_bus_type; - dev_set_name(&h->dev, "%s", h->devname); - h->dev.parent = &h->pdev->dev; - - return device_add(&h->dev); -} - -/* - * Remove sysfs entries for an hba. - */ -static void cciss_destroy_hba_sysfs_entry(struct ctlr_info *h) -{ - device_del(&h->dev); - put_device(&h->dev); /* final put. */ -} - -/* cciss_device_release is called when the reference count - * of h->drv[x]dev goes to zero. - */ -static void cciss_device_release(struct device *dev) -{ - drive_info_struct *drv = to_drv(dev); - kfree(drv); -} - -/* - * Initialize sysfs for each logical drive. This sets up and registers - * the 'c#d#' directory for each individual logical drive under - * /sys/bus/pci/devices/<dev/ccis#/. We also create a link from - * /sys/block/cciss!c#d# to this entry. - */ -static long cciss_create_ld_sysfs_entry(struct ctlr_info *h, - int drv_index) -{ - struct device *dev; - - if (h->drv[drv_index]->device_initialized) - return 0; - - dev = &h->drv[drv_index]->dev; - device_initialize(dev); - dev->type = &cciss_dev_type; - dev->bus = &cciss_bus_type; - dev_set_name(dev, "c%dd%d", h->ctlr, drv_index); - dev->parent = &h->dev; - h->drv[drv_index]->device_initialized = 1; - return device_add(dev); -} - -/* - * Remove sysfs entries for a logical drive. - */ -static void cciss_destroy_ld_sysfs_entry(struct ctlr_info *h, int drv_index, - int ctlr_exiting) -{ - struct device *dev = &h->drv[drv_index]->dev; - - /* special case for c*d0, we only destroy it on controller exit */ - if (drv_index == 0 && !ctlr_exiting) - return; - - device_del(dev); - put_device(dev); /* the "final" put. */ - h->drv[drv_index] = NULL; -} - -/* - * For operations that cannot sleep, a command block is allocated at init, - * and managed by cmd_alloc() and cmd_free() using a simple bitmap to track - * which ones are free or in use. - */ -static CommandList_struct *cmd_alloc(ctlr_info_t *h) -{ - CommandList_struct *c; - int i; - u64bit temp64; - dma_addr_t cmd_dma_handle, err_dma_handle; - - do { - i = find_first_zero_bit(h->cmd_pool_bits, h->nr_cmds); - if (i == h->nr_cmds) - return NULL; - } while (test_and_set_bit(i, h->cmd_pool_bits) != 0); - c = h->cmd_pool + i; - memset(c, 0, sizeof(CommandList_struct)); - cmd_dma_handle = h->cmd_pool_dhandle + i * sizeof(CommandList_struct); - c->err_info = h->errinfo_pool + i; - memset(c->err_info, 0, sizeof(ErrorInfo_struct)); - err_dma_handle = h->errinfo_pool_dhandle - + i * sizeof(ErrorInfo_struct); - h->nr_allocs++; - - c->cmdindex = i; - - INIT_LIST_HEAD(&c->list); - c->busaddr = (__u32) cmd_dma_handle; - temp64.val = (__u64) err_dma_handle; - c->ErrDesc.Addr.lower = temp64.val32.lower; - c->ErrDesc.Addr.upper = temp64.val32.upper; - c->ErrDesc.Len = sizeof(ErrorInfo_struct); - - c->ctlr = h->ctlr; - return c; -} - -/* allocate a command using pci_alloc_consistent, used for ioctls, - * etc., not for the main i/o path. - */ -static CommandList_struct *cmd_special_alloc(ctlr_info_t *h) -{ - CommandList_struct *c; - u64bit temp64; - dma_addr_t cmd_dma_handle, err_dma_handle; - - c = pci_zalloc_consistent(h->pdev, sizeof(CommandList_struct), - &cmd_dma_handle); - if (c == NULL) - return NULL; - - c->cmdindex = -1; - - c->err_info = pci_zalloc_consistent(h->pdev, sizeof(ErrorInfo_struct), - &err_dma_handle); - - if (c->err_info == NULL) { - pci_free_consistent(h->pdev, - sizeof(CommandList_struct), c, cmd_dma_handle); - return NULL; - } - - INIT_LIST_HEAD(&c->list); - c->busaddr = (__u32) cmd_dma_handle; - temp64.val = (__u64) err_dma_handle; - c->ErrDesc.Addr.lower = temp64.val32.lower; - c->ErrDesc.Addr.upper = temp64.val32.upper; - c->ErrDesc.Len = sizeof(ErrorInfo_struct); - - c->ctlr = h->ctlr; - return c; -} - -static void cmd_free(ctlr_info_t *h, CommandList_struct *c) -{ - int i; - - i = c - h->cmd_pool; - clear_bit(i, h->cmd_pool_bits); - h->nr_frees++; -} - -static void cmd_special_free(ctlr_info_t *h, CommandList_struct *c) -{ - u64bit temp64; - - temp64.val32.lower = c->ErrDesc.Addr.lower; - temp64.val32.upper = c->ErrDesc.Addr.upper; - pci_free_consistent(h->pdev, sizeof(ErrorInfo_struct), - c->err_info, (dma_addr_t) temp64.val); - pci_free_consistent(h->pdev, sizeof(CommandList_struct), c, - (dma_addr_t) cciss_tag_discard_error_bits(h, (u32) c->busaddr)); -} - -static inline ctlr_info_t *get_host(struct gendisk *disk) -{ - return disk->queue->queuedata; -} - -static inline drive_info_struct *get_drv(struct gendisk *disk) -{ - return disk->private_data; -} - -/* - * Open. Make sure the device is really there. - */ -static int cciss_open(struct block_device *bdev, fmode_t mode) -{ - ctlr_info_t *h = get_host(bdev->bd_disk); - drive_info_struct *drv = get_drv(bdev->bd_disk); - - dev_dbg(&h->pdev->dev, "cciss_open %s\n", bdev->bd_disk->disk_name); - if (drv->busy_configuring) - return -EBUSY; - /* - * Root is allowed to open raw volume zero even if it's not configured - * so array config can still work. Root is also allowed to open any - * volume that has a LUN ID, so it can issue IOCTL to reread the - * disk information. I don't think I really like this - * but I'm already using way to many device nodes to claim another one - * for "raw controller". - */ - if (drv->heads == 0) { - if (MINOR(bdev->bd_dev) != 0) { /* not node 0? */ - /* if not node 0 make sure it is a partition = 0 */ - if (MINOR(bdev->bd_dev) & 0x0f) { - return -ENXIO; - /* if it is, make sure we have a LUN ID */ - } else if (memcmp(drv->LunID, CTLR_LUNID, - sizeof(drv->LunID))) { - return -ENXIO; - } - } - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - } - drv->usage_count++; - h->usage_count++; - return 0; -} - -static int cciss_unlocked_open(struct block_device *bdev, fmode_t mode) -{ - int ret; - - mutex_lock(&cciss_mutex); - ret = cciss_open(bdev, mode); - mutex_unlock(&cciss_mutex); - - return ret; -} - -/* - * Close. Sync first. - */ -static void cciss_release(struct gendisk *disk, fmode_t mode) -{ - ctlr_info_t *h; - drive_info_struct *drv; - - mutex_lock(&cciss_mutex); - h = get_host(disk); - drv = get_drv(disk); - dev_dbg(&h->pdev->dev, "cciss_release %s\n", disk->disk_name); - drv->usage_count--; - h->usage_count--; - mutex_unlock(&cciss_mutex); -} - -#ifdef CONFIG_COMPAT - -static int cciss_ioctl32_passthru(struct block_device *bdev, fmode_t mode, - unsigned cmd, unsigned long arg); -static int cciss_ioctl32_big_passthru(struct block_device *bdev, fmode_t mode, - unsigned cmd, unsigned long arg); - -static int cciss_compat_ioctl(struct block_device *bdev, fmode_t mode, - unsigned cmd, unsigned long arg) -{ - switch (cmd) { - case CCISS_GETPCIINFO: - case CCISS_GETINTINFO: - case CCISS_SETINTINFO: - case CCISS_GETNODENAME: - case CCISS_SETNODENAME: - case CCISS_GETHEARTBEAT: - case CCISS_GETBUSTYPES: - case CCISS_GETFIRMVER: - case CCISS_GETDRIVVER: - case CCISS_REVALIDVOLS: - case CCISS_DEREGDISK: - case CCISS_REGNEWDISK: - case CCISS_REGNEWD: - case CCISS_RESCANDISK: - case CCISS_GETLUNINFO: - return cciss_ioctl(bdev, mode, cmd, arg); - - case CCISS_PASSTHRU32: - return cciss_ioctl32_passthru(bdev, mode, cmd, arg); - case CCISS_BIG_PASSTHRU32: - return cciss_ioctl32_big_passthru(bdev, mode, cmd, arg); - - default: - return -ENOIOCTLCMD; - } -} - -static int cciss_ioctl32_passthru(struct block_device *bdev, fmode_t mode, - unsigned cmd, unsigned long arg) -{ - IOCTL32_Command_struct __user *arg32 = - (IOCTL32_Command_struct __user *) arg; - IOCTL_Command_struct arg64; - IOCTL_Command_struct __user *p = compat_alloc_user_space(sizeof(arg64)); - int err; - u32 cp; - - memset(&arg64, 0, sizeof(arg64)); - err = 0; - err |= - copy_from_user(&arg64.LUN_info, &arg32->LUN_info, - sizeof(arg64.LUN_info)); - err |= - copy_from_user(&arg64.Request, &arg32->Request, - sizeof(arg64.Request)); - err |= - copy_from_user(&arg64.error_info, &arg32->error_info, - sizeof(arg64.error_info)); - err |= get_user(arg64.buf_size, &arg32->buf_size); - err |= get_user(cp, &arg32->buf); - arg64.buf = compat_ptr(cp); - err |= copy_to_user(p, &arg64, sizeof(arg64)); - - if (err) - return -EFAULT; - - err = cciss_ioctl(bdev, mode, CCISS_PASSTHRU, (unsigned long)p); - if (err) - return err; - err |= - copy_in_user(&arg32->error_info, &p->error_info, - sizeof(arg32->error_info)); - if (err) - return -EFAULT; - return err; -} - -static int cciss_ioctl32_big_passthru(struct block_device *bdev, fmode_t mode, - unsigned cmd, unsigned long arg) -{ - BIG_IOCTL32_Command_struct __user *arg32 = - (BIG_IOCTL32_Command_struct __user *) arg; - BIG_IOCTL_Command_struct arg64; - BIG_IOCTL_Command_struct __user *p = - compat_alloc_user_space(sizeof(arg64)); - int err; - u32 cp; - - memset(&arg64, 0, sizeof(arg64)); - err = 0; - err |= - copy_from_user(&arg64.LUN_info, &arg32->LUN_info, - sizeof(arg64.LUN_info)); - err |= - copy_from_user(&arg64.Request, &arg32->Request, - sizeof(arg64.Request)); - err |= - copy_from_user(&arg64.error_info, &arg32->error_info, - sizeof(arg64.error_info)); - err |= get_user(arg64.buf_size, &arg32->buf_size); - err |= get_user(arg64.malloc_size, &arg32->malloc_size); - err |= get_user(cp, &arg32->buf); - arg64.buf = compat_ptr(cp); - err |= copy_to_user(p, &arg64, sizeof(arg64)); - - if (err) - return -EFAULT; - - err = cciss_ioctl(bdev, mode, CCISS_BIG_PASSTHRU, (unsigned long)p); - if (err) - return err; - err |= - copy_in_user(&arg32->error_info, &p->error_info, - sizeof(arg32->error_info)); - if (err) - return -EFAULT; - return err; -} -#endif - -static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo) -{ - drive_info_struct *drv = get_drv(bdev->bd_disk); - - if (!drv->cylinders) - return -ENXIO; - - geo->heads = drv->heads; - geo->sectors = drv->sectors; - geo->cylinders = drv->cylinders; - return 0; -} - -static void check_ioctl_unit_attention(ctlr_info_t *h, CommandList_struct *c) -{ - if (c->err_info->CommandStatus == CMD_TARGET_STATUS && - c->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION) - (void)check_for_unit_attention(h, c); -} - -static int cciss_getpciinfo(ctlr_info_t *h, void __user *argp) -{ - cciss_pci_info_struct pciinfo; - - if (!argp) - return -EINVAL; - pciinfo.domain = pci_domain_nr(h->pdev->bus); - pciinfo.bus = h->pdev->bus->number; - pciinfo.dev_fn = h->pdev->devfn; - pciinfo.board_id = h->board_id; - if (copy_to_user(argp, &pciinfo, sizeof(cciss_pci_info_struct))) - return -EFAULT; - return 0; -} - -static int cciss_getintinfo(ctlr_info_t *h, void __user *argp) -{ - cciss_coalint_struct intinfo; - unsigned long flags; - - if (!argp) - return -EINVAL; - spin_lock_irqsave(&h->lock, flags); - intinfo.delay = readl(&h->cfgtable->HostWrite.CoalIntDelay); - intinfo.count = readl(&h->cfgtable->HostWrite.CoalIntCount); - spin_unlock_irqrestore(&h->lock, flags); - if (copy_to_user - (argp, &intinfo, sizeof(cciss_coalint_struct))) - return -EFAULT; - return 0; -} - -static int cciss_setintinfo(ctlr_info_t *h, void __user *argp) -{ - cciss_coalint_struct intinfo; - unsigned long flags; - int i; - - if (!argp) - return -EINVAL; - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - if (copy_from_user(&intinfo, argp, sizeof(intinfo))) - return -EFAULT; - if ((intinfo.delay == 0) && (intinfo.count == 0)) - return -EINVAL; - spin_lock_irqsave(&h->lock, flags); - /* Update the field, and then ring the doorbell */ - writel(intinfo.delay, &(h->cfgtable->HostWrite.CoalIntDelay)); - writel(intinfo.count, &(h->cfgtable->HostWrite.CoalIntCount)); - writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); - - for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) { - if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq)) - break; - udelay(1000); /* delay and try again */ - } - spin_unlock_irqrestore(&h->lock, flags); - if (i >= MAX_IOCTL_CONFIG_WAIT) - return -EAGAIN; - return 0; -} - -static int cciss_getnodename(ctlr_info_t *h, void __user *argp) -{ - NodeName_type NodeName; - unsigned long flags; - int i; - - if (!argp) - return -EINVAL; - spin_lock_irqsave(&h->lock, flags); - for (i = 0; i < 16; i++) - NodeName[i] = readb(&h->cfgtable->ServerName[i]); - spin_unlock_irqrestore(&h->lock, flags); - if (copy_to_user(argp, NodeName, sizeof(NodeName_type))) - return -EFAULT; - return 0; -} - -static int cciss_setnodename(ctlr_info_t *h, void __user *argp) -{ - NodeName_type NodeName; - unsigned long flags; - int i; - - if (!argp) - return -EINVAL; - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - if (copy_from_user(NodeName, argp, sizeof(NodeName_type))) - return -EFAULT; - spin_lock_irqsave(&h->lock, flags); - /* Update the field, and then ring the doorbell */ - for (i = 0; i < 16; i++) - writeb(NodeName[i], &h->cfgtable->ServerName[i]); - writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); - for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) { - if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq)) - break; - udelay(1000); /* delay and try again */ - } - spin_unlock_irqrestore(&h->lock, flags); - if (i >= MAX_IOCTL_CONFIG_WAIT) - return -EAGAIN; - return 0; -} - -static int cciss_getheartbeat(ctlr_info_t *h, void __user *argp) -{ - Heartbeat_type heartbeat; - unsigned long flags; - - if (!argp) - return -EINVAL; - spin_lock_irqsave(&h->lock, flags); - heartbeat = readl(&h->cfgtable->HeartBeat); - spin_unlock_irqrestore(&h->lock, flags); - if (copy_to_user(argp, &heartbeat, sizeof(Heartbeat_type))) - return -EFAULT; - return 0; -} - -static int cciss_getbustypes(ctlr_info_t *h, void __user *argp) -{ - BusTypes_type BusTypes; - unsigned long flags; - - if (!argp) - return -EINVAL; - spin_lock_irqsave(&h->lock, flags); - BusTypes = readl(&h->cfgtable->BusTypes); - spin_unlock_irqrestore(&h->lock, flags); - if (copy_to_user(argp, &BusTypes, sizeof(BusTypes_type))) - return -EFAULT; - return 0; -} - -static int cciss_getfirmver(ctlr_info_t *h, void __user *argp) -{ - FirmwareVer_type firmware; - - if (!argp) - return -EINVAL; - memcpy(firmware, h->firm_ver, 4); - - if (copy_to_user - (argp, firmware, sizeof(FirmwareVer_type))) - return -EFAULT; - return 0; -} - -static int cciss_getdrivver(ctlr_info_t *h, void __user *argp) -{ - DriverVer_type DriverVer = DRIVER_VERSION; - - if (!argp) - return -EINVAL; - if (copy_to_user(argp, &DriverVer, sizeof(DriverVer_type))) - return -EFAULT; - return 0; -} - -static int cciss_getluninfo(ctlr_info_t *h, - struct gendisk *disk, void __user *argp) -{ - LogvolInfo_struct luninfo; - drive_info_struct *drv = get_drv(disk); - - if (!argp) - return -EINVAL; - memcpy(&luninfo.LunID, drv->LunID, sizeof(luninfo.LunID)); - luninfo.num_opens = drv->usage_count; - luninfo.num_parts = 0; - if (copy_to_user(argp, &luninfo, sizeof(LogvolInfo_struct))) - return -EFAULT; - return 0; -} - -static int cciss_passthru(ctlr_info_t *h, void __user *argp) -{ - IOCTL_Command_struct iocommand; - CommandList_struct *c; - char *buff = NULL; - u64bit temp64; - DECLARE_COMPLETION_ONSTACK(wait); - - if (!argp) - return -EINVAL; - - if (!capable(CAP_SYS_RAWIO)) - return -EPERM; - - if (copy_from_user - (&iocommand, argp, sizeof(IOCTL_Command_struct))) - return -EFAULT; - if ((iocommand.buf_size < 1) && - (iocommand.Request.Type.Direction != XFER_NONE)) { - return -EINVAL; - } - if (iocommand.buf_size > 0) { - buff = kmalloc(iocommand.buf_size, GFP_KERNEL); - if (buff == NULL) - return -EFAULT; - } - if (iocommand.Request.Type.Direction == XFER_WRITE) { - /* Copy the data into the buffer we created */ - if (copy_from_user(buff, iocommand.buf, iocommand.buf_size)) { - kfree(buff); - return -EFAULT; - } - } else { - memset(buff, 0, iocommand.buf_size); - } - c = cmd_special_alloc(h); - if (!c) { - kfree(buff); - return -ENOMEM; - } - /* Fill in the command type */ - c->cmd_type = CMD_IOCTL_PEND; - /* Fill in Command Header */ - c->Header.ReplyQueue = 0; /* unused in simple mode */ - if (iocommand.buf_size > 0) { /* buffer to fill */ - c->Header.SGList = 1; - c->Header.SGTotal = 1; - } else { /* no buffers to fill */ - c->Header.SGList = 0; - c->Header.SGTotal = 0; - } - c->Header.LUN = iocommand.LUN_info; - /* use the kernel address the cmd block for tag */ - c->Header.Tag.lower = c->busaddr; - - /* Fill in Request block */ - c->Request = iocommand.Request; - - /* Fill in the scatter gather information */ - if (iocommand.buf_size > 0) { - temp64.val = pci_map_single(h->pdev, buff, - iocommand.buf_size, PCI_DMA_BIDIRECTIONAL); - c->SG[0].Addr.lower = temp64.val32.lower; - c->SG[0].Addr.upper = temp64.val32.upper; - c->SG[0].Len = iocommand.buf_size; - c->SG[0].Ext = 0; /* we are not chaining */ - } - c->waiting = &wait; - - enqueue_cmd_and_start_io(h, c); - wait_for_completion(&wait); - - /* unlock the buffers from DMA */ - temp64.val32.lower = c->SG[0].Addr.lower; - temp64.val32.upper = c->SG[0].Addr.upper; - pci_unmap_single(h->pdev, (dma_addr_t) temp64.val, iocommand.buf_size, - PCI_DMA_BIDIRECTIONAL); - check_ioctl_unit_attention(h, c); - - /* Copy the error information out */ - iocommand.error_info = *(c->err_info); - if (copy_to_user(argp, &iocommand, sizeof(IOCTL_Command_struct))) { - kfree(buff); - cmd_special_free(h, c); - return -EFAULT; - } - - if (iocommand.Request.Type.Direction == XFER_READ) { - /* Copy the data out of the buffer we created */ - if (copy_to_user(iocommand.buf, buff, iocommand.buf_size)) { - kfree(buff); - cmd_special_free(h, c); - return -EFAULT; - } - } - kfree(buff); - cmd_special_free(h, c); - return 0; -} - -static int cciss_bigpassthru(ctlr_info_t *h, void __user *argp) -{ - BIG_IOCTL_Command_struct *ioc; - CommandList_struct *c; - unsigned char **buff = NULL; - int *buff_size = NULL; - u64bit temp64; - BYTE sg_used = 0; - int status = 0; - int i; - DECLARE_COMPLETION_ONSTACK(wait); - __u32 left; - __u32 sz; - BYTE __user *data_ptr; - - if (!argp) - return -EINVAL; - if (!capable(CAP_SYS_RAWIO)) - return -EPERM; - ioc = kmalloc(sizeof(*ioc), GFP_KERNEL); - if (!ioc) { - status = -ENOMEM; - goto cleanup1; - } - if (copy_from_user(ioc, argp, sizeof(*ioc))) { - status = -EFAULT; - goto cleanup1; - } - if ((ioc->buf_size < 1) && - (ioc->Request.Type.Direction != XFER_NONE)) { - status = -EINVAL; - goto cleanup1; - } - /* Check kmalloc limits using all SGs */ - if (ioc->malloc_size > MAX_KMALLOC_SIZE) { - status = -EINVAL; - goto cleanup1; - } - if (ioc->buf_size > ioc->malloc_size * MAXSGENTRIES) { - status = -EINVAL; - goto cleanup1; - } - buff = kzalloc(MAXSGENTRIES * sizeof(char *), GFP_KERNEL); - if (!buff) { - status = -ENOMEM; - goto cleanup1; - } - buff_size = kmalloc(MAXSGENTRIES * sizeof(int), GFP_KERNEL); - if (!buff_size) { - status = -ENOMEM; - goto cleanup1; - } - left = ioc->buf_size; - data_ptr = ioc->buf; - while (left) { - sz = (left > ioc->malloc_size) ? ioc->malloc_size : left; - buff_size[sg_used] = sz; - buff[sg_used] = kmalloc(sz, GFP_KERNEL); - if (buff[sg_used] == NULL) { - status = -ENOMEM; - goto cleanup1; - } - if (ioc->Request.Type.Direction == XFER_WRITE) { - if (copy_from_user(buff[sg_used], data_ptr, sz)) { - status = -EFAULT; - goto cleanup1; - } - } else { - memset(buff[sg_used], 0, sz); - } - left -= sz; - data_ptr += sz; - sg_used++; - } - c = cmd_special_alloc(h); - if (!c) { - status = -ENOMEM; - goto cleanup1; - } - c->cmd_type = CMD_IOCTL_PEND; - c->Header.ReplyQueue = 0; - c->Header.SGList = sg_used; - c->Header.SGTotal = sg_used; - c->Header.LUN = ioc->LUN_info; - c->Header.Tag.lower = c->busaddr; - - c->Request = ioc->Request; - for (i = 0; i < sg_used; i++) { - temp64.val = pci_map_single(h->pdev, buff[i], buff_size[i], - PCI_DMA_BIDIRECTIONAL); - c->SG[i].Addr.lower = temp64.val32.lower; - c->SG[i].Addr.upper = temp64.val32.upper; - c->SG[i].Len = buff_size[i]; - c->SG[i].Ext = 0; /* we are not chaining */ - } - c->waiting = &wait; - enqueue_cmd_and_start_io(h, c); - wait_for_completion(&wait); - /* unlock the buffers from DMA */ - for (i = 0; i < sg_used; i++) { - temp64.val32.lower = c->SG[i].Addr.lower; - temp64.val32.upper = c->SG[i].Addr.upper; - pci_unmap_single(h->pdev, - (dma_addr_t) temp64.val, buff_size[i], - PCI_DMA_BIDIRECTIONAL); - } - check_ioctl_unit_attention(h, c); - /* Copy the error information out */ - ioc->error_info = *(c->err_info); - if (copy_to_user(argp, ioc, sizeof(*ioc))) { - cmd_special_free(h, c); - status = -EFAULT; - goto cleanup1; - } - if (ioc->Request.Type.Direction == XFER_READ) { - /* Copy the data out of the buffer we created */ - BYTE __user *ptr = ioc->buf; - for (i = 0; i < sg_used; i++) { - if (copy_to_user(ptr, buff[i], buff_size[i])) { - cmd_special_free(h, c); - status = -EFAULT; - goto cleanup1; - } - ptr += buff_size[i]; - } - } - cmd_special_free(h, c); - status = 0; -cleanup1: - if (buff) { - for (i = 0; i < sg_used; i++) - kfree(buff[i]); - kfree(buff); - } - kfree(buff_size); - kfree(ioc); - return status; -} - -static int cciss_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long arg) -{ - struct gendisk *disk = bdev->bd_disk; - ctlr_info_t *h = get_host(disk); - void __user *argp = (void __user *)arg; - - dev_dbg(&h->pdev->dev, "cciss_ioctl: Called with cmd=%x %lx\n", - cmd, arg); - switch (cmd) { - case CCISS_GETPCIINFO: - return cciss_getpciinfo(h, argp); - case CCISS_GETINTINFO: - return cciss_getintinfo(h, argp); - case CCISS_SETINTINFO: - return cciss_setintinfo(h, argp); - case CCISS_GETNODENAME: - return cciss_getnodename(h, argp); - case CCISS_SETNODENAME: - return cciss_setnodename(h, argp); - case CCISS_GETHEARTBEAT: - return cciss_getheartbeat(h, argp); - case CCISS_GETBUSTYPES: - return cciss_getbustypes(h, argp); - case CCISS_GETFIRMVER: - return cciss_getfirmver(h, argp); - case CCISS_GETDRIVVER: - return cciss_getdrivver(h, argp); - case CCISS_DEREGDISK: - case CCISS_REGNEWD: - case CCISS_REVALIDVOLS: - return rebuild_lun_table(h, 0, 1); - case CCISS_GETLUNINFO: - return cciss_getluninfo(h, disk, argp); - case CCISS_PASSTHRU: - return cciss_passthru(h, argp); - case CCISS_BIG_PASSTHRU: - return cciss_bigpassthru(h, argp); - - /* scsi_cmd_blk_ioctl handles these, below, though some are not */ - /* very meaningful for cciss. SG_IO is the main one people want. */ - - case SG_GET_VERSION_NUM: - case SG_SET_TIMEOUT: - case SG_GET_TIMEOUT: - case SG_GET_RESERVED_SIZE: - case SG_SET_RESERVED_SIZE: - case SG_EMULATED_HOST: - case SG_IO: - case SCSI_IOCTL_SEND_COMMAND: - return scsi_cmd_blk_ioctl(bdev, mode, cmd, argp); - - /* scsi_cmd_blk_ioctl would normally handle these, below, but */ - /* they aren't a good fit for cciss, as CD-ROMs are */ - /* not supported, and we don't have any bus/target/lun */ - /* which we present to the kernel. */ - - case CDROM_SEND_PACKET: - case CDROMCLOSETRAY: - case CDROMEJECT: - case SCSI_IOCTL_GET_IDLUN: - case SCSI_IOCTL_GET_BUS_NUMBER: - default: - return -ENOTTY; - } -} - -static void cciss_check_queues(ctlr_info_t *h) -{ - int start_queue = h->next_to_run; - int i; - - /* check to see if we have maxed out the number of commands that can - * be placed on the queue. If so then exit. We do this check here - * in case the interrupt we serviced was from an ioctl and did not - * free any new commands. - */ - if ((find_first_zero_bit(h->cmd_pool_bits, h->nr_cmds)) == h->nr_cmds) - return; - - /* We have room on the queue for more commands. Now we need to queue - * them up. We will also keep track of the next queue to run so - * that every queue gets a chance to be started first. - */ - for (i = 0; i < h->highest_lun + 1; i++) { - int curr_queue = (start_queue + i) % (h->highest_lun + 1); - /* make sure the disk has been added and the drive is real - * because this can be called from the middle of init_one. - */ - if (!h->drv[curr_queue]) - continue; - if (!(h->drv[curr_queue]->queue) || - !(h->drv[curr_queue]->heads)) - continue; - blk_start_queue(h->gendisk[curr_queue]->queue); - - /* check to see if we have maxed out the number of commands - * that can be placed on the queue. - */ - if ((find_first_zero_bit(h->cmd_pool_bits, h->nr_cmds)) == h->nr_cmds) { - if (curr_queue == start_queue) { - h->next_to_run = - (start_queue + 1) % (h->highest_lun + 1); - break; - } else { - h->next_to_run = curr_queue; - break; - } - } - } -} - -static void cciss_softirq_done(struct request *rq) -{ - CommandList_struct *c = rq->completion_data; - ctlr_info_t *h = hba[c->ctlr]; - SGDescriptor_struct *curr_sg = c->SG; - u64bit temp64; - unsigned long flags; - int i, ddir; - int sg_index = 0; - - if (c->Request.Type.Direction == XFER_READ) - ddir = PCI_DMA_FROMDEVICE; - else - ddir = PCI_DMA_TODEVICE; - - /* command did not need to be retried */ - /* unmap the DMA mapping for all the scatter gather elements */ - for (i = 0; i < c->Header.SGList; i++) { - if (curr_sg[sg_index].Ext == CCISS_SG_CHAIN) { - cciss_unmap_sg_chain_block(h, c); - /* Point to the next block */ - curr_sg = h->cmd_sg_list[c->cmdindex]; - sg_index = 0; - } - temp64.val32.lower = curr_sg[sg_index].Addr.lower; - temp64.val32.upper = curr_sg[sg_index].Addr.upper; - pci_unmap_page(h->pdev, temp64.val, curr_sg[sg_index].Len, - ddir); - ++sg_index; - } - - dev_dbg(&h->pdev->dev, "Done with %p\n", rq); - - /* set the residual count for pc requests */ - if (blk_rq_is_passthrough(rq)) - scsi_req(rq)->resid_len = c->err_info->ResidualCnt; - blk_end_request_all(rq, scsi_req(rq)->result ? - BLK_STS_IOERR : BLK_STS_OK); - - spin_lock_irqsave(&h->lock, flags); - cmd_free(h, c); - cciss_check_queues(h); - spin_unlock_irqrestore(&h->lock, flags); -} - -static inline void log_unit_to_scsi3addr(ctlr_info_t *h, - unsigned char scsi3addr[], uint32_t log_unit) -{ - memcpy(scsi3addr, h->drv[log_unit]->LunID, - sizeof(h->drv[log_unit]->LunID)); -} - -/* This function gets the SCSI vendor, model, and revision of a logical drive - * via the inquiry page 0. Model, vendor, and rev are set to empty strings if - * they cannot be read. - */ -static void cciss_get_device_descr(ctlr_info_t *h, int logvol, - char *vendor, char *model, char *rev) -{ - int rc; - InquiryData_struct *inq_buf; - unsigned char scsi3addr[8]; - - *vendor = '\0'; - *model = '\0'; - *rev = '\0'; - - inq_buf = kzalloc(sizeof(InquiryData_struct), GFP_KERNEL); - if (!inq_buf) - return; - - log_unit_to_scsi3addr(h, scsi3addr, logvol); - rc = sendcmd_withirq(h, CISS_INQUIRY, inq_buf, sizeof(*inq_buf), 0, - scsi3addr, TYPE_CMD); - if (rc == IO_OK) { - memcpy(vendor, &inq_buf->data_byte[8], VENDOR_LEN); - vendor[VENDOR_LEN] = '\0'; - memcpy(model, &inq_buf->data_byte[16], MODEL_LEN); - model[MODEL_LEN] = '\0'; - memcpy(rev, &inq_buf->data_byte[32], REV_LEN); - rev[REV_LEN] = '\0'; - } - - kfree(inq_buf); - return; -} - -/* This function gets the serial number of a logical drive via - * inquiry page 0x83. Serial no. is 16 bytes. If the serial - * number cannot be had, for whatever reason, 16 bytes of 0xff - * are returned instead. - */ -static void cciss_get_serial_no(ctlr_info_t *h, int logvol, - unsigned char *serial_no, int buflen) -{ -#define PAGE_83_INQ_BYTES 64 - int rc; - unsigned char *buf; - unsigned char scsi3addr[8]; - - if (buflen > 16) - buflen = 16; - memset(serial_no, 0xff, buflen); - buf = kzalloc(PAGE_83_INQ_BYTES, GFP_KERNEL); - if (!buf) - return; - memset(serial_no, 0, buflen); - log_unit_to_scsi3addr(h, scsi3addr, logvol); - rc = sendcmd_withirq(h, CISS_INQUIRY, buf, - PAGE_83_INQ_BYTES, 0x83, scsi3addr, TYPE_CMD); - if (rc == IO_OK) - memcpy(serial_no, &buf[8], buflen); - kfree(buf); - return; -} - -static void cciss_initialize_rq(struct request *rq) -{ - struct scsi_request *sreq = blk_mq_rq_to_pdu(rq); - - scsi_req_init(sreq); -} - -/* - * cciss_add_disk sets up the block device queue for a logical drive - */ -static int cciss_add_disk(ctlr_info_t *h, struct gendisk *disk, - int drv_index) -{ - disk->queue = blk_alloc_queue(GFP_KERNEL); - if (!disk->queue) - goto init_queue_failure; - - disk->queue->cmd_size = sizeof(struct scsi_request); - disk->queue->request_fn = do_cciss_request; - disk->queue->initialize_rq_fn = cciss_initialize_rq; - disk->queue->queue_lock = &h->lock; - queue_flag_set_unlocked(QUEUE_FLAG_SCSI_PASSTHROUGH, disk->queue); - if (blk_init_allocated_queue(disk->queue) < 0) - goto cleanup_queue; - - sprintf(disk->disk_name, "cciss/c%dd%d", h->ctlr, drv_index); - disk->major = h->major; - disk->first_minor = drv_index << NWD_SHIFT; - disk->fops = &cciss_fops; - if (cciss_create_ld_sysfs_entry(h, drv_index)) - goto cleanup_queue; - disk->private_data = h->drv[drv_index]; - - /* Set up queue information */ - blk_queue_bounce_limit(disk->queue, h->pdev->dma_mask); - - /* This is a hardware imposed limit. */ - blk_queue_max_segments(disk->queue, h->maxsgentries); - - blk_queue_max_hw_sectors(disk->queue, h->cciss_max_sectors); - - blk_queue_softirq_done(disk->queue, cciss_softirq_done); - - disk->queue->queuedata = h; - - blk_queue_logical_block_size(disk->queue, - h->drv[drv_index]->block_size); - - /* Make sure all queue data is written out before */ - /* setting h->drv[drv_index]->queue, as setting this */ - /* allows the interrupt handler to start the queue */ - wmb(); - h->drv[drv_index]->queue = disk->queue; - device_add_disk(&h->drv[drv_index]->dev, disk); - return 0; - -cleanup_queue: - blk_cleanup_queue(disk->queue); - disk->queue = NULL; -init_queue_failure: - return -1; -} - -/* This function will check the usage_count of the drive to be updated/added. - * If the usage_count is zero and it is a heretofore unknown drive, or, - * the drive's capacity, geometry, or serial number has changed, - * then the drive information will be updated and the disk will be - * re-registered with the kernel. If these conditions don't hold, - * then it will be left alone for the next reboot. The exception to this - * is disk 0 which will always be left registered with the kernel since it - * is also the controller node. Any changes to disk 0 will show up on - * the next reboot. - */ -static void cciss_update_drive_info(ctlr_info_t *h, int drv_index, - int first_time, int via_ioctl) -{ - struct gendisk *disk; - InquiryData_struct *inq_buff = NULL; - unsigned int block_size; - sector_t total_size; - unsigned long flags = 0; - int ret = 0; - drive_info_struct *drvinfo; - - /* Get information about the disk and modify the driver structure */ - inq_buff = kmalloc(sizeof(InquiryData_struct), GFP_KERNEL); - drvinfo = kzalloc(sizeof(*drvinfo), GFP_KERNEL); - if (inq_buff == NULL || drvinfo == NULL) - goto mem_msg; - - /* testing to see if 16-byte CDBs are already being used */ - if (h->cciss_read == CCISS_READ_16) { - cciss_read_capacity_16(h, drv_index, - &total_size, &block_size); - - } else { - cciss_read_capacity(h, drv_index, &total_size, &block_size); - /* if read_capacity returns all F's this volume is >2TB */ - /* in size so we switch to 16-byte CDB's for all */ - /* read/write ops */ - if (total_size == 0xFFFFFFFFULL) { - cciss_read_capacity_16(h, drv_index, - &total_size, &block_size); - h->cciss_read = CCISS_READ_16; - h->cciss_write = CCISS_WRITE_16; - } else { - h->cciss_read = CCISS_READ_10; - h->cciss_write = CCISS_WRITE_10; - } - } - - cciss_geometry_inquiry(h, drv_index, total_size, block_size, - inq_buff, drvinfo); - drvinfo->block_size = block_size; - drvinfo->nr_blocks = total_size + 1; - - cciss_get_device_descr(h, drv_index, drvinfo->vendor, - drvinfo->model, drvinfo->rev); - cciss_get_serial_no(h, drv_index, drvinfo->serial_no, - sizeof(drvinfo->serial_no)); - /* Save the lunid in case we deregister the disk, below. */ - memcpy(drvinfo->LunID, h->drv[drv_index]->LunID, - sizeof(drvinfo->LunID)); - - /* Is it the same disk we already know, and nothing's changed? */ - if (h->drv[drv_index]->raid_level != -1 && - ((memcmp(drvinfo->serial_no, - h->drv[drv_index]->serial_no, 16) == 0) && - drvinfo->block_size == h->drv[drv_index]->block_size && - drvinfo->nr_blocks == h->drv[drv_index]->nr_blocks && - drvinfo->heads == h->drv[drv_index]->heads && - drvinfo->sectors == h->drv[drv_index]->sectors && - drvinfo->cylinders == h->drv[drv_index]->cylinders)) - /* The disk is unchanged, nothing to update */ - goto freeret; - - /* If we get here it's not the same disk, or something's changed, - * so we need to * deregister it, and re-register it, if it's not - * in use. - * If the disk already exists then deregister it before proceeding - * (unless it's the first disk (for the controller node). - */ - if (h->drv[drv_index]->raid_level != -1 && drv_index != 0) { - dev_warn(&h->pdev->dev, "disk %d has changed.\n", drv_index); - spin_lock_irqsave(&h->lock, flags); - h->drv[drv_index]->busy_configuring = 1; - spin_unlock_irqrestore(&h->lock, flags); - - /* deregister_disk sets h->drv[drv_index]->queue = NULL - * which keeps the interrupt handler from starting - * the queue. - */ - ret = deregister_disk(h, drv_index, 0, via_ioctl); - } - - /* If the disk is in use return */ - if (ret) - goto freeret; - - /* Save the new information from cciss_geometry_inquiry - * and serial number inquiry. If the disk was deregistered - * above, then h->drv[drv_index] will be NULL. - */ - if (h->drv[drv_index] == NULL) { - drvinfo->device_initialized = 0; - h->drv[drv_index] = drvinfo; - drvinfo = NULL; /* so it won't be freed below. */ - } else { - /* special case for cxd0 */ - h->drv[drv_index]->block_size = drvinfo->block_size; - h->drv[drv_index]->nr_blocks = drvinfo->nr_blocks; - h->drv[drv_index]->heads = drvinfo->heads; - h->drv[drv_index]->sectors = drvinfo->sectors; - h->drv[drv_index]->cylinders = drvinfo->cylinders; - h->drv[drv_index]->raid_level = drvinfo->raid_level; - memcpy(h->drv[drv_index]->serial_no, drvinfo->serial_no, 16); - memcpy(h->drv[drv_index]->vendor, drvinfo->vendor, - VENDOR_LEN + 1); - memcpy(h->drv[drv_index]->model, drvinfo->model, MODEL_LEN + 1); - memcpy(h->drv[drv_index]->rev, drvinfo->rev, REV_LEN + 1); - } - - ++h->num_luns; - disk = h->gendisk[drv_index]; - set_capacity(disk, h->drv[drv_index]->nr_blocks); - - /* If it's not disk 0 (drv_index != 0) - * or if it was disk 0, but there was previously - * no actual corresponding configured logical drive - * (raid_leve == -1) then we want to update the - * logical drive's information. - */ - if (drv_index || first_time) { - if (cciss_add_disk(h, disk, drv_index) != 0) { - cciss_free_gendisk(h, drv_index); - cciss_free_drive_info(h, drv_index); - dev_warn(&h->pdev->dev, "could not update disk %d\n", - drv_index); - --h->num_luns; - } - } - -freeret: - kfree(inq_buff); - kfree(drvinfo); - return; -mem_msg: - dev_err(&h->pdev->dev, "out of memory\n"); - goto freeret; -} - -/* This function will find the first index of the controllers drive array - * that has a null drv pointer and allocate the drive info struct and - * will return that index This is where new drives will be added. - * If the index to be returned is greater than the highest_lun index for - * the controller then highest_lun is set * to this new index. - * If there are no available indexes or if tha allocation fails, then -1 - * is returned. * "controller_node" is used to know if this is a real - * logical drive, or just the controller node, which determines if this - * counts towards highest_lun. - */ -static int cciss_alloc_drive_info(ctlr_info_t *h, int controller_node) -{ - int i; - drive_info_struct *drv; - - /* Search for an empty slot for our drive info */ - for (i = 0; i < CISS_MAX_LUN; i++) { - - /* if not cxd0 case, and it's occupied, skip it. */ - if (h->drv[i] && i != 0) - continue; - /* - * If it's cxd0 case, and drv is alloc'ed already, and a - * disk is configured there, skip it. - */ - if (i == 0 && h->drv[i] && h->drv[i]->raid_level != -1) - continue; - - /* - * We've found an empty slot. Update highest_lun - * provided this isn't just the fake cxd0 controller node. - */ - if (i > h->highest_lun && !controller_node) - h->highest_lun = i; - - /* If adding a real disk at cxd0, and it's already alloc'ed */ - if (i == 0 && h->drv[i] != NULL) - return i; - - /* - * Found an empty slot, not already alloc'ed. Allocate it. - * Mark it with raid_level == -1, so we know it's new later on. - */ - drv = kzalloc(sizeof(*drv), GFP_KERNEL); - if (!drv) - return -1; - drv->raid_level = -1; /* so we know it's new */ - h->drv[i] = drv; - return i; - } - return -1; -} - -static void cciss_free_drive_info(ctlr_info_t *h, int drv_index) -{ - kfree(h->drv[drv_index]); - h->drv[drv_index] = NULL; -} - -static void cciss_free_gendisk(ctlr_info_t *h, int drv_index) -{ - put_disk(h->gendisk[drv_index]); - h->gendisk[drv_index] = NULL; -} - -/* cciss_add_gendisk finds a free hba[]->drv structure - * and allocates a gendisk if needed, and sets the lunid - * in the drvinfo structure. It returns the index into - * the ->drv[] array, or -1 if none are free. - * is_controller_node indicates whether highest_lun should - * count this disk, or if it's only being added to provide - * a means to talk to the controller in case no logical - * drives have yet been configured. - */ -static int cciss_add_gendisk(ctlr_info_t *h, unsigned char lunid[], - int controller_node) -{ - int drv_index; - - drv_index = cciss_alloc_drive_info(h, controller_node); - if (drv_index == -1) - return -1; - - /*Check if the gendisk needs to be allocated */ - if (!h->gendisk[drv_index]) { - h->gendisk[drv_index] = - alloc_disk(1 << NWD_SHIFT); - if (!h->gendisk[drv_index]) { - dev_err(&h->pdev->dev, - "could not allocate a new disk %d\n", - drv_index); - goto err_free_drive_info; - } - } - memcpy(h->drv[drv_index]->LunID, lunid, - sizeof(h->drv[drv_index]->LunID)); - if (cciss_create_ld_sysfs_entry(h, drv_index)) - goto err_free_disk; - /* Don't need to mark this busy because nobody */ - /* else knows about this disk yet to contend */ - /* for access to it. */ - h->drv[drv_index]->busy_configuring = 0; - wmb(); - return drv_index; - -err_free_disk: - cciss_free_gendisk(h, drv_index); -err_free_drive_info: - cciss_free_drive_info(h, drv_index); - return -1; -} - -/* This is for the special case of a controller which - * has no logical drives. In this case, we still need - * to register a disk so the controller can be accessed - * by the Array Config Utility. - */ -static void cciss_add_controller_node(ctlr_info_t *h) -{ - struct gendisk *disk; - int drv_index; - - if (h->gendisk[0] != NULL) /* already did this? Then bail. */ - return; - - drv_index = cciss_add_gendisk(h, CTLR_LUNID, 1); - if (drv_index == -1) - goto error; - h->drv[drv_index]->block_size = 512; - h->drv[drv_index]->nr_blocks = 0; - h->drv[drv_index]->heads = 0; - h->drv[drv_index]->sectors = 0; - h->drv[drv_index]->cylinders = 0; - h->drv[drv_index]->raid_level = -1; - memset(h->drv[drv_index]->serial_no, 0, 16); - disk = h->gendisk[drv_index]; - if (cciss_add_disk(h, disk, drv_index) == 0) - return; - cciss_free_gendisk(h, drv_index); - cciss_free_drive_info(h, drv_index); -error: - dev_warn(&h->pdev->dev, "could not add disk 0.\n"); - return; -} - -/* This function will add and remove logical drives from the Logical - * drive array of the controller and maintain persistency of ordering - * so that mount points are preserved until the next reboot. This allows - * for the removal of logical drives in the middle of the drive array - * without a re-ordering of those drives. - * INPUT - * h = The controller to perform the operations on - */ -static int rebuild_lun_table(ctlr_info_t *h, int first_time, - int via_ioctl) -{ - int num_luns; - ReportLunData_struct *ld_buff = NULL; - int return_code; - int listlength = 0; - int i; - int drv_found; - int drv_index = 0; - unsigned char lunid[8] = CTLR_LUNID; - unsigned long flags; - - if (!capable(CAP_SYS_RAWIO)) - return -EPERM; - - /* Set busy_configuring flag for this operation */ - spin_lock_irqsave(&h->lock, flags); - if (h->busy_configuring) { - spin_unlock_irqrestore(&h->lock, flags); - return -EBUSY; - } - h->busy_configuring = 1; - spin_unlock_irqrestore(&h->lock, flags); - - ld_buff = kzalloc(sizeof(ReportLunData_struct), GFP_KERNEL); - if (ld_buff == NULL) - goto mem_msg; - - return_code = sendcmd_withirq(h, CISS_REPORT_LOG, ld_buff, - sizeof(ReportLunData_struct), - 0, CTLR_LUNID, TYPE_CMD); - - if (return_code == IO_OK) - listlength = be32_to_cpu(*(__be32 *) ld_buff->LUNListLength); - else { /* reading number of logical volumes failed */ - dev_warn(&h->pdev->dev, - "report logical volume command failed\n"); - listlength = 0; - goto freeret; - } - - num_luns = listlength / 8; /* 8 bytes per entry */ - if (num_luns > CISS_MAX_LUN) { - num_luns = CISS_MAX_LUN; - dev_warn(&h->pdev->dev, "more luns configured" - " on controller than can be handled by" - " this driver.\n"); - } - - if (num_luns == 0) - cciss_add_controller_node(h); - - /* Compare controller drive array to driver's drive array - * to see if any drives are missing on the controller due - * to action of Array Config Utility (user deletes drive) - * and deregister logical drives which have disappeared. - */ - for (i = 0; i <= h->highest_lun; i++) { - int j; - drv_found = 0; - - /* skip holes in the array from already deleted drives */ - if (h->drv[i] == NULL) - continue; - - for (j = 0; j < num_luns; j++) { - memcpy(lunid, &ld_buff->LUN[j][0], sizeof(lunid)); - if (memcmp(h->drv[i]->LunID, lunid, - sizeof(lunid)) == 0) { - drv_found = 1; - break; - } - } - if (!drv_found) { - /* Deregister it from the OS, it's gone. */ - spin_lock_irqsave(&h->lock, flags); - h->drv[i]->busy_configuring = 1; - spin_unlock_irqrestore(&h->lock, flags); - return_code = deregister_disk(h, i, 1, via_ioctl); - if (h->drv[i] != NULL) - h->drv[i]->busy_configuring = 0; - } - } - - /* Compare controller drive array to driver's drive array. - * Check for updates in the drive information and any new drives - * on the controller due to ACU adding logical drives, or changing - * a logical drive's size, etc. Reregister any new/changed drives - */ - for (i = 0; i < num_luns; i++) { - int j; - - drv_found = 0; - - memcpy(lunid, &ld_buff->LUN[i][0], sizeof(lunid)); - /* Find if the LUN is already in the drive array - * of the driver. If so then update its info - * if not in use. If it does not exist then find - * the first free index and add it. - */ - for (j = 0; j <= h->highest_lun; j++) { - if (h->drv[j] != NULL && - memcmp(h->drv[j]->LunID, lunid, - sizeof(h->drv[j]->LunID)) == 0) { - drv_index = j; - drv_found = 1; - break; - } - } - - /* check if the drive was found already in the array */ - if (!drv_found) { - drv_index = cciss_add_gendisk(h, lunid, 0); - if (drv_index == -1) - goto freeret; - } - cciss_update_drive_info(h, drv_index, first_time, via_ioctl); - } /* end for */ - -freeret: - kfree(ld_buff); - h->busy_configuring = 0; - /* We return -1 here to tell the ACU that we have registered/updated - * all of the drives that we can and to keep it from calling us - * additional times. - */ - return -1; -mem_msg: - dev_err(&h->pdev->dev, "out of memory\n"); - h->busy_configuring = 0; - goto freeret; -} - -static void cciss_clear_drive_info(drive_info_struct *drive_info) -{ - /* zero out the disk size info */ - drive_info->nr_blocks = 0; - drive_info->block_size = 0; - drive_info->heads = 0; - drive_info->sectors = 0; - drive_info->cylinders = 0; - drive_info->raid_level = -1; - memset(drive_info->serial_no, 0, sizeof(drive_info->serial_no)); - memset(drive_info->model, 0, sizeof(drive_info->model)); - memset(drive_info->rev, 0, sizeof(drive_info->rev)); - memset(drive_info->vendor, 0, sizeof(drive_info->vendor)); - /* - * don't clear the LUNID though, we need to remember which - * one this one is. - */ -} - -/* This function will deregister the disk and it's queue from the - * kernel. It must be called with the controller lock held and the - * drv structures busy_configuring flag set. It's parameters are: - * - * disk = This is the disk to be deregistered - * drv = This is the drive_info_struct associated with the disk to be - * deregistered. It contains information about the disk used - * by the driver. - * clear_all = This flag determines whether or not the disk information - * is going to be completely cleared out and the highest_lun - * reset. Sometimes we want to clear out information about - * the disk in preparation for re-adding it. In this case - * the highest_lun should be left unchanged and the LunID - * should not be cleared. - * via_ioctl - * This indicates whether we've reached this path via ioctl. - * This affects the maximum usage count allowed for c0d0 to be messed with. - * If this path is reached via ioctl(), then the max_usage_count will - * be 1, as the process calling ioctl() has got to have the device open. - * If we get here via sysfs, then the max usage count will be zero. -*/ -static int deregister_disk(ctlr_info_t *h, int drv_index, - int clear_all, int via_ioctl) -{ - int i; - struct gendisk *disk; - drive_info_struct *drv; - int recalculate_highest_lun; - - if (!capable(CAP_SYS_RAWIO)) - return -EPERM; - - drv = h->drv[drv_index]; - disk = h->gendisk[drv_index]; - - /* make sure logical volume is NOT is use */ - if (clear_all || (h->gendisk[0] == disk)) { - if (drv->usage_count > via_ioctl) - return -EBUSY; - } else if (drv->usage_count > 0) - return -EBUSY; - - recalculate_highest_lun = (drv == h->drv[h->highest_lun]); - - /* invalidate the devices and deregister the disk. If it is disk - * zero do not deregister it but just zero out it's values. This - * allows us to delete disk zero but keep the controller registered. - */ - if (h->gendisk[0] != disk) { - struct request_queue *q = disk->queue; - if (disk->flags & GENHD_FL_UP) { - cciss_destroy_ld_sysfs_entry(h, drv_index, 0); - del_gendisk(disk); - } - if (q) - blk_cleanup_queue(q); - /* If clear_all is set then we are deleting the logical - * drive, not just refreshing its info. For drives - * other than disk 0 we will call put_disk. We do not - * do this for disk 0 as we need it to be able to - * configure the controller. - */ - if (clear_all){ - /* This isn't pretty, but we need to find the - * disk in our array and NULL our the pointer. - * This is so that we will call alloc_disk if - * this index is used again later. - */ - for (i=0; i < CISS_MAX_LUN; i++){ - if (h->gendisk[i] == disk) { - h->gendisk[i] = NULL; - break; - } - } - put_disk(disk); - } - } else { - set_capacity(disk, 0); - cciss_clear_drive_info(drv); - } - - --h->num_luns; - - /* if it was the last disk, find the new hightest lun */ - if (clear_all && recalculate_highest_lun) { - int newhighest = -1; - for (i = 0; i <= h->highest_lun; i++) { - /* if the disk has size > 0, it is available */ - if (h->drv[i] && h->drv[i]->heads) - newhighest = i; - } - h->highest_lun = newhighest; - } - return 0; -} - -static int fill_cmd(ctlr_info_t *h, CommandList_struct *c, __u8 cmd, void *buff, - size_t size, __u8 page_code, unsigned char *scsi3addr, - int cmd_type) -{ - u64bit buff_dma_handle; - int status = IO_OK; - - c->cmd_type = CMD_IOCTL_PEND; - c->Header.ReplyQueue = 0; - if (buff != NULL) { - c->Header.SGList = 1; - c->Header.SGTotal = 1; - } else { - c->Header.SGList = 0; - c->Header.SGTotal = 0; - } - c->Header.Tag.lower = c->busaddr; - memcpy(c->Header.LUN.LunAddrBytes, scsi3addr, 8); - - c->Request.Type.Type = cmd_type; - if (cmd_type == TYPE_CMD) { - switch (cmd) { - case CISS_INQUIRY: - /* are we trying to read a vital product page */ - if (page_code != 0) { - c->Request.CDB[1] = 0x01; - c->Request.CDB[2] = page_code; - } - c->Request.CDBLen = 6; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_READ; - c->Request.Timeout = 0; - c->Request.CDB[0] = CISS_INQUIRY; - c->Request.CDB[4] = size & 0xFF; - break; - case CISS_REPORT_LOG: - case CISS_REPORT_PHYS: - /* Talking to controller so It's a physical command - mode = 00 target = 0. Nothing to write. - */ - c->Request.CDBLen = 12; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_READ; - c->Request.Timeout = 0; - c->Request.CDB[0] = cmd; - c->Request.CDB[6] = (size >> 24) & 0xFF; /* MSB */ - c->Request.CDB[7] = (size >> 16) & 0xFF; - c->Request.CDB[8] = (size >> 8) & 0xFF; - c->Request.CDB[9] = size & 0xFF; - break; - - case CCISS_READ_CAPACITY: - c->Request.CDBLen = 10; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_READ; - c->Request.Timeout = 0; - c->Request.CDB[0] = cmd; - break; - case CCISS_READ_CAPACITY_16: - c->Request.CDBLen = 16; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_READ; - c->Request.Timeout = 0; - c->Request.CDB[0] = cmd; - c->Request.CDB[1] = 0x10; - c->Request.CDB[10] = (size >> 24) & 0xFF; - c->Request.CDB[11] = (size >> 16) & 0xFF; - c->Request.CDB[12] = (size >> 8) & 0xFF; - c->Request.CDB[13] = size & 0xFF; - c->Request.Timeout = 0; - c->Request.CDB[0] = cmd; - break; - case CCISS_CACHE_FLUSH: - c->Request.CDBLen = 12; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_WRITE; - c->Request.Timeout = 0; - c->Request.CDB[0] = BMIC_WRITE; - c->Request.CDB[6] = BMIC_CACHE_FLUSH; - c->Request.CDB[7] = (size >> 8) & 0xFF; - c->Request.CDB[8] = size & 0xFF; - break; - case TEST_UNIT_READY: - c->Request.CDBLen = 6; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_NONE; - c->Request.Timeout = 0; - break; - default: - dev_warn(&h->pdev->dev, "Unknown Command 0x%c\n", cmd); - return IO_ERROR; - } - } else if (cmd_type == TYPE_MSG) { - switch (cmd) { - case CCISS_ABORT_MSG: - c->Request.CDBLen = 12; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_WRITE; - c->Request.Timeout = 0; - c->Request.CDB[0] = cmd; /* abort */ - c->Request.CDB[1] = 0; /* abort a command */ - /* buff contains the tag of the command to abort */ - memcpy(&c->Request.CDB[4], buff, 8); - break; - case CCISS_RESET_MSG: - c->Request.CDBLen = 16; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_NONE; - c->Request.Timeout = 0; - memset(&c->Request.CDB[0], 0, sizeof(c->Request.CDB)); - c->Request.CDB[0] = cmd; /* reset */ - c->Request.CDB[1] = CCISS_RESET_TYPE_TARGET; - break; - case CCISS_NOOP_MSG: - c->Request.CDBLen = 1; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = XFER_WRITE; - c->Request.Timeout = 0; - c->Request.CDB[0] = cmd; - break; - default: - dev_warn(&h->pdev->dev, - "unknown message type %d\n", cmd); - return IO_ERROR; - } - } else { - dev_warn(&h->pdev->dev, "unknown command type %d\n", cmd_type); - return IO_ERROR; - } - /* Fill in the scatter gather information */ - if (size > 0) { - buff_dma_handle.val = (__u64) pci_map_single(h->pdev, - buff, size, - PCI_DMA_BIDIRECTIONAL); - c->SG[0].Addr.lower = buff_dma_handle.val32.lower; - c->SG[0].Addr.upper = buff_dma_handle.val32.upper; - c->SG[0].Len = size; - c->SG[0].Ext = 0; /* we are not chaining */ - } - return status; -} - -static int cciss_send_reset(ctlr_info_t *h, unsigned char *scsi3addr, - u8 reset_type) -{ - CommandList_struct *c; - int return_status; - - c = cmd_alloc(h); - if (!c) - return -ENOMEM; - return_status = fill_cmd(h, c, CCISS_RESET_MSG, NULL, 0, 0, - CTLR_LUNID, TYPE_MSG); - c->Request.CDB[1] = reset_type; /* fill_cmd defaults to target reset */ - if (return_status != IO_OK) { - cmd_special_free(h, c); - return return_status; - } - c->waiting = NULL; - enqueue_cmd_and_start_io(h, c); - /* Don't wait for completion, the reset won't complete. Don't free - * the command either. This is the last command we will send before - * re-initializing everything, so it doesn't matter and won't leak. - */ - return 0; -} - -static int check_target_status(ctlr_info_t *h, CommandList_struct *c) -{ - switch (c->err_info->ScsiStatus) { - case SAM_STAT_GOOD: - return IO_OK; - case SAM_STAT_CHECK_CONDITION: - switch (0xf & c->err_info->SenseInfo[2]) { - case 0: return IO_OK; /* no sense */ - case 1: return IO_OK; /* recovered error */ - default: - if (check_for_unit_attention(h, c)) - return IO_NEEDS_RETRY; - dev_warn(&h->pdev->dev, "cmd 0x%02x " - "check condition, sense key = 0x%02x\n", - c->Request.CDB[0], c->err_info->SenseInfo[2]); - } - break; - default: - dev_warn(&h->pdev->dev, "cmd 0x%02x" - "scsi status = 0x%02x\n", - c->Request.CDB[0], c->err_info->ScsiStatus); - break; - } - return IO_ERROR; -} - -static int process_sendcmd_error(ctlr_info_t *h, CommandList_struct *c) -{ - int return_status = IO_OK; - - if (c->err_info->CommandStatus == CMD_SUCCESS) - return IO_OK; - - switch (c->err_info->CommandStatus) { - case CMD_TARGET_STATUS: - return_status = check_target_status(h, c); - break; - case CMD_DATA_UNDERRUN: - case CMD_DATA_OVERRUN: - /* expected for inquiry and report lun commands */ - break; - case CMD_INVALID: - dev_warn(&h->pdev->dev, "cmd 0x%02x is " - "reported invalid\n", c->Request.CDB[0]); - return_status = IO_ERROR; - break; - case CMD_PROTOCOL_ERR: - dev_warn(&h->pdev->dev, "cmd 0x%02x has " - "protocol error\n", c->Request.CDB[0]); - return_status = IO_ERROR; - break; - case CMD_HARDWARE_ERR: - dev_warn(&h->pdev->dev, "cmd 0x%02x had " - " hardware error\n", c->Request.CDB[0]); - return_status = IO_ERROR; - break; - case CMD_CONNECTION_LOST: - dev_warn(&h->pdev->dev, "cmd 0x%02x had " - "connection lost\n", c->Request.CDB[0]); - return_status = IO_ERROR; - break; - case CMD_ABORTED: - dev_warn(&h->pdev->dev, "cmd 0x%02x was " - "aborted\n", c->Request.CDB[0]); - return_status = IO_ERROR; - break; - case CMD_ABORT_FAILED: - dev_warn(&h->pdev->dev, "cmd 0x%02x reports " - "abort failed\n", c->Request.CDB[0]); - return_status = IO_ERROR; - break; - case CMD_UNSOLICITED_ABORT: - dev_warn(&h->pdev->dev, "unsolicited abort 0x%02x\n", - c->Request.CDB[0]); - return_status = IO_NEEDS_RETRY; - break; - case CMD_UNABORTABLE: - dev_warn(&h->pdev->dev, "cmd unabortable\n"); - return_status = IO_ERROR; - break; - default: - dev_warn(&h->pdev->dev, "cmd 0x%02x returned " - "unknown status %x\n", c->Request.CDB[0], - c->err_info->CommandStatus); - return_status = IO_ERROR; - } - return return_status; -} - -static int sendcmd_withirq_core(ctlr_info_t *h, CommandList_struct *c, - int attempt_retry) -{ - DECLARE_COMPLETION_ONSTACK(wait); - u64bit buff_dma_handle; - int return_status = IO_OK; - -resend_cmd2: - c->waiting = &wait; - enqueue_cmd_and_start_io(h, c); - - wait_for_completion(&wait); - - if (c->err_info->CommandStatus == 0 || !attempt_retry) - goto command_done; - - return_status = process_sendcmd_error(h, c); - - if (return_status == IO_NEEDS_RETRY && - c->retry_count < MAX_CMD_RETRIES) { - dev_warn(&h->pdev->dev, "retrying 0x%02x\n", - c->Request.CDB[0]); - c->retry_count++; - /* erase the old error information */ - memset(c->err_info, 0, sizeof(ErrorInfo_struct)); - return_status = IO_OK; - reinit_completion(&wait); - goto resend_cmd2; - } - -command_done: - /* unlock the buffers from DMA */ - buff_dma_handle.val32.lower = c->SG[0].Addr.lower; - buff_dma_handle.val32.upper = c->SG[0].Addr.upper; - pci_unmap_single(h->pdev, (dma_addr_t) buff_dma_handle.val, - c->SG[0].Len, PCI_DMA_BIDIRECTIONAL); - return return_status; -} - -static int sendcmd_withirq(ctlr_info_t *h, __u8 cmd, void *buff, size_t size, - __u8 page_code, unsigned char scsi3addr[], - int cmd_type) -{ - CommandList_struct *c; - int return_status; - - c = cmd_special_alloc(h); - if (!c) - return -ENOMEM; - return_status = fill_cmd(h, c, cmd, buff, size, page_code, - scsi3addr, cmd_type); - if (return_status == IO_OK) - return_status = sendcmd_withirq_core(h, c, 1); - - cmd_special_free(h, c); - return return_status; -} - -static void cciss_geometry_inquiry(ctlr_info_t *h, int logvol, - sector_t total_size, - unsigned int block_size, - InquiryData_struct *inq_buff, - drive_info_struct *drv) -{ - int return_code; - unsigned long t; - unsigned char scsi3addr[8]; - - memset(inq_buff, 0, sizeof(InquiryData_struct)); - log_unit_to_scsi3addr(h, scsi3addr, logvol); - return_code = sendcmd_withirq(h, CISS_INQUIRY, inq_buff, - sizeof(*inq_buff), 0xC1, scsi3addr, TYPE_CMD); - if (return_code == IO_OK) { - if (inq_buff->data_byte[8] == 0xFF) { - dev_warn(&h->pdev->dev, - "reading geometry failed, volume " - "does not support reading geometry\n"); - drv->heads = 255; - drv->sectors = 32; /* Sectors per track */ - drv->cylinders = total_size + 1; - drv->raid_level = RAID_UNKNOWN; - } else { - drv->heads = inq_buff->data_byte[6]; - drv->sectors = inq_buff->data_byte[7]; - drv->cylinders = (inq_buff->data_byte[4] & 0xff) << 8; - drv->cylinders += inq_buff->data_byte[5]; - drv->raid_level = inq_buff->data_byte[8]; - } - drv->block_size = block_size; - drv->nr_blocks = total_size + 1; - t = drv->heads * drv->sectors; - if (t > 1) { - sector_t real_size = total_size + 1; - unsigned long rem = sector_div(real_size, t); - if (rem) - real_size++; - drv->cylinders = real_size; - } - } else { /* Get geometry failed */ - dev_warn(&h->pdev->dev, "reading geometry failed\n"); - } -} - -static void -cciss_read_capacity(ctlr_info_t *h, int logvol, sector_t *total_size, - unsigned int *block_size) -{ - ReadCapdata_struct *buf; - int return_code; - unsigned char scsi3addr[8]; - - buf = kzalloc(sizeof(ReadCapdata_struct), GFP_KERNEL); - if (!buf) { - dev_warn(&h->pdev->dev, "out of memory\n"); - return; - } - - log_unit_to_scsi3addr(h, scsi3addr, logvol); - return_code = sendcmd_withirq(h, CCISS_READ_CAPACITY, buf, - sizeof(ReadCapdata_struct), 0, scsi3addr, TYPE_CMD); - if (return_code == IO_OK) { - *total_size = be32_to_cpu(*(__be32 *) buf->total_size); - *block_size = be32_to_cpu(*(__be32 *) buf->block_size); - } else { /* read capacity command failed */ - dev_warn(&h->pdev->dev, "read capacity failed\n"); - *total_size = 0; - *block_size = BLOCK_SIZE; - } - kfree(buf); -} - -static void cciss_read_capacity_16(ctlr_info_t *h, int logvol, - sector_t *total_size, unsigned int *block_size) -{ - ReadCapdata_struct_16 *buf; - int return_code; - unsigned char scsi3addr[8]; - - buf = kzalloc(sizeof(ReadCapdata_struct_16), GFP_KERNEL); - if (!buf) { - dev_warn(&h->pdev->dev, "out of memory\n"); - return; - } - - log_unit_to_scsi3addr(h, scsi3addr, logvol); - return_code = sendcmd_withirq(h, CCISS_READ_CAPACITY_16, - buf, sizeof(ReadCapdata_struct_16), - 0, scsi3addr, TYPE_CMD); - if (return_code == IO_OK) { - *total_size = be64_to_cpu(*(__be64 *) buf->total_size); - *block_size = be32_to_cpu(*(__be32 *) buf->block_size); - } else { /* read capacity command failed */ - dev_warn(&h->pdev->dev, "read capacity failed\n"); - *total_size = 0; - *block_size = BLOCK_SIZE; - } - dev_info(&h->pdev->dev, " blocks= %llu block_size= %d\n", - (unsigned long long)*total_size+1, *block_size); - kfree(buf); -} - -static int cciss_revalidate(struct gendisk *disk) -{ - ctlr_info_t *h = get_host(disk); - drive_info_struct *drv = get_drv(disk); - int logvol; - int FOUND = 0; - unsigned int block_size; - sector_t total_size; - InquiryData_struct *inq_buff = NULL; - - for (logvol = 0; logvol <= h->highest_lun; logvol++) { - if (!h->drv[logvol]) - continue; - if (memcmp(h->drv[logvol]->LunID, drv->LunID, - sizeof(drv->LunID)) == 0) { - FOUND = 1; - break; - } - } - - if (!FOUND) - return 1; - - inq_buff = kmalloc(sizeof(InquiryData_struct), GFP_KERNEL); - if (inq_buff == NULL) { - dev_warn(&h->pdev->dev, "out of memory\n"); - return 1; - } - if (h->cciss_read == CCISS_READ_10) { - cciss_read_capacity(h, logvol, - &total_size, &block_size); - } else { - cciss_read_capacity_16(h, logvol, - &total_size, &block_size); - } - cciss_geometry_inquiry(h, logvol, total_size, block_size, - inq_buff, drv); - - blk_queue_logical_block_size(drv->queue, drv->block_size); - set_capacity(disk, drv->nr_blocks); - - kfree(inq_buff); - return 0; -} - -/* - * Map (physical) PCI mem into (virtual) kernel space - */ -static void __iomem *remap_pci_mem(ulong base, ulong size) -{ - ulong page_base = ((ulong) base) & PAGE_MASK; - ulong page_offs = ((ulong) base) - page_base; - void __iomem *page_remapped = ioremap(page_base, page_offs + size); - - return page_remapped ? (page_remapped + page_offs) : NULL; -} - -/* - * Takes jobs of the Q and sends them to the hardware, then puts it on - * the Q to wait for completion. - */ -static void start_io(ctlr_info_t *h) -{ - CommandList_struct *c; - - while (!list_empty(&h->reqQ)) { - c = list_entry(h->reqQ.next, CommandList_struct, list); - /* can't do anything if fifo is full */ - if ((h->access.fifo_full(h))) { - dev_warn(&h->pdev->dev, "fifo full\n"); - break; - } - - /* Get the first entry from the Request Q */ - removeQ(c); - h->Qdepth--; - - /* Tell the controller execute command */ - h->access.submit_command(h, c); - - /* Put job onto the completed Q */ - addQ(&h->cmpQ, c); - } -} - -/* Assumes that h->lock is held. */ -/* Zeros out the error record and then resends the command back */ -/* to the controller */ -static inline void resend_cciss_cmd(ctlr_info_t *h, CommandList_struct *c) -{ - /* erase the old error information */ - memset(c->err_info, 0, sizeof(ErrorInfo_struct)); - - /* add it to software queue and then send it to the controller */ - addQ(&h->reqQ, c); - h->Qdepth++; - if (h->Qdepth > h->maxQsinceinit) - h->maxQsinceinit = h->Qdepth; - - start_io(h); -} - -static inline unsigned int make_status_bytes(unsigned int scsi_status_byte, - unsigned int msg_byte, unsigned int host_byte, - unsigned int driver_byte) -{ - /* inverse of macros in scsi.h */ - return (scsi_status_byte & 0xff) | - ((msg_byte & 0xff) << 8) | - ((host_byte & 0xff) << 16) | - ((driver_byte & 0xff) << 24); -} - -static inline int evaluate_target_status(ctlr_info_t *h, - CommandList_struct *cmd, int *retry_cmd) -{ - unsigned char sense_key; - unsigned char status_byte, msg_byte, host_byte, driver_byte; - int error_value; - - *retry_cmd = 0; - /* If we get in here, it means we got "target status", that is, scsi status */ - status_byte = cmd->err_info->ScsiStatus; - driver_byte = DRIVER_OK; - msg_byte = cmd->err_info->CommandStatus; /* correct? seems too device specific */ - - if (blk_rq_is_passthrough(cmd->rq)) - host_byte = DID_PASSTHROUGH; - else - host_byte = DID_OK; - - error_value = make_status_bytes(status_byte, msg_byte, - host_byte, driver_byte); - - if (cmd->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION) { - if (!blk_rq_is_passthrough(cmd->rq)) - dev_warn(&h->pdev->dev, "cmd %p " - "has SCSI Status 0x%x\n", - cmd, cmd->err_info->ScsiStatus); - return error_value; - } - - /* check the sense key */ - sense_key = 0xf & cmd->err_info->SenseInfo[2]; - /* no status or recovered error */ - if (((sense_key == 0x0) || (sense_key == 0x1)) && - !blk_rq_is_passthrough(cmd->rq)) - error_value = 0; - - if (check_for_unit_attention(h, cmd)) { - *retry_cmd = !blk_rq_is_passthrough(cmd->rq); - return 0; - } - - /* Not SG_IO or similar? */ - if (!blk_rq_is_passthrough(cmd->rq)) { - if (error_value != 0) - dev_warn(&h->pdev->dev, "cmd %p has CHECK CONDITION" - " sense key = 0x%x\n", cmd, sense_key); - return error_value; - } - - scsi_req(cmd->rq)->sense_len = cmd->err_info->SenseLen; - return error_value; -} - -/* checks the status of the job and calls complete buffers to mark all - * buffers for the completed job. Note that this function does not need - * to hold the hba/queue lock. - */ -static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd, - int timeout) -{ - int retry_cmd = 0; - struct request *rq = cmd->rq; - struct scsi_request *sreq = scsi_req(rq); - - sreq->result = 0; - - if (timeout) - sreq->result = make_status_bytes(0, 0, 0, DRIVER_TIMEOUT); - - if (cmd->err_info->CommandStatus == 0) /* no error has occurred */ - goto after_error_processing; - - switch (cmd->err_info->CommandStatus) { - case CMD_TARGET_STATUS: - sreq->result = evaluate_target_status(h, cmd, &retry_cmd); - break; - case CMD_DATA_UNDERRUN: - if (!blk_rq_is_passthrough(cmd->rq)) { - dev_warn(&h->pdev->dev, "cmd %p has" - " completed with data underrun " - "reported\n", cmd); - } - break; - case CMD_DATA_OVERRUN: - if (!blk_rq_is_passthrough(cmd->rq)) - dev_warn(&h->pdev->dev, "cciss: cmd %p has" - " completed with data overrun " - "reported\n", cmd); - break; - case CMD_INVALID: - dev_warn(&h->pdev->dev, "cciss: cmd %p is " - "reported invalid\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - break; - case CMD_PROTOCOL_ERR: - dev_warn(&h->pdev->dev, "cciss: cmd %p has " - "protocol error\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - break; - case CMD_HARDWARE_ERR: - dev_warn(&h->pdev->dev, "cciss: cmd %p had " - " hardware error\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - break; - case CMD_CONNECTION_LOST: - dev_warn(&h->pdev->dev, "cciss: cmd %p had " - "connection lost\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - break; - case CMD_ABORTED: - dev_warn(&h->pdev->dev, "cciss: cmd %p was " - "aborted\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ABORT); - break; - case CMD_ABORT_FAILED: - dev_warn(&h->pdev->dev, "cciss: cmd %p reports " - "abort failed\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - break; - case CMD_UNSOLICITED_ABORT: - dev_warn(&h->pdev->dev, "cciss%d: unsolicited " - "abort %p\n", h->ctlr, cmd); - if (cmd->retry_count < MAX_CMD_RETRIES) { - retry_cmd = 1; - dev_warn(&h->pdev->dev, "retrying %p\n", cmd); - cmd->retry_count++; - } else - dev_warn(&h->pdev->dev, - "%p retried too many times\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ABORT); - break; - case CMD_TIMEOUT: - dev_warn(&h->pdev->dev, "cmd %p timedout\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - break; - case CMD_UNABORTABLE: - dev_warn(&h->pdev->dev, "cmd %p unabortable\n", cmd); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - break; - default: - dev_warn(&h->pdev->dev, "cmd %p returned " - "unknown status %x\n", cmd, - cmd->err_info->CommandStatus); - sreq->result = make_status_bytes(SAM_STAT_GOOD, - cmd->err_info->CommandStatus, DRIVER_OK, - blk_rq_is_passthrough(cmd->rq) ? - DID_PASSTHROUGH : DID_ERROR); - } - -after_error_processing: - - /* We need to return this command */ - if (retry_cmd) { - resend_cciss_cmd(h, cmd); - return; - } - cmd->rq->completion_data = cmd; - blk_complete_request(cmd->rq); -} - -static inline u32 cciss_tag_contains_index(u32 tag) -{ -#define DIRECT_LOOKUP_BIT 0x10 - return tag & DIRECT_LOOKUP_BIT; -} - -static inline u32 cciss_tag_to_index(u32 tag) -{ -#define DIRECT_LOOKUP_SHIFT 5 - return tag >> DIRECT_LOOKUP_SHIFT; -} - -static inline u32 cciss_tag_discard_error_bits(ctlr_info_t *h, u32 tag) -{ -#define CCISS_PERF_ERROR_BITS ((1 << DIRECT_LOOKUP_SHIFT) - 1) -#define CCISS_SIMPLE_ERROR_BITS 0x03 - if (likely(h->transMethod & CFGTBL_Trans_Performant)) - return tag & ~CCISS_PERF_ERROR_BITS; - return tag & ~CCISS_SIMPLE_ERROR_BITS; -} - -static inline void cciss_mark_tag_indexed(u32 *tag) -{ - *tag |= DIRECT_LOOKUP_BIT; -} - -static inline void cciss_set_tag_index(u32 *tag, u32 index) -{ - *tag |= (index << DIRECT_LOOKUP_SHIFT); -} - -/* - * Get a request and submit it to the controller. - */ -static void do_cciss_request(struct request_queue *q) -{ - ctlr_info_t *h = q->queuedata; - CommandList_struct *c; - sector_t start_blk; - int seg; - struct request *creq; - u64bit temp64; - struct scatterlist *tmp_sg; - SGDescriptor_struct *curr_sg; - drive_info_struct *drv; - int i, dir; - int sg_index = 0; - int chained = 0; - - queue: - creq = blk_peek_request(q); - if (!creq) - goto startio; - - BUG_ON(creq->nr_phys_segments > h->maxsgentries); - - c = cmd_alloc(h); - if (!c) - goto full; - - blk_start_request(creq); - - tmp_sg = h->scatter_list[c->cmdindex]; - spin_unlock_irq(q->queue_lock); - - c->cmd_type = CMD_RWREQ; - c->rq = creq; - - /* fill in the request */ - drv = creq->rq_disk->private_data; - c->Header.ReplyQueue = 0; /* unused in simple mode */ - /* got command from pool, so use the command block index instead */ - /* for direct lookups. */ - /* The first 2 bits are reserved for controller error reporting. */ - cciss_set_tag_index(&c->Header.Tag.lower, c->cmdindex); - cciss_mark_tag_indexed(&c->Header.Tag.lower); - memcpy(&c->Header.LUN, drv->LunID, sizeof(drv->LunID)); - c->Request.CDBLen = 10; /* 12 byte commands not in FW yet; */ - c->Request.Type.Type = TYPE_CMD; /* It is a command. */ - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = - (rq_data_dir(creq) == READ) ? XFER_READ : XFER_WRITE; - c->Request.Timeout = 0; /* Don't time out */ - c->Request.CDB[0] = - (rq_data_dir(creq) == READ) ? h->cciss_read : h->cciss_write; - start_blk = blk_rq_pos(creq); - dev_dbg(&h->pdev->dev, "sector =%d nr_sectors=%d\n", - (int)blk_rq_pos(creq), (int)blk_rq_sectors(creq)); - sg_init_table(tmp_sg, h->maxsgentries); - seg = blk_rq_map_sg(q, creq, tmp_sg); - - /* get the DMA records for the setup */ - if (c->Request.Type.Direction == XFER_READ) - dir = PCI_DMA_FROMDEVICE; - else - dir = PCI_DMA_TODEVICE; - - curr_sg = c->SG; - sg_index = 0; - chained = 0; - - for (i = 0; i < seg; i++) { - if (((sg_index+1) == (h->max_cmd_sgentries)) && - !chained && ((seg - i) > 1)) { - /* Point to next chain block. */ - curr_sg = h->cmd_sg_list[c->cmdindex]; - sg_index = 0; - chained = 1; - } - curr_sg[sg_index].Len = tmp_sg[i].length; - temp64.val = (__u64) pci_map_page(h->pdev, sg_page(&tmp_sg[i]), - tmp_sg[i].offset, - tmp_sg[i].length, dir); - if (dma_mapping_error(&h->pdev->dev, temp64.val)) { - dev_warn(&h->pdev->dev, - "%s: error mapping page for DMA\n", __func__); - scsi_req(creq)->result = - make_status_bytes(SAM_STAT_GOOD, 0, DRIVER_OK, - DID_SOFT_ERROR); - cmd_free(h, c); - return; - } - curr_sg[sg_index].Addr.lower = temp64.val32.lower; - curr_sg[sg_index].Addr.upper = temp64.val32.upper; - curr_sg[sg_index].Ext = 0; /* we are not chaining */ - ++sg_index; - } - if (chained) { - if (cciss_map_sg_chain_block(h, c, h->cmd_sg_list[c->cmdindex], - (seg - (h->max_cmd_sgentries - 1)) * - sizeof(SGDescriptor_struct))) { - scsi_req(creq)->result = - make_status_bytes(SAM_STAT_GOOD, 0, DRIVER_OK, - DID_SOFT_ERROR); - cmd_free(h, c); - return; - } - } - - /* track how many SG entries we are using */ - if (seg > h->maxSG) - h->maxSG = seg; - - dev_dbg(&h->pdev->dev, "Submitting %u sectors in %d segments " - "chained[%d]\n", - blk_rq_sectors(creq), seg, chained); - - c->Header.SGTotal = seg + chained; - if (seg <= h->max_cmd_sgentries) - c->Header.SGList = c->Header.SGTotal; - else - c->Header.SGList = h->max_cmd_sgentries; - set_performant_mode(h, c); - - switch (req_op(creq)) { - case REQ_OP_READ: - case REQ_OP_WRITE: - if(h->cciss_read == CCISS_READ_10) { - c->Request.CDB[1] = 0; - c->Request.CDB[2] = (start_blk >> 24) & 0xff; /* MSB */ - c->Request.CDB[3] = (start_blk >> 16) & 0xff; - c->Request.CDB[4] = (start_blk >> 8) & 0xff; - c->Request.CDB[5] = start_blk & 0xff; - c->Request.CDB[6] = 0; /* (sect >> 24) & 0xff; MSB */ - c->Request.CDB[7] = (blk_rq_sectors(creq) >> 8) & 0xff; - c->Request.CDB[8] = blk_rq_sectors(creq) & 0xff; - c->Request.CDB[9] = c->Request.CDB[11] = c->Request.CDB[12] = 0; - } else { - u32 upper32 = upper_32_bits(start_blk); - - c->Request.CDBLen = 16; - c->Request.CDB[1]= 0; - c->Request.CDB[2]= (upper32 >> 24) & 0xff; /* MSB */ - c->Request.CDB[3]= (upper32 >> 16) & 0xff; - c->Request.CDB[4]= (upper32 >> 8) & 0xff; - c->Request.CDB[5]= upper32 & 0xff; - c->Request.CDB[6]= (start_blk >> 24) & 0xff; - c->Request.CDB[7]= (start_blk >> 16) & 0xff; - c->Request.CDB[8]= (start_blk >> 8) & 0xff; - c->Request.CDB[9]= start_blk & 0xff; - c->Request.CDB[10]= (blk_rq_sectors(creq) >> 24) & 0xff; - c->Request.CDB[11]= (blk_rq_sectors(creq) >> 16) & 0xff; - c->Request.CDB[12]= (blk_rq_sectors(creq) >> 8) & 0xff; - c->Request.CDB[13]= blk_rq_sectors(creq) & 0xff; - c->Request.CDB[14] = c->Request.CDB[15] = 0; - } - break; - case REQ_OP_SCSI_IN: - case REQ_OP_SCSI_OUT: - c->Request.CDBLen = scsi_req(creq)->cmd_len; - memcpy(c->Request.CDB, scsi_req(creq)->cmd, BLK_MAX_CDB); - scsi_req(creq)->sense = c->err_info->SenseInfo; - break; - default: - dev_warn(&h->pdev->dev, "bad request type %d\n", - creq->cmd_flags); - BUG(); - } - - spin_lock_irq(q->queue_lock); - - addQ(&h->reqQ, c); - h->Qdepth++; - if (h->Qdepth > h->maxQsinceinit) - h->maxQsinceinit = h->Qdepth; - - goto queue; -full: - blk_stop_queue(q); -startio: - /* We will already have the driver lock here so not need - * to lock it. - */ - start_io(h); -} - -static inline unsigned long get_next_completion(ctlr_info_t *h) -{ - return h->access.command_completed(h); -} - -static inline int interrupt_pending(ctlr_info_t *h) -{ - return h->access.intr_pending(h); -} - -static inline long interrupt_not_for_us(ctlr_info_t *h) -{ - return ((h->access.intr_pending(h) == 0) || - (h->interrupts_enabled == 0)); -} - -static inline int bad_tag(ctlr_info_t *h, u32 tag_index, - u32 raw_tag) -{ - if (unlikely(tag_index >= h->nr_cmds)) { - dev_warn(&h->pdev->dev, "bad tag 0x%08x ignored.\n", raw_tag); - return 1; - } - return 0; -} - -static inline void finish_cmd(ctlr_info_t *h, CommandList_struct *c, - u32 raw_tag) -{ - removeQ(c); - if (likely(c->cmd_type == CMD_RWREQ)) - complete_command(h, c, 0); - else if (c->cmd_type == CMD_IOCTL_PEND) - complete(c->waiting); -#ifdef CONFIG_CISS_SCSI_TAPE - else if (c->cmd_type == CMD_SCSI) - complete_scsi_command(c, 0, raw_tag); -#endif -} - -static inline u32 next_command(ctlr_info_t *h) -{ - u32 a; - - if (unlikely(!(h->transMethod & CFGTBL_Trans_Performant))) - return h->access.command_completed(h); - - if ((*(h->reply_pool_head) & 1) == (h->reply_pool_wraparound)) { - a = *(h->reply_pool_head); /* Next cmd in ring buffer */ - (h->reply_pool_head)++; - h->commands_outstanding--; - } else { - a = FIFO_EMPTY; - } - /* Check for wraparound */ - if (h->reply_pool_head == (h->reply_pool + h->max_commands)) { - h->reply_pool_head = h->reply_pool; - h->reply_pool_wraparound ^= 1; - } - return a; -} - -/* process completion of an indexed ("direct lookup") command */ -static inline u32 process_indexed_cmd(ctlr_info_t *h, u32 raw_tag) -{ - u32 tag_index; - CommandList_struct *c; - - tag_index = cciss_tag_to_index(raw_tag); - if (bad_tag(h, tag_index, raw_tag)) - return next_command(h); - c = h->cmd_pool + tag_index; - finish_cmd(h, c, raw_tag); - return next_command(h); -} - -/* process completion of a non-indexed command */ -static inline u32 process_nonindexed_cmd(ctlr_info_t *h, u32 raw_tag) -{ - CommandList_struct *c = NULL; - __u32 busaddr_masked, tag_masked; - - tag_masked = cciss_tag_discard_error_bits(h, raw_tag); - list_for_each_entry(c, &h->cmpQ, list) { - busaddr_masked = cciss_tag_discard_error_bits(h, c->busaddr); - if (busaddr_masked == tag_masked) { - finish_cmd(h, c, raw_tag); - return next_command(h); - } - } - bad_tag(h, h->nr_cmds + 1, raw_tag); - return next_command(h); -} - -/* Some controllers, like p400, will give us one interrupt - * after a soft reset, even if we turned interrupts off. - * Only need to check for this in the cciss_xxx_discard_completions - * functions. - */ -static int ignore_bogus_interrupt(ctlr_info_t *h) -{ - if (likely(!reset_devices)) - return 0; - - if (likely(h->interrupts_enabled)) - return 0; - - dev_info(&h->pdev->dev, "Received interrupt while interrupts disabled " - "(known firmware bug.) Ignoring.\n"); - - return 1; -} - -static irqreturn_t cciss_intx_discard_completions(int irq, void *dev_id) -{ - ctlr_info_t *h = dev_id; - unsigned long flags; - u32 raw_tag; - - if (ignore_bogus_interrupt(h)) - return IRQ_NONE; - - if (interrupt_not_for_us(h)) - return IRQ_NONE; - spin_lock_irqsave(&h->lock, flags); - while (interrupt_pending(h)) { - raw_tag = get_next_completion(h); - while (raw_tag != FIFO_EMPTY) - raw_tag = next_command(h); - } - spin_unlock_irqrestore(&h->lock, flags); - return IRQ_HANDLED; -} - -static irqreturn_t cciss_msix_discard_completions(int irq, void *dev_id) -{ - ctlr_info_t *h = dev_id; - unsigned long flags; - u32 raw_tag; - - if (ignore_bogus_interrupt(h)) - return IRQ_NONE; - - spin_lock_irqsave(&h->lock, flags); - raw_tag = get_next_completion(h); - while (raw_tag != FIFO_EMPTY) - raw_tag = next_command(h); - spin_unlock_irqrestore(&h->lock, flags); - return IRQ_HANDLED; -} - -static irqreturn_t do_cciss_intx(int irq, void *dev_id) -{ - ctlr_info_t *h = dev_id; - unsigned long flags; - u32 raw_tag; - - if (interrupt_not_for_us(h)) - return IRQ_NONE; - spin_lock_irqsave(&h->lock, flags); - while (interrupt_pending(h)) { - raw_tag = get_next_completion(h); - while (raw_tag != FIFO_EMPTY) { - if (cciss_tag_contains_index(raw_tag)) - raw_tag = process_indexed_cmd(h, raw_tag); - else - raw_tag = process_nonindexed_cmd(h, raw_tag); - } - } - spin_unlock_irqrestore(&h->lock, flags); - return IRQ_HANDLED; -} - -/* Add a second interrupt handler for MSI/MSI-X mode. In this mode we never - * check the interrupt pending register because it is not set. - */ -static irqreturn_t do_cciss_msix_intr(int irq, void *dev_id) -{ - ctlr_info_t *h = dev_id; - unsigned long flags; - u32 raw_tag; - - spin_lock_irqsave(&h->lock, flags); - raw_tag = get_next_completion(h); - while (raw_tag != FIFO_EMPTY) { - if (cciss_tag_contains_index(raw_tag)) - raw_tag = process_indexed_cmd(h, raw_tag); - else - raw_tag = process_nonindexed_cmd(h, raw_tag); - } - spin_unlock_irqrestore(&h->lock, flags); - return IRQ_HANDLED; -} - -/** - * add_to_scan_list() - add controller to rescan queue - * @h: Pointer to the controller. - * - * Adds the controller to the rescan queue if not already on the queue. - * - * returns 1 if added to the queue, 0 if skipped (could be on the - * queue already, or the controller could be initializing or shutting - * down). - **/ -static int add_to_scan_list(struct ctlr_info *h) -{ - struct ctlr_info *test_h; - int found = 0; - int ret = 0; - - if (h->busy_initializing) - return 0; - - if (!mutex_trylock(&h->busy_shutting_down)) - return 0; - - mutex_lock(&scan_mutex); - list_for_each_entry(test_h, &scan_q, scan_list) { - if (test_h == h) { - found = 1; - break; - } - } - if (!found && !h->busy_scanning) { - reinit_completion(&h->scan_wait); - list_add_tail(&h->scan_list, &scan_q); - ret = 1; - } - mutex_unlock(&scan_mutex); - mutex_unlock(&h->busy_shutting_down); - - return ret; -} - -/** - * remove_from_scan_list() - remove controller from rescan queue - * @h: Pointer to the controller. - * - * Removes the controller from the rescan queue if present. Blocks if - * the controller is currently conducting a rescan. The controller - * can be in one of three states: - * 1. Doesn't need a scan - * 2. On the scan list, but not scanning yet (we remove it) - * 3. Busy scanning (and not on the list). In this case we want to wait for - * the scan to complete to make sure the scanning thread for this - * controller is completely idle. - **/ -static void remove_from_scan_list(struct ctlr_info *h) -{ - struct ctlr_info *test_h, *tmp_h; - - mutex_lock(&scan_mutex); - list_for_each_entry_safe(test_h, tmp_h, &scan_q, scan_list) { - if (test_h == h) { /* state 2. */ - list_del(&h->scan_list); - complete_all(&h->scan_wait); - mutex_unlock(&scan_mutex); - return; - } - } - if (h->busy_scanning) { /* state 3. */ - mutex_unlock(&scan_mutex); - wait_for_completion(&h->scan_wait); - } else { /* state 1, nothing to do. */ - mutex_unlock(&scan_mutex); - } -} - -/** - * scan_thread() - kernel thread used to rescan controllers - * @data: Ignored. - * - * A kernel thread used scan for drive topology changes on - * controllers. The thread processes only one controller at a time - * using a queue. Controllers are added to the queue using - * add_to_scan_list() and removed from the queue either after done - * processing or using remove_from_scan_list(). - * - * returns 0. - **/ -static int scan_thread(void *data) -{ - struct ctlr_info *h; - - while (1) { - set_current_state(TASK_INTERRUPTIBLE); - schedule(); - if (kthread_should_stop()) - break; - - while (1) { - mutex_lock(&scan_mutex); - if (list_empty(&scan_q)) { - mutex_unlock(&scan_mutex); - break; - } - - h = list_entry(scan_q.next, - struct ctlr_info, - scan_list); - list_del(&h->scan_list); - h->busy_scanning = 1; - mutex_unlock(&scan_mutex); - - rebuild_lun_table(h, 0, 0); - complete_all(&h->scan_wait); - mutex_lock(&scan_mutex); - h->busy_scanning = 0; - mutex_unlock(&scan_mutex); - } - } - - return 0; -} - -static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c) -{ - if (c->err_info->SenseInfo[2] != UNIT_ATTENTION) - return 0; - - switch (c->err_info->SenseInfo[12]) { - case STATE_CHANGED: - dev_warn(&h->pdev->dev, "a state change " - "detected, command retried\n"); - return 1; - break; - case LUN_FAILED: - dev_warn(&h->pdev->dev, "LUN failure " - "detected, action required\n"); - return 1; - break; - case REPORT_LUNS_CHANGED: - dev_warn(&h->pdev->dev, "report LUN data changed\n"); - /* - * Here, we could call add_to_scan_list and wake up the scan thread, - * except that it's quite likely that we will get more than one - * REPORT_LUNS_CHANGED condition in quick succession, which means - * that those which occur after the first one will likely happen - * *during* the scan_thread's rescan. And the rescan code is not - * robust enough to restart in the middle, undoing what it has already - * done, and it's not clear that it's even possible to do this, since - * part of what it does is notify the block layer, which starts - * doing it's own i/o to read partition tables and so on, and the - * driver doesn't have visibility to know what might need undoing. - * In any event, if possible, it is horribly complicated to get right - * so we just don't do it for now. - * - * Note: this REPORT_LUNS_CHANGED condition only occurs on the MSA2012. - */ - return 1; - break; - case POWER_OR_RESET: - dev_warn(&h->pdev->dev, - "a power on or device reset detected\n"); - return 1; - break; - case UNIT_ATTENTION_CLEARED: - dev_warn(&h->pdev->dev, - "unit attention cleared by another initiator\n"); - return 1; - break; - default: - dev_warn(&h->pdev->dev, "unknown unit attention detected\n"); - return 1; - } -} - -/* - * We cannot read the structure directly, for portability we must use - * the io functions. - * This is for debug only. - */ -static void print_cfg_table(ctlr_info_t *h) -{ - int i; - char temp_name[17]; - CfgTable_struct *tb = h->cfgtable; - - dev_dbg(&h->pdev->dev, "Controller Configuration information\n"); - dev_dbg(&h->pdev->dev, "------------------------------------\n"); - for (i = 0; i < 4; i++) - temp_name[i] = readb(&(tb->Signature[i])); - temp_name[4] = '\0'; - dev_dbg(&h->pdev->dev, " Signature = %s\n", temp_name); - dev_dbg(&h->pdev->dev, " Spec Number = %d\n", - readl(&(tb->SpecValence))); - dev_dbg(&h->pdev->dev, " Transport methods supported = 0x%x\n", - readl(&(tb->TransportSupport))); - dev_dbg(&h->pdev->dev, " Transport methods active = 0x%x\n", - readl(&(tb->TransportActive))); - dev_dbg(&h->pdev->dev, " Requested transport Method = 0x%x\n", - readl(&(tb->HostWrite.TransportRequest))); - dev_dbg(&h->pdev->dev, " Coalesce Interrupt Delay = 0x%x\n", - readl(&(tb->HostWrite.CoalIntDelay))); - dev_dbg(&h->pdev->dev, " Coalesce Interrupt Count = 0x%x\n", - readl(&(tb->HostWrite.CoalIntCount))); - dev_dbg(&h->pdev->dev, " Max outstanding commands = 0x%x\n", - readl(&(tb->CmdsOutMax))); - dev_dbg(&h->pdev->dev, " Bus Types = 0x%x\n", - readl(&(tb->BusTypes))); - for (i = 0; i < 16; i++) - temp_name[i] = readb(&(tb->ServerName[i])); - temp_name[16] = '\0'; - dev_dbg(&h->pdev->dev, " Server Name = %s\n", temp_name); - dev_dbg(&h->pdev->dev, " Heartbeat Counter = 0x%x\n\n\n", - readl(&(tb->HeartBeat))); -} - -static int find_PCI_BAR_index(struct pci_dev *pdev, unsigned long pci_bar_addr) -{ - int i, offset, mem_type, bar_type; - if (pci_bar_addr == PCI_BASE_ADDRESS_0) /* looking for BAR zero? */ - return 0; - offset = 0; - for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) { - bar_type = pci_resource_flags(pdev, i) & PCI_BASE_ADDRESS_SPACE; - if (bar_type == PCI_BASE_ADDRESS_SPACE_IO) - offset += 4; - else { - mem_type = pci_resource_flags(pdev, i) & - PCI_BASE_ADDRESS_MEM_TYPE_MASK; - switch (mem_type) { - case PCI_BASE_ADDRESS_MEM_TYPE_32: - case PCI_BASE_ADDRESS_MEM_TYPE_1M: - offset += 4; /* 32 bit */ - break; - case PCI_BASE_ADDRESS_MEM_TYPE_64: - offset += 8; - break; - default: /* reserved in PCI 2.2 */ - dev_warn(&pdev->dev, - "Base address is invalid\n"); - return -1; - break; - } - } - if (offset == pci_bar_addr - PCI_BASE_ADDRESS_0) - return i + 1; - } - return -1; -} - -/* Fill in bucket_map[], given nsgs (the max number of - * scatter gather elements supported) and bucket[], - * which is an array of 8 integers. The bucket[] array - * contains 8 different DMA transfer sizes (in 16 - * byte increments) which the controller uses to fetch - * commands. This function fills in bucket_map[], which - * maps a given number of scatter gather elements to one of - * the 8 DMA transfer sizes. The point of it is to allow the - * controller to only do as much DMA as needed to fetch the - * command, with the DMA transfer size encoded in the lower - * bits of the command address. - */ -static void calc_bucket_map(int bucket[], int num_buckets, - int nsgs, int *bucket_map) -{ - int i, j, b, size; - - /* even a command with 0 SGs requires 4 blocks */ -#define MINIMUM_TRANSFER_BLOCKS 4 -#define NUM_BUCKETS 8 - /* Note, bucket_map must have nsgs+1 entries. */ - for (i = 0; i <= nsgs; i++) { - /* Compute size of a command with i SG entries */ - size = i + MINIMUM_TRANSFER_BLOCKS; - b = num_buckets; /* Assume the biggest bucket */ - /* Find the bucket that is just big enough */ - for (j = 0; j < 8; j++) { - if (bucket[j] >= size) { - b = j; - break; - } - } - /* for a command with i SG entries, use bucket b. */ - bucket_map[i] = b; - } -} - -static void cciss_wait_for_mode_change_ack(ctlr_info_t *h) -{ - int i; - - /* under certain very rare conditions, this can take awhile. - * (e.g.: hot replace a failed 144GB drive in a RAID 5 set right - * as we enter this code.) */ - for (i = 0; i < MAX_CONFIG_WAIT; i++) { - if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq)) - break; - usleep_range(10000, 20000); - } -} - -static void cciss_enter_performant_mode(ctlr_info_t *h, u32 use_short_tags) -{ - /* This is a bit complicated. There are 8 registers on - * the controller which we write to to tell it 8 different - * sizes of commands which there may be. It's a way of - * reducing the DMA done to fetch each command. Encoded into - * each command's tag are 3 bits which communicate to the controller - * which of the eight sizes that command fits within. The size of - * each command depends on how many scatter gather entries there are. - * Each SG entry requires 16 bytes. The eight registers are programmed - * with the number of 16-byte blocks a command of that size requires. - * The smallest command possible requires 5 such 16 byte blocks. - * the largest command possible requires MAXSGENTRIES + 4 16-byte - * blocks. Note, this only extends to the SG entries contained - * within the command block, and does not extend to chained blocks - * of SG elements. bft[] contains the eight values we write to - * the registers. They are not evenly distributed, but have more - * sizes for small commands, and fewer sizes for larger commands. - */ - __u32 trans_offset; - int bft[8] = { 5, 6, 8, 10, 12, 20, 28, MAXSGENTRIES + 4}; - /* - * 5 = 1 s/g entry or 4k - * 6 = 2 s/g entry or 8k - * 8 = 4 s/g entry or 16k - * 10 = 6 s/g entry or 24k - */ - unsigned long register_value; - BUILD_BUG_ON(28 > MAXSGENTRIES + 4); - - h->reply_pool_wraparound = 1; /* spec: init to 1 */ - - /* Controller spec: zero out this buffer. */ - memset(h->reply_pool, 0, h->max_commands * sizeof(__u64)); - h->reply_pool_head = h->reply_pool; - - trans_offset = readl(&(h->cfgtable->TransMethodOffset)); - calc_bucket_map(bft, ARRAY_SIZE(bft), h->maxsgentries, - h->blockFetchTable); - writel(bft[0], &h->transtable->BlockFetch0); - writel(bft[1], &h->transtable->BlockFetch1); - writel(bft[2], &h->transtable->BlockFetch2); - writel(bft[3], &h->transtable->BlockFetch3); - writel(bft[4], &h->transtable->BlockFetch4); - writel(bft[5], &h->transtable->BlockFetch5); - writel(bft[6], &h->transtable->BlockFetch6); - writel(bft[7], &h->transtable->BlockFetch7); - - /* size of controller ring buffer */ - writel(h->max_commands, &h->transtable->RepQSize); - writel(1, &h->transtable->RepQCount); - writel(0, &h->transtable->RepQCtrAddrLow32); - writel(0, &h->transtable->RepQCtrAddrHigh32); - writel(h->reply_pool_dhandle, &h->transtable->RepQAddr0Low32); - writel(0, &h->transtable->RepQAddr0High32); - writel(CFGTBL_Trans_Performant | use_short_tags, - &(h->cfgtable->HostWrite.TransportRequest)); - - writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); - cciss_wait_for_mode_change_ack(h); - register_value = readl(&(h->cfgtable->TransportActive)); - if (!(register_value & CFGTBL_Trans_Performant)) - dev_warn(&h->pdev->dev, "cciss: unable to get board into" - " performant mode\n"); -} - -static void cciss_put_controller_into_performant_mode(ctlr_info_t *h) -{ - __u32 trans_support; - - if (cciss_simple_mode) - return; - - dev_dbg(&h->pdev->dev, "Trying to put board into Performant mode\n"); - /* Attempt to put controller into performant mode if supported */ - /* Does board support performant mode? */ - trans_support = readl(&(h->cfgtable->TransportSupport)); - if (!(trans_support & PERFORMANT_MODE)) - return; - - dev_dbg(&h->pdev->dev, "Placing controller into performant mode\n"); - /* Performant mode demands commands on a 32 byte boundary - * pci_alloc_consistent aligns on page boundarys already. - * Just need to check if divisible by 32 - */ - if ((sizeof(CommandList_struct) % 32) != 0) { - dev_warn(&h->pdev->dev, "%s %d %s\n", - "cciss info: command size[", - (int)sizeof(CommandList_struct), - "] not divisible by 32, no performant mode..\n"); - return; - } - - /* Performant mode ring buffer and supporting data structures */ - h->reply_pool = (__u64 *)pci_alloc_consistent( - h->pdev, h->max_commands * sizeof(__u64), - &(h->reply_pool_dhandle)); - - /* Need a block fetch table for performant mode */ - h->blockFetchTable = kmalloc(((h->maxsgentries+1) * - sizeof(__u32)), GFP_KERNEL); - - if ((h->reply_pool == NULL) || (h->blockFetchTable == NULL)) - goto clean_up; - - cciss_enter_performant_mode(h, - trans_support & CFGTBL_Trans_use_short_tags); - - /* Change the access methods to the performant access methods */ - h->access = SA5_performant_access; - h->transMethod = CFGTBL_Trans_Performant; - - return; -clean_up: - kfree(h->blockFetchTable); - if (h->reply_pool) - pci_free_consistent(h->pdev, - h->max_commands * sizeof(__u64), - h->reply_pool, - h->reply_pool_dhandle); - return; - -} /* cciss_put_controller_into_performant_mode */ - -/* If MSI/MSI-X is supported by the kernel we will try to enable it on - * controllers that are capable. If not, we use IO-APIC mode. - */ - -static void cciss_interrupt_mode(ctlr_info_t *h) -{ - int ret; - - /* Some boards advertise MSI but don't really support it */ - if ((h->board_id == 0x40700E11) || (h->board_id == 0x40800E11) || - (h->board_id == 0x40820E11) || (h->board_id == 0x40830E11)) - goto default_int_mode; - - ret = pci_alloc_irq_vectors(h->pdev, 4, 4, PCI_IRQ_MSIX); - if (ret >= 0) { - h->intr[0] = pci_irq_vector(h->pdev, 0); - h->intr[1] = pci_irq_vector(h->pdev, 1); - h->intr[2] = pci_irq_vector(h->pdev, 2); - h->intr[3] = pci_irq_vector(h->pdev, 3); - return; - } - - ret = pci_alloc_irq_vectors(h->pdev, 1, 1, PCI_IRQ_MSI); - -default_int_mode: - /* if we get here we're going to use the default interrupt mode */ - h->intr[h->intr_mode] = pci_irq_vector(h->pdev, 0); - return; -} - -static int cciss_lookup_board_id(struct pci_dev *pdev, u32 *board_id) -{ - int i; - u32 subsystem_vendor_id, subsystem_device_id; - - subsystem_vendor_id = pdev->subsystem_vendor; - subsystem_device_id = pdev->subsystem_device; - *board_id = ((subsystem_device_id << 16) & 0xffff0000) | - subsystem_vendor_id; - - for (i = 0; i < ARRAY_SIZE(products); i++) { - /* Stand aside for hpsa driver on request */ - if (cciss_allow_hpsa) - return -ENODEV; - if (*board_id == products[i].board_id) - return i; - } - dev_warn(&pdev->dev, "unrecognized board ID: 0x%08x, ignoring.\n", - *board_id); - return -ENODEV; -} - -static inline bool cciss_board_disabled(ctlr_info_t *h) -{ - u16 command; - - (void) pci_read_config_word(h->pdev, PCI_COMMAND, &command); - return ((command & PCI_COMMAND_MEMORY) == 0); -} - -static int cciss_pci_find_memory_BAR(struct pci_dev *pdev, - unsigned long *memory_bar) -{ - int i; - - for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) - if (pci_resource_flags(pdev, i) & IORESOURCE_MEM) { - /* addressing mode bits already removed */ - *memory_bar = pci_resource_start(pdev, i); - dev_dbg(&pdev->dev, "memory BAR = %lx\n", - *memory_bar); - return 0; - } - dev_warn(&pdev->dev, "no memory BAR found\n"); - return -ENODEV; -} - -static int cciss_wait_for_board_state(struct pci_dev *pdev, - void __iomem *vaddr, int wait_for_ready) -#define BOARD_READY 1 -#define BOARD_NOT_READY 0 -{ - int i, iterations; - u32 scratchpad; - - if (wait_for_ready) - iterations = CCISS_BOARD_READY_ITERATIONS; - else - iterations = CCISS_BOARD_NOT_READY_ITERATIONS; - - for (i = 0; i < iterations; i++) { - scratchpad = readl(vaddr + SA5_SCRATCHPAD_OFFSET); - if (wait_for_ready) { - if (scratchpad == CCISS_FIRMWARE_READY) - return 0; - } else { - if (scratchpad != CCISS_FIRMWARE_READY) - return 0; - } - msleep(CCISS_BOARD_READY_POLL_INTERVAL_MSECS); - } - dev_warn(&pdev->dev, "board not ready, timed out.\n"); - return -ENODEV; -} - -static int cciss_find_cfg_addrs(struct pci_dev *pdev, void __iomem *vaddr, - u32 *cfg_base_addr, u64 *cfg_base_addr_index, - u64 *cfg_offset) -{ - *cfg_base_addr = readl(vaddr + SA5_CTCFG_OFFSET); - *cfg_offset = readl(vaddr + SA5_CTMEM_OFFSET); - *cfg_base_addr &= (u32) 0x0000ffff; - *cfg_base_addr_index = find_PCI_BAR_index(pdev, *cfg_base_addr); - if (*cfg_base_addr_index == -1) { - dev_warn(&pdev->dev, "cannot find cfg_base_addr_index, " - "*cfg_base_addr = 0x%08x\n", *cfg_base_addr); - return -ENODEV; - } - return 0; -} - -static int cciss_find_cfgtables(ctlr_info_t *h) -{ - u64 cfg_offset; - u32 cfg_base_addr; - u64 cfg_base_addr_index; - u32 trans_offset; - int rc; - - rc = cciss_find_cfg_addrs(h->pdev, h->vaddr, &cfg_base_addr, - &cfg_base_addr_index, &cfg_offset); - if (rc) - return rc; - h->cfgtable = remap_pci_mem(pci_resource_start(h->pdev, - cfg_base_addr_index) + cfg_offset, sizeof(*h->cfgtable)); - if (!h->cfgtable) - return -ENOMEM; - rc = write_driver_ver_to_cfgtable(h->cfgtable); - if (rc) - return rc; - /* Find performant mode table. */ - trans_offset = readl(&h->cfgtable->TransMethodOffset); - h->transtable = remap_pci_mem(pci_resource_start(h->pdev, - cfg_base_addr_index)+cfg_offset+trans_offset, - sizeof(*h->transtable)); - if (!h->transtable) - return -ENOMEM; - return 0; -} - -static void cciss_get_max_perf_mode_cmds(struct ctlr_info *h) -{ - h->max_commands = readl(&(h->cfgtable->MaxPerformantModeCommands)); - - /* Limit commands in memory limited kdump scenario. */ - if (reset_devices && h->max_commands > 32) - h->max_commands = 32; - - if (h->max_commands < 16) { - dev_warn(&h->pdev->dev, "Controller reports " - "max supported commands of %d, an obvious lie. " - "Using 16. Ensure that firmware is up to date.\n", - h->max_commands); - h->max_commands = 16; - } -} - -/* Interrogate the hardware for some limits: - * max commands, max SG elements without chaining, and with chaining, - * SG chain block size, etc. - */ -static void cciss_find_board_params(ctlr_info_t *h) -{ - cciss_get_max_perf_mode_cmds(h); - h->nr_cmds = h->max_commands - 4 - cciss_tape_cmds; - h->maxsgentries = readl(&(h->cfgtable->MaxSGElements)); - /* - * The P600 may exhibit poor performnace under some workloads - * if we use the value in the configuration table. Limit this - * controller to MAXSGENTRIES (32) instead. - */ - if (h->board_id == 0x3225103C) - h->maxsgentries = MAXSGENTRIES; - /* - * Limit in-command s/g elements to 32 save dma'able memory. - * Howvever spec says if 0, use 31 - */ - h->max_cmd_sgentries = 31; - if (h->maxsgentries > 512) { - h->max_cmd_sgentries = 32; - h->chainsize = h->maxsgentries - h->max_cmd_sgentries + 1; - h->maxsgentries--; /* save one for chain pointer */ - } else { - h->maxsgentries = 31; /* default to traditional values */ - h->chainsize = 0; - } -} - -static inline bool CISS_signature_present(ctlr_info_t *h) -{ - if (!check_signature(h->cfgtable->Signature, "CISS", 4)) { - dev_warn(&h->pdev->dev, "not a valid CISS config table\n"); - return false; - } - return true; -} - -/* Need to enable prefetch in the SCSI core for 6400 in x86 */ -static inline void cciss_enable_scsi_prefetch(ctlr_info_t *h) -{ -#ifdef CONFIG_X86 - u32 prefetch; - - prefetch = readl(&(h->cfgtable->SCSI_Prefetch)); - prefetch |= 0x100; - writel(prefetch, &(h->cfgtable->SCSI_Prefetch)); -#endif -} - -/* Disable DMA prefetch for the P600. Otherwise an ASIC bug may result - * in a prefetch beyond physical memory. - */ -static inline void cciss_p600_dma_prefetch_quirk(ctlr_info_t *h) -{ - u32 dma_prefetch; - __u32 dma_refetch; - - if (h->board_id != 0x3225103C) - return; - dma_prefetch = readl(h->vaddr + I2O_DMA1_CFG); - dma_prefetch |= 0x8000; - writel(dma_prefetch, h->vaddr + I2O_DMA1_CFG); - pci_read_config_dword(h->pdev, PCI_COMMAND_PARITY, &dma_refetch); - dma_refetch |= 0x1; - pci_write_config_dword(h->pdev, PCI_COMMAND_PARITY, dma_refetch); -} - -static int cciss_pci_init(ctlr_info_t *h) -{ - int prod_index, err; - - prod_index = cciss_lookup_board_id(h->pdev, &h->board_id); - if (prod_index < 0) - return -ENODEV; - h->product_name = products[prod_index].product_name; - h->access = *(products[prod_index].access); - - if (cciss_board_disabled(h)) { - dev_warn(&h->pdev->dev, "controller appears to be disabled\n"); - return -ENODEV; - } - - pci_disable_link_state(h->pdev, PCIE_LINK_STATE_L0S | - PCIE_LINK_STATE_L1 | PCIE_LINK_STATE_CLKPM); - - err = pci_enable_device(h->pdev); - if (err) { - dev_warn(&h->pdev->dev, "Unable to Enable PCI device\n"); - return err; - } - - err = pci_request_regions(h->pdev, "cciss"); - if (err) { - dev_warn(&h->pdev->dev, - "Cannot obtain PCI resources, aborting\n"); - return err; - } - - dev_dbg(&h->pdev->dev, "irq = %x\n", h->pdev->irq); - dev_dbg(&h->pdev->dev, "board_id = %x\n", h->board_id); - -/* If the kernel supports MSI/MSI-X we will try to enable that functionality, - * else we use the IO-APIC interrupt assigned to us by system ROM. - */ - cciss_interrupt_mode(h); - err = cciss_pci_find_memory_BAR(h->pdev, &h->paddr); - if (err) - goto err_out_free_res; - h->vaddr = remap_pci_mem(h->paddr, 0x250); - if (!h->vaddr) { - err = -ENOMEM; - goto err_out_free_res; - } - err = cciss_wait_for_board_state(h->pdev, h->vaddr, BOARD_READY); - if (err) - goto err_out_free_res; - err = cciss_find_cfgtables(h); - if (err) - goto err_out_free_res; - print_cfg_table(h); - cciss_find_board_params(h); - - if (!CISS_signature_present(h)) { - err = -ENODEV; - goto err_out_free_res; - } - cciss_enable_scsi_prefetch(h); - cciss_p600_dma_prefetch_quirk(h); - err = cciss_enter_simple_mode(h); - if (err) - goto err_out_free_res; - cciss_put_controller_into_performant_mode(h); - return 0; - -err_out_free_res: - /* - * Deliberately omit pci_disable_device(): it does something nasty to - * Smart Array controllers that pci_enable_device does not undo - */ - if (h->transtable) - iounmap(h->transtable); - if (h->cfgtable) - iounmap(h->cfgtable); - if (h->vaddr) - iounmap(h->vaddr); - pci_release_regions(h->pdev); - return err; -} - -/* Function to find the first free pointer into our hba[] array - * Returns -1 if no free entries are left. - */ -static int alloc_cciss_hba(struct pci_dev *pdev) -{ - int i; - - for (i = 0; i < MAX_CTLR; i++) { - if (!hba[i]) { - ctlr_info_t *h; - - h = kzalloc(sizeof(ctlr_info_t), GFP_KERNEL); - if (!h) - goto Enomem; - hba[i] = h; - return i; - } - } - dev_warn(&pdev->dev, "This driver supports a maximum" - " of %d controllers.\n", MAX_CTLR); - return -1; -Enomem: - dev_warn(&pdev->dev, "out of memory.\n"); - return -1; -} - -static void free_hba(ctlr_info_t *h) -{ - int i; - - hba[h->ctlr] = NULL; - for (i = 0; i < h->highest_lun + 1; i++) - if (h->gendisk[i] != NULL) - put_disk(h->gendisk[i]); - kfree(h); -} - -/* Send a message CDB to the firmware. */ -static int cciss_message(struct pci_dev *pdev, unsigned char opcode, - unsigned char type) -{ - typedef struct { - CommandListHeader_struct CommandHeader; - RequestBlock_struct Request; - ErrDescriptor_struct ErrorDescriptor; - } Command; - static const size_t cmd_sz = sizeof(Command) + sizeof(ErrorInfo_struct); - Command *cmd; - dma_addr_t paddr64; - uint32_t paddr32, tag; - void __iomem *vaddr; - int i, err; - - vaddr = ioremap_nocache(pci_resource_start(pdev, 0), pci_resource_len(pdev, 0)); - if (vaddr == NULL) - return -ENOMEM; - - /* The Inbound Post Queue only accepts 32-bit physical addresses for the - CCISS commands, so they must be allocated from the lower 4GiB of - memory. */ - err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32)); - if (err) { - iounmap(vaddr); - return -ENOMEM; - } - - cmd = pci_alloc_consistent(pdev, cmd_sz, &paddr64); - if (cmd == NULL) { - iounmap(vaddr); - return -ENOMEM; - } - - /* This must fit, because of the 32-bit consistent DMA mask. Also, - although there's no guarantee, we assume that the address is at - least 4-byte aligned (most likely, it's page-aligned). */ - paddr32 = paddr64; - - cmd->CommandHeader.ReplyQueue = 0; - cmd->CommandHeader.SGList = 0; - cmd->CommandHeader.SGTotal = 0; - cmd->CommandHeader.Tag.lower = paddr32; - cmd->CommandHeader.Tag.upper = 0; - memset(&cmd->CommandHeader.LUN.LunAddrBytes, 0, 8); - - cmd->Request.CDBLen = 16; - cmd->Request.Type.Type = TYPE_MSG; - cmd->Request.Type.Attribute = ATTR_HEADOFQUEUE; - cmd->Request.Type.Direction = XFER_NONE; - cmd->Request.Timeout = 0; /* Don't time out */ - cmd->Request.CDB[0] = opcode; - cmd->Request.CDB[1] = type; - memset(&cmd->Request.CDB[2], 0, 14); /* the rest of the CDB is reserved */ - - cmd->ErrorDescriptor.Addr.lower = paddr32 + sizeof(Command); - cmd->ErrorDescriptor.Addr.upper = 0; - cmd->ErrorDescriptor.Len = sizeof(ErrorInfo_struct); - - writel(paddr32, vaddr + SA5_REQUEST_PORT_OFFSET); - - for (i = 0; i < 10; i++) { - tag = readl(vaddr + SA5_REPLY_PORT_OFFSET); - if ((tag & ~3) == paddr32) - break; - msleep(CCISS_POST_RESET_NOOP_TIMEOUT_MSECS); - } - - iounmap(vaddr); - - /* we leak the DMA buffer here ... no choice since the controller could - still complete the command. */ - if (i == 10) { - dev_err(&pdev->dev, - "controller message %02x:%02x timed out\n", - opcode, type); - return -ETIMEDOUT; - } - - pci_free_consistent(pdev, cmd_sz, cmd, paddr64); - - if (tag & 2) { - dev_err(&pdev->dev, "controller message %02x:%02x failed\n", - opcode, type); - return -EIO; - } - - dev_info(&pdev->dev, "controller message %02x:%02x succeeded\n", - opcode, type); - return 0; -} - -#define cciss_noop(p) cciss_message(p, 3, 0) - -static int cciss_controller_hard_reset(struct pci_dev *pdev, - void * __iomem vaddr, u32 use_doorbell) -{ - u16 pmcsr; - int pos; - - if (use_doorbell) { - /* For everything after the P600, the PCI power state method - * of resetting the controller doesn't work, so we have this - * other way using the doorbell register. - */ - dev_info(&pdev->dev, "using doorbell to reset controller\n"); - writel(use_doorbell, vaddr + SA5_DOORBELL); - } else { /* Try to do it the PCI power state way */ - - /* Quoting from the Open CISS Specification: "The Power - * Management Control/Status Register (CSR) controls the power - * state of the device. The normal operating state is D0, - * CSR=00h. The software off state is D3, CSR=03h. To reset - * the controller, place the interface device in D3 then to D0, - * this causes a secondary PCI reset which will reset the - * controller." */ - - pos = pci_find_capability(pdev, PCI_CAP_ID_PM); - if (pos == 0) { - dev_err(&pdev->dev, - "cciss_controller_hard_reset: " - "PCI PM not supported\n"); - return -ENODEV; - } - dev_info(&pdev->dev, "using PCI PM to reset controller\n"); - /* enter the D3hot power management state */ - pci_read_config_word(pdev, pos + PCI_PM_CTRL, &pmcsr); - pmcsr &= ~PCI_PM_CTRL_STATE_MASK; - pmcsr |= PCI_D3hot; - pci_write_config_word(pdev, pos + PCI_PM_CTRL, pmcsr); - - msleep(500); - - /* enter the D0 power management state */ - pmcsr &= ~PCI_PM_CTRL_STATE_MASK; - pmcsr |= PCI_D0; - pci_write_config_word(pdev, pos + PCI_PM_CTRL, pmcsr); - - /* - * The P600 requires a small delay when changing states. - * Otherwise we may think the board did not reset and we bail. - * This for kdump only and is particular to the P600. - */ - msleep(500); - } - return 0; -} - -static void init_driver_version(char *driver_version, int len) -{ - memset(driver_version, 0, len); - strncpy(driver_version, "cciss " DRIVER_NAME, len - 1); -} - -static int write_driver_ver_to_cfgtable(CfgTable_struct __iomem *cfgtable) -{ - char *driver_version; - int i, size = sizeof(cfgtable->driver_version); - - driver_version = kmalloc(size, GFP_KERNEL); - if (!driver_version) - return -ENOMEM; - - init_driver_version(driver_version, size); - for (i = 0; i < size; i++) - writeb(driver_version[i], &cfgtable->driver_version[i]); - kfree(driver_version); - return 0; -} - -static void read_driver_ver_from_cfgtable(CfgTable_struct __iomem *cfgtable, - unsigned char *driver_ver) -{ - int i; - - for (i = 0; i < sizeof(cfgtable->driver_version); i++) - driver_ver[i] = readb(&cfgtable->driver_version[i]); -} - -static int controller_reset_failed(CfgTable_struct __iomem *cfgtable) -{ - - char *driver_ver, *old_driver_ver; - int rc, size = sizeof(cfgtable->driver_version); - - old_driver_ver = kmalloc(2 * size, GFP_KERNEL); - if (!old_driver_ver) - return -ENOMEM; - driver_ver = old_driver_ver + size; - - /* After a reset, the 32 bytes of "driver version" in the cfgtable - * should have been changed, otherwise we know the reset failed. - */ - init_driver_version(old_driver_ver, size); - read_driver_ver_from_cfgtable(cfgtable, driver_ver); - rc = !memcmp(driver_ver, old_driver_ver, size); - kfree(old_driver_ver); - return rc; -} - -/* This does a hard reset of the controller using PCI power management - * states or using the doorbell register. */ -static int cciss_kdump_hard_reset_controller(struct pci_dev *pdev) -{ - u64 cfg_offset; - u32 cfg_base_addr; - u64 cfg_base_addr_index; - void __iomem *vaddr; - unsigned long paddr; - u32 misc_fw_support; - int rc; - CfgTable_struct __iomem *cfgtable; - u32 use_doorbell; - u32 board_id; - u16 command_register; - - /* For controllers as old a the p600, this is very nearly - * the same thing as - * - * pci_save_state(pci_dev); - * pci_set_power_state(pci_dev, PCI_D3hot); - * pci_set_power_state(pci_dev, PCI_D0); - * pci_restore_state(pci_dev); - * - * For controllers newer than the P600, the pci power state - * method of resetting doesn't work so we have another way - * using the doorbell register. - */ - - /* Exclude 640x boards. These are two pci devices in one slot - * which share a battery backed cache module. One controls the - * cache, the other accesses the cache through the one that controls - * it. If we reset the one controlling the cache, the other will - * likely not be happy. Just forbid resetting this conjoined mess. - */ - cciss_lookup_board_id(pdev, &board_id); - if (!ctlr_is_resettable(board_id)) { - dev_warn(&pdev->dev, "Controller not resettable\n"); - return -ENODEV; - } - - /* if controller is soft- but not hard resettable... */ - if (!ctlr_is_hard_resettable(board_id)) - return -ENOTSUPP; /* try soft reset later. */ - - /* Save the PCI command register */ - pci_read_config_word(pdev, 4, &command_register); - /* Turn the board off. This is so that later pci_restore_state() - * won't turn the board on before the rest of config space is ready. - */ - pci_disable_device(pdev); - pci_save_state(pdev); - - /* find the first memory BAR, so we can find the cfg table */ - rc = cciss_pci_find_memory_BAR(pdev, &paddr); - if (rc) - return rc; - vaddr = remap_pci_mem(paddr, 0x250); - if (!vaddr) - return -ENOMEM; - - /* find cfgtable in order to check if reset via doorbell is supported */ - rc = cciss_find_cfg_addrs(pdev, vaddr, &cfg_base_addr, - &cfg_base_addr_index, &cfg_offset); - if (rc) - goto unmap_vaddr; - cfgtable = remap_pci_mem(pci_resource_start(pdev, - cfg_base_addr_index) + cfg_offset, sizeof(*cfgtable)); - if (!cfgtable) { - rc = -ENOMEM; - goto unmap_vaddr; - } - rc = write_driver_ver_to_cfgtable(cfgtable); - if (rc) - goto unmap_vaddr; - - /* If reset via doorbell register is supported, use that. - * There are two such methods. Favor the newest method. - */ - misc_fw_support = readl(&cfgtable->misc_fw_support); - use_doorbell = misc_fw_support & MISC_FW_DOORBELL_RESET2; - if (use_doorbell) { - use_doorbell = DOORBELL_CTLR_RESET2; - } else { - use_doorbell = misc_fw_support & MISC_FW_DOORBELL_RESET; - if (use_doorbell) { - dev_warn(&pdev->dev, "Controller claims that " - "'Bit 2 doorbell reset' is " - "supported, but not 'bit 5 doorbell reset'. " - "Firmware update is recommended.\n"); - rc = -ENOTSUPP; /* use the soft reset */ - goto unmap_cfgtable; - } - } - - rc = cciss_controller_hard_reset(pdev, vaddr, use_doorbell); - if (rc) - goto unmap_cfgtable; - pci_restore_state(pdev); - rc = pci_enable_device(pdev); - if (rc) { - dev_warn(&pdev->dev, "failed to enable device.\n"); - goto unmap_cfgtable; - } - pci_write_config_word(pdev, 4, command_register); - - /* Some devices (notably the HP Smart Array 5i Controller) - need a little pause here */ - msleep(CCISS_POST_RESET_PAUSE_MSECS); - - /* Wait for board to become not ready, then ready. */ - dev_info(&pdev->dev, "Waiting for board to reset.\n"); - rc = cciss_wait_for_board_state(pdev, vaddr, BOARD_NOT_READY); - if (rc) { - dev_warn(&pdev->dev, "Failed waiting for board to hard reset." - " Will try soft reset.\n"); - rc = -ENOTSUPP; /* Not expected, but try soft reset later */ - goto unmap_cfgtable; - } - rc = cciss_wait_for_board_state(pdev, vaddr, BOARD_READY); - if (rc) { - dev_warn(&pdev->dev, - "failed waiting for board to become ready " - "after hard reset\n"); - goto unmap_cfgtable; - } - - rc = controller_reset_failed(vaddr); - if (rc < 0) - goto unmap_cfgtable; - if (rc) { - dev_warn(&pdev->dev, "Unable to successfully hard reset " - "controller. Will try soft reset.\n"); - rc = -ENOTSUPP; /* Not expected, but try soft reset later */ - } else { - dev_info(&pdev->dev, "Board ready after hard reset.\n"); - } - -unmap_cfgtable: - iounmap(cfgtable); - -unmap_vaddr: - iounmap(vaddr); - return rc; -} - -static int cciss_init_reset_devices(struct pci_dev *pdev) -{ - int rc, i; - - if (!reset_devices) - return 0; - - /* Reset the controller with a PCI power-cycle or via doorbell */ - rc = cciss_kdump_hard_reset_controller(pdev); - - /* -ENOTSUPP here means we cannot reset the controller - * but it's already (and still) up and running in - * "performant mode". Or, it might be 640x, which can't reset - * due to concerns about shared bbwc between 6402/6404 pair. - */ - if (rc == -ENOTSUPP) - return rc; /* just try to do the kdump anyhow. */ - if (rc) - return -ENODEV; - - /* Now try to get the controller to respond to a no-op */ - dev_warn(&pdev->dev, "Waiting for controller to respond to no-op\n"); - for (i = 0; i < CCISS_POST_RESET_NOOP_RETRIES; i++) { - if (cciss_noop(pdev) == 0) - break; - else - dev_warn(&pdev->dev, "no-op failed%s\n", - (i < CCISS_POST_RESET_NOOP_RETRIES - 1 ? - "; re-trying" : "")); - msleep(CCISS_POST_RESET_NOOP_INTERVAL_MSECS); - } - return 0; -} - -static int cciss_allocate_cmd_pool(ctlr_info_t *h) -{ - h->cmd_pool_bits = kmalloc(BITS_TO_LONGS(h->nr_cmds) * - sizeof(unsigned long), GFP_KERNEL); - h->cmd_pool = pci_alloc_consistent(h->pdev, - h->nr_cmds * sizeof(CommandList_struct), - &(h->cmd_pool_dhandle)); - h->errinfo_pool = pci_alloc_consistent(h->pdev, - h->nr_cmds * sizeof(ErrorInfo_struct), - &(h->errinfo_pool_dhandle)); - if ((h->cmd_pool_bits == NULL) - || (h->cmd_pool == NULL) - || (h->errinfo_pool == NULL)) { - dev_err(&h->pdev->dev, "out of memory"); - return -ENOMEM; - } - return 0; -} - -static int cciss_allocate_scatterlists(ctlr_info_t *h) -{ - int i; - - /* zero it, so that on free we need not know how many were alloc'ed */ - h->scatter_list = kzalloc(h->max_commands * - sizeof(struct scatterlist *), GFP_KERNEL); - if (!h->scatter_list) - return -ENOMEM; - - for (i = 0; i < h->nr_cmds; i++) { - h->scatter_list[i] = kmalloc(sizeof(struct scatterlist) * - h->maxsgentries, GFP_KERNEL); - if (h->scatter_list[i] == NULL) { - dev_err(&h->pdev->dev, "could not allocate " - "s/g lists\n"); - return -ENOMEM; - } - } - return 0; -} - -static void cciss_free_scatterlists(ctlr_info_t *h) -{ - int i; - - if (h->scatter_list) { - for (i = 0; i < h->nr_cmds; i++) - kfree(h->scatter_list[i]); - kfree(h->scatter_list); - } -} - -static void cciss_free_cmd_pool(ctlr_info_t *h) -{ - kfree(h->cmd_pool_bits); - if (h->cmd_pool) - pci_free_consistent(h->pdev, - h->nr_cmds * sizeof(CommandList_struct), - h->cmd_pool, h->cmd_pool_dhandle); - if (h->errinfo_pool) - pci_free_consistent(h->pdev, - h->nr_cmds * sizeof(ErrorInfo_struct), - h->errinfo_pool, h->errinfo_pool_dhandle); -} - -static int cciss_request_irq(ctlr_info_t *h, - irqreturn_t (*msixhandler)(int, void *), - irqreturn_t (*intxhandler)(int, void *)) -{ - if (h->pdev->msi_enabled || h->pdev->msix_enabled) { - if (!request_irq(h->intr[h->intr_mode], msixhandler, - 0, h->devname, h)) - return 0; - dev_err(&h->pdev->dev, "Unable to get msi irq %d" - " for %s\n", h->intr[h->intr_mode], - h->devname); - return -1; - } - - if (!request_irq(h->intr[h->intr_mode], intxhandler, - IRQF_SHARED, h->devname, h)) - return 0; - dev_err(&h->pdev->dev, "Unable to get irq %d for %s\n", - h->intr[h->intr_mode], h->devname); - return -1; -} - -static int cciss_kdump_soft_reset(ctlr_info_t *h) -{ - if (cciss_send_reset(h, CTLR_LUNID, CCISS_RESET_TYPE_CONTROLLER)) { - dev_warn(&h->pdev->dev, "Resetting array controller failed.\n"); - return -EIO; - } - - dev_info(&h->pdev->dev, "Waiting for board to soft reset.\n"); - if (cciss_wait_for_board_state(h->pdev, h->vaddr, BOARD_NOT_READY)) { - dev_warn(&h->pdev->dev, "Soft reset had no effect.\n"); - return -1; - } - - dev_info(&h->pdev->dev, "Board reset, awaiting READY status.\n"); - if (cciss_wait_for_board_state(h->pdev, h->vaddr, BOARD_READY)) { - dev_warn(&h->pdev->dev, "Board failed to become ready " - "after soft reset.\n"); - return -1; - } - - return 0; -} - -static void cciss_undo_allocations_after_kdump_soft_reset(ctlr_info_t *h) -{ - int ctlr = h->ctlr; - - free_irq(h->intr[h->intr_mode], h); - pci_free_irq_vectors(h->pdev); - cciss_free_sg_chain_blocks(h->cmd_sg_list, h->nr_cmds); - cciss_free_scatterlists(h); - cciss_free_cmd_pool(h); - kfree(h->blockFetchTable); - if (h->reply_pool) - pci_free_consistent(h->pdev, h->max_commands * sizeof(__u64), - h->reply_pool, h->reply_pool_dhandle); - if (h->transtable) - iounmap(h->transtable); - if (h->cfgtable) - iounmap(h->cfgtable); - if (h->vaddr) - iounmap(h->vaddr); - unregister_blkdev(h->major, h->devname); - cciss_destroy_hba_sysfs_entry(h); - pci_release_regions(h->pdev); - kfree(h); - hba[ctlr] = NULL; -} - -/* - * This is it. Find all the controllers and register them. I really hate - * stealing all these major device numbers. - * returns the number of block devices registered. - */ -static int cciss_init_one(struct pci_dev *pdev, const struct pci_device_id *ent) -{ - int i; - int j = 0; - int rc; - int try_soft_reset = 0; - int dac, return_code; - InquiryData_struct *inq_buff; - ctlr_info_t *h; - unsigned long flags; - - /* - * By default the cciss driver is used for all older HP Smart Array - * controllers. There are module paramaters that allow a user to - * override this behavior and instead use the hpsa SCSI driver. If - * this is the case cciss may be loaded first from the kdump initrd - * image and cause a kernel panic. So if reset_devices is true and - * cciss_allow_hpsa is set just bail. - */ - if ((reset_devices) && (cciss_allow_hpsa == 1)) - return -ENODEV; - rc = cciss_init_reset_devices(pdev); - if (rc) { - if (rc != -ENOTSUPP) - return rc; - /* If the reset fails in a particular way (it has no way to do - * a proper hard reset, so returns -ENOTSUPP) we can try to do - * a soft reset once we get the controller configured up to the - * point that it can accept a command. - */ - try_soft_reset = 1; - rc = 0; - } - -reinit_after_soft_reset: - - i = alloc_cciss_hba(pdev); - if (i < 0) - return -ENOMEM; - - h = hba[i]; - h->pdev = pdev; - h->busy_initializing = 1; - h->intr_mode = cciss_simple_mode ? SIMPLE_MODE_INT : PERF_MODE_INT; - INIT_LIST_HEAD(&h->cmpQ); - INIT_LIST_HEAD(&h->reqQ); - mutex_init(&h->busy_shutting_down); - - if (cciss_pci_init(h) != 0) - goto clean_no_release_regions; - - sprintf(h->devname, "cciss%d", i); - h->ctlr = i; - - if (cciss_tape_cmds < 2) - cciss_tape_cmds = 2; - if (cciss_tape_cmds > 16) - cciss_tape_cmds = 16; - - init_completion(&h->scan_wait); - - if (cciss_create_hba_sysfs_entry(h)) - goto clean0; - - /* configure PCI DMA stuff */ - if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(64))) - dac = 1; - else if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(32))) - dac = 0; - else { - dev_err(&h->pdev->dev, "no suitable DMA available\n"); - goto clean1; - } - - /* - * register with the major number, or get a dynamic major number - * by passing 0 as argument. This is done for greater than - * 8 controller support. - */ - if (i < MAX_CTLR_ORIG) - h->major = COMPAQ_CISS_MAJOR + i; - rc = register_blkdev(h->major, h->devname); - if (rc == -EBUSY || rc == -EINVAL) { - dev_err(&h->pdev->dev, - "Unable to get major number %d for %s " - "on hba %d\n", h->major, h->devname, i); - goto clean1; - } else { - if (i >= MAX_CTLR_ORIG) - h->major = rc; - } - - /* make sure the board interrupts are off */ - h->access.set_intr_mask(h, CCISS_INTR_OFF); - rc = cciss_request_irq(h, do_cciss_msix_intr, do_cciss_intx); - if (rc) - goto clean2; - - dev_info(&h->pdev->dev, "%s: <0x%x> at PCI %s IRQ %d%s using DAC\n", - h->devname, pdev->device, pci_name(pdev), - h->intr[h->intr_mode], dac ? "" : " not"); - - if (cciss_allocate_cmd_pool(h)) - goto clean4; - - if (cciss_allocate_scatterlists(h)) - goto clean4; - - h->cmd_sg_list = cciss_allocate_sg_chain_blocks(h, - h->chainsize, h->nr_cmds); - if (!h->cmd_sg_list && h->chainsize > 0) - goto clean4; - - spin_lock_init(&h->lock); - - /* Initialize the pdev driver private data. - have it point to h. */ - pci_set_drvdata(pdev, h); - /* command and error info recs zeroed out before - they are used */ - bitmap_zero(h->cmd_pool_bits, h->nr_cmds); - - h->num_luns = 0; - h->highest_lun = -1; - for (j = 0; j < CISS_MAX_LUN; j++) { - h->drv[j] = NULL; - h->gendisk[j] = NULL; - } - - /* At this point, the controller is ready to take commands. - * Now, if reset_devices and the hard reset didn't work, try - * the soft reset and see if that works. - */ - if (try_soft_reset) { - - /* This is kind of gross. We may or may not get a completion - * from the soft reset command, and if we do, then the value - * from the fifo may or may not be valid. So, we wait 10 secs - * after the reset throwing away any completions we get during - * that time. Unregister the interrupt handler and register - * fake ones to scoop up any residual completions. - */ - spin_lock_irqsave(&h->lock, flags); - h->access.set_intr_mask(h, CCISS_INTR_OFF); - spin_unlock_irqrestore(&h->lock, flags); - free_irq(h->intr[h->intr_mode], h); - rc = cciss_request_irq(h, cciss_msix_discard_completions, - cciss_intx_discard_completions); - if (rc) { - dev_warn(&h->pdev->dev, "Failed to request_irq after " - "soft reset.\n"); - goto clean4; - } - - rc = cciss_kdump_soft_reset(h); - if (rc) { - dev_warn(&h->pdev->dev, "Soft reset failed.\n"); - goto clean4; - } - - dev_info(&h->pdev->dev, "Board READY.\n"); - dev_info(&h->pdev->dev, - "Waiting for stale completions to drain.\n"); - h->access.set_intr_mask(h, CCISS_INTR_ON); - msleep(10000); - h->access.set_intr_mask(h, CCISS_INTR_OFF); - - rc = controller_reset_failed(h->cfgtable); - if (rc) - dev_info(&h->pdev->dev, - "Soft reset appears to have failed.\n"); - - /* since the controller's reset, we have to go back and re-init - * everything. Easiest to just forget what we've done and do it - * all over again. - */ - cciss_undo_allocations_after_kdump_soft_reset(h); - try_soft_reset = 0; - if (rc) - /* don't go to clean4, we already unallocated */ - return -ENODEV; - - goto reinit_after_soft_reset; - } - - cciss_scsi_setup(h); - - /* Turn the interrupts on so we can service requests */ - h->access.set_intr_mask(h, CCISS_INTR_ON); - - /* Get the firmware version */ - inq_buff = kzalloc(sizeof(InquiryData_struct), GFP_KERNEL); - if (inq_buff == NULL) { - dev_err(&h->pdev->dev, "out of memory\n"); - goto clean4; - } - - return_code = sendcmd_withirq(h, CISS_INQUIRY, inq_buff, - sizeof(InquiryData_struct), 0, CTLR_LUNID, TYPE_CMD); - if (return_code == IO_OK) { - h->firm_ver[0] = inq_buff->data_byte[32]; - h->firm_ver[1] = inq_buff->data_byte[33]; - h->firm_ver[2] = inq_buff->data_byte[34]; - h->firm_ver[3] = inq_buff->data_byte[35]; - } else { /* send command failed */ - dev_warn(&h->pdev->dev, "unable to determine firmware" - " version of controller\n"); - } - kfree(inq_buff); - - cciss_procinit(h); - - h->cciss_max_sectors = 8192; - - rebuild_lun_table(h, 1, 0); - cciss_engage_scsi(h); - h->busy_initializing = 0; - return 0; - -clean4: - cciss_free_cmd_pool(h); - cciss_free_scatterlists(h); - cciss_free_sg_chain_blocks(h->cmd_sg_list, h->nr_cmds); - free_irq(h->intr[h->intr_mode], h); -clean2: - unregister_blkdev(h->major, h->devname); -clean1: - cciss_destroy_hba_sysfs_entry(h); -clean0: - pci_release_regions(pdev); -clean_no_release_regions: - h->busy_initializing = 0; - - /* - * Deliberately omit pci_disable_device(): it does something nasty to - * Smart Array controllers that pci_enable_device does not undo - */ - pci_set_drvdata(pdev, NULL); - free_hba(h); - return -ENODEV; -} - -static void cciss_shutdown(struct pci_dev *pdev) -{ - ctlr_info_t *h; - char *flush_buf; - int return_code; - - h = pci_get_drvdata(pdev); - flush_buf = kzalloc(4, GFP_KERNEL); - if (!flush_buf) { - dev_warn(&h->pdev->dev, "cache not flushed, out of memory.\n"); - return; - } - /* write all data in the battery backed cache to disk */ - return_code = sendcmd_withirq(h, CCISS_CACHE_FLUSH, flush_buf, - 4, 0, CTLR_LUNID, TYPE_CMD); - kfree(flush_buf); - if (return_code != IO_OK) - dev_warn(&h->pdev->dev, "Error flushing cache\n"); - h->access.set_intr_mask(h, CCISS_INTR_OFF); - free_irq(h->intr[h->intr_mode], h); -} - -static int cciss_enter_simple_mode(struct ctlr_info *h) -{ - u32 trans_support; - - trans_support = readl(&(h->cfgtable->TransportSupport)); - if (!(trans_support & SIMPLE_MODE)) - return -ENOTSUPP; - - h->max_commands = readl(&(h->cfgtable->CmdsOutMax)); - writel(CFGTBL_Trans_Simple, &(h->cfgtable->HostWrite.TransportRequest)); - writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL); - cciss_wait_for_mode_change_ack(h); - print_cfg_table(h); - if (!(readl(&(h->cfgtable->TransportActive)) & CFGTBL_Trans_Simple)) { - dev_warn(&h->pdev->dev, "unable to get board into simple mode\n"); - return -ENODEV; - } - h->transMethod = CFGTBL_Trans_Simple; - return 0; -} - - -static void cciss_remove_one(struct pci_dev *pdev) -{ - ctlr_info_t *h; - int i, j; - - if (pci_get_drvdata(pdev) == NULL) { - dev_err(&pdev->dev, "Unable to remove device\n"); - return; - } - - h = pci_get_drvdata(pdev); - i = h->ctlr; - if (hba[i] == NULL) { - dev_err(&pdev->dev, "device appears to already be removed\n"); - return; - } - - mutex_lock(&h->busy_shutting_down); - - remove_from_scan_list(h); - remove_proc_entry(h->devname, proc_cciss); - unregister_blkdev(h->major, h->devname); - - /* remove it from the disk list */ - for (j = 0; j < CISS_MAX_LUN; j++) { - struct gendisk *disk = h->gendisk[j]; - if (disk) { - struct request_queue *q = disk->queue; - - if (disk->flags & GENHD_FL_UP) { - cciss_destroy_ld_sysfs_entry(h, j, 1); - del_gendisk(disk); - } - if (q) - blk_cleanup_queue(q); - } - } - -#ifdef CONFIG_CISS_SCSI_TAPE - cciss_unregister_scsi(h); /* unhook from SCSI subsystem */ -#endif - - cciss_shutdown(pdev); - - pci_free_irq_vectors(h->pdev); - - iounmap(h->transtable); - iounmap(h->cfgtable); - iounmap(h->vaddr); - - cciss_free_cmd_pool(h); - /* Free up sg elements */ - for (j = 0; j < h->nr_cmds; j++) - kfree(h->scatter_list[j]); - kfree(h->scatter_list); - cciss_free_sg_chain_blocks(h->cmd_sg_list, h->nr_cmds); - kfree(h->blockFetchTable); - if (h->reply_pool) - pci_free_consistent(h->pdev, h->max_commands * sizeof(__u64), - h->reply_pool, h->reply_pool_dhandle); - /* - * Deliberately omit pci_disable_device(): it does something nasty to - * Smart Array controllers that pci_enable_device does not undo - */ - pci_release_regions(pdev); - pci_set_drvdata(pdev, NULL); - cciss_destroy_hba_sysfs_entry(h); - mutex_unlock(&h->busy_shutting_down); - free_hba(h); -} - -static struct pci_driver cciss_pci_driver = { - .name = "cciss", - .probe = cciss_init_one, - .remove = cciss_remove_one, - .id_table = cciss_pci_device_id, /* id_table */ - .shutdown = cciss_shutdown, -}; - -/* - * This is it. Register the PCI driver information for the cards we control - * the OS will call our registered routines when it finds one of our cards. - */ -static int __init cciss_init(void) -{ - int err; - - /* - * The hardware requires that commands are aligned on a 64-bit - * boundary. Given that we use pci_alloc_consistent() to allocate an - * array of them, the size must be a multiple of 8 bytes. - */ - BUILD_BUG_ON(sizeof(CommandList_struct) % COMMANDLIST_ALIGNMENT); - printk(KERN_INFO DRIVER_NAME "\n"); - - err = bus_register(&cciss_bus_type); - if (err) - return err; - - /* Start the scan thread */ - cciss_scan_thread = kthread_run(scan_thread, NULL, "cciss_scan"); - if (IS_ERR(cciss_scan_thread)) { - err = PTR_ERR(cciss_scan_thread); - goto err_bus_unregister; - } - - /* Register for our PCI devices */ - err = pci_register_driver(&cciss_pci_driver); - if (err) - goto err_thread_stop; - - return err; - -err_thread_stop: - kthread_stop(cciss_scan_thread); -err_bus_unregister: - bus_unregister(&cciss_bus_type); - - return err; -} - -static void __exit cciss_cleanup(void) -{ - int i; - - pci_unregister_driver(&cciss_pci_driver); - /* double check that all controller entrys have been removed */ - for (i = 0; i < MAX_CTLR; i++) { - if (hba[i] != NULL) { - dev_warn(&hba[i]->pdev->dev, - "had to remove controller\n"); - cciss_remove_one(hba[i]->pdev); - } - } - kthread_stop(cciss_scan_thread); - if (proc_cciss) - remove_proc_entry("driver/cciss", NULL); - bus_unregister(&cciss_bus_type); -} - -module_init(cciss_init); -module_exit(cciss_cleanup); diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h deleted file mode 100644 index 24b5fd75501a..000000000000 --- a/drivers/block/cciss.h +++ /dev/null @@ -1,433 +0,0 @@ -#ifndef CCISS_H -#define CCISS_H - -#include <linux/genhd.h> -#include <linux/mutex.h> - -#include "cciss_cmd.h" - - -#define NWD_SHIFT 4 -#define MAX_PART (1 << NWD_SHIFT) - -#define IO_OK 0 -#define IO_ERROR 1 -#define IO_NEEDS_RETRY 3 - -#define VENDOR_LEN 8 -#define MODEL_LEN 16 -#define REV_LEN 4 - -struct ctlr_info; -typedef struct ctlr_info ctlr_info_t; - -struct access_method { - void (*submit_command)(ctlr_info_t *h, CommandList_struct *c); - void (*set_intr_mask)(ctlr_info_t *h, unsigned long val); - unsigned long (*fifo_full)(ctlr_info_t *h); - bool (*intr_pending)(ctlr_info_t *h); - unsigned long (*command_completed)(ctlr_info_t *h); -}; -typedef struct _drive_info_struct -{ - unsigned char LunID[8]; - int usage_count; - struct request_queue *queue; - sector_t nr_blocks; - int block_size; - int heads; - int sectors; - int cylinders; - int raid_level; /* set to -1 to indicate that - * the drive is not in use/configured - */ - int busy_configuring; /* This is set when a drive is being removed - * to prevent it from being opened or it's - * queue from being started. - */ - struct device dev; - __u8 serial_no[16]; /* from inquiry page 0x83, - * not necc. null terminated. - */ - char vendor[VENDOR_LEN + 1]; /* SCSI vendor string */ - char model[MODEL_LEN + 1]; /* SCSI model string */ - char rev[REV_LEN + 1]; /* SCSI revision string */ - char device_initialized; /* indicates whether dev is initialized */ -} drive_info_struct; - -struct ctlr_info -{ - int ctlr; - char devname[8]; - char *product_name; - char firm_ver[4]; /* Firmware version */ - struct pci_dev *pdev; - __u32 board_id; - void __iomem *vaddr; - unsigned long paddr; - int nr_cmds; /* Number of commands allowed on this controller */ - CfgTable_struct __iomem *cfgtable; - int interrupts_enabled; - int major; - int max_commands; - int commands_outstanding; - int max_outstanding; /* Debug */ - int num_luns; - int highest_lun; - int usage_count; /* number of opens all all minor devices */ - /* Need space for temp sg list - * number of scatter/gathers supported - * number of scatter/gathers in chained block - */ - struct scatterlist **scatter_list; - int maxsgentries; - int chainsize; - int max_cmd_sgentries; - SGDescriptor_struct **cmd_sg_list; - -# define PERF_MODE_INT 0 -# define DOORBELL_INT 1 -# define SIMPLE_MODE_INT 2 -# define MEMQ_MODE_INT 3 - unsigned int intr[4]; - int intr_mode; - int cciss_max_sectors; - BYTE cciss_read; - BYTE cciss_write; - BYTE cciss_read_capacity; - - /* information about each logical volume */ - drive_info_struct *drv[CISS_MAX_LUN]; - - struct access_method access; - - /* queue and queue Info */ - struct list_head reqQ; - struct list_head cmpQ; - unsigned int Qdepth; - unsigned int maxQsinceinit; - unsigned int maxSG; - spinlock_t lock; - - /* pointers to command and error info pool */ - CommandList_struct *cmd_pool; - dma_addr_t cmd_pool_dhandle; - ErrorInfo_struct *errinfo_pool; - dma_addr_t errinfo_pool_dhandle; - unsigned long *cmd_pool_bits; - int nr_allocs; - int nr_frees; - int busy_configuring; - int busy_initializing; - int busy_scanning; - struct mutex busy_shutting_down; - - /* This element holds the zero based queue number of the last - * queue to be started. It is used for fairness. - */ - int next_to_run; - - /* Disk structures we need to pass back */ - struct gendisk *gendisk[CISS_MAX_LUN]; -#ifdef CONFIG_CISS_SCSI_TAPE - struct cciss_scsi_adapter_data_t *scsi_ctlr; -#endif - unsigned char alive; - struct list_head scan_list; - struct completion scan_wait; - struct device dev; - /* - * Performant mode tables. - */ - u32 trans_support; - u32 trans_offset; - struct TransTable_struct *transtable; - unsigned long transMethod; - - /* - * Performant mode completion buffer - */ - u64 *reply_pool; - dma_addr_t reply_pool_dhandle; - u64 *reply_pool_head; - size_t reply_pool_size; - unsigned char reply_pool_wraparound; - u32 *blockFetchTable; -}; - -/* Defining the diffent access_methods - * - * Memory mapped FIFO interface (SMART 53xx cards) - */ -#define SA5_DOORBELL 0x20 -#define SA5_REQUEST_PORT_OFFSET 0x40 -#define SA5_REPLY_INTR_MASK_OFFSET 0x34 -#define SA5_REPLY_PORT_OFFSET 0x44 -#define SA5_INTR_STATUS 0x30 -#define SA5_SCRATCHPAD_OFFSET 0xB0 - -#define SA5_CTCFG_OFFSET 0xB4 -#define SA5_CTMEM_OFFSET 0xB8 - -#define SA5_INTR_OFF 0x08 -#define SA5B_INTR_OFF 0x04 -#define SA5_INTR_PENDING 0x08 -#define SA5B_INTR_PENDING 0x04 -#define FIFO_EMPTY 0xffffffff -#define CCISS_FIRMWARE_READY 0xffff0000 /* value in scratchpad register */ -/* Perf. mode flags */ -#define SA5_PERF_INTR_PENDING 0x04 -#define SA5_PERF_INTR_OFF 0x05 -#define SA5_OUTDB_STATUS_PERF_BIT 0x01 -#define SA5_OUTDB_CLEAR_PERF_BIT 0x01 -#define SA5_OUTDB_CLEAR 0xA0 -#define SA5_OUTDB_CLEAR_PERF_BIT 0x01 -#define SA5_OUTDB_STATUS 0x9C - - -#define CISS_ERROR_BIT 0x02 - -#define CCISS_INTR_ON 1 -#define CCISS_INTR_OFF 0 - - -/* CCISS_BOARD_READY_WAIT_SECS is how long to wait for a board - * to become ready, in seconds, before giving up on it. - * CCISS_BOARD_READY_POLL_INTERVAL_MSECS * is how long to wait - * between polling the board to see if it is ready, in - * milliseconds. CCISS_BOARD_READY_ITERATIONS is derived - * the above. - */ -#define CCISS_BOARD_READY_WAIT_SECS (120) -#define CCISS_BOARD_NOT_READY_WAIT_SECS (100) -#define CCISS_BOARD_READY_POLL_INTERVAL_MSECS (100) -#define CCISS_BOARD_READY_ITERATIONS \ - ((CCISS_BOARD_READY_WAIT_SECS * 1000) / \ - CCISS_BOARD_READY_POLL_INTERVAL_MSECS) -#define CCISS_BOARD_NOT_READY_ITERATIONS \ - ((CCISS_BOARD_NOT_READY_WAIT_SECS * 1000) / \ - CCISS_BOARD_READY_POLL_INTERVAL_MSECS) -#define CCISS_POST_RESET_PAUSE_MSECS (3000) -#define CCISS_POST_RESET_NOOP_INTERVAL_MSECS (4000) -#define CCISS_POST_RESET_NOOP_RETRIES (12) -#define CCISS_POST_RESET_NOOP_TIMEOUT_MSECS (10000) - -/* - Send the command to the hardware -*/ -static void SA5_submit_command( ctlr_info_t *h, CommandList_struct *c) -{ -#ifdef CCISS_DEBUG - printk(KERN_WARNING "cciss%d: Sending %08x - down to controller\n", - h->ctlr, c->busaddr); -#endif /* CCISS_DEBUG */ - writel(c->busaddr, h->vaddr + SA5_REQUEST_PORT_OFFSET); - readl(h->vaddr + SA5_SCRATCHPAD_OFFSET); - h->commands_outstanding++; - if ( h->commands_outstanding > h->max_outstanding) - h->max_outstanding = h->commands_outstanding; -} - -/* - * This card is the opposite of the other cards. - * 0 turns interrupts on... - * 0x08 turns them off... - */ -static void SA5_intr_mask(ctlr_info_t *h, unsigned long val) -{ - if (val) - { /* Turn interrupts on */ - h->interrupts_enabled = 1; - writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - } else /* Turn them off */ - { - h->interrupts_enabled = 0; - writel( SA5_INTR_OFF, - h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - } -} -/* - * This card is the opposite of the other cards. - * 0 turns interrupts on... - * 0x04 turns them off... - */ -static void SA5B_intr_mask(ctlr_info_t *h, unsigned long val) -{ - if (val) - { /* Turn interrupts on */ - h->interrupts_enabled = 1; - writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - } else /* Turn them off */ - { - h->interrupts_enabled = 0; - writel( SA5B_INTR_OFF, - h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - } -} - -/* Performant mode intr_mask */ -static void SA5_performant_intr_mask(ctlr_info_t *h, unsigned long val) -{ - if (val) { /* turn on interrupts */ - h->interrupts_enabled = 1; - writel(0, h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - } else { - h->interrupts_enabled = 0; - writel(SA5_PERF_INTR_OFF, - h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - (void) readl(h->vaddr + SA5_REPLY_INTR_MASK_OFFSET); - } -} - -/* - * Returns true if fifo is full. - * - */ -static unsigned long SA5_fifo_full(ctlr_info_t *h) -{ - if( h->commands_outstanding >= h->max_commands) - return(1); - else - return(0); - -} -/* - * returns value read from hardware. - * returns FIFO_EMPTY if there is nothing to read - */ -static unsigned long SA5_completed(ctlr_info_t *h) -{ - unsigned long register_value - = readl(h->vaddr + SA5_REPLY_PORT_OFFSET); - if(register_value != FIFO_EMPTY) - { - h->commands_outstanding--; -#ifdef CCISS_DEBUG - printk("cciss: Read %lx back from board\n", register_value); -#endif /* CCISS_DEBUG */ - } -#ifdef CCISS_DEBUG - else - { - printk("cciss: FIFO Empty read\n"); - } -#endif - return ( register_value); - -} - -/* Performant mode command completed */ -static unsigned long SA5_performant_completed(ctlr_info_t *h) -{ - unsigned long register_value = FIFO_EMPTY; - - /* flush the controller write of the reply queue by reading - * outbound doorbell status register. - */ - register_value = readl(h->vaddr + SA5_OUTDB_STATUS); - /* msi auto clears the interrupt pending bit. */ - if (!(h->pdev->msi_enabled || h->pdev->msix_enabled)) { - writel(SA5_OUTDB_CLEAR_PERF_BIT, h->vaddr + SA5_OUTDB_CLEAR); - /* Do a read in order to flush the write to the controller - * (as per spec.) - */ - register_value = readl(h->vaddr + SA5_OUTDB_STATUS); - } - - if ((*(h->reply_pool_head) & 1) == (h->reply_pool_wraparound)) { - register_value = *(h->reply_pool_head); - (h->reply_pool_head)++; - h->commands_outstanding--; - } else { - register_value = FIFO_EMPTY; - } - /* Check for wraparound */ - if (h->reply_pool_head == (h->reply_pool + h->max_commands)) { - h->reply_pool_head = h->reply_pool; - h->reply_pool_wraparound ^= 1; - } - - return register_value; -} -/* - * Returns true if an interrupt is pending.. - */ -static bool SA5_intr_pending(ctlr_info_t *h) -{ - unsigned long register_value = - readl(h->vaddr + SA5_INTR_STATUS); -#ifdef CCISS_DEBUG - printk("cciss: intr_pending %lx\n", register_value); -#endif /* CCISS_DEBUG */ - if( register_value & SA5_INTR_PENDING) - return 1; - return 0 ; -} - -/* - * Returns true if an interrupt is pending.. - */ -static bool SA5B_intr_pending(ctlr_info_t *h) -{ - unsigned long register_value = - readl(h->vaddr + SA5_INTR_STATUS); -#ifdef CCISS_DEBUG - printk("cciss: intr_pending %lx\n", register_value); -#endif /* CCISS_DEBUG */ - if( register_value & SA5B_INTR_PENDING) - return 1; - return 0 ; -} - -static bool SA5_performant_intr_pending(ctlr_info_t *h) -{ - unsigned long register_value = readl(h->vaddr + SA5_INTR_STATUS); - - if (!register_value) - return false; - - if (h->pdev->msi_enabled || h->pdev->msix_enabled) - return true; - - /* Read outbound doorbell to flush */ - register_value = readl(h->vaddr + SA5_OUTDB_STATUS); - return register_value & SA5_OUTDB_STATUS_PERF_BIT; -} - -static struct access_method SA5_access = { - .submit_command = SA5_submit_command, - .set_intr_mask = SA5_intr_mask, - .fifo_full = SA5_fifo_full, - .intr_pending = SA5_intr_pending, - .command_completed = SA5_completed, -}; - -static struct access_method SA5B_access = { - .submit_command = SA5_submit_command, - .set_intr_mask = SA5B_intr_mask, - .fifo_full = SA5_fifo_full, - .intr_pending = SA5B_intr_pending, - .command_completed = SA5_completed, -}; - -static struct access_method SA5_performant_access = { - .submit_command = SA5_submit_command, - .set_intr_mask = SA5_performant_intr_mask, - .fifo_full = SA5_fifo_full, - .intr_pending = SA5_performant_intr_pending, - .command_completed = SA5_performant_completed, -}; - -struct board_type { - __u32 board_id; - char *product_name; - struct access_method *access; - int nr_cmds; /* Max cmds this kind of ctlr can handle. */ -}; - -#endif /* CCISS_H */ diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h deleted file mode 100644 index d9be6b4d49a6..000000000000 --- a/drivers/block/cciss_cmd.h +++ /dev/null @@ -1,269 +0,0 @@ -#ifndef CCISS_CMD_H -#define CCISS_CMD_H - -#include <linux/cciss_defs.h> - -/* DEFINES */ -#define CISS_VERSION "1.00" - -/* general boundary definitions */ -#define MAXSGENTRIES 32 -#define CCISS_SG_CHAIN 0x80000000 -#define MAXREPLYQS 256 - -/* Unit Attentions ASC's as defined for the MSA2012sa */ -#define POWER_OR_RESET 0x29 -#define STATE_CHANGED 0x2a -#define UNIT_ATTENTION_CLEARED 0x2f -#define LUN_FAILED 0x3e -#define REPORT_LUNS_CHANGED 0x3f - -/* Unit Attentions ASCQ's as defined for the MSA2012sa */ - - /* These ASCQ's defined for ASC = POWER_OR_RESET */ -#define POWER_ON_RESET 0x00 -#define POWER_ON_REBOOT 0x01 -#define SCSI_BUS_RESET 0x02 -#define MSA_TARGET_RESET 0x03 -#define CONTROLLER_FAILOVER 0x04 -#define TRANSCEIVER_SE 0x05 -#define TRANSCEIVER_LVD 0x06 - - /* These ASCQ's defined for ASC = STATE_CHANGED */ -#define RESERVATION_PREEMPTED 0x03 -#define ASYM_ACCESS_CHANGED 0x06 -#define LUN_CAPACITY_CHANGED 0x09 - -/* config space register offsets */ -#define CFG_VENDORID 0x00 -#define CFG_DEVICEID 0x02 -#define CFG_I2OBAR 0x10 -#define CFG_MEM1BAR 0x14 - -/* i2o space register offsets */ -#define I2O_IBDB_SET 0x20 -#define I2O_IBDB_CLEAR 0x70 -#define I2O_INT_STATUS 0x30 -#define I2O_INT_MASK 0x34 -#define I2O_IBPOST_Q 0x40 -#define I2O_OBPOST_Q 0x44 -#define I2O_DMA1_CFG 0x214 - -/* Configuration Table */ -#define CFGTBL_ChangeReq 0x00000001l -#define CFGTBL_AccCmds 0x00000001l -#define DOORBELL_CTLR_RESET 0x00000004l -#define DOORBELL_CTLR_RESET2 0x00000020l - -#define CFGTBL_Trans_Simple 0x00000002l -#define CFGTBL_Trans_Performant 0x00000004l -#define CFGTBL_Trans_use_short_tags 0x20000000l - -#define CFGTBL_BusType_Ultra2 0x00000001l -#define CFGTBL_BusType_Ultra3 0x00000002l -#define CFGTBL_BusType_Fibre1G 0x00000100l -#define CFGTBL_BusType_Fibre2G 0x00000200l -typedef struct _vals32 -{ - __u32 lower; - __u32 upper; -} vals32; - -typedef union _u64bit -{ - vals32 val32; - __u64 val; -} u64bit; - -/* Type defs used in the following structs */ -#define QWORD vals32 - -/* STRUCTURES */ -#define CISS_MAX_PHYS_LUN 1024 -/* SCSI-3 Cmmands */ - -#pragma pack(1) - -#define CISS_INQUIRY 0x12 -/* Date returned */ -typedef struct _InquiryData_struct -{ - BYTE data_byte[36]; -} InquiryData_struct; - -#define CISS_REPORT_LOG 0xc2 /* Report Logical LUNs */ -#define CISS_REPORT_PHYS 0xc3 /* Report Physical LUNs */ -/* Data returned */ -typedef struct _ReportLUNdata_struct -{ - BYTE LUNListLength[4]; - DWORD reserved; - BYTE LUN[CISS_MAX_LUN][8]; -} ReportLunData_struct; - -#define CCISS_READ_CAPACITY 0x25 /* Read Capacity */ -typedef struct _ReadCapdata_struct -{ - BYTE total_size[4]; /* Total size in blocks */ - BYTE block_size[4]; /* Size of blocks in bytes */ -} ReadCapdata_struct; - -#define CCISS_READ_CAPACITY_16 0x9e /* Read Capacity 16 */ - -/* service action to differentiate a 16 byte read capacity from - other commands that use the 0x9e SCSI op code */ - -#define CCISS_READ_CAPACITY_16_SERVICE_ACT 0x10 - -typedef struct _ReadCapdata_struct_16 -{ - BYTE total_size[8]; /* Total size in blocks */ - BYTE block_size[4]; /* Size of blocks in bytes */ - BYTE prot_en:1; /* protection enable bit */ - BYTE rto_en:1; /* reference tag own enable bit */ - BYTE reserved:6; /* reserved bits */ - BYTE reserved2[18]; /* reserved bytes per spec */ -} ReadCapdata_struct_16; - -/* Define the supported read/write commands for cciss based controllers */ - -#define CCISS_READ_10 0x28 /* Read(10) */ -#define CCISS_WRITE_10 0x2a /* Write(10) */ -#define CCISS_READ_16 0x88 /* Read(16) */ -#define CCISS_WRITE_16 0x8a /* Write(16) */ - -/* Define the CDB lengths supported by cciss based controllers */ - -#define CDB_LEN10 10 -#define CDB_LEN16 16 - -/* BMIC commands */ -#define BMIC_READ 0x26 -#define BMIC_WRITE 0x27 -#define BMIC_CACHE_FLUSH 0xc2 -#define CCISS_CACHE_FLUSH 0x01 /* C2 was already being used by CCISS */ - -#define CCISS_ABORT_MSG 0x00 -#define CCISS_RESET_MSG 0x01 -#define CCISS_RESET_TYPE_CONTROLLER 0x00 -#define CCISS_RESET_TYPE_BUS 0x01 -#define CCISS_RESET_TYPE_TARGET 0x03 -#define CCISS_RESET_TYPE_LUN 0x04 -#define CCISS_NOOP_MSG 0x03 - -/* Command List Structure */ -#define CTLR_LUNID "\0\0\0\0\0\0\0\0" - -typedef struct _CommandListHeader_struct { - BYTE ReplyQueue; - BYTE SGList; - HWORD SGTotal; - QWORD Tag; - LUNAddr_struct LUN; -} CommandListHeader_struct; -typedef struct _ErrDescriptor_struct { - QWORD Addr; - DWORD Len; -} ErrDescriptor_struct; -typedef struct _SGDescriptor_struct { - QWORD Addr; - DWORD Len; - DWORD Ext; -} SGDescriptor_struct; - -/* Command types */ -#define CMD_RWREQ 0x00 -#define CMD_IOCTL_PEND 0x01 -#define CMD_SCSI 0x03 -#define CMD_MSG_DONE 0x04 -#define CMD_MSG_TIMEOUT 0x05 -#define CMD_MSG_STALE 0xff - -/* This structure needs to be divisible by COMMANDLIST_ALIGNMENT - * because low bits of the address are used to to indicate that - * whether the tag contains an index or an address. PAD_32 and - * PAD_64 can be adjusted independently as needed for 32-bit - * and 64-bits systems. - */ -#define COMMANDLIST_ALIGNMENT (32) -#define IS_64_BIT ((sizeof(long) - 4)/4) -#define IS_32_BIT (!IS_64_BIT) -#define PAD_32 (0) -#define PAD_64 (4) -#define PADSIZE (IS_32_BIT * PAD_32 + IS_64_BIT * PAD_64) -#define DIRECT_LOOKUP_BIT 0x10 -#define DIRECT_LOOKUP_SHIFT 5 - -typedef struct _CommandList_struct { - CommandListHeader_struct Header; - RequestBlock_struct Request; - ErrDescriptor_struct ErrDesc; - SGDescriptor_struct SG[MAXSGENTRIES]; - /* information associated with the command */ - __u32 busaddr; /* physical address of this record */ - ErrorInfo_struct * err_info; /* pointer to the allocated mem */ - int ctlr; - int cmd_type; - long cmdindex; - struct list_head list; - struct request * rq; - struct completion *waiting; - int retry_count; - void * scsi_cmd; - char pad[PADSIZE]; -} CommandList_struct; - -/* Configuration Table Structure */ -typedef struct _HostWrite_struct { - DWORD TransportRequest; - DWORD Reserved; - DWORD CoalIntDelay; - DWORD CoalIntCount; -} HostWrite_struct; - -typedef struct _CfgTable_struct { - BYTE Signature[4]; - DWORD SpecValence; -#define SIMPLE_MODE 0x02 -#define PERFORMANT_MODE 0x04 -#define MEMQ_MODE 0x08 - DWORD TransportSupport; - DWORD TransportActive; - HostWrite_struct HostWrite; - DWORD CmdsOutMax; - DWORD BusTypes; - DWORD TransMethodOffset; - BYTE ServerName[16]; - DWORD HeartBeat; - DWORD SCSI_Prefetch; - DWORD MaxSGElements; - DWORD MaxLogicalUnits; - DWORD MaxPhysicalDrives; - DWORD MaxPhysicalDrivesPerLogicalUnit; - DWORD MaxPerformantModeCommands; - u8 reserved[0x78 - 0x58]; - u32 misc_fw_support; /* offset 0x78 */ -#define MISC_FW_DOORBELL_RESET (0x02) -#define MISC_FW_DOORBELL_RESET2 (0x10) - u8 driver_version[32]; -} CfgTable_struct; - -struct TransTable_struct { - u32 BlockFetch0; - u32 BlockFetch1; - u32 BlockFetch2; - u32 BlockFetch3; - u32 BlockFetch4; - u32 BlockFetch5; - u32 BlockFetch6; - u32 BlockFetch7; - u32 RepQSize; - u32 RepQCount; - u32 RepQCtrAddrLow32; - u32 RepQCtrAddrHigh32; - u32 RepQAddr0Low32; - u32 RepQAddr0High32; -}; - -#pragma pack() -#endif /* CCISS_CMD_H */ diff --git a/drivers/block/cciss_scsi.c b/drivers/block/cciss_scsi.c deleted file mode 100644 index 01a1f7e24978..000000000000 --- a/drivers/block/cciss_scsi.c +++ /dev/null @@ -1,1653 +0,0 @@ -/* - * Disk Array driver for HP Smart Array controllers, SCSI Tape module. - * (C) Copyright 2001, 2007 Hewlett-Packard Development Company, L.P. - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; version 2 of the License. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 300, Boston, MA - * 02111-1307, USA. - * - * Questions/Comments/Bugfixes to iss_storagedev@hp.com - * - * Author: Stephen M. Cameron - */ -#ifdef CONFIG_CISS_SCSI_TAPE - -/* Here we have code to present the driver as a scsi driver - as it is simultaneously presented as a block driver. The - reason for doing this is to allow access to SCSI tape drives - through the array controller. Note in particular, neither - physical nor logical disks are presented through the scsi layer. */ - -#include <linux/timer.h> -#include <linux/completion.h> -#include <linux/slab.h> -#include <linux/string.h> - -#include <linux/atomic.h> - -#include <scsi/scsi_cmnd.h> -#include <scsi/scsi_device.h> -#include <scsi/scsi_host.h> - -#include "cciss_scsi.h" - -#define CCISS_ABORT_MSG 0x00 -#define CCISS_RESET_MSG 0x01 - -static int fill_cmd(ctlr_info_t *h, CommandList_struct *c, __u8 cmd, void *buff, - size_t size, - __u8 page_code, unsigned char *scsi3addr, - int cmd_type); - -static CommandList_struct *cmd_alloc(ctlr_info_t *h); -static CommandList_struct *cmd_special_alloc(ctlr_info_t *h); -static void cmd_free(ctlr_info_t *h, CommandList_struct *c); -static void cmd_special_free(ctlr_info_t *h, CommandList_struct *c); - -static int cciss_scsi_write_info(struct Scsi_Host *sh, - char *buffer, /* data buffer */ - int length); /* length of data in buffer */ -static int cciss_scsi_show_info(struct seq_file *m, - struct Scsi_Host *sh); - -static int cciss_scsi_queue_command (struct Scsi_Host *h, - struct scsi_cmnd *cmd); -static int cciss_eh_device_reset_handler(struct scsi_cmnd *); -static int cciss_eh_abort_handler(struct scsi_cmnd *); - -static struct cciss_scsi_hba_t ccissscsi[MAX_CTLR] = { - { .name = "cciss0", .ndevices = 0 }, - { .name = "cciss1", .ndevices = 0 }, - { .name = "cciss2", .ndevices = 0 }, - { .name = "cciss3", .ndevices = 0 }, - { .name = "cciss4", .ndevices = 0 }, - { .name = "cciss5", .ndevices = 0 }, - { .name = "cciss6", .ndevices = 0 }, - { .name = "cciss7", .ndevices = 0 }, -}; - -static struct scsi_host_template cciss_driver_template = { - .module = THIS_MODULE, - .name = "cciss", - .proc_name = "cciss", - .write_info = cciss_scsi_write_info, - .show_info = cciss_scsi_show_info, - .queuecommand = cciss_scsi_queue_command, - .this_id = 7, - .use_clustering = DISABLE_CLUSTERING, - /* Can't have eh_bus_reset_handler or eh_host_reset_handler for cciss */ - .eh_device_reset_handler= cciss_eh_device_reset_handler, - .eh_abort_handler = cciss_eh_abort_handler, -}; - -#pragma pack(1) - -#define SCSI_PAD_32 8 -#define SCSI_PAD_64 8 - -struct cciss_scsi_cmd_stack_elem_t { - CommandList_struct cmd; - ErrorInfo_struct Err; - __u32 busaddr; - int cmdindex; - u8 pad[IS_32_BIT * SCSI_PAD_32 + IS_64_BIT * SCSI_PAD_64]; -}; - -#pragma pack() - -#pragma pack(1) -struct cciss_scsi_cmd_stack_t { - struct cciss_scsi_cmd_stack_elem_t *pool; - struct cciss_scsi_cmd_stack_elem_t **elem; - dma_addr_t cmd_pool_handle; - int top; - int nelems; -}; -#pragma pack() - -struct cciss_scsi_adapter_data_t { - struct Scsi_Host *scsi_host; - struct cciss_scsi_cmd_stack_t cmd_stack; - SGDescriptor_struct **cmd_sg_list; - int registered; - spinlock_t lock; // to protect ccissscsi[ctlr]; -}; - -#define CPQ_TAPE_LOCK(h, flags) spin_lock_irqsave( \ - &h->scsi_ctlr->lock, flags); -#define CPQ_TAPE_UNLOCK(h, flags) spin_unlock_irqrestore( \ - &h->scsi_ctlr->lock, flags); - -static CommandList_struct * -scsi_cmd_alloc(ctlr_info_t *h) -{ - /* assume only one process in here at a time, locking done by caller. */ - /* use h->lock */ - /* might be better to rewrite how we allocate scsi commands in a way that */ - /* needs no locking at all. */ - - /* take the top memory chunk off the stack and return it, if any. */ - struct cciss_scsi_cmd_stack_elem_t *c; - struct cciss_scsi_adapter_data_t *sa; - struct cciss_scsi_cmd_stack_t *stk; - u64bit temp64; - - sa = h->scsi_ctlr; - stk = &sa->cmd_stack; - - if (stk->top < 0) - return NULL; - c = stk->elem[stk->top]; - /* memset(c, 0, sizeof(*c)); */ - memset(&c->cmd, 0, sizeof(c->cmd)); - memset(&c->Err, 0, sizeof(c->Err)); - /* set physical addr of cmd and addr of scsi parameters */ - c->cmd.busaddr = c->busaddr; - c->cmd.cmdindex = c->cmdindex; - /* (__u32) (stk->cmd_pool_handle + - (sizeof(struct cciss_scsi_cmd_stack_elem_t)*stk->top)); */ - - temp64.val = (__u64) (c->busaddr + sizeof(CommandList_struct)); - /* (__u64) (stk->cmd_pool_handle + - (sizeof(struct cciss_scsi_cmd_stack_elem_t)*stk->top) + - sizeof(CommandList_struct)); */ - stk->top--; - c->cmd.ErrDesc.Addr.lower = temp64.val32.lower; - c->cmd.ErrDesc.Addr.upper = temp64.val32.upper; - c->cmd.ErrDesc.Len = sizeof(ErrorInfo_struct); - - c->cmd.ctlr = h->ctlr; - c->cmd.err_info = &c->Err; - - return (CommandList_struct *) c; -} - -static void -scsi_cmd_free(ctlr_info_t *h, CommandList_struct *c) -{ - /* assume only one process in here at a time, locking done by caller. */ - /* use h->lock */ - /* drop the free memory chunk on top of the stack. */ - - struct cciss_scsi_adapter_data_t *sa; - struct cciss_scsi_cmd_stack_t *stk; - - sa = h->scsi_ctlr; - stk = &sa->cmd_stack; - stk->top++; - if (stk->top >= stk->nelems) { - dev_err(&h->pdev->dev, - "scsi_cmd_free called too many times.\n"); - BUG(); - } - stk->elem[stk->top] = (struct cciss_scsi_cmd_stack_elem_t *) c; -} - -static int -scsi_cmd_stack_setup(ctlr_info_t *h, struct cciss_scsi_adapter_data_t *sa) -{ - int i; - struct cciss_scsi_cmd_stack_t *stk; - size_t size; - - stk = &sa->cmd_stack; - stk->nelems = cciss_tape_cmds + 2; - sa->cmd_sg_list = cciss_allocate_sg_chain_blocks(h, - h->chainsize, stk->nelems); - if (!sa->cmd_sg_list && h->chainsize > 0) - return -ENOMEM; - - size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * stk->nelems; - - /* Check alignment, see cciss_cmd.h near CommandList_struct def. */ - BUILD_BUG_ON((sizeof(*stk->pool) % COMMANDLIST_ALIGNMENT) != 0); - /* pci_alloc_consistent guarantees 32-bit DMA address will be used */ - stk->pool = (struct cciss_scsi_cmd_stack_elem_t *) - pci_alloc_consistent(h->pdev, size, &stk->cmd_pool_handle); - - if (stk->pool == NULL) { - cciss_free_sg_chain_blocks(sa->cmd_sg_list, stk->nelems); - sa->cmd_sg_list = NULL; - return -ENOMEM; - } - stk->elem = kmalloc(sizeof(stk->elem[0]) * stk->nelems, GFP_KERNEL); - if (!stk->elem) { - pci_free_consistent(h->pdev, size, stk->pool, - stk->cmd_pool_handle); - return -1; - } - for (i = 0; i < stk->nelems; i++) { - stk->elem[i] = &stk->pool[i]; - stk->elem[i]->busaddr = (__u32) (stk->cmd_pool_handle + - (sizeof(struct cciss_scsi_cmd_stack_elem_t) * i)); - stk->elem[i]->cmdindex = i; - } - stk->top = stk->nelems-1; - return 0; -} - -static void -scsi_cmd_stack_free(ctlr_info_t *h) -{ - struct cciss_scsi_adapter_data_t *sa; - struct cciss_scsi_cmd_stack_t *stk; - size_t size; - - sa = h->scsi_ctlr; - stk = &sa->cmd_stack; - if (stk->top != stk->nelems-1) { - dev_warn(&h->pdev->dev, - "bug: %d scsi commands are still outstanding.\n", - stk->nelems - stk->top); - } - size = sizeof(struct cciss_scsi_cmd_stack_elem_t) * stk->nelems; - - pci_free_consistent(h->pdev, size, stk->pool, stk->cmd_pool_handle); - stk->pool = NULL; - cciss_free_sg_chain_blocks(sa->cmd_sg_list, stk->nelems); - kfree(stk->elem); - stk->elem = NULL; -} - -#if 0 -static void -print_cmd(CommandList_struct *cp) -{ - printk("queue:%d\n", cp->Header.ReplyQueue); - printk("sglist:%d\n", cp->Header.SGList); - printk("sgtot:%d\n", cp->Header.SGTotal); - printk("Tag:0x%08x/0x%08x\n", cp->Header.Tag.upper, - cp->Header.Tag.lower); - printk("LUN:0x%8phN\n", cp->Header.LUN.LunAddrBytes); - printk("CDBLen:%d\n", cp->Request.CDBLen); - printk("Type:%d\n",cp->Request.Type.Type); - printk("Attr:%d\n",cp->Request.Type.Attribute); - printk(" Dir:%d\n",cp->Request.Type.Direction); - printk("Timeout:%d\n",cp->Request.Timeout); - printk("CDB: %16ph\n", cp->Request.CDB); - printk("edesc.Addr: 0x%08x/0%08x, Len = %d\n", - cp->ErrDesc.Addr.upper, cp->ErrDesc.Addr.lower, - cp->ErrDesc.Len); - printk("sgs..........Errorinfo:\n"); - printk("scsistatus:%d\n", cp->err_info->ScsiStatus); - printk("senselen:%d\n", cp->err_info->SenseLen); - printk("cmd status:%d\n", cp->err_info->CommandStatus); - printk("resid cnt:%d\n", cp->err_info->ResidualCnt); - printk("offense size:%d\n", cp->err_info->MoreErrInfo.Invalid_Cmd.offense_size); - printk("offense byte:%d\n", cp->err_info->MoreErrInfo.Invalid_Cmd.offense_num); - printk("offense value:%d\n", cp->err_info->MoreErrInfo.Invalid_Cmd.offense_value); -} -#endif - -static int -find_bus_target_lun(ctlr_info_t *h, int *bus, int *target, int *lun) -{ - /* finds an unused bus, target, lun for a new device */ - /* assumes h->scsi_ctlr->lock is held */ - int i, found=0; - unsigned char target_taken[CCISS_MAX_SCSI_DEVS_PER_HBA]; - - memset(&target_taken[0], 0, CCISS_MAX_SCSI_DEVS_PER_HBA); - - target_taken[SELF_SCSI_ID] = 1; - for (i = 0; i < ccissscsi[h->ctlr].ndevices; i++) - target_taken[ccissscsi[h->ctlr].dev[i].target] = 1; - - for (i = 0; i < CCISS_MAX_SCSI_DEVS_PER_HBA; i++) { - if (!target_taken[i]) { - *bus = 0; *target=i; *lun = 0; found=1; - break; - } - } - return (!found); -} -struct scsi2map { - char scsi3addr[8]; - int bus, target, lun; -}; - -static int -cciss_scsi_add_entry(ctlr_info_t *h, int hostno, - struct cciss_scsi_dev_t *device, - struct scsi2map *added, int *nadded) -{ - /* assumes h->scsi_ctlr->lock is held */ - int n = ccissscsi[h->ctlr].ndevices; - struct cciss_scsi_dev_t *sd; - int i, bus, target, lun; - unsigned char addr1[8], addr2[8]; - - if (n >= CCISS_MAX_SCSI_DEVS_PER_HBA) { - dev_warn(&h->pdev->dev, "Too many devices, " - "some will be inaccessible.\n"); - return -1; - } - - bus = target = -1; - lun = 0; - /* Is this device a non-zero lun of a multi-lun device */ - /* byte 4 of the 8-byte LUN addr will contain the logical unit no. */ - if (device->scsi3addr[4] != 0) { - /* Search through our list and find the device which */ - /* has the same 8 byte LUN address, excepting byte 4. */ - /* Assign the same bus and target for this new LUN. */ - /* Use the logical unit number from the firmware. */ - memcpy(addr1, device->scsi3addr, 8); - addr1[4] = 0; - for (i = 0; i < n; i++) { - sd = &ccissscsi[h->ctlr].dev[i]; - memcpy(addr2, sd->scsi3addr, 8); - addr2[4] = 0; - /* differ only in byte 4? */ - if (memcmp(addr1, addr2, 8) == 0) { - bus = sd->bus; - target = sd->target; - lun = device->scsi3addr[4]; - break; - } - } - } - - sd = &ccissscsi[h->ctlr].dev[n]; - if (lun == 0) { - if (find_bus_target_lun(h, - &sd->bus, &sd->target, &sd->lun) != 0) - return -1; - } else { - sd->bus = bus; - sd->target = target; - sd->lun = lun; - } - added[*nadded].bus = sd->bus; - added[*nadded].target = sd->target; - added[*nadded].lun = sd->lun; - (*nadded)++; - - memcpy(sd->scsi3addr, device->scsi3addr, 8); - memcpy(sd->vendor, device->vendor, sizeof(sd->vendor)); - memcpy(sd->revision, device->revision, sizeof(sd->revision)); - memcpy(sd->device_id, device->device_id, sizeof(sd->device_id)); - sd->devtype = device->devtype; - - ccissscsi[h->ctlr].ndevices++; - - /* initially, (before registering with scsi layer) we don't - know our hostno and we don't want to print anything first - time anyway (the scsi layer's inquiries will show that info) */ - if (hostno != -1) - dev_info(&h->pdev->dev, "%s device c%db%dt%dl%d added.\n", - scsi_device_type(sd->devtype), hostno, - sd->bus, sd->target, sd->lun); - return 0; -} - -static void -cciss_scsi_remove_entry(ctlr_info_t *h, int hostno, int entry, - struct scsi2map *removed, int *nremoved) -{ - /* assumes h->ctlr]->scsi_ctlr->lock is held */ - int i; - struct cciss_scsi_dev_t sd; - - if (entry < 0 || entry >= CCISS_MAX_SCSI_DEVS_PER_HBA) return; - sd = ccissscsi[h->ctlr].dev[entry]; - removed[*nremoved].bus = sd.bus; - removed[*nremoved].target = sd.target; - removed[*nremoved].lun = sd.lun; - (*nremoved)++; - for (i = entry; i < ccissscsi[h->ctlr].ndevices-1; i++) - ccissscsi[h->ctlr].dev[i] = ccissscsi[h->ctlr].dev[i+1]; - ccissscsi[h->ctlr].ndevices--; - dev_info(&h->pdev->dev, "%s device c%db%dt%dl%d removed.\n", - scsi_device_type(sd.devtype), hostno, - sd.bus, sd.target, sd.lun); -} - - -#define SCSI3ADDR_EQ(a,b) ( \ - (a)[7] == (b)[7] && \ - (a)[6] == (b)[6] && \ - (a)[5] == (b)[5] && \ - (a)[4] == (b)[4] && \ - (a)[3] == (b)[3] && \ - (a)[2] == (b)[2] && \ - (a)[1] == (b)[1] && \ - (a)[0] == (b)[0]) - -static void fixup_botched_add(ctlr_info_t *h, char *scsi3addr) -{ - /* called when scsi_add_device fails in order to re-adjust */ - /* ccissscsi[] to match the mid layer's view. */ - unsigned long flags; - int i, j; - CPQ_TAPE_LOCK(h, flags); - for (i = 0; i < ccissscsi[h->ctlr].ndevices; i++) { - if (memcmp(scsi3addr, - ccissscsi[h->ctlr].dev[i].scsi3addr, 8) == 0) { - for (j = i; j < ccissscsi[h->ctlr].ndevices-1; j++) - ccissscsi[h->ctlr].dev[j] = - ccissscsi[h->ctlr].dev[j+1]; - ccissscsi[h->ctlr].ndevices--; - break; - } - } - CPQ_TAPE_UNLOCK(h, flags); -} - -static int device_is_the_same(struct cciss_scsi_dev_t *dev1, - struct cciss_scsi_dev_t *dev2) -{ - return dev1->devtype == dev2->devtype && - memcmp(dev1->scsi3addr, dev2->scsi3addr, - sizeof(dev1->scsi3addr)) == 0 && - memcmp(dev1->device_id, dev2->device_id, - sizeof(dev1->device_id)) == 0 && - memcmp(dev1->vendor, dev2->vendor, - sizeof(dev1->vendor)) == 0 && - memcmp(dev1->model, dev2->model, - sizeof(dev1->model)) == 0 && - memcmp(dev1->revision, dev2->revision, - sizeof(dev1->revision)) == 0; -} - -static int -adjust_cciss_scsi_table(ctlr_info_t *h, int hostno, - struct cciss_scsi_dev_t sd[], int nsds) -{ - /* sd contains scsi3 addresses and devtypes, but - bus target and lun are not filled in. This funciton - takes what's in sd to be the current and adjusts - ccissscsi[] to be in line with what's in sd. */ - - int i,j, found, changes=0; - struct cciss_scsi_dev_t *csd; - unsigned long flags; - struct scsi2map *added, *removed; - int nadded, nremoved; - struct Scsi_Host *sh = NULL; - - added = kzalloc(sizeof(*added) * CCISS_MAX_SCSI_DEVS_PER_HBA, - GFP_KERNEL); - removed = kzalloc(sizeof(*removed) * CCISS_MAX_SCSI_DEVS_PER_HBA, - GFP_KERNEL); - - if (!added || !removed) { - dev_warn(&h->pdev->dev, - "Out of memory in adjust_cciss_scsi_table\n"); - goto free_and_out; - } - - CPQ_TAPE_LOCK(h, flags); - - if (hostno != -1) /* if it's not the first time... */ - sh = h->scsi_ctlr->scsi_host; - - /* find any devices in ccissscsi[] that are not in - sd[] and remove them from ccissscsi[] */ - - i = 0; - nremoved = 0; - nadded = 0; - while (i < ccissscsi[h->ctlr].ndevices) { - csd = &ccissscsi[h->ctlr].dev[i]; - found=0; - for (j=0;j<nsds;j++) { - if (SCSI3ADDR_EQ(sd[j].scsi3addr, - csd->scsi3addr)) { - if (device_is_the_same(&sd[j], csd)) - found=2; - else - found=1; - break; - } - } - - if (found == 0) { /* device no longer present. */ - changes++; - cciss_scsi_remove_entry(h, hostno, i, - removed, &nremoved); - /* remove ^^^, hence i not incremented */ - } else if (found == 1) { /* device is different in some way */ - changes++; - dev_info(&h->pdev->dev, - "device c%db%dt%dl%d has changed.\n", - hostno, csd->bus, csd->target, csd->lun); - cciss_scsi_remove_entry(h, hostno, i, - removed, &nremoved); - /* remove ^^^, hence i not incremented */ - if (cciss_scsi_add_entry(h, hostno, &sd[j], - added, &nadded) != 0) - /* we just removed one, so add can't fail. */ - BUG(); - csd->devtype = sd[j].devtype; - memcpy(csd->device_id, sd[j].device_id, - sizeof(csd->device_id)); - memcpy(csd->vendor, sd[j].vendor, - sizeof(csd->vendor)); - memcpy(csd->model, sd[j].model, - sizeof(csd->model)); - memcpy(csd->revision, sd[j].revision, - sizeof(csd->revision)); - } else /* device is same as it ever was, */ - i++; /* so just move along. */ - } - - /* Now, make sure every device listed in sd[] is also - listed in ccissscsi[], adding them if they aren't found */ - - for (i=0;i<nsds;i++) { - found=0; - for (j = 0; j < ccissscsi[h->ctlr].ndevices; j++) { - csd = &ccissscsi[h->ctlr].dev[j]; - if (SCSI3ADDR_EQ(sd[i].scsi3addr, - csd->scsi3addr)) { - if (device_is_the_same(&sd[i], csd)) - found=2; /* found device */ - else - found=1; /* found a bug. */ - break; - } - } - if (!found) { - changes++; - if (cciss_scsi_add_entry(h, hostno, &sd[i], - added, &nadded) != 0) - break; - } else if (found == 1) { - /* should never happen... */ - changes++; - dev_warn(&h->pdev->dev, - "device unexpectedly changed\n"); - /* but if it does happen, we just ignore that device */ - } - } - CPQ_TAPE_UNLOCK(h, flags); - - /* Don't notify scsi mid layer of any changes the first time through */ - /* (or if there are no changes) scsi_scan_host will do it later the */ - /* first time through. */ - if (hostno == -1 || !changes) - goto free_and_out; - - /* Notify scsi mid layer of any removed devices */ - for (i = 0; i < nremoved; i++) { - struct scsi_device *sdev = - scsi_device_lookup(sh, removed[i].bus, - removed[i].target, removed[i].lun); - if (sdev != NULL) { - scsi_remove_device(sdev); - scsi_device_put(sdev); - } else { - /* We don't expect to get here. */ - /* future cmds to this device will get selection */ - /* timeout as if the device was gone. */ - dev_warn(&h->pdev->dev, "didn't find " - "c%db%dt%dl%d\n for removal.", - hostno, removed[i].bus, - removed[i].target, removed[i].lun); - } - } - - /* Notify scsi mid layer of any added devices */ - for (i = 0; i < nadded; i++) { - int rc; - rc = scsi_add_device(sh, added[i].bus, - added[i].target, added[i].lun); - if (rc == 0) - continue; - dev_warn(&h->pdev->dev, "scsi_add_device " - "c%db%dt%dl%d failed, device not added.\n", - hostno, added[i].bus, added[i].target, added[i].lun); - /* now we have to remove it from ccissscsi, */ - /* since it didn't get added to scsi mid layer */ - fixup_botched_add(h, added[i].scsi3addr); - } - -free_and_out: - kfree(added); - kfree(removed); - return 0; -} - -static int -lookup_scsi3addr(ctlr_info_t *h, int bus, int target, int lun, char *scsi3addr) -{ - int i; - struct cciss_scsi_dev_t *sd; - unsigned long flags; - - CPQ_TAPE_LOCK(h, flags); - for (i = 0; i < ccissscsi[h->ctlr].ndevices; i++) { - sd = &ccissscsi[h->ctlr].dev[i]; - if (sd->bus == bus && - sd->target == target && - sd->lun == lun) { - memcpy(scsi3addr, &sd->scsi3addr[0], 8); - CPQ_TAPE_UNLOCK(h, flags); - return 0; - } - } - CPQ_TAPE_UNLOCK(h, flags); - return -1; -} - -static void -cciss_scsi_setup(ctlr_info_t *h) -{ - struct cciss_scsi_adapter_data_t * shba; - - ccissscsi[h->ctlr].ndevices = 0; - shba = kmalloc(sizeof(*shba), GFP_KERNEL); - if (shba == NULL) - return; - shba->scsi_host = NULL; - spin_lock_init(&shba->lock); - shba->registered = 0; - if (scsi_cmd_stack_setup(h, shba) != 0) { - kfree(shba); - shba = NULL; - } - h->scsi_ctlr = shba; - return; -} - -static void complete_scsi_command(CommandList_struct *c, int timeout, - __u32 tag) -{ - struct scsi_cmnd *cmd; - ctlr_info_t *h; - ErrorInfo_struct *ei; - - ei = c->err_info; - - /* First, see if it was a message rather than a command */ - if (c->Request.Type.Type == TYPE_MSG) { - c->cmd_type = CMD_MSG_DONE; - return; - } - - cmd = (struct scsi_cmnd *) c->scsi_cmd; - h = hba[c->ctlr]; - - scsi_dma_unmap(cmd); - if (c->Header.SGTotal > h->max_cmd_sgentries) - cciss_unmap_sg_chain_block(h, c); - - cmd->result = (DID_OK << 16); /* host byte */ - cmd->result |= (COMMAND_COMPLETE << 8); /* msg byte */ - /* cmd->result |= (GOOD < 1); */ /* status byte */ - - cmd->result |= (ei->ScsiStatus); - /* printk("Scsistatus is 0x%02x\n", ei->ScsiStatus); */ - - /* copy the sense data whether we need to or not. */ - - memcpy(cmd->sense_buffer, ei->SenseInfo, - ei->SenseLen > SCSI_SENSE_BUFFERSIZE ? - SCSI_SENSE_BUFFERSIZE : - ei->SenseLen); - scsi_set_resid(cmd, ei->ResidualCnt); - - if (ei->CommandStatus != 0) { /* an error has occurred */ - switch (ei->CommandStatus) { - case CMD_TARGET_STATUS: - /* Pass it up to the upper layers... */ - if (!ei->ScsiStatus) { - - /* Ordinarily, this case should never happen, but there is a bug - in some released firmware revisions that allows it to happen - if, for example, a 4100 backplane loses power and the tape - drive is in it. We assume that it's a fatal error of some - kind because we can't show that it wasn't. We will make it - look like selection timeout since that is the most common - reason for this to occur, and it's severe enough. */ - - cmd->result = DID_NO_CONNECT << 16; - } - break; - case CMD_DATA_UNDERRUN: /* let mid layer handle it. */ - break; - case CMD_DATA_OVERRUN: - dev_warn(&h->pdev->dev, "%p has" - " completed with data overrun " - "reported\n", c); - break; - case CMD_INVALID: { - /* - print_hex_dump(KERN_INFO, "", DUMP_PREFIX_OFFSET, 16, 1, c, sizeof(*c), false); - print_cmd(c); - */ - /* We get CMD_INVALID if you address a non-existent tape drive instead - of a selection timeout (no response). You will see this if you yank - out a tape drive, then try to access it. This is kind of a shame - because it means that any other CMD_INVALID (e.g. driver bug) will - get interpreted as a missing target. */ - cmd->result = DID_NO_CONNECT << 16; - } - break; - case CMD_PROTOCOL_ERR: - cmd->result = DID_ERROR << 16; - dev_warn(&h->pdev->dev, - "%p has protocol error\n", c); - break; - case CMD_HARDWARE_ERR: - cmd->result = DID_ERROR << 16; - dev_warn(&h->pdev->dev, - "%p had hardware error\n", c); - break; - case CMD_CONNECTION_LOST: - cmd->result = DID_ERROR << 16; - dev_warn(&h->pdev->dev, - "%p had connection lost\n", c); - break; - case CMD_ABORTED: - cmd->result = DID_ABORT << 16; - dev_warn(&h->pdev->dev, "%p was aborted\n", c); - break; - case CMD_ABORT_FAILED: - cmd->result = DID_ERROR << 16; - dev_warn(&h->pdev->dev, - "%p reports abort failed\n", c); - break; - case CMD_UNSOLICITED_ABORT: - cmd->result = DID_ABORT << 16; - dev_warn(&h->pdev->dev, "%p aborted due to an " - "unsolicited abort\n", c); - break; - case CMD_TIMEOUT: - cmd->result = DID_TIME_OUT << 16; - dev_warn(&h->pdev->dev, "%p timedout\n", c); - break; - case CMD_UNABORTABLE: - cmd->result = DID_ERROR << 16; - dev_warn(&h->pdev->dev, "c %p command " - "unabortable\n", c); - break; - default: - cmd->result = DID_ERROR << 16; - dev_warn(&h->pdev->dev, - "%p returned unknown status %x\n", c, - ei->CommandStatus); - } - } - cmd->scsi_done(cmd); - scsi_cmd_free(h, c); -} - -static int -cciss_scsi_detect(ctlr_info_t *h) -{ - struct Scsi_Host *sh; - int error; - - sh = scsi_host_alloc(&cciss_driver_template, sizeof(struct ctlr_info *)); - if (sh == NULL) - goto fail; - sh->io_port = 0; // good enough? FIXME, - sh->n_io_port = 0; // I don't think we use these two... - sh->this_id = SELF_SCSI_ID; - sh->can_queue = cciss_tape_cmds; - sh->sg_tablesize = h->maxsgentries; - sh->max_cmd_len = MAX_COMMAND_SIZE; - sh->max_sectors = h->cciss_max_sectors; - - ((struct cciss_scsi_adapter_data_t *) - h->scsi_ctlr)->scsi_host = sh; - sh->hostdata[0] = (unsigned long) h; - sh->irq = h->intr[SIMPLE_MODE_INT]; - sh->unique_id = sh->irq; - error = scsi_add_host(sh, &h->pdev->dev); - if (error) - goto fail_host_put; - scsi_scan_host(sh); - return 1; - - fail_host_put: - scsi_host_put(sh); - fail: - return 0; -} - -static void -cciss_unmap_one(struct pci_dev *pdev, - CommandList_struct *c, - size_t buflen, - int data_direction) -{ - u64bit addr64; - - addr64.val32.lower = c->SG[0].Addr.lower; - addr64.val32.upper = c->SG[0].Addr.upper; - pci_unmap_single(pdev, (dma_addr_t) addr64.val, buflen, data_direction); -} - -static void -cciss_map_one(struct pci_dev *pdev, - CommandList_struct *c, - unsigned char *buf, - size_t buflen, - int data_direction) -{ - __u64 addr64; - - addr64 = (__u64) pci_map_single(pdev, buf, buflen, data_direction); - c->SG[0].Addr.lower = - (__u32) (addr64 & (__u64) 0x00000000FFFFFFFF); - c->SG[0].Addr.upper = - (__u32) ((addr64 >> 32) & (__u64) 0x00000000FFFFFFFF); - c->SG[0].Len = buflen; - c->Header.SGList = (__u8) 1; /* no. SGs contig in this cmd */ - c->Header.SGTotal = (__u16) 1; /* total sgs in this cmd list */ -} - -static int -cciss_scsi_do_simple_cmd(ctlr_info_t *h, - CommandList_struct *c, - unsigned char *scsi3addr, - unsigned char *cdb, - unsigned char cdblen, - unsigned char *buf, int bufsize, - int direction) -{ - DECLARE_COMPLETION_ONSTACK(wait); - - c->cmd_type = CMD_IOCTL_PEND; /* treat this like an ioctl */ - c->scsi_cmd = NULL; - c->Header.ReplyQueue = 0; /* unused in simple mode */ - memcpy(&c->Header.LUN, scsi3addr, sizeof(c->Header.LUN)); - c->Header.Tag.lower = c->busaddr; /* Use k. address of cmd as tag */ - // Fill in the request block... - - /* printk("Using scsi3addr 0x%02x%0x2%0x2%0x2%0x2%0x2%0x2%0x2\n", - scsi3addr[0], scsi3addr[1], scsi3addr[2], scsi3addr[3], - scsi3addr[4], scsi3addr[5], scsi3addr[6], scsi3addr[7]); */ - - memset(c->Request.CDB, 0, sizeof(c->Request.CDB)); - memcpy(c->Request.CDB, cdb, cdblen); - c->Request.Timeout = 0; - c->Request.CDBLen = cdblen; - c->Request.Type.Type = TYPE_CMD; - c->Request.Type.Attribute = ATTR_SIMPLE; - c->Request.Type.Direction = direction; - - /* Fill in the SG list and do dma mapping */ - cciss_map_one(h->pdev, c, (unsigned char *) buf, - bufsize, DMA_FROM_DEVICE); - - c->waiting = &wait; - enqueue_cmd_and_start_io(h, c); - wait_for_completion(&wait); - - /* undo the dma mapping */ - cciss_unmap_one(h->pdev, c, bufsize, DMA_FROM_DEVICE); - return(0); -} - -static void -cciss_scsi_interpret_error(ctlr_info_t *h, CommandList_struct *c) -{ - ErrorInfo_struct *ei; - - ei = c->err_info; - switch (ei->CommandStatus) { - case CMD_TARGET_STATUS: - dev_warn(&h->pdev->dev, - "cmd %p has completed with errors\n", c); - dev_warn(&h->pdev->dev, - "cmd %p has SCSI Status = %x\n", - c, ei->ScsiStatus); - if (ei->ScsiStatus == 0) - dev_warn(&h->pdev->dev, - "SCSI status is abnormally zero. " - "(probably indicates selection timeout " - "reported incorrectly due to a known " - "firmware bug, circa July, 2001.)\n"); - break; - case CMD_DATA_UNDERRUN: /* let mid layer handle it. */ - dev_info(&h->pdev->dev, "UNDERRUN\n"); - break; - case CMD_DATA_OVERRUN: - dev_warn(&h->pdev->dev, "%p has" - " completed with data overrun " - "reported\n", c); - break; - case CMD_INVALID: { - /* controller unfortunately reports SCSI passthru's */ - /* to non-existent targets as invalid commands. */ - dev_warn(&h->pdev->dev, - "%p is reported invalid (probably means " - "target device no longer present)\n", c); - /* - print_hex_dump(KERN_INFO, "", DUMP_PREFIX_OFFSET, 16, 1, c, sizeof(*c), false); - print_cmd(c); - */ - } - break; - case CMD_PROTOCOL_ERR: - dev_warn(&h->pdev->dev, "%p has protocol error\n", c); - break; - case CMD_HARDWARE_ERR: - /* cmd->result = DID_ERROR << 16; */ - dev_warn(&h->pdev->dev, "%p had hardware error\n", c); - break; - case CMD_CONNECTION_LOST: - dev_warn(&h->pdev->dev, "%p had connection lost\n", c); - break; - case CMD_ABORTED: - dev_warn(&h->pdev->dev, "%p was aborted\n", c); - break; - case CMD_ABORT_FAILED: - dev_warn(&h->pdev->dev, - "%p reports abort failed\n", c); - break; - case CMD_UNSOLICITED_ABORT: - dev_warn(&h->pdev->dev, - "%p aborted due to an unsolicited abort\n", c); - break; - case CMD_TIMEOUT: - dev_warn(&h->pdev->dev, "%p timedout\n", c); - break; - case CMD_UNABORTABLE: - dev_warn(&h->pdev->dev, - "%p unabortable\n", c); - break; - default: - dev_warn(&h->pdev->dev, - "%p returned unknown status %x\n", - c, ei->CommandStatus); - } -} - -static int -cciss_scsi_do_inquiry(ctlr_info_t *h, unsigned char *scsi3addr, - unsigned char page, unsigned char *buf, - unsigned char bufsize) -{ - int rc; - CommandList_struct *c; - char cdb[6]; - ErrorInfo_struct *ei; - unsigned long flags; - - spin_lock_irqsave(&h->lock, flags); - c = scsi_cmd_alloc(h); - spin_unlock_irqrestore(&h->lock, flags); - - if (c == NULL) { /* trouble... */ - printk("cmd_alloc returned NULL!\n"); - return -1; - } - - ei = c->err_info; - - cdb[0] = CISS_INQUIRY; - cdb[1] = (page != 0); - cdb[2] = page; - cdb[3] = 0; - cdb[4] = bufsize; - cdb[5] = 0; - rc = cciss_scsi_do_simple_cmd(h, c, scsi3addr, cdb, - 6, buf, bufsize, XFER_READ); - - if (rc != 0) return rc; /* something went wrong */ - - if (ei->CommandStatus != 0 && - ei->CommandStatus != CMD_DATA_UNDERRUN) { - cciss_scsi_interpret_error(h, c); - rc = -1; - } - spin_lock_irqsave(&h->lock, flags); - scsi_cmd_free(h, c); - spin_unlock_irqrestore(&h->lock, flags); - return rc; -} - -/* Get the device id from inquiry page 0x83 */ -static int cciss_scsi_get_device_id(ctlr_info_t *h, unsigned char *scsi3addr, - unsigned char *device_id, int buflen) -{ - int rc; - unsigned char *buf; - - if (buflen > 16) - buflen = 16; - buf = kzalloc(64, GFP_KERNEL); - if (!buf) - return -1; - rc = cciss_scsi_do_inquiry(h, scsi3addr, 0x83, buf, 64); - if (rc == 0) - memcpy(device_id, &buf[8], buflen); - kfree(buf); - return rc != 0; -} - -static int -cciss_scsi_do_report_phys_luns(ctlr_info_t *h, - ReportLunData_struct *buf, int bufsize) -{ - int rc; - CommandList_struct *c; - unsigned char cdb[12]; - unsigned char scsi3addr[8]; - ErrorInfo_struct *ei; - unsigned long flags; - - spin_lock_irqsave(&h->lock, flags); - c = scsi_cmd_alloc(h); - spin_unlock_irqrestore(&h->lock, flags); - if (c == NULL) { /* trouble... */ - printk("cmd_alloc returned NULL!\n"); - return -1; - } - - memset(&scsi3addr[0], 0, 8); /* address the controller */ - cdb[0] = CISS_REPORT_PHYS; - cdb[1] = 0; - cdb[2] = 0; - cdb[3] = 0; - cdb[4] = 0; - cdb[5] = 0; - cdb[6] = (bufsize >> 24) & 0xFF; //MSB - cdb[7] = (bufsize >> 16) & 0xFF; - cdb[8] = (bufsize >> 8) & 0xFF; - cdb[9] = bufsize & 0xFF; - cdb[10] = 0; - cdb[11] = 0; - - rc = cciss_scsi_do_simple_cmd(h, c, scsi3addr, - cdb, 12, - (unsigned char *) buf, - bufsize, XFER_READ); - - if (rc != 0) return rc; /* something went wrong */ - - ei = c->err_info; - if (ei->CommandStatus != 0 && - ei->CommandStatus != CMD_DATA_UNDERRUN) { - cciss_scsi_interpret_error(h, c); - rc = -1; - } - spin_lock_irqsave(&h->lock, flags); - scsi_cmd_free(h, c); - spin_unlock_irqrestore(&h->lock, flags); - return rc; -} - -static void -cciss_update_non_disk_devices(ctlr_info_t *h, int hostno) -{ - /* the idea here is we could get notified from /proc - that some devices have changed, so we do a report - physical luns cmd, and adjust our list of devices - accordingly. (We can't rely on the scsi-mid layer just - doing inquiries, because the "busses" that the scsi - mid-layer probes are totally fabricated by this driver, - so new devices wouldn't show up. - - the scsi3addr's of devices won't change so long as the - adapter is not reset. That means we can rescan and - tell which devices we already know about, vs. new - devices, vs. disappearing devices. - - Also, if you yank out a tape drive, then put in a disk - in it's place, (say, a configured volume from another - array controller for instance) _don't_ poke this driver - (so it thinks it's still a tape, but _do_ poke the scsi - mid layer, so it does an inquiry... the scsi mid layer - will see the physical disk. This would be bad. Need to - think about how to prevent that. One idea would be to - snoop all scsi responses and if an inquiry repsonse comes - back that reports a disk, chuck it an return selection - timeout instead and adjust our table... Not sure i like - that though. - - */ -#define OBDR_TAPE_INQ_SIZE 49 -#define OBDR_TAPE_SIG "$DR-10" - ReportLunData_struct *ld_buff; - unsigned char *inq_buff; - unsigned char scsi3addr[8]; - __u32 num_luns=0; - unsigned char *ch; - struct cciss_scsi_dev_t *currentsd, *this_device; - int ncurrent=0; - int reportlunsize = sizeof(*ld_buff) + CISS_MAX_PHYS_LUN * 8; - int i; - - ld_buff = kzalloc(reportlunsize, GFP_KERNEL); - inq_buff = kmalloc(OBDR_TAPE_INQ_SIZE, GFP_KERNEL); - currentsd = kzalloc(sizeof(*currentsd) * - (CCISS_MAX_SCSI_DEVS_PER_HBA+1), GFP_KERNEL); - if (ld_buff == NULL || inq_buff == NULL || currentsd == NULL) { - printk(KERN_ERR "cciss: out of memory\n"); - goto out; - } - this_device = ¤tsd[CCISS_MAX_SCSI_DEVS_PER_HBA]; - if (cciss_scsi_do_report_phys_luns(h, ld_buff, reportlunsize) == 0) { - ch = &ld_buff->LUNListLength[0]; - num_luns = ((ch[0]<<24) | (ch[1]<<16) | (ch[2]<<8) | ch[3]) / 8; - if (num_luns > CISS_MAX_PHYS_LUN) { - printk(KERN_WARNING - "cciss: Maximum physical LUNs (%d) exceeded. " - "%d LUNs ignored.\n", CISS_MAX_PHYS_LUN, - num_luns - CISS_MAX_PHYS_LUN); - num_luns = CISS_MAX_PHYS_LUN; - } - } - else { - printk(KERN_ERR "cciss: Report physical LUNs failed.\n"); - goto out; - } - - - /* adjust our table of devices */ - for (i = 0; i < num_luns; i++) { - /* for each physical lun, do an inquiry */ - if (ld_buff->LUN[i][3] & 0xC0) continue; - memset(inq_buff, 0, OBDR_TAPE_INQ_SIZE); - memcpy(&scsi3addr[0], &ld_buff->LUN[i][0], 8); - - if (cciss_scsi_do_inquiry(h, scsi3addr, 0, inq_buff, - (unsigned char) OBDR_TAPE_INQ_SIZE) != 0) - /* Inquiry failed (msg printed already) */ - continue; /* so we will skip this device. */ - - this_device->devtype = (inq_buff[0] & 0x1f); - this_device->bus = -1; - this_device->target = -1; - this_device->lun = -1; - memcpy(this_device->scsi3addr, scsi3addr, 8); - memcpy(this_device->vendor, &inq_buff[8], - sizeof(this_device->vendor)); - memcpy(this_device->model, &inq_buff[16], - sizeof(this_device->model)); - memcpy(this_device->revision, &inq_buff[32], - sizeof(this_device->revision)); - memset(this_device->device_id, 0, - sizeof(this_device->device_id)); - cciss_scsi_get_device_id(h, scsi3addr, - this_device->device_id, sizeof(this_device->device_id)); - - switch (this_device->devtype) { - case 0x05: /* CD-ROM */ { - - /* We don't *really* support actual CD-ROM devices, - * just this "One Button Disaster Recovery" tape drive - * which temporarily pretends to be a CD-ROM drive. - * So we check that the device is really an OBDR tape - * device by checking for "$DR-10" in bytes 43-48 of - * the inquiry data. - */ - char obdr_sig[7]; - - strncpy(obdr_sig, &inq_buff[43], 6); - obdr_sig[6] = '\0'; - if (strncmp(obdr_sig, OBDR_TAPE_SIG, 6) != 0) - /* Not OBDR device, ignore it. */ - break; - } - /* fall through . . . */ - case 0x01: /* sequential access, (tape) */ - case 0x08: /* medium changer */ - if (ncurrent >= CCISS_MAX_SCSI_DEVS_PER_HBA) { - printk(KERN_INFO "cciss%d: %s ignored, " - "too many devices.\n", h->ctlr, - scsi_device_type(this_device->devtype)); - break; - } - currentsd[ncurrent] = *this_device; - ncurrent++; - break; - default: - break; - } - } - - adjust_cciss_scsi_table(h, hostno, currentsd, ncurrent); -out: - kfree(inq_buff); - kfree(ld_buff); - kfree(currentsd); - return; -} - -static int -is_keyword(char *ptr, int len, char *verb) // Thanks to ncr53c8xx.c -{ - int verb_len = strlen(verb); - if (len >= verb_len && !memcmp(verb,ptr,verb_len)) - return verb_len; - else - return 0; -} - -static int -cciss_scsi_user_command(ctlr_info_t *h, int hostno, char *buffer, int length) -{ - int arg_len; - - if ((arg_len = is_keyword(buffer, length, "rescan")) != 0) - cciss_update_non_disk_devices(h, hostno); - else - return -EINVAL; - return length; -} - -static int -cciss_scsi_write_info(struct Scsi_Host *sh, - char *buffer, /* data buffer */ - int length) /* length of data in buffer */ -{ - ctlr_info_t *h = (ctlr_info_t *) sh->hostdata[0]; - if (h == NULL) /* This really shouldn't ever happen. */ - return -EINVAL; - - return cciss_scsi_user_command(h, sh->host_no, - buffer, length); -} - -static int -cciss_scsi_show_info(struct seq_file *m, struct Scsi_Host *sh) -{ - - ctlr_info_t *h = (ctlr_info_t *) sh->hostdata[0]; - int i; - - if (h == NULL) /* This really shouldn't ever happen. */ - return -EINVAL; - - seq_printf(m, "cciss%d: SCSI host: %d\n", - h->ctlr, sh->host_no); - - /* this information is needed by apps to know which cciss - device corresponds to which scsi host number without - having to open a scsi target device node. The device - information is not a duplicate of /proc/scsi/scsi because - the two may be out of sync due to scsi hotplug, rather - this info is for an app to be able to use to know how to - get them back in sync. */ - - for (i = 0; i < ccissscsi[h->ctlr].ndevices; i++) { - struct cciss_scsi_dev_t *sd = - &ccissscsi[h->ctlr].dev[i]; - seq_printf(m, "c%db%dt%dl%d %02d " - "0x%02x%02x%02x%02x%02x%02x%02x%02x\n", - sh->host_no, sd->bus, sd->target, sd->lun, - sd->devtype, - sd->scsi3addr[0], sd->scsi3addr[1], - sd->scsi3addr[2], sd->scsi3addr[3], - sd->scsi3addr[4], sd->scsi3addr[5], - sd->scsi3addr[6], sd->scsi3addr[7]); - } - return 0; -} - -/* cciss_scatter_gather takes a struct scsi_cmnd, (cmd), and does the pci - dma mapping and fills in the scatter gather entries of the - cciss command, c. */ - -static void cciss_scatter_gather(ctlr_info_t *h, CommandList_struct *c, - struct scsi_cmnd *cmd) -{ - unsigned int len; - struct scatterlist *sg; - __u64 addr64; - int request_nsgs, i, chained, sg_index; - struct cciss_scsi_adapter_data_t *sa = h->scsi_ctlr; - SGDescriptor_struct *curr_sg; - - BUG_ON(scsi_sg_count(cmd) > h->maxsgentries); - - chained = 0; - sg_index = 0; - curr_sg = c->SG; - request_nsgs = scsi_dma_map(cmd); - if (request_nsgs) { - scsi_for_each_sg(cmd, sg, request_nsgs, i) { - if (sg_index + 1 == h->max_cmd_sgentries && - !chained && request_nsgs - i > 1) { - chained = 1; - sg_index = 0; - curr_sg = sa->cmd_sg_list[c->cmdindex]; - } - addr64 = (__u64) sg_dma_address(sg); - len = sg_dma_len(sg); - curr_sg[sg_index].Addr.lower = - (__u32) (addr64 & 0x0FFFFFFFFULL); - curr_sg[sg_index].Addr.upper = - (__u32) ((addr64 >> 32) & 0x0FFFFFFFFULL); - curr_sg[sg_index].Len = len; - curr_sg[sg_index].Ext = 0; - ++sg_index; - } - if (chained) - cciss_map_sg_chain_block(h, c, - sa->cmd_sg_list[c->cmdindex], - (request_nsgs - (h->max_cmd_sgentries - 1)) * - sizeof(SGDescriptor_struct)); - } - /* track how many SG entries we are using */ - if (request_nsgs > h->maxSG) - h->maxSG = request_nsgs; - c->Header.SGTotal = (u16) request_nsgs + chained; - if (request_nsgs > h->max_cmd_sgentries) - c->Header.SGList = h->max_cmd_sgentries; - else - c->Header.SGList = c->Header.SGTotal; - return; -} - - -static int -cciss_scsi_queue_command_lck(struct scsi_cmnd *cmd, void (*done)(struct scsi_cmnd *)) -{ - ctlr_info_t *h; - int rc; - unsigned char scsi3addr[8]; - CommandList_struct *c; - unsigned long flags; - - // Get the ptr to our adapter structure (hba[i]) out of cmd->host. - // We violate cmd->host privacy here. (Is there another way?) - h = (ctlr_info_t *) cmd->device->host->hostdata[0]; - - rc = lookup_scsi3addr(h, cmd->device->channel, cmd->device->id, - cmd->device->lun, scsi3addr); - if (rc != 0) { - /* the scsi nexus does not match any that we presented... */ - /* pretend to mid layer that we got selection timeout */ - cmd->result = DID_NO_CONNECT << 16; - done(cmd); - /* we might want to think about registering controller itself - as a processor device on the bus so sg binds to it. */ - return 0; - } - - /* Ok, we have a reasonable scsi nexus, so send the cmd down, and - see what the device thinks of it. */ - - spin_lock_irqsave(&h->lock, flags); - c = scsi_cmd_alloc(h); - spin_unlock_irqrestore(&h->lock, flags); - if (c == NULL) { /* trouble... */ - dev_warn(&h->pdev->dev, "scsi_cmd_alloc returned NULL!\n"); - /* FIXME: next 3 lines are -> BAD! <- */ - cmd->result = DID_NO_CONNECT << 16; - done(cmd); - return 0; - } - - // Fill in the command list header - - cmd->scsi_done = done; // save this for use by completion code - - /* save c in case we have to abort it */ - cmd->host_scribble = (unsigned char *) c; - - c->cmd_type = CMD_SCSI; - c->scsi_cmd = cmd; - c->Header.ReplyQueue = 0; /* unused in simple mode */ - memcpy(&c->Header.LUN.LunAddrBytes[0], &scsi3addr[0], 8); - c->Header.Tag.lower = c->busaddr; /* Use k. address of cmd as tag */ - - // Fill in the request block... - - c->Request.Timeout = 0; - memset(c->Request.CDB, 0, sizeof(c->Request.CDB)); - BUG_ON(cmd->cmd_len > sizeof(c->Request.CDB)); - c->Request.CDBLen = cmd->cmd_len; - memcpy(c->Request.CDB, cmd->cmnd, cmd->cmd_len); - c->Request.Type.Type = TYPE_CMD; - c->Request.Type.Attribute = ATTR_SIMPLE; - switch (cmd->sc_data_direction) { - case DMA_TO_DEVICE: - c->Request.Type.Direction = XFER_WRITE; - break; - case DMA_FROM_DEVICE: - c->Request.Type.Direction = XFER_READ; - break; - case DMA_NONE: - c->Request.Type.Direction = XFER_NONE; - break; - case DMA_BIDIRECTIONAL: - // This can happen if a buggy application does a scsi passthru - // and sets both inlen and outlen to non-zero. ( see - // ../scsi/scsi_ioctl.c:scsi_ioctl_send_command() ) - - c->Request.Type.Direction = XFER_RSVD; - // This is technically wrong, and cciss controllers should - // reject it with CMD_INVALID, which is the most correct - // response, but non-fibre backends appear to let it - // slide by, and give the same results as if this field - // were set correctly. Either way is acceptable for - // our purposes here. - - break; - - default: - dev_warn(&h->pdev->dev, "unknown data direction: %d\n", - cmd->sc_data_direction); - BUG(); - break; - } - cciss_scatter_gather(h, c, cmd); - enqueue_cmd_and_start_io(h, c); - /* the cmd'll come back via intr handler in complete_scsi_command() */ - return 0; -} - -static DEF_SCSI_QCMD(cciss_scsi_queue_command) - -static void cciss_unregister_scsi(ctlr_info_t *h) -{ - struct cciss_scsi_adapter_data_t *sa; - struct cciss_scsi_cmd_stack_t *stk; - unsigned long flags; - - /* we are being forcibly unloaded, and may not refuse. */ - - spin_lock_irqsave(&h->lock, flags); - sa = h->scsi_ctlr; - stk = &sa->cmd_stack; - - /* if we weren't ever actually registered, don't unregister */ - if (sa->registered) { - spin_unlock_irqrestore(&h->lock, flags); - scsi_remove_host(sa->scsi_host); - scsi_host_put(sa->scsi_host); - spin_lock_irqsave(&h->lock, flags); - } - - /* set scsi_host to NULL so our detect routine will - find us on register */ - sa->scsi_host = NULL; - spin_unlock_irqrestore(&h->lock, flags); - scsi_cmd_stack_free(h); - kfree(sa); -} - -static int cciss_engage_scsi(ctlr_info_t *h) -{ - struct cciss_scsi_adapter_data_t *sa; - struct cciss_scsi_cmd_stack_t *stk; - unsigned long flags; - - spin_lock_irqsave(&h->lock, flags); - sa = h->scsi_ctlr; - stk = &sa->cmd_stack; - - if (sa->registered) { - dev_info(&h->pdev->dev, "SCSI subsystem already engaged.\n"); - spin_unlock_irqrestore(&h->lock, flags); - return -ENXIO; - } - sa->registered = 1; - spin_unlock_irqrestore(&h->lock, flags); - cciss_update_non_disk_devices(h, -1); - cciss_scsi_detect(h); - return 0; -} - -static void -cciss_seq_tape_report(struct seq_file *seq, ctlr_info_t *h) -{ - unsigned long flags; - - CPQ_TAPE_LOCK(h, flags); - seq_printf(seq, - "Sequential access devices: %d\n\n", - ccissscsi[h->ctlr].ndevices); - CPQ_TAPE_UNLOCK(h, flags); -} - -static int wait_for_device_to_become_ready(ctlr_info_t *h, - unsigned char lunaddr[]) -{ - int rc; - int count = 0; - int waittime = HZ; - CommandList_struct *c; - - c = cmd_alloc(h); - if (!c) { - dev_warn(&h->pdev->dev, "out of memory in " - "wait_for_device_to_become_ready.\n"); - return IO_ERROR; - } - - /* Send test unit ready until device ready, or give up. */ - while (count < 20) { - - /* Wait for a bit. do this first, because if we send - * the TUR right away, the reset will just abort it. - */ - schedule_timeout_uninterruptible(waittime); - count++; - - /* Increase wait time with each try, up to a point. */ - if (waittime < (HZ * 30)) - waittime = waittime * 2; - - /* Send the Test Unit Ready */ - rc = fill_cmd(h, c, TEST_UNIT_READY, NULL, 0, 0, - lunaddr, TYPE_CMD); - if (rc == 0) - rc = sendcmd_withirq_core(h, c, 0); - - (void) process_sendcmd_error(h, c); - - if (rc != 0) - goto retry_tur; - - if (c->err_info->CommandStatus == CMD_SUCCESS) - break; - - if (c->err_info->CommandStatus == CMD_TARGET_STATUS && - c->err_info->ScsiStatus == SAM_STAT_CHECK_CONDITION) { - if (c->err_info->SenseInfo[2] == NO_SENSE) - break; - if (c->err_info->SenseInfo[2] == UNIT_ATTENTION) { - unsigned char asc; - asc = c->err_info->SenseInfo[12]; - check_for_unit_attention(h, c); - if (asc == POWER_OR_RESET) - break; - } - } -retry_tur: - dev_warn(&h->pdev->dev, "Waiting %d secs " - "for device to become ready.\n", - waittime / HZ); - rc = 1; /* device not ready. */ - } - - if (rc) - dev_warn(&h->pdev->dev, "giving up on device.\n"); - else - dev_warn(&h->pdev->dev, "device is ready.\n"); - - cmd_free(h, c); - return rc; -} - -/* Need at least one of these error handlers to keep ../scsi/hosts.c from - * complaining. Doing a host- or bus-reset can't do anything good here. - * Despite what it might say in scsi_error.c, there may well be commands - * on the controller, as the cciss driver registers twice, once as a block - * device for the logical drives, and once as a scsi device, for any tape - * drives. So we know there are no commands out on the tape drives, but we - * don't know there are no commands on the controller, and it is likely - * that there probably are, as the cciss block device is most commonly used - * as a boot device (embedded controller on HP/Compaq systems.) -*/ - -static int cciss_eh_device_reset_handler(struct scsi_cmnd *scsicmd) -{ - int rc; - CommandList_struct *cmd_in_trouble; - unsigned char lunaddr[8]; - ctlr_info_t *h; - - /* find the controller to which the command to be aborted was sent */ - h = (ctlr_info_t *) scsicmd->device->host->hostdata[0]; - if (h == NULL) /* paranoia */ - return FAILED; - dev_warn(&h->pdev->dev, "resetting tape drive or medium changer.\n"); - /* find the command that's giving us trouble */ - cmd_in_trouble = (CommandList_struct *) scsicmd->host_scribble; - if (cmd_in_trouble == NULL) /* paranoia */ - return FAILED; - memcpy(lunaddr, &cmd_in_trouble->Header.LUN.LunAddrBytes[0], 8); - /* send a reset to the SCSI LUN which the command was sent to */ - rc = sendcmd_withirq(h, CCISS_RESET_MSG, NULL, 0, 0, lunaddr, - TYPE_MSG); - if (rc == 0 && wait_for_device_to_become_ready(h, lunaddr) == 0) - return SUCCESS; - dev_warn(&h->pdev->dev, "resetting device failed.\n"); - return FAILED; -} - -static int cciss_eh_abort_handler(struct scsi_cmnd *scsicmd) -{ - int rc; - CommandList_struct *cmd_to_abort; - unsigned char lunaddr[8]; - ctlr_info_t *h; - - /* find the controller to which the command to be aborted was sent */ - h = (ctlr_info_t *) scsicmd->device->host->hostdata[0]; - if (h == NULL) /* paranoia */ - return FAILED; - dev_warn(&h->pdev->dev, "aborting tardy SCSI cmd\n"); - - /* find the command to be aborted */ - cmd_to_abort = (CommandList_struct *) scsicmd->host_scribble; - if (cmd_to_abort == NULL) /* paranoia */ - return FAILED; - memcpy(lunaddr, &cmd_to_abort->Header.LUN.LunAddrBytes[0], 8); - rc = sendcmd_withirq(h, CCISS_ABORT_MSG, &cmd_to_abort->Header.Tag, - 0, 0, lunaddr, TYPE_MSG); - if (rc == 0) - return SUCCESS; - return FAILED; - -} - -#else /* no CONFIG_CISS_SCSI_TAPE */ - -/* If no tape support, then these become defined out of existence */ - -#define cciss_scsi_setup(cntl_num) -#define cciss_engage_scsi(h) - -#endif /* CONFIG_CISS_SCSI_TAPE */ diff --git a/drivers/block/cciss_scsi.h b/drivers/block/cciss_scsi.h deleted file mode 100644 index e71d986727ca..000000000000 --- a/drivers/block/cciss_scsi.h +++ /dev/null @@ -1,79 +0,0 @@ -/* - * Disk Array driver for HP Smart Array controllers, SCSI Tape module. - * (C) Copyright 2001, 2007 Hewlett-Packard Development Company, L.P. - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; version 2 of the License. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 300, Boston, MA - * 02111-1307, USA. - * - * Questions/Comments/Bugfixes to iss_storagedev@hp.com - * - */ -#ifdef CONFIG_CISS_SCSI_TAPE -#ifndef _CCISS_SCSI_H_ -#define _CCISS_SCSI_H_ - -#include <scsi/scsicam.h> /* possibly irrelevant, since we don't show disks */ - - /* the scsi id of the adapter... */ -#define SELF_SCSI_ID 15 - /* 15 is somewhat arbitrary, since the scsi-2 bus - that's presented by the driver to the OS is - fabricated. The "real" scsi-3 bus the - hardware presents is fabricated too. - The actual, honest-to-goodness physical - bus that the devices are attached to is not - addressible natively, and may in fact turn - out to be not scsi at all. */ - - -/* - -If the upper scsi layer tries to track how many commands we have -outstanding, it will be operating under the misapprehension that it is -the only one sending us requests. We also have the block interface, -which is where most requests must surely come from, so the upper layer's -notion of how many requests we have outstanding will be wrong most or -all of the time. - -Note, the normal SCSI mid-layer error handling doesn't work well -for this driver because 1) it takes the io_request_lock before -calling error handlers and uses a local variable to store flags, -so the io_request_lock cannot be released and interrupts enabled -inside the error handlers, and, the error handlers cannot poll -for command completion because they might get commands from the -block half of the driver completing, and not know what to do -with them. That's what we get for making a hybrid scsi/block -driver, I suppose. - -*/ - -struct cciss_scsi_dev_t { - int devtype; - int bus, target, lun; /* as presented to the OS */ - unsigned char scsi3addr[8]; /* as presented to the HW */ - unsigned char device_id[16]; /* from inquiry pg. 0x83 */ - unsigned char vendor[8]; /* bytes 8-15 of inquiry data */ - unsigned char model[16]; /* bytes 16-31 of inquiry data */ - unsigned char revision[4]; /* bytes 32-35 of inquiry data */ -}; - -struct cciss_scsi_hba_t { - char *name; - int ndevices; -#define CCISS_MAX_SCSI_DEVS_PER_HBA 16 - struct cciss_scsi_dev_t dev[CCISS_MAX_SCSI_DEVS_PER_HBA]; -}; - -#endif /* _CCISS_SCSI_H_ */ -#endif /* CONFIG_CISS_SCSI_TAPE */ diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index e02c45cd3c5a..5f0eaee8c8a7 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -151,7 +151,7 @@ static int _drbd_md_sync_page_io(struct drbd_device *device, op_flags |= REQ_SYNC; bio = bio_alloc_drbd(GFP_NOIO); - bio->bi_bdev = bdev->md_bdev; + bio_set_dev(bio, bdev->md_bdev); bio->bi_iter.bi_sector = sector; err = -EIO; if (bio_add_page(bio, device->md_io.page, size, 0) != size) diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c index 809fd245c3dc..bd97908c766f 100644 --- a/drivers/block/drbd/drbd_bitmap.c +++ b/drivers/block/drbd/drbd_bitmap.c @@ -1019,7 +1019,7 @@ static void bm_page_io_async(struct drbd_bm_aio_ctx *ctx, int page_nr) __must_ho bm_store_page_idx(page, page_nr); } else page = b->bm_pages[page_nr]; - bio->bi_bdev = device->ldev->md_bdev; + bio_set_dev(bio, device->ldev->md_bdev); bio->bi_iter.bi_sector = on_disk_sector; /* bio_add_page of a single page to an empty bio will always succeed, * according to api. Do we want to assert that? */ diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index d17b6e6393c7..7e8589ce631c 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -63,19 +63,15 @@ # define __must_hold(x) #endif -/* module parameter, defined in drbd_main.c */ -extern unsigned int minor_count; -extern bool disable_sendpage; -extern bool allow_oos; -void tl_abort_disk_io(struct drbd_device *device); - +/* shared module parameters, defined in drbd_main.c */ #ifdef CONFIG_DRBD_FAULT_INJECTION -extern int enable_faults; -extern int fault_rate; -extern int fault_devs; +extern int drbd_enable_faults; +extern int drbd_fault_rate; #endif -extern char usermode_helper[]; +extern unsigned int drbd_minor_count; +extern char drbd_usermode_helper[]; +extern int drbd_proc_details; /* This is used to stop/restart our threads. @@ -181,8 +177,8 @@ _drbd_insert_fault(struct drbd_device *device, unsigned int type); static inline int drbd_insert_fault(struct drbd_device *device, unsigned int type) { #ifdef CONFIG_DRBD_FAULT_INJECTION - return fault_rate && - (enable_faults & (1<<type)) && + return drbd_fault_rate && + (drbd_enable_faults & (1<<type)) && _drbd_insert_fault(device, type); #else return 0; @@ -745,6 +741,8 @@ struct drbd_connection { unsigned current_tle_writes; /* writes seen within this tl epoch */ unsigned long last_reconnect_jif; + /* empty member on older kernels without blk_start_plug() */ + struct blk_plug receiver_plug; struct drbd_thread receiver; struct drbd_thread worker; struct drbd_thread ack_receiver; @@ -1131,7 +1129,8 @@ extern void conn_send_sr_reply(struct drbd_connection *connection, enum drbd_sta extern int drbd_send_rs_deallocated(struct drbd_peer_device *, struct drbd_peer_request *); extern void drbd_backing_dev_free(struct drbd_device *device, struct drbd_backing_dev *ldev); extern void drbd_device_cleanup(struct drbd_device *device); -void drbd_print_uuids(struct drbd_device *device, const char *text); +extern void drbd_print_uuids(struct drbd_device *device, const char *text); +extern void drbd_queue_unplug(struct drbd_device *device); extern void conn_md_sync(struct drbd_connection *connection); extern void drbd_md_write(struct drbd_device *device, void *buffer); @@ -1463,8 +1462,6 @@ extern struct drbd_resource *drbd_find_resource(const char *name); extern void drbd_destroy_resource(struct kref *kref); extern void conn_free_crypto(struct drbd_connection *connection); -extern int proc_details; - /* drbd_req */ extern void do_submit(struct work_struct *ws); extern void __drbd_make_request(struct drbd_device *, struct bio *, unsigned long); @@ -1628,8 +1625,8 @@ static inline void drbd_generic_make_request(struct drbd_device *device, int fault_type, struct bio *bio) { __release(local); - if (!bio->bi_bdev) { - drbd_err(device, "drbd_generic_make_request: bio->bi_bdev == NULL\n"); + if (!bio->bi_disk) { + drbd_err(device, "drbd_generic_make_request: bio->bi_disk == NULL\n"); bio->bi_status = BLK_STS_IOERR; bio_endio(bio); return; diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index e2ed28d45ce1..8cb3791898ae 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -77,41 +77,41 @@ MODULE_PARM_DESC(minor_count, "Approximate number of drbd devices (" MODULE_ALIAS_BLOCKDEV_MAJOR(DRBD_MAJOR); #include <linux/moduleparam.h> -/* allow_open_on_secondary */ -MODULE_PARM_DESC(allow_oos, "DONT USE!"); /* thanks to these macros, if compiled into the kernel (not-module), - * this becomes the boot parameter drbd.minor_count */ -module_param(minor_count, uint, 0444); -module_param(disable_sendpage, bool, 0644); -module_param(allow_oos, bool, 0); -module_param(proc_details, int, 0644); + * these become boot parameters (e.g., drbd.minor_count) */ #ifdef CONFIG_DRBD_FAULT_INJECTION -int enable_faults; -int fault_rate; -static int fault_count; -int fault_devs; +int drbd_enable_faults; +int drbd_fault_rate; +static int drbd_fault_count; +static int drbd_fault_devs; /* bitmap of enabled faults */ -module_param(enable_faults, int, 0664); +module_param_named(enable_faults, drbd_enable_faults, int, 0664); /* fault rate % value - applies to all enabled faults */ -module_param(fault_rate, int, 0664); +module_param_named(fault_rate, drbd_fault_rate, int, 0664); /* count of faults inserted */ -module_param(fault_count, int, 0664); +module_param_named(fault_count, drbd_fault_count, int, 0664); /* bitmap of devices to insert faults on */ -module_param(fault_devs, int, 0644); +module_param_named(fault_devs, drbd_fault_devs, int, 0644); #endif -/* module parameter, defined */ -unsigned int minor_count = DRBD_MINOR_COUNT_DEF; -bool disable_sendpage; -bool allow_oos; -int proc_details; /* Detail level in proc drbd*/ - +/* module parameters we can keep static */ +static bool drbd_allow_oos; /* allow_open_on_secondary */ +static bool drbd_disable_sendpage; +MODULE_PARM_DESC(allow_oos, "DONT USE!"); +module_param_named(allow_oos, drbd_allow_oos, bool, 0); +module_param_named(disable_sendpage, drbd_disable_sendpage, bool, 0644); + +/* module parameters we share */ +int drbd_proc_details; /* Detail level in proc drbd*/ +module_param_named(proc_details, drbd_proc_details, int, 0644); +/* module parameters shared with defaults */ +unsigned int drbd_minor_count = DRBD_MINOR_COUNT_DEF; /* Module parameter for setting the user mode helper program * to run. Default is /sbin/drbdadm */ -char usermode_helper[80] = "/sbin/drbdadm"; - -module_param_string(usermode_helper, usermode_helper, sizeof(usermode_helper), 0644); +char drbd_usermode_helper[80] = "/sbin/drbdadm"; +module_param_named(minor_count, drbd_minor_count, uint, 0444); +module_param_string(usermode_helper, drbd_usermode_helper, sizeof(drbd_usermode_helper), 0644); /* in 2.6.x, our device mapping and config info contains our virtual gendisks * as member "struct gendisk *vdisk;" @@ -923,7 +923,9 @@ void drbd_gen_and_send_sync_uuid(struct drbd_peer_device *peer_device) } /* communicated if (agreed_features & DRBD_FF_WSAME) */ -void assign_p_sizes_qlim(struct drbd_device *device, struct p_sizes *p, struct request_queue *q) +static void +assign_p_sizes_qlim(struct drbd_device *device, struct p_sizes *p, + struct request_queue *q) { if (q) { p->qlim->physical_block_size = cpu_to_be32(queue_physical_block_size(q)); @@ -1560,7 +1562,7 @@ static int _drbd_send_page(struct drbd_peer_device *peer_device, struct page *pa * put_page(); and would cause either a VM_BUG directly, or * __page_cache_release a page that would actually still be referenced * by someone, leading to some obscure delayed Oops somewhere else. */ - if (disable_sendpage || (page_count(page) < 1) || PageSlab(page)) + if (drbd_disable_sendpage || (page_count(page) < 1) || PageSlab(page)) return _drbd_no_send_page(peer_device, page, offset, size, msg_flags); msg_flags |= MSG_NOSIGNAL; @@ -1932,7 +1934,7 @@ static int drbd_open(struct block_device *bdev, fmode_t mode) if (device->state.role != R_PRIMARY) { if (mode & FMODE_WRITE) rv = -EROFS; - else if (!allow_oos) + else if (!drbd_allow_oos) rv = -EMEDIUMTYPE; } @@ -1952,6 +1954,19 @@ static void drbd_release(struct gendisk *gd, fmode_t mode) mutex_unlock(&drbd_main_mutex); } +/* need to hold resource->req_lock */ +void drbd_queue_unplug(struct drbd_device *device) +{ + if (device->state.pdsk >= D_INCONSISTENT && device->state.conn >= C_CONNECTED) { + D_ASSERT(device, device->state.role == R_PRIMARY); + if (test_and_clear_bit(UNPLUG_REMOTE, &device->flags)) { + drbd_queue_work_if_unqueued( + &first_peer_device(device)->connection->sender_work, + &device->unplug_work); + } + } +} + static void drbd_set_defaults(struct drbd_device *device) { /* Beware! The actual layout differs @@ -2008,18 +2023,14 @@ void drbd_init_set_defaults(struct drbd_device *device) device->unplug_work.cb = w_send_write_hint; device->bm_io_work.w.cb = w_bitmap_io; - init_timer(&device->resync_timer); - init_timer(&device->md_sync_timer); - init_timer(&device->start_resync_timer); - init_timer(&device->request_timer); - device->resync_timer.function = resync_timer_fn; - device->resync_timer.data = (unsigned long) device; - device->md_sync_timer.function = md_sync_timer_fn; - device->md_sync_timer.data = (unsigned long) device; - device->start_resync_timer.function = start_resync_timer_fn; - device->start_resync_timer.data = (unsigned long) device; - device->request_timer.function = request_timer_fn; - device->request_timer.data = (unsigned long) device; + setup_timer(&device->resync_timer, resync_timer_fn, + (unsigned long)device); + setup_timer(&device->md_sync_timer, md_sync_timer_fn, + (unsigned long)device); + setup_timer(&device->start_resync_timer, start_resync_timer_fn, + (unsigned long)device); + setup_timer(&device->request_timer, request_timer_fn, + (unsigned long)device); init_waitqueue_head(&device->misc_wait); init_waitqueue_head(&device->state_wait); @@ -2131,7 +2142,7 @@ static void drbd_destroy_mempools(void) static int drbd_create_mempools(void) { struct page *page; - const int number = (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count; + const int number = (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * drbd_minor_count; int i; /* prepare our caches and mempools */ @@ -2167,13 +2178,12 @@ static int drbd_create_mempools(void) goto Enomem; /* mempools */ - drbd_io_bio_set = bioset_create(BIO_POOL_SIZE, 0, BIOSET_NEED_RESCUER); + drbd_io_bio_set = bioset_create(BIO_POOL_SIZE, 0, 0); if (drbd_io_bio_set == NULL) goto Enomem; drbd_md_io_bio_set = bioset_create(DRBD_MIN_POOL_PAGES, 0, - BIOSET_NEED_BVECS | - BIOSET_NEED_RESCUER); + BIOSET_NEED_BVECS); if (drbd_md_io_bio_set == NULL) goto Enomem; @@ -2409,7 +2419,6 @@ static void drbd_cleanup(void) destroy_workqueue(retry.wq); drbd_genl_unregister(); - drbd_debugfs_cleanup(); idr_for_each_entry(&drbd_devices, device, i) drbd_delete_device(device); @@ -2420,6 +2429,8 @@ static void drbd_cleanup(void) drbd_free_resource(resource); } + drbd_debugfs_cleanup(); + drbd_destroy_mempools(); unregister_blkdev(DRBD_MAJOR, "drbd"); @@ -2972,12 +2983,12 @@ static int __init drbd_init(void) { int err; - if (minor_count < DRBD_MINOR_COUNT_MIN || minor_count > DRBD_MINOR_COUNT_MAX) { - pr_err("invalid minor_count (%d)\n", minor_count); + if (drbd_minor_count < DRBD_MINOR_COUNT_MIN || drbd_minor_count > DRBD_MINOR_COUNT_MAX) { + pr_err("invalid minor_count (%d)\n", drbd_minor_count); #ifdef MODULE return -EINVAL; #else - minor_count = DRBD_MINOR_COUNT_DEF; + drbd_minor_count = DRBD_MINOR_COUNT_DEF; #endif } @@ -3900,12 +3911,12 @@ _drbd_insert_fault(struct drbd_device *device, unsigned int type) static struct fault_random_state rrs = {0, 0}; unsigned int ret = ( - (fault_devs == 0 || - ((1 << device_to_minor(device)) & fault_devs) != 0) && - (((_drbd_fault_random(&rrs) % 100) + 1) <= fault_rate)); + (drbd_fault_devs == 0 || + ((1 << device_to_minor(device)) & drbd_fault_devs) != 0) && + (((_drbd_fault_random(&rrs) % 100) + 1) <= drbd_fault_rate)); if (ret) { - fault_count++; + drbd_fault_count++; if (__ratelimit(&drbd_ratelimit_state)) drbd_warn(device, "***Simulating %s failure\n", diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index ad0fcb43e45c..a12f77e6891e 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -344,7 +344,7 @@ int drbd_khelper(struct drbd_device *device, char *cmd) (char[60]) { }, /* address */ NULL }; char mb[14]; - char *argv[] = {usermode_helper, cmd, mb, NULL }; + char *argv[] = {drbd_usermode_helper, cmd, mb, NULL }; struct drbd_connection *connection = first_peer_device(device)->connection; struct sib_info sib; int ret; @@ -359,19 +359,19 @@ int drbd_khelper(struct drbd_device *device, char *cmd) * write out any unsynced meta data changes now */ drbd_md_sync(device); - drbd_info(device, "helper command: %s %s %s\n", usermode_helper, cmd, mb); + drbd_info(device, "helper command: %s %s %s\n", drbd_usermode_helper, cmd, mb); sib.sib_reason = SIB_HELPER_PRE; sib.helper_name = cmd; drbd_bcast_event(device, &sib); notify_helper(NOTIFY_CALL, device, connection, cmd, 0); - ret = call_usermodehelper(usermode_helper, argv, envp, UMH_WAIT_PROC); + ret = call_usermodehelper(drbd_usermode_helper, argv, envp, UMH_WAIT_PROC); if (ret) drbd_warn(device, "helper command: %s %s %s exit code %u (0x%x)\n", - usermode_helper, cmd, mb, + drbd_usermode_helper, cmd, mb, (ret >> 8) & 0xff, ret); else drbd_info(device, "helper command: %s %s %s exit code %u (0x%x)\n", - usermode_helper, cmd, mb, + drbd_usermode_helper, cmd, mb, (ret >> 8) & 0xff, ret); sib.sib_reason = SIB_HELPER_POST; sib.helper_exit_code = ret; @@ -396,24 +396,24 @@ enum drbd_peer_state conn_khelper(struct drbd_connection *connection, char *cmd) (char[60]) { }, /* address */ NULL }; char *resource_name = connection->resource->name; - char *argv[] = {usermode_helper, cmd, resource_name, NULL }; + char *argv[] = {drbd_usermode_helper, cmd, resource_name, NULL }; int ret; setup_khelper_env(connection, envp); conn_md_sync(connection); - drbd_info(connection, "helper command: %s %s %s\n", usermode_helper, cmd, resource_name); + drbd_info(connection, "helper command: %s %s %s\n", drbd_usermode_helper, cmd, resource_name); /* TODO: conn_bcast_event() ?? */ notify_helper(NOTIFY_CALL, NULL, connection, cmd, 0); - ret = call_usermodehelper(usermode_helper, argv, envp, UMH_WAIT_PROC); + ret = call_usermodehelper(drbd_usermode_helper, argv, envp, UMH_WAIT_PROC); if (ret) drbd_warn(connection, "helper command: %s %s %s exit code %u (0x%x)\n", - usermode_helper, cmd, resource_name, + drbd_usermode_helper, cmd, resource_name, (ret >> 8) & 0xff, ret); else drbd_info(connection, "helper command: %s %s %s exit code %u (0x%x)\n", - usermode_helper, cmd, resource_name, + drbd_usermode_helper, cmd, resource_name, (ret >> 8) & 0xff, ret); /* TODO: conn_bcast_event() ?? */ notify_helper(NOTIFY_RESPONSE, NULL, connection, cmd, ret); @@ -1236,12 +1236,18 @@ static void fixup_discard_if_not_supported(struct request_queue *q) static void decide_on_write_same_support(struct drbd_device *device, struct request_queue *q, - struct request_queue *b, struct o_qlim *o) + struct request_queue *b, struct o_qlim *o, + bool disable_write_same) { struct drbd_peer_device *peer_device = first_peer_device(device); struct drbd_connection *connection = peer_device->connection; bool can_do = b ? b->limits.max_write_same_sectors : true; + if (can_do && disable_write_same) { + can_do = false; + drbd_info(peer_device, "WRITE_SAME disabled by config\n"); + } + if (can_do && connection->cstate >= C_CONNECTED && !(connection->agreed_features & DRBD_FF_WSAME)) { can_do = false; drbd_info(peer_device, "peer does not support WRITE_SAME\n"); @@ -1302,6 +1308,7 @@ static void drbd_setup_queue_param(struct drbd_device *device, struct drbd_backi struct request_queue *b = NULL; struct disk_conf *dc; bool discard_zeroes_if_aligned = true; + bool disable_write_same = false; if (bdev) { b = bdev->backing_bdev->bd_disk->queue; @@ -1311,6 +1318,7 @@ static void drbd_setup_queue_param(struct drbd_device *device, struct drbd_backi dc = rcu_dereference(device->ldev->disk_conf); max_segments = dc->max_bio_bvecs; discard_zeroes_if_aligned = dc->discard_zeroes_if_aligned; + disable_write_same = dc->disable_write_same; rcu_read_unlock(); blk_set_stacking_limits(&q->limits); @@ -1321,7 +1329,7 @@ static void drbd_setup_queue_param(struct drbd_device *device, struct drbd_backi blk_queue_max_segments(q, max_segments ? max_segments : BLK_MAX_SEGMENTS); blk_queue_segment_boundary(q, PAGE_SIZE-1); decide_on_discard_support(device, q, b, discard_zeroes_if_aligned); - decide_on_write_same_support(device, q, b, o); + decide_on_write_same_support(device, q, b, o, disable_write_same); if (b) { blk_queue_stack_limits(q, b); @@ -1612,7 +1620,8 @@ int drbd_adm_disk_opts(struct sk_buff *skb, struct genl_info *info) if (write_ordering_changed(old_disk_conf, new_disk_conf)) drbd_bump_write_ordering(device->resource, NULL, WO_BDEV_FLUSH); - if (old_disk_conf->discard_zeroes_if_aligned != new_disk_conf->discard_zeroes_if_aligned) + if (old_disk_conf->discard_zeroes_if_aligned != new_disk_conf->discard_zeroes_if_aligned + || old_disk_conf->disable_write_same != new_disk_conf->disable_write_same) drbd_reconsider_queue_parameters(device, device->ldev, NULL); drbd_md_sync(device); @@ -2140,34 +2149,13 @@ int drbd_adm_attach(struct sk_buff *skb, struct genl_info *info) static int adm_detach(struct drbd_device *device, int force) { - enum drbd_state_rv retcode; - void *buffer; - int ret; - if (force) { set_bit(FORCE_DETACH, &device->flags); drbd_force_state(device, NS(disk, D_FAILED)); - retcode = SS_SUCCESS; - goto out; + return SS_SUCCESS; } - drbd_suspend_io(device); /* so no-one is stuck in drbd_al_begin_io */ - buffer = drbd_md_get_buffer(device, __func__); /* make sure there is no in-flight meta-data IO */ - if (buffer) { - retcode = drbd_request_state(device, NS(disk, D_FAILED)); - drbd_md_put_buffer(device); - } else /* already <= D_FAILED */ - retcode = SS_NOTHING_TO_DO; - /* D_FAILED will transition to DISKLESS. */ - drbd_resume_io(device); - ret = wait_event_interruptible(device->misc_wait, - device->state.disk != D_FAILED); - if ((int)retcode == (int)SS_IS_DISKLESS) - retcode = SS_NOTHING_TO_DO; - if (ret) - retcode = ERR_INTR; -out: - return retcode; + return drbd_request_detach_interruptible(device); } /* Detaching the disk is a process in multiple stages. First we need to lock diff --git a/drivers/block/drbd/drbd_proc.c b/drivers/block/drbd/drbd_proc.c index 8378142f7a55..582caeb0de86 100644 --- a/drivers/block/drbd/drbd_proc.c +++ b/drivers/block/drbd/drbd_proc.c @@ -127,7 +127,7 @@ static void drbd_syncer_progress(struct drbd_device *device, struct seq_file *se seq_putc(seq, '='); seq_putc(seq, '>'); for (i = 0; i < y; i++) - seq_printf(seq, "."); + seq_putc(seq, '.'); seq_puts(seq, "] "); if (state.conn == C_VERIFY_S || state.conn == C_VERIFY_T) @@ -179,7 +179,7 @@ static void drbd_syncer_progress(struct drbd_device *device, struct seq_file *se seq_printf_with_thousands_grouping(seq, dbdt); seq_puts(seq, " ("); /* ------------------------- ~3s average ------------------------ */ - if (proc_details >= 1) { + if (drbd_proc_details >= 1) { /* this is what drbd_rs_should_slow_down() uses */ i = (device->rs_last_mark + DRBD_SYNC_MARKS-1) % DRBD_SYNC_MARKS; dt = (jiffies - device->rs_mark_time[i]) / HZ; @@ -209,7 +209,7 @@ static void drbd_syncer_progress(struct drbd_device *device, struct seq_file *se } seq_printf(seq, " K/sec%s\n", stalled ? " (stalled)" : ""); - if (proc_details >= 1) { + if (drbd_proc_details >= 1) { /* 64 bit: * we convert to sectors in the display below. */ unsigned long bm_bits = drbd_bm_bits(device); @@ -332,13 +332,13 @@ static int drbd_seq_show(struct seq_file *seq, void *v) state.conn == C_VERIFY_T) drbd_syncer_progress(device, seq, state); - if (proc_details >= 1 && get_ldev_if_state(device, D_FAILED)) { + if (drbd_proc_details >= 1 && get_ldev_if_state(device, D_FAILED)) { lc_seq_printf_stats(seq, device->resync); lc_seq_printf_stats(seq, device->act_log); put_ldev(device); } - if (proc_details >= 2) + if (drbd_proc_details >= 2) seq_printf(seq, "\tblocked on activity log: %d\n", atomic_read(&device->ap_actlog_cnt)); } rcu_read_unlock(); diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index c7e95e6380fb..796eaf347dc0 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -332,7 +332,7 @@ static void drbd_free_pages(struct drbd_device *device, struct page *page, int i if (page == NULL) return; - if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count) + if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * drbd_minor_count) i = page_chain_free(page); else { struct page *tmp; @@ -1100,7 +1100,10 @@ randomize: idr_for_each_entry(&connection->peer_devices, peer_device, vnr) mutex_lock(peer_device->device->state_mutex); + /* avoid a race with conn_request_state( C_DISCONNECTING ) */ + spin_lock_irq(&connection->resource->req_lock); set_bit(STATE_SENT, &connection->flags); + spin_unlock_irq(&connection->resource->req_lock); idr_for_each_entry(&connection->peer_devices, peer_device, vnr) mutex_unlock(peer_device->device->state_mutex); @@ -1194,6 +1197,14 @@ static int decode_header(struct drbd_connection *connection, void *header, struc return 0; } +static void drbd_unplug_all_devices(struct drbd_connection *connection) +{ + if (current->plug == &connection->receiver_plug) { + blk_finish_plug(&connection->receiver_plug); + blk_start_plug(&connection->receiver_plug); + } /* else: maybe just schedule() ?? */ +} + static int drbd_recv_header(struct drbd_connection *connection, struct packet_info *pi) { void *buffer = connection->data.rbuf; @@ -1209,6 +1220,36 @@ static int drbd_recv_header(struct drbd_connection *connection, struct packet_in return err; } +static int drbd_recv_header_maybe_unplug(struct drbd_connection *connection, struct packet_info *pi) +{ + void *buffer = connection->data.rbuf; + unsigned int size = drbd_header_size(connection); + int err; + + err = drbd_recv_short(connection->data.socket, buffer, size, MSG_NOSIGNAL|MSG_DONTWAIT); + if (err != size) { + /* If we have nothing in the receive buffer now, to reduce + * application latency, try to drain the backend queues as + * quickly as possible, and let remote TCP know what we have + * received so far. */ + if (err == -EAGAIN) { + drbd_tcp_quickack(connection->data.socket); + drbd_unplug_all_devices(connection); + } + if (err > 0) { + buffer += err; + size -= err; + } + err = drbd_recv_all_warn(connection, buffer, size); + if (err) + return err; + } + + err = decode_header(connection, connection->data.rbuf, pi); + connection->last_received = jiffies; + + return err; +} /* This is blkdev_issue_flush, but asynchronous. * We want to submit to all component volumes in parallel, * then wait for all completions. @@ -1223,7 +1264,7 @@ struct one_flush_context { struct issue_flush_context *ctx; }; -void one_flush_endio(struct bio *bio) +static void one_flush_endio(struct bio *bio) { struct one_flush_context *octx = bio->bi_private; struct drbd_device *device = octx->device; @@ -1265,7 +1306,7 @@ static void submit_one_flush(struct drbd_device *device, struct issue_flush_cont octx->device = device; octx->ctx = ctx; - bio->bi_bdev = device->ldev->backing_bdev; + bio_set_dev(bio, device->ldev->backing_bdev); bio->bi_private = octx; bio->bi_end_io = one_flush_endio; bio->bi_opf = REQ_OP_FLUSH | REQ_PREFLUSH; @@ -1548,7 +1589,7 @@ next_bio: } /* > peer_req->i.sector, unless this is the first bio */ bio->bi_iter.bi_sector = sector; - bio->bi_bdev = device->ldev->backing_bdev; + bio_set_dev(bio, device->ldev->backing_bdev); bio_set_op_attrs(bio, op, op_flags); bio->bi_private = peer_req; bio->bi_end_io = drbd_peer_request_endio; @@ -4085,7 +4126,7 @@ static int receive_uuids(struct drbd_connection *connection, struct packet_info return config_unknown_volume(connection, pi); device = peer_device->device; - p_uuid = kmalloc(sizeof(u64)*UI_EXTENDED_SIZE, GFP_NOIO); + p_uuid = kmalloc_array(UI_EXTENDED_SIZE, sizeof(*p_uuid), GFP_NOIO); if (!p_uuid) { drbd_err(device, "kmalloc of p_uuid failed\n"); return false; @@ -4882,8 +4923,8 @@ static void drbdd(struct drbd_connection *connection) struct data_cmd const *cmd; drbd_thread_current_set_cpu(&connection->receiver); - update_receiver_timing_details(connection, drbd_recv_header); - if (drbd_recv_header(connection, &pi)) + update_receiver_timing_details(connection, drbd_recv_header_maybe_unplug); + if (drbd_recv_header_maybe_unplug(connection, &pi)) goto err_out; cmd = &drbd_cmd_handler[pi.cmd]; @@ -5375,8 +5416,11 @@ int drbd_receiver(struct drbd_thread *thi) } } while (h == 0); - if (h > 0) + if (h > 0) { + blk_start_plug(&connection->receiver_plug); drbdd(connection); + blk_finish_plug(&connection->receiver_plug); + } conn_disconnect(connection); diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index f6e865b2d543..de8566e55334 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -36,14 +36,18 @@ static bool drbd_may_do_local_read(struct drbd_device *device, sector_t sector, /* Update disk stats at start of I/O request */ static void _drbd_start_io_acct(struct drbd_device *device, struct drbd_request *req) { - generic_start_io_acct(bio_data_dir(req->master_bio), req->i.size >> 9, - &device->vdisk->part0); + struct request_queue *q = device->rq_queue; + + generic_start_io_acct(q, bio_data_dir(req->master_bio), + req->i.size >> 9, &device->vdisk->part0); } /* Update disk stats when completing request upwards */ static void _drbd_end_io_acct(struct drbd_device *device, struct drbd_request *req) { - generic_end_io_acct(bio_data_dir(req->master_bio), + struct request_queue *q = device->rq_queue; + + generic_end_io_acct(q, bio_data_dir(req->master_bio), &device->vdisk->part0, req->start_jif); } @@ -1175,7 +1179,7 @@ drbd_submit_req_private_bio(struct drbd_request *req) else type = DRBD_FAULT_DT_RD; - bio->bi_bdev = device->ldev->backing_bdev; + bio_set_dev(bio, device->ldev->backing_bdev); /* State may have changed since we grabbed our reference on the * ->ldev member. Double check, and short-circuit to endio. @@ -1275,6 +1279,57 @@ static bool may_do_writes(struct drbd_device *device) return s.disk == D_UP_TO_DATE || s.pdsk == D_UP_TO_DATE; } +struct drbd_plug_cb { + struct blk_plug_cb cb; + struct drbd_request *most_recent_req; + /* do we need more? */ +}; + +static void drbd_unplug(struct blk_plug_cb *cb, bool from_schedule) +{ + struct drbd_plug_cb *plug = container_of(cb, struct drbd_plug_cb, cb); + struct drbd_resource *resource = plug->cb.data; + struct drbd_request *req = plug->most_recent_req; + + kfree(cb); + if (!req) + return; + + spin_lock_irq(&resource->req_lock); + /* In case the sender did not process it yet, raise the flag to + * have it followed with P_UNPLUG_REMOTE just after. */ + req->rq_state |= RQ_UNPLUG; + /* but also queue a generic unplug */ + drbd_queue_unplug(req->device); + kref_put(&req->kref, drbd_req_destroy); + spin_unlock_irq(&resource->req_lock); +} + +static struct drbd_plug_cb* drbd_check_plugged(struct drbd_resource *resource) +{ + /* A lot of text to say + * return (struct drbd_plug_cb*)blk_check_plugged(); */ + struct drbd_plug_cb *plug; + struct blk_plug_cb *cb = blk_check_plugged(drbd_unplug, resource, sizeof(*plug)); + + if (cb) + plug = container_of(cb, struct drbd_plug_cb, cb); + else + plug = NULL; + return plug; +} + +static void drbd_update_plug(struct drbd_plug_cb *plug, struct drbd_request *req) +{ + struct drbd_request *tmp = plug->most_recent_req; + /* Will be sent to some peer. + * Remember to tag it with UNPLUG_REMOTE on unplug */ + kref_get(&req->kref); + plug->most_recent_req = req; + if (tmp) + kref_put(&tmp->kref, drbd_req_destroy); +} + static void drbd_send_and_submit(struct drbd_device *device, struct drbd_request *req) { struct drbd_resource *resource = device->resource; @@ -1347,6 +1402,12 @@ static void drbd_send_and_submit(struct drbd_device *device, struct drbd_request no_remote = true; } + if (no_remote == false) { + struct drbd_plug_cb *plug = drbd_check_plugged(resource); + if (plug) + drbd_update_plug(plug, req); + } + /* If it took the fast path in drbd_request_prepare, add it here. * The slow path has added it already. */ if (list_empty(&req->req_pending_master_completion)) @@ -1395,7 +1456,10 @@ void __drbd_make_request(struct drbd_device *device, struct bio *bio, unsigned l static void submit_fast_path(struct drbd_device *device, struct list_head *incoming) { + struct blk_plug plug; struct drbd_request *req, *tmp; + + blk_start_plug(&plug); list_for_each_entry_safe(req, tmp, incoming, tl_requests) { const int rw = bio_data_dir(req->master_bio); @@ -1413,6 +1477,7 @@ static void submit_fast_path(struct drbd_device *device, struct list_head *incom list_del_init(&req->tl_requests); drbd_send_and_submit(device, req); } + blk_finish_plug(&plug); } static bool prepare_al_transaction_nonblock(struct drbd_device *device, @@ -1420,12 +1485,12 @@ static bool prepare_al_transaction_nonblock(struct drbd_device *device, struct list_head *pending, struct list_head *later) { - struct drbd_request *req, *tmp; + struct drbd_request *req; int wake = 0; int err; spin_lock_irq(&device->al_lock); - list_for_each_entry_safe(req, tmp, incoming, tl_requests) { + while ((req = list_first_entry_or_null(incoming, struct drbd_request, tl_requests))) { err = drbd_al_begin_io_nonblock(device, &req->i); if (err == -ENOBUFS) break; @@ -1442,17 +1507,20 @@ static bool prepare_al_transaction_nonblock(struct drbd_device *device, return !list_empty(pending); } -void send_and_submit_pending(struct drbd_device *device, struct list_head *pending) +static void send_and_submit_pending(struct drbd_device *device, struct list_head *pending) { - struct drbd_request *req, *tmp; + struct blk_plug plug; + struct drbd_request *req; - list_for_each_entry_safe(req, tmp, pending, tl_requests) { + blk_start_plug(&plug); + while ((req = list_first_entry_or_null(pending, struct drbd_request, tl_requests))) { req->rq_state |= RQ_IN_ACT_LOG; req->in_actlog_jif = jiffies; atomic_dec(&device->ap_actlog_cnt); list_del_init(&req->tl_requests); drbd_send_and_submit(device, req); } + blk_finish_plug(&plug); } void do_submit(struct work_struct *ws) diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h index 9e1866ab238f..a2254f825601 100644 --- a/drivers/block/drbd/drbd_req.h +++ b/drivers/block/drbd/drbd_req.h @@ -212,6 +212,11 @@ enum drbd_req_state_bits { /* Should call drbd_al_complete_io() for this request... */ __RQ_IN_ACT_LOG, + /* This was the most recent request during some blk_finish_plug() + * or its implicit from-schedule equivalent. + * We may use it as hint to send a P_UNPLUG_REMOTE */ + __RQ_UNPLUG, + /* The peer has sent a retry ACK */ __RQ_POSTPONED, @@ -249,6 +254,7 @@ enum drbd_req_state_bits { #define RQ_WSAME (1UL << __RQ_WSAME) #define RQ_UNMAP (1UL << __RQ_UNMAP) #define RQ_IN_ACT_LOG (1UL << __RQ_IN_ACT_LOG) +#define RQ_UNPLUG (1UL << __RQ_UNPLUG) #define RQ_POSTPONED (1UL << __RQ_POSTPONED) #define RQ_COMPLETION_SUSP (1UL << __RQ_COMPLETION_SUSP) #define RQ_EXP_RECEIVE_ACK (1UL << __RQ_EXP_RECEIVE_ACK) diff --git a/drivers/block/drbd/drbd_state.c b/drivers/block/drbd/drbd_state.c index eea0c4aec978..0813c654c893 100644 --- a/drivers/block/drbd/drbd_state.c +++ b/drivers/block/drbd/drbd_state.c @@ -346,7 +346,7 @@ static enum drbd_role min_role(enum drbd_role role1, enum drbd_role role2) enum drbd_role conn_highest_role(struct drbd_connection *connection) { - enum drbd_role role = R_UNKNOWN; + enum drbd_role role = R_SECONDARY; struct drbd_peer_device *peer_device; int vnr; @@ -579,11 +579,14 @@ drbd_req_state(struct drbd_device *device, union drbd_state mask, unsigned long flags; union drbd_state os, ns; enum drbd_state_rv rv; + void *buffer = NULL; init_completion(&done); if (f & CS_SERIALIZE) mutex_lock(device->state_mutex); + if (f & CS_INHIBIT_MD_IO) + buffer = drbd_md_get_buffer(device, __func__); spin_lock_irqsave(&device->resource->req_lock, flags); os = drbd_read_state(device); @@ -636,6 +639,8 @@ drbd_req_state(struct drbd_device *device, union drbd_state mask, } abort: + if (buffer) + drbd_md_put_buffer(device); if (f & CS_SERIALIZE) mutex_unlock(device->state_mutex); @@ -664,6 +669,47 @@ _drbd_request_state(struct drbd_device *device, union drbd_state mask, return rv; } +/* + * We grab drbd_md_get_buffer(), because we don't want to "fail" the disk while + * there is IO in-flight: the transition into D_FAILED for detach purposes + * may get misinterpreted as actual IO error in a confused endio function. + * + * We wrap it all into wait_event(), to retry in case the drbd_req_state() + * returns SS_IN_TRANSIENT_STATE. + * + * To avoid potential deadlock with e.g. the receiver thread trying to grab + * drbd_md_get_buffer() while trying to get out of the "transient state", we + * need to grab and release the meta data buffer inside of that wait_event loop. + */ +static enum drbd_state_rv +request_detach(struct drbd_device *device) +{ + return drbd_req_state(device, NS(disk, D_FAILED), + CS_VERBOSE | CS_ORDERED | CS_INHIBIT_MD_IO); +} + +enum drbd_state_rv +drbd_request_detach_interruptible(struct drbd_device *device) +{ + enum drbd_state_rv rv; + int ret; + + drbd_suspend_io(device); /* so no-one is stuck in drbd_al_begin_io */ + wait_event_interruptible(device->state_wait, + (rv = request_detach(device)) != SS_IN_TRANSIENT_STATE); + drbd_resume_io(device); + + ret = wait_event_interruptible(device->misc_wait, + device->state.disk != D_FAILED); + + if (rv == SS_IS_DISKLESS) + rv = SS_NOTHING_TO_DO; + if (ret) + rv = ERR_INTR; + + return rv; +} + enum drbd_state_rv _drbd_request_state_holding_state_mutex(struct drbd_device *device, union drbd_state mask, union drbd_state val, enum chg_state_flags f) diff --git a/drivers/block/drbd/drbd_state.h b/drivers/block/drbd/drbd_state.h index 6c9d5d4a8a75..0276c98fbbdd 100644 --- a/drivers/block/drbd/drbd_state.h +++ b/drivers/block/drbd/drbd_state.h @@ -71,6 +71,10 @@ enum chg_state_flags { CS_DC_SUSP = 1 << 10, CS_DC_MASK = CS_DC_ROLE + CS_DC_PEER + CS_DC_CONN + CS_DC_DISK + CS_DC_PDSK, CS_IGN_OUTD_FAIL = 1 << 11, + + /* Make sure no meta data IO is in flight, by calling + * drbd_md_get_buffer(). Used for graceful detach. */ + CS_INHIBIT_MD_IO = 1 << 12, }; /* drbd_dev_state and drbd_state are different types. This is to stress the @@ -156,6 +160,10 @@ static inline int drbd_request_state(struct drbd_device *device, return _drbd_request_state(device, mask, val, CS_VERBOSE + CS_ORDERED); } +/* for use in adm_detach() (drbd_adm_detach(), drbd_adm_down()) */ +enum drbd_state_rv +drbd_request_detach_interruptible(struct drbd_device *device); + enum drbd_role conn_highest_role(struct drbd_connection *connection); enum drbd_role conn_highest_peer(struct drbd_connection *connection); enum drbd_disk_state conn_highest_disk(struct drbd_connection *connection); diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 1d8726a8df34..03471b3fce86 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -65,6 +65,11 @@ void drbd_md_endio(struct bio *bio) device = bio->bi_private; device->md_io.error = blk_status_to_errno(bio->bi_status); + /* special case: drbd_md_read() during drbd_adm_attach() */ + if (device->ldev) + put_ldev(device); + bio_put(bio); + /* We grabbed an extra reference in _drbd_md_sync_page_io() to be able * to timeout on the lower level device, and eventually detach from it. * If this io completion runs after that timeout expired, this @@ -79,9 +84,6 @@ void drbd_md_endio(struct bio *bio) drbd_md_put_buffer(device); device->md_io.done = 1; wake_up(&device->misc_wait); - bio_put(bio); - if (device->ldev) /* special case: drbd_md_read() during drbd_adm_attach() */ - put_ldev(device); } /* reads on behalf of the partner, @@ -128,6 +130,14 @@ void drbd_endio_write_sec_final(struct drbd_peer_request *peer_req) __releases(l block_id = peer_req->block_id; peer_req->flags &= ~EE_CALL_AL_COMPLETE_IO; + if (peer_req->flags & EE_WAS_ERROR) { + /* In protocol != C, we usually do not send write acks. + * In case of a write error, send the neg ack anyways. */ + if (!__test_and_set_bit(__EE_SEND_WRITE_ACK, &peer_req->flags)) + inc_unacked(device); + drbd_set_out_of_sync(device, peer_req->i.sector, peer_req->i.size); + } + spin_lock_irqsave(&device->resource->req_lock, flags); device->writ_cnt += peer_req->i.size >> 9; list_move_tail(&peer_req->w.list, &device->done_ee); @@ -195,7 +205,8 @@ void drbd_peer_request_endio(struct bio *bio) } } -void drbd_panic_after_delayed_completion_of_aborted_request(struct drbd_device *device) +static void +drbd_panic_after_delayed_completion_of_aborted_request(struct drbd_device *device) { panic("drbd%u %s/%u potential random memory corruption caused by delayed completion of aborted local request\n", device->minor, device->resource->name, device->vnr); @@ -1382,18 +1393,22 @@ static int drbd_send_barrier(struct drbd_connection *connection) return conn_send_command(connection, sock, P_BARRIER, sizeof(*p), NULL, 0); } +static int pd_send_unplug_remote(struct drbd_peer_device *pd) +{ + struct drbd_socket *sock = &pd->connection->data; + if (!drbd_prepare_command(pd, sock)) + return -EIO; + return drbd_send_command(pd, sock, P_UNPLUG_REMOTE, 0, NULL, 0); +} + int w_send_write_hint(struct drbd_work *w, int cancel) { struct drbd_device *device = container_of(w, struct drbd_device, unplug_work); - struct drbd_socket *sock; if (cancel) return 0; - sock = &first_peer_device(device)->connection->data; - if (!drbd_prepare_command(first_peer_device(device), sock)) - return -EIO; - return drbd_send_command(first_peer_device(device), sock, P_UNPLUG_REMOTE, 0, NULL, 0); + return pd_send_unplug_remote(first_peer_device(device)); } static void re_init_if_first_write(struct drbd_connection *connection, unsigned int epoch) @@ -1455,6 +1470,7 @@ int w_send_dblock(struct drbd_work *w, int cancel) struct drbd_device *device = req->device; struct drbd_peer_device *const peer_device = first_peer_device(device); struct drbd_connection *connection = peer_device->connection; + bool do_send_unplug = req->rq_state & RQ_UNPLUG; int err; if (unlikely(cancel)) { @@ -1470,6 +1486,9 @@ int w_send_dblock(struct drbd_work *w, int cancel) err = drbd_send_dblock(peer_device, req); req_mod(req, err ? SEND_FAILED : HANDED_OVER_TO_NETWORK); + if (do_send_unplug && !err) + pd_send_unplug_remote(peer_device); + return err; } @@ -1484,6 +1503,7 @@ int w_send_read_req(struct drbd_work *w, int cancel) struct drbd_device *device = req->device; struct drbd_peer_device *const peer_device = first_peer_device(device); struct drbd_connection *connection = peer_device->connection; + bool do_send_unplug = req->rq_state & RQ_UNPLUG; int err; if (unlikely(cancel)) { @@ -1501,6 +1521,9 @@ int w_send_read_req(struct drbd_work *w, int cancel) req_mod(req, err ? SEND_FAILED : HANDED_OVER_TO_NETWORK); + if (do_send_unplug && !err) + pd_send_unplug_remote(peer_device); + return err; } @@ -1513,7 +1536,7 @@ int w_restart_disk_io(struct drbd_work *w, int cancel) drbd_al_begin_io(device, &req->i); drbd_req_make_private_bio(req, req->master_bio); - req->private_bio->bi_bdev = device->ldev->backing_bdev; + bio_set_dev(req->private_bio, device->ldev->backing_bdev); generic_make_request(req->private_bio); return 0; @@ -1733,6 +1756,11 @@ void drbd_start_resync(struct drbd_device *device, enum drbd_conns side) return; } + if (!connection) { + drbd_err(device, "No connection to peer, aborting!\n"); + return; + } + if (!test_bit(B_RS_H_DONE, &device->flags)) { if (side == C_SYNC_TARGET) { /* Since application IO was locked out during C_WF_BITMAP_T and diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index 9c00f29e40c1..60c086a53609 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -4134,7 +4134,7 @@ static int __floppy_read_block_0(struct block_device *bdev, int drive) cbdata.drive = drive; bio_init(&bio, &bio_vec, 1); - bio.bi_bdev = bdev; + bio_set_dev(&bio, bdev); bio_add_page(&bio, page, size, 0); bio.bi_iter.bi_sector = 0; diff --git a/drivers/block/loop.c b/drivers/block/loop.c index ef8334949b42..85de67334695 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -213,16 +213,18 @@ static void __loop_update_dio(struct loop_device *lo, bool dio) */ blk_mq_freeze_queue(lo->lo_queue); lo->use_dio = use_dio; - if (use_dio) + if (use_dio) { + queue_flag_clear_unlocked(QUEUE_FLAG_NOMERGES, lo->lo_queue); lo->lo_flags |= LO_FLAGS_DIRECT_IO; - else + } else { + queue_flag_set_unlocked(QUEUE_FLAG_NOMERGES, lo->lo_queue); lo->lo_flags &= ~LO_FLAGS_DIRECT_IO; + } blk_mq_unfreeze_queue(lo->lo_queue); } static int -figure_loop_size(struct loop_device *lo, loff_t offset, loff_t sizelimit, - loff_t logical_blocksize) +figure_loop_size(struct loop_device *lo, loff_t offset, loff_t sizelimit) { loff_t size = get_size(offset, sizelimit, lo->lo_backing_file); sector_t x = (sector_t)size; @@ -234,12 +236,6 @@ figure_loop_size(struct loop_device *lo, loff_t offset, loff_t sizelimit, lo->lo_offset = offset; if (lo->lo_sizelimit != sizelimit) lo->lo_sizelimit = sizelimit; - if (lo->lo_flags & LO_FLAGS_BLOCKSIZE) { - lo->lo_logical_blocksize = logical_blocksize; - blk_queue_physical_block_size(lo->lo_queue, lo->lo_blocksize); - blk_queue_logical_block_size(lo->lo_queue, - lo->lo_logical_blocksize); - } set_capacity(lo->lo_disk, x); bd_set_size(bdev, (loff_t)get_capacity(bdev->bd_disk) << 9); /* let user-space know about the new size */ @@ -467,12 +463,21 @@ static void lo_complete_rq(struct request *rq) blk_mq_end_request(rq, cmd->ret < 0 ? BLK_STS_IOERR : BLK_STS_OK); } +static void lo_rw_aio_do_completion(struct loop_cmd *cmd) +{ + if (!atomic_dec_and_test(&cmd->ref)) + return; + kfree(cmd->bvec); + cmd->bvec = NULL; + blk_mq_complete_request(cmd->rq); +} + static void lo_rw_aio_complete(struct kiocb *iocb, long ret, long ret2) { struct loop_cmd *cmd = container_of(iocb, struct loop_cmd, iocb); cmd->ret = ret; - blk_mq_complete_request(cmd->rq); + lo_rw_aio_do_completion(cmd); } static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd, @@ -480,22 +485,51 @@ static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd, { struct iov_iter iter; struct bio_vec *bvec; - struct bio *bio = cmd->rq->bio; + struct request *rq = cmd->rq; + struct bio *bio = rq->bio; struct file *file = lo->lo_backing_file; + unsigned int offset; + int segments = 0; int ret; - /* nomerge for loop request queue */ - WARN_ON(cmd->rq->bio != cmd->rq->biotail); + if (rq->bio != rq->biotail) { + struct req_iterator iter; + struct bio_vec tmp; + + __rq_for_each_bio(bio, rq) + segments += bio_segments(bio); + bvec = kmalloc(sizeof(struct bio_vec) * segments, GFP_NOIO); + if (!bvec) + return -EIO; + cmd->bvec = bvec; + + /* + * The bios of the request may be started from the middle of + * the 'bvec' because of bio splitting, so we can't directly + * copy bio->bi_iov_vec to new bvec. The rq_for_each_segment + * API will take care of all details for us. + */ + rq_for_each_segment(tmp, rq, iter) { + *bvec = tmp; + bvec++; + } + bvec = cmd->bvec; + offset = 0; + } else { + /* + * Same here, this bio may be started from the middle of the + * 'bvec' because of bio splitting, so offset from the bvec + * must be passed to iov iterator + */ + offset = bio->bi_iter.bi_bvec_done; + bvec = __bvec_iter_bvec(bio->bi_io_vec, bio->bi_iter); + segments = bio_segments(bio); + } + atomic_set(&cmd->ref, 2); - bvec = __bvec_iter_bvec(bio->bi_io_vec, bio->bi_iter); iov_iter_bvec(&iter, ITER_BVEC | rw, bvec, - bio_segments(bio), blk_rq_bytes(cmd->rq)); - /* - * This bio may be started from the middle of the 'bvec' - * because of bio splitting, so offset from the bvec must - * be passed to iov iterator - */ - iter.iov_offset = bio->bi_iter.bi_bvec_done; + segments, blk_rq_bytes(rq)); + iter.iov_offset = offset; cmd->iocb.ki_pos = pos; cmd->iocb.ki_filp = file; @@ -507,6 +541,8 @@ static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd, else ret = call_read_iter(file, &cmd->iocb, &iter); + lo_rw_aio_do_completion(cmd); + if (ret != -EIOCBQUEUED) cmd->iocb.ki_complete(&cmd->iocb, ret, 0); return 0; @@ -553,74 +589,12 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq) } } -struct switch_request { - struct file *file; - struct completion wait; -}; - static inline void loop_update_dio(struct loop_device *lo) { __loop_update_dio(lo, io_is_direct(lo->lo_backing_file) | lo->use_dio); } -/* - * Do the actual switch; called from the BIO completion routine - */ -static void do_loop_switch(struct loop_device *lo, struct switch_request *p) -{ - struct file *file = p->file; - struct file *old_file = lo->lo_backing_file; - struct address_space *mapping; - - /* if no new file, only flush of queued bios requested */ - if (!file) - return; - - mapping = file->f_mapping; - mapping_set_gfp_mask(old_file->f_mapping, lo->old_gfp_mask); - lo->lo_backing_file = file; - lo->lo_blocksize = S_ISBLK(mapping->host->i_mode) ? - mapping->host->i_bdev->bd_block_size : PAGE_SIZE; - lo->old_gfp_mask = mapping_gfp_mask(mapping); - mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS)); - loop_update_dio(lo); -} - -/* - * loop_switch performs the hard work of switching a backing store. - * First it needs to flush existing IO, it does this by sending a magic - * BIO down the pipe. The completion of this BIO does the actual switch. - */ -static int loop_switch(struct loop_device *lo, struct file *file) -{ - struct switch_request w; - - w.file = file; - - /* freeze queue and wait for completion of scheduled requests */ - blk_mq_freeze_queue(lo->lo_queue); - - /* do the switch action */ - do_loop_switch(lo, &w); - - /* unfreeze */ - blk_mq_unfreeze_queue(lo->lo_queue); - - return 0; -} - -/* - * Helper to flush the IOs in loop, but keeping loop thread running - */ -static int loop_flush(struct loop_device *lo) -{ - /* loop not yet configured, no running thread, nothing to flush */ - if (lo->lo_state != Lo_bound) - return 0; - return loop_switch(lo, NULL); -} - static void loop_reread_partitions(struct loop_device *lo, struct block_device *bdev) { @@ -685,9 +659,14 @@ static int loop_change_fd(struct loop_device *lo, struct block_device *bdev, goto out_putf; /* and ... switch */ - error = loop_switch(lo, file); - if (error) - goto out_putf; + blk_mq_freeze_queue(lo->lo_queue); + mapping_set_gfp_mask(old_file->f_mapping, lo->old_gfp_mask); + lo->lo_backing_file = file; + lo->old_gfp_mask = mapping_gfp_mask(file->f_mapping); + mapping_set_gfp_mask(file->f_mapping, + lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS)); + loop_update_dio(lo); + blk_mq_unfreeze_queue(lo->lo_queue); fput(old_file); if (lo->lo_flags & LO_FLAGS_PARTSCAN) @@ -820,7 +799,6 @@ static void loop_config_discard(struct loop_device *lo) struct file *file = lo->lo_backing_file; struct inode *inode = file->f_mapping->host; struct request_queue *q = lo->lo_queue; - int lo_bits = 9; /* * We use punch hole to reclaim the free space used by the @@ -840,11 +818,9 @@ static void loop_config_discard(struct loop_device *lo) q->limits.discard_granularity = inode->i_sb->s_blocksize; q->limits.discard_alignment = 0; - if (lo->lo_flags & LO_FLAGS_BLOCKSIZE) - lo_bits = blksize_bits(lo->lo_logical_blocksize); - blk_queue_max_discard_sectors(q, UINT_MAX >> lo_bits); - blk_queue_max_write_zeroes_sectors(q, UINT_MAX >> lo_bits); + blk_queue_max_discard_sectors(q, UINT_MAX >> 9); + blk_queue_max_write_zeroes_sectors(q, UINT_MAX >> 9); queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q); } @@ -877,7 +853,6 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, struct file *file, *f; struct inode *inode; struct address_space *mapping; - unsigned lo_blocksize; int lo_flags = 0; int error; loff_t size; @@ -921,9 +896,6 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, !file->f_op->write_iter) lo_flags |= LO_FLAGS_READ_ONLY; - lo_blocksize = S_ISBLK(inode->i_mode) ? - inode->i_bdev->bd_block_size : PAGE_SIZE; - error = -EFBIG; size = get_loop_size(lo, file); if ((loff_t)(sector_t)size != size) @@ -937,8 +909,6 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, set_device_ro(bdev, (lo_flags & LO_FLAGS_READ_ONLY) != 0); lo->use_dio = false; - lo->lo_blocksize = lo_blocksize; - lo->lo_logical_blocksize = 512; lo->lo_device = bdev; lo->lo_flags = lo_flags; lo->lo_backing_file = file; @@ -958,7 +928,8 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, /* let user-space know about the new size */ kobject_uevent(&disk_to_dev(bdev->bd_disk)->kobj, KOBJ_CHANGE); - set_blocksize(bdev, lo_blocksize); + set_blocksize(bdev, S_ISBLK(inode->i_mode) ? + block_size(inode->i_bdev) : PAGE_SIZE); lo->lo_state = Lo_bound; if (part_shift) @@ -1064,6 +1035,9 @@ static int loop_clr_fd(struct loop_device *lo) memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE); memset(lo->lo_crypt_name, 0, LO_NAME_SIZE); memset(lo->lo_file_name, 0, LO_NAME_SIZE); + blk_queue_logical_block_size(lo->lo_queue, 512); + blk_queue_physical_block_size(lo->lo_queue, 512); + blk_queue_io_min(lo->lo_queue, 512); if (bdev) { bdput(bdev); invalidate_bdev(bdev); @@ -1104,7 +1078,6 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info) int err; struct loop_func_table *xfer; kuid_t uid = current_uid(); - int lo_flags = lo->lo_flags; if (lo->lo_encrypt_key_size && !uid_eq(lo->lo_key_owner, uid) && @@ -1137,26 +1110,9 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info) if (err) goto exit; - if (info->lo_flags & LO_FLAGS_BLOCKSIZE) { - if (!(lo->lo_flags & LO_FLAGS_BLOCKSIZE)) - lo->lo_logical_blocksize = 512; - lo->lo_flags |= LO_FLAGS_BLOCKSIZE; - if (LO_INFO_BLOCKSIZE(info) != 512 && - LO_INFO_BLOCKSIZE(info) != 1024 && - LO_INFO_BLOCKSIZE(info) != 2048 && - LO_INFO_BLOCKSIZE(info) != 4096) - return -EINVAL; - if (LO_INFO_BLOCKSIZE(info) > lo->lo_blocksize) - return -EINVAL; - } - if (lo->lo_offset != info->lo_offset || - lo->lo_sizelimit != info->lo_sizelimit || - lo->lo_flags != lo_flags || - ((lo->lo_flags & LO_FLAGS_BLOCKSIZE) && - lo->lo_logical_blocksize != LO_INFO_BLOCKSIZE(info))) { - if (figure_loop_size(lo, info->lo_offset, info->lo_sizelimit, - LO_INFO_BLOCKSIZE(info))) { + lo->lo_sizelimit != info->lo_sizelimit) { + if (figure_loop_size(lo, info->lo_offset, info->lo_sizelimit)) { err = -EFBIG; goto exit; } @@ -1348,8 +1304,7 @@ static int loop_set_capacity(struct loop_device *lo) if (unlikely(lo->lo_state != Lo_bound)) return -ENXIO; - return figure_loop_size(lo, lo->lo_offset, lo->lo_sizelimit, - lo->lo_logical_blocksize); + return figure_loop_size(lo, lo->lo_offset, lo->lo_sizelimit); } static int loop_set_dio(struct loop_device *lo, unsigned long arg) @@ -1366,6 +1321,26 @@ static int loop_set_dio(struct loop_device *lo, unsigned long arg) return error; } +static int loop_set_block_size(struct loop_device *lo, unsigned long arg) +{ + if (lo->lo_state != Lo_bound) + return -ENXIO; + + if (arg < 512 || arg > PAGE_SIZE || !is_power_of_2(arg)) + return -EINVAL; + + blk_mq_freeze_queue(lo->lo_queue); + + blk_queue_logical_block_size(lo->lo_queue, arg); + blk_queue_physical_block_size(lo->lo_queue, arg); + blk_queue_io_min(lo->lo_queue, arg); + loop_update_dio(lo); + + blk_mq_unfreeze_queue(lo->lo_queue); + + return 0; +} + static int lo_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, unsigned long arg) { @@ -1414,6 +1389,11 @@ static int lo_ioctl(struct block_device *bdev, fmode_t mode, if ((mode & FMODE_WRITE) || capable(CAP_SYS_ADMIN)) err = loop_set_dio(lo, arg); break; + case LOOP_SET_BLOCK_SIZE: + err = -EPERM; + if ((mode & FMODE_WRITE) || capable(CAP_SYS_ADMIN)) + err = loop_set_block_size(lo, arg); + break; default: err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL; } @@ -1613,12 +1593,13 @@ static void lo_release(struct gendisk *disk, fmode_t mode) err = loop_clr_fd(lo); if (!err) return; - } else { + } else if (lo->lo_state == Lo_bound) { /* * Otherwise keep thread (if running) and config, * but flush possible ongoing bios in thread. */ - loop_flush(lo); + blk_mq_freeze_queue(lo->lo_queue); + blk_mq_unfreeze_queue(lo->lo_queue); } mutex_unlock(&lo->lo_ctl_mutex); @@ -1800,9 +1781,13 @@ static int loop_add(struct loop_device **l, int i) } lo->lo_queue->queuedata = lo; + blk_queue_max_hw_sectors(lo->lo_queue, BLK_DEF_MAX_SECTORS); + /* - * It doesn't make sense to enable merge because the I/O - * submitted to backing file is handled page by page. + * By default, we do buffer IO, so it doesn't make sense to enable + * merge because the I/O submitted to backing file is handled page by + * page. For directio mode, merge does help to dispatch bigger request + * to underlayer disk. We will enable merge once directio is enabled. */ queue_flag_set_unlocked(QUEUE_FLAG_NOMERGES, lo->lo_queue); @@ -1996,10 +1981,6 @@ static int __init loop_init(void) struct loop_device *lo; int err; - err = misc_register(&loop_misc); - if (err < 0) - return err; - part_shift = 0; if (max_part > 0) { part_shift = fls(max_part); @@ -2017,12 +1998,12 @@ static int __init loop_init(void) if ((1UL << part_shift) > DISK_MAX_PARTS) { err = -EINVAL; - goto misc_out; + goto err_out; } if (max_loop > 1UL << (MINORBITS - part_shift)) { err = -EINVAL; - goto misc_out; + goto err_out; } /* @@ -2041,6 +2022,11 @@ static int __init loop_init(void) range = 1UL << MINORBITS; } + err = misc_register(&loop_misc); + if (err < 0) + goto err_out; + + if (register_blkdev(LOOP_MAJOR, "loop")) { err = -EIO; goto misc_out; @@ -2060,6 +2046,7 @@ static int __init loop_init(void) misc_out: misc_deregister(&loop_misc); +err_out: return err; } diff --git a/drivers/block/loop.h b/drivers/block/loop.h index 2c096b9a17b8..f68c1d50802f 100644 --- a/drivers/block/loop.h +++ b/drivers/block/loop.h @@ -48,8 +48,6 @@ struct loop_device { struct file * lo_backing_file; struct block_device *lo_device; - unsigned lo_blocksize; - unsigned lo_logical_blocksize; void *key_data; gfp_t old_gfp_mask; @@ -69,10 +67,13 @@ struct loop_device { struct loop_cmd { struct kthread_work work; struct request *rq; - struct list_head list; - bool use_aio; /* use AIO interface to handle I/O */ + union { + bool use_aio; /* use AIO interface to handle I/O */ + atomic_t ref; /* only for aio */ + }; long ret; struct kiocb iocb; + struct bio_vec *bvec; }; /* Support for loadable transfer modules */ diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c index 5bdf923294a5..2aa87cbdede0 100644 --- a/drivers/block/nbd.c +++ b/drivers/block/nbd.c @@ -128,7 +128,7 @@ static struct dentry *nbd_dbg_dir; #define NBD_MAGIC 0x68797548 static unsigned int nbds_max = 16; -static int max_part; +static int max_part = 16; static struct workqueue_struct *recv_workqueue; static int part_shift; @@ -165,7 +165,7 @@ static ssize_t pid_show(struct device *dev, return sprintf(buf, "%d\n", task_pid_nr(nbd->task_recv)); } -static struct device_attribute pid_attr = { +static const struct device_attribute pid_attr = { .attr = { .name = "pid", .mode = S_IRUGO}, .show = pid_show, }; @@ -1584,6 +1584,15 @@ again: } } else { nbd = idr_find(&nbd_index_idr, index); + if (!nbd) { + ret = nbd_dev_add(index); + if (ret < 0) { + mutex_unlock(&nbd_index_mutex); + printk(KERN_ERR "nbd: failed to add new device\n"); + return ret; + } + nbd = idr_find(&nbd_index_idr, index); + } } if (!nbd) { printk(KERN_ERR "nbd: couldn't find device at index %d\n", @@ -2137,4 +2146,4 @@ MODULE_LICENSE("GPL"); module_param(nbds_max, int, 0444); MODULE_PARM_DESC(nbds_max, "number of network block devices to initialize (default: 16)"); module_param(max_part, int, 0444); -MODULE_PARM_DESC(max_part, "number of partitions per device (default: 0)"); +MODULE_PARM_DESC(max_part, "number of partitions per device (default: 16)"); diff --git a/drivers/block/null_blk.c b/drivers/block/null_blk.c index 85c24cace973..8042c26ea9e6 100644 --- a/drivers/block/null_blk.c +++ b/drivers/block/null_blk.c @@ -1,3 +1,7 @@ +/* + * Add configfs and memory store: Kyungchan Koh <kkc6196@fb.com> and + * Shaohua Li <shli@fb.com> + */ #include <linux/module.h> #include <linux/moduleparam.h> @@ -9,27 +13,110 @@ #include <linux/blk-mq.h> #include <linux/hrtimer.h> #include <linux/lightnvm.h> +#include <linux/configfs.h> +#include <linux/badblocks.h> + +#define SECTOR_SHIFT 9 +#define PAGE_SECTORS_SHIFT (PAGE_SHIFT - SECTOR_SHIFT) +#define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT) +#define SECTOR_SIZE (1 << SECTOR_SHIFT) +#define SECTOR_MASK (PAGE_SECTORS - 1) + +#define FREE_BATCH 16 + +#define TICKS_PER_SEC 50ULL +#define TIMER_INTERVAL (NSEC_PER_SEC / TICKS_PER_SEC) + +static inline u64 mb_per_tick(int mbps) +{ + return (1 << 20) / TICKS_PER_SEC * ((u64) mbps); +} struct nullb_cmd { struct list_head list; struct llist_node ll_list; - struct call_single_data csd; + call_single_data_t csd; struct request *rq; struct bio *bio; unsigned int tag; struct nullb_queue *nq; struct hrtimer timer; + blk_status_t error; }; struct nullb_queue { unsigned long *tag_map; wait_queue_head_t wait; unsigned int queue_depth; + struct nullb_device *dev; struct nullb_cmd *cmds; }; +/* + * Status flags for nullb_device. + * + * CONFIGURED: Device has been configured and turned on. Cannot reconfigure. + * UP: Device is currently on and visible in userspace. + * THROTTLED: Device is being throttled. + * CACHE: Device is using a write-back cache. + */ +enum nullb_device_flags { + NULLB_DEV_FL_CONFIGURED = 0, + NULLB_DEV_FL_UP = 1, + NULLB_DEV_FL_THROTTLED = 2, + NULLB_DEV_FL_CACHE = 3, +}; + +/* + * nullb_page is a page in memory for nullb devices. + * + * @page: The page holding the data. + * @bitmap: The bitmap represents which sector in the page has data. + * Each bit represents one block size. For example, sector 8 + * will use the 7th bit + * The highest 2 bits of bitmap are for special purpose. LOCK means the cache + * page is being flushing to storage. FREE means the cache page is freed and + * should be skipped from flushing to storage. Please see + * null_make_cache_space + */ +struct nullb_page { + struct page *page; + unsigned long bitmap; +}; +#define NULLB_PAGE_LOCK (sizeof(unsigned long) * 8 - 1) +#define NULLB_PAGE_FREE (sizeof(unsigned long) * 8 - 2) + +struct nullb_device { + struct nullb *nullb; + struct config_item item; + struct radix_tree_root data; /* data stored in the disk */ + struct radix_tree_root cache; /* disk cache data */ + unsigned long flags; /* device flags */ + unsigned int curr_cache; + struct badblocks badblocks; + + unsigned long size; /* device size in MB */ + unsigned long completion_nsec; /* time in ns to complete a request */ + unsigned long cache_size; /* disk cache size in MB */ + unsigned int submit_queues; /* number of submission queues */ + unsigned int home_node; /* home node for the device */ + unsigned int queue_mode; /* block interface */ + unsigned int blocksize; /* block size */ + unsigned int irqmode; /* IRQ completion handler */ + unsigned int hw_queue_depth; /* queue depth */ + unsigned int index; /* index of the disk, only valid with a disk */ + unsigned int mbps; /* Bandwidth throttle cap (in MB/s) */ + bool use_lightnvm; /* register as a LightNVM device */ + bool blocking; /* blocking blk-mq device */ + bool use_per_node_hctx; /* use per-node allocation for hardware context */ + bool power; /* power on/off the device */ + bool memory_backed; /* if data is stored in memory */ + bool discard; /* if support discard */ +}; + struct nullb { + struct nullb_device *dev; struct list_head list; unsigned int index; struct request_queue *q; @@ -37,8 +124,10 @@ struct nullb { struct nvm_dev *ndev; struct blk_mq_tag_set *tag_set; struct blk_mq_tag_set __tag_set; - struct hrtimer timer; unsigned int queue_depth; + atomic_long_t cur_bytes; + struct hrtimer bw_timer; + unsigned long cache_flush_pos; spinlock_t lock; struct nullb_queue *queues; @@ -49,7 +138,7 @@ struct nullb { static LIST_HEAD(nullb_list); static struct mutex lock; static int null_major; -static int nullb_indexes; +static DEFINE_IDA(nullb_indexes); static struct kmem_cache *ppa_cache; static struct blk_mq_tag_set tag_set; @@ -65,15 +154,15 @@ enum { NULL_Q_MQ = 2, }; -static int submit_queues; -module_param(submit_queues, int, S_IRUGO); +static int g_submit_queues = 1; +module_param_named(submit_queues, g_submit_queues, int, S_IRUGO); MODULE_PARM_DESC(submit_queues, "Number of submission queues"); -static int home_node = NUMA_NO_NODE; -module_param(home_node, int, S_IRUGO); +static int g_home_node = NUMA_NO_NODE; +module_param_named(home_node, g_home_node, int, S_IRUGO); MODULE_PARM_DESC(home_node, "Home node for the device"); -static int queue_mode = NULL_Q_MQ; +static int g_queue_mode = NULL_Q_MQ; static int null_param_store_val(const char *str, int *val, int min, int max) { @@ -92,7 +181,7 @@ static int null_param_store_val(const char *str, int *val, int min, int max) static int null_set_queue_mode(const char *str, const struct kernel_param *kp) { - return null_param_store_val(str, &queue_mode, NULL_Q_BIO, NULL_Q_MQ); + return null_param_store_val(str, &g_queue_mode, NULL_Q_BIO, NULL_Q_MQ); } static const struct kernel_param_ops null_queue_mode_param_ops = { @@ -100,38 +189,38 @@ static const struct kernel_param_ops null_queue_mode_param_ops = { .get = param_get_int, }; -device_param_cb(queue_mode, &null_queue_mode_param_ops, &queue_mode, S_IRUGO); +device_param_cb(queue_mode, &null_queue_mode_param_ops, &g_queue_mode, S_IRUGO); MODULE_PARM_DESC(queue_mode, "Block interface to use (0=bio,1=rq,2=multiqueue)"); -static int gb = 250; -module_param(gb, int, S_IRUGO); +static int g_gb = 250; +module_param_named(gb, g_gb, int, S_IRUGO); MODULE_PARM_DESC(gb, "Size in GB"); -static int bs = 512; -module_param(bs, int, S_IRUGO); +static int g_bs = 512; +module_param_named(bs, g_bs, int, S_IRUGO); MODULE_PARM_DESC(bs, "Block size (in bytes)"); static int nr_devices = 1; module_param(nr_devices, int, S_IRUGO); MODULE_PARM_DESC(nr_devices, "Number of devices to register"); -static bool use_lightnvm; -module_param(use_lightnvm, bool, S_IRUGO); +static bool g_use_lightnvm; +module_param_named(use_lightnvm, g_use_lightnvm, bool, S_IRUGO); MODULE_PARM_DESC(use_lightnvm, "Register as a LightNVM device"); -static bool blocking; -module_param(blocking, bool, S_IRUGO); +static bool g_blocking; +module_param_named(blocking, g_blocking, bool, S_IRUGO); MODULE_PARM_DESC(blocking, "Register as a blocking blk-mq driver device"); static bool shared_tags; module_param(shared_tags, bool, S_IRUGO); MODULE_PARM_DESC(shared_tags, "Share tag set between devices for blk-mq"); -static int irqmode = NULL_IRQ_SOFTIRQ; +static int g_irqmode = NULL_IRQ_SOFTIRQ; static int null_set_irqmode(const char *str, const struct kernel_param *kp) { - return null_param_store_val(str, &irqmode, NULL_IRQ_NONE, + return null_param_store_val(str, &g_irqmode, NULL_IRQ_NONE, NULL_IRQ_TIMER); } @@ -140,21 +229,358 @@ static const struct kernel_param_ops null_irqmode_param_ops = { .get = param_get_int, }; -device_param_cb(irqmode, &null_irqmode_param_ops, &irqmode, S_IRUGO); +device_param_cb(irqmode, &null_irqmode_param_ops, &g_irqmode, S_IRUGO); MODULE_PARM_DESC(irqmode, "IRQ completion handler. 0-none, 1-softirq, 2-timer"); -static unsigned long completion_nsec = 10000; -module_param(completion_nsec, ulong, S_IRUGO); +static unsigned long g_completion_nsec = 10000; +module_param_named(completion_nsec, g_completion_nsec, ulong, S_IRUGO); MODULE_PARM_DESC(completion_nsec, "Time in ns to complete a request in hardware. Default: 10,000ns"); -static int hw_queue_depth = 64; -module_param(hw_queue_depth, int, S_IRUGO); +static int g_hw_queue_depth = 64; +module_param_named(hw_queue_depth, g_hw_queue_depth, int, S_IRUGO); MODULE_PARM_DESC(hw_queue_depth, "Queue depth for each hardware queue. Default: 64"); -static bool use_per_node_hctx = false; -module_param(use_per_node_hctx, bool, S_IRUGO); +static bool g_use_per_node_hctx; +module_param_named(use_per_node_hctx, g_use_per_node_hctx, bool, S_IRUGO); MODULE_PARM_DESC(use_per_node_hctx, "Use per-node allocation for hardware context queues. Default: false"); +static struct nullb_device *null_alloc_dev(void); +static void null_free_dev(struct nullb_device *dev); +static void null_del_dev(struct nullb *nullb); +static int null_add_dev(struct nullb_device *dev); +static void null_free_device_storage(struct nullb_device *dev, bool is_cache); + +static inline struct nullb_device *to_nullb_device(struct config_item *item) +{ + return item ? container_of(item, struct nullb_device, item) : NULL; +} + +static inline ssize_t nullb_device_uint_attr_show(unsigned int val, char *page) +{ + return snprintf(page, PAGE_SIZE, "%u\n", val); +} + +static inline ssize_t nullb_device_ulong_attr_show(unsigned long val, + char *page) +{ + return snprintf(page, PAGE_SIZE, "%lu\n", val); +} + +static inline ssize_t nullb_device_bool_attr_show(bool val, char *page) +{ + return snprintf(page, PAGE_SIZE, "%u\n", val); +} + +static ssize_t nullb_device_uint_attr_store(unsigned int *val, + const char *page, size_t count) +{ + unsigned int tmp; + int result; + + result = kstrtouint(page, 0, &tmp); + if (result) + return result; + + *val = tmp; + return count; +} + +static ssize_t nullb_device_ulong_attr_store(unsigned long *val, + const char *page, size_t count) +{ + int result; + unsigned long tmp; + + result = kstrtoul(page, 0, &tmp); + if (result) + return result; + + *val = tmp; + return count; +} + +static ssize_t nullb_device_bool_attr_store(bool *val, const char *page, + size_t count) +{ + bool tmp; + int result; + + result = kstrtobool(page, &tmp); + if (result) + return result; + + *val = tmp; + return count; +} + +/* The following macro should only be used with TYPE = {uint, ulong, bool}. */ +#define NULLB_DEVICE_ATTR(NAME, TYPE) \ +static ssize_t \ +nullb_device_##NAME##_show(struct config_item *item, char *page) \ +{ \ + return nullb_device_##TYPE##_attr_show( \ + to_nullb_device(item)->NAME, page); \ +} \ +static ssize_t \ +nullb_device_##NAME##_store(struct config_item *item, const char *page, \ + size_t count) \ +{ \ + if (test_bit(NULLB_DEV_FL_CONFIGURED, &to_nullb_device(item)->flags)) \ + return -EBUSY; \ + return nullb_device_##TYPE##_attr_store( \ + &to_nullb_device(item)->NAME, page, count); \ +} \ +CONFIGFS_ATTR(nullb_device_, NAME); + +NULLB_DEVICE_ATTR(size, ulong); +NULLB_DEVICE_ATTR(completion_nsec, ulong); +NULLB_DEVICE_ATTR(submit_queues, uint); +NULLB_DEVICE_ATTR(home_node, uint); +NULLB_DEVICE_ATTR(queue_mode, uint); +NULLB_DEVICE_ATTR(blocksize, uint); +NULLB_DEVICE_ATTR(irqmode, uint); +NULLB_DEVICE_ATTR(hw_queue_depth, uint); +NULLB_DEVICE_ATTR(index, uint); +NULLB_DEVICE_ATTR(use_lightnvm, bool); +NULLB_DEVICE_ATTR(blocking, bool); +NULLB_DEVICE_ATTR(use_per_node_hctx, bool); +NULLB_DEVICE_ATTR(memory_backed, bool); +NULLB_DEVICE_ATTR(discard, bool); +NULLB_DEVICE_ATTR(mbps, uint); +NULLB_DEVICE_ATTR(cache_size, ulong); + +static ssize_t nullb_device_power_show(struct config_item *item, char *page) +{ + return nullb_device_bool_attr_show(to_nullb_device(item)->power, page); +} + +static ssize_t nullb_device_power_store(struct config_item *item, + const char *page, size_t count) +{ + struct nullb_device *dev = to_nullb_device(item); + bool newp = false; + ssize_t ret; + + ret = nullb_device_bool_attr_store(&newp, page, count); + if (ret < 0) + return ret; + + if (!dev->power && newp) { + if (test_and_set_bit(NULLB_DEV_FL_UP, &dev->flags)) + return count; + if (null_add_dev(dev)) { + clear_bit(NULLB_DEV_FL_UP, &dev->flags); + return -ENOMEM; + } + + set_bit(NULLB_DEV_FL_CONFIGURED, &dev->flags); + dev->power = newp; + } else if (dev->power && !newp) { + mutex_lock(&lock); + dev->power = newp; + null_del_dev(dev->nullb); + mutex_unlock(&lock); + clear_bit(NULLB_DEV_FL_UP, &dev->flags); + } + + return count; +} + +CONFIGFS_ATTR(nullb_device_, power); + +static ssize_t nullb_device_badblocks_show(struct config_item *item, char *page) +{ + struct nullb_device *t_dev = to_nullb_device(item); + + return badblocks_show(&t_dev->badblocks, page, 0); +} + +static ssize_t nullb_device_badblocks_store(struct config_item *item, + const char *page, size_t count) +{ + struct nullb_device *t_dev = to_nullb_device(item); + char *orig, *buf, *tmp; + u64 start, end; + int ret; + + orig = kstrndup(page, count, GFP_KERNEL); + if (!orig) + return -ENOMEM; + + buf = strstrip(orig); + + ret = -EINVAL; + if (buf[0] != '+' && buf[0] != '-') + goto out; + tmp = strchr(&buf[1], '-'); + if (!tmp) + goto out; + *tmp = '\0'; + ret = kstrtoull(buf + 1, 0, &start); + if (ret) + goto out; + ret = kstrtoull(tmp + 1, 0, &end); + if (ret) + goto out; + ret = -EINVAL; + if (start > end) + goto out; + /* enable badblocks */ + cmpxchg(&t_dev->badblocks.shift, -1, 0); + if (buf[0] == '+') + ret = badblocks_set(&t_dev->badblocks, start, + end - start + 1, 1); + else + ret = badblocks_clear(&t_dev->badblocks, start, + end - start + 1); + if (ret == 0) + ret = count; +out: + kfree(orig); + return ret; +} +CONFIGFS_ATTR(nullb_device_, badblocks); + +static struct configfs_attribute *nullb_device_attrs[] = { + &nullb_device_attr_size, + &nullb_device_attr_completion_nsec, + &nullb_device_attr_submit_queues, + &nullb_device_attr_home_node, + &nullb_device_attr_queue_mode, + &nullb_device_attr_blocksize, + &nullb_device_attr_irqmode, + &nullb_device_attr_hw_queue_depth, + &nullb_device_attr_index, + &nullb_device_attr_use_lightnvm, + &nullb_device_attr_blocking, + &nullb_device_attr_use_per_node_hctx, + &nullb_device_attr_power, + &nullb_device_attr_memory_backed, + &nullb_device_attr_discard, + &nullb_device_attr_mbps, + &nullb_device_attr_cache_size, + &nullb_device_attr_badblocks, + NULL, +}; + +static void nullb_device_release(struct config_item *item) +{ + struct nullb_device *dev = to_nullb_device(item); + + badblocks_exit(&dev->badblocks); + null_free_device_storage(dev, false); + null_free_dev(dev); +} + +static struct configfs_item_operations nullb_device_ops = { + .release = nullb_device_release, +}; + +static struct config_item_type nullb_device_type = { + .ct_item_ops = &nullb_device_ops, + .ct_attrs = nullb_device_attrs, + .ct_owner = THIS_MODULE, +}; + +static struct +config_item *nullb_group_make_item(struct config_group *group, const char *name) +{ + struct nullb_device *dev; + + dev = null_alloc_dev(); + if (!dev) + return ERR_PTR(-ENOMEM); + + config_item_init_type_name(&dev->item, name, &nullb_device_type); + + return &dev->item; +} + +static void +nullb_group_drop_item(struct config_group *group, struct config_item *item) +{ + struct nullb_device *dev = to_nullb_device(item); + + if (test_and_clear_bit(NULLB_DEV_FL_UP, &dev->flags)) { + mutex_lock(&lock); + dev->power = false; + null_del_dev(dev->nullb); + mutex_unlock(&lock); + } + + config_item_put(item); +} + +static ssize_t memb_group_features_show(struct config_item *item, char *page) +{ + return snprintf(page, PAGE_SIZE, "memory_backed,discard,bandwidth,cache,badblocks\n"); +} + +CONFIGFS_ATTR_RO(memb_group_, features); + +static struct configfs_attribute *nullb_group_attrs[] = { + &memb_group_attr_features, + NULL, +}; + +static struct configfs_group_operations nullb_group_ops = { + .make_item = nullb_group_make_item, + .drop_item = nullb_group_drop_item, +}; + +static struct config_item_type nullb_group_type = { + .ct_group_ops = &nullb_group_ops, + .ct_attrs = nullb_group_attrs, + .ct_owner = THIS_MODULE, +}; + +static struct configfs_subsystem nullb_subsys = { + .su_group = { + .cg_item = { + .ci_namebuf = "nullb", + .ci_type = &nullb_group_type, + }, + }, +}; + +static inline int null_cache_active(struct nullb *nullb) +{ + return test_bit(NULLB_DEV_FL_CACHE, &nullb->dev->flags); +} + +static struct nullb_device *null_alloc_dev(void) +{ + struct nullb_device *dev; + + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (!dev) + return NULL; + INIT_RADIX_TREE(&dev->data, GFP_ATOMIC); + INIT_RADIX_TREE(&dev->cache, GFP_ATOMIC); + if (badblocks_init(&dev->badblocks, 0)) { + kfree(dev); + return NULL; + } + + dev->size = g_gb * 1024; + dev->completion_nsec = g_completion_nsec; + dev->submit_queues = g_submit_queues; + dev->home_node = g_home_node; + dev->queue_mode = g_queue_mode; + dev->blocksize = g_bs; + dev->irqmode = g_irqmode; + dev->hw_queue_depth = g_hw_queue_depth; + dev->use_lightnvm = g_use_lightnvm; + dev->blocking = g_blocking; + dev->use_per_node_hctx = g_use_per_node_hctx; + return dev; +} + +static void null_free_dev(struct nullb_device *dev) +{ + kfree(dev); +} + static void put_tag(struct nullb_queue *nq, unsigned int tag) { clear_bit_unlock(tag, nq->tag_map); @@ -193,7 +619,7 @@ static struct nullb_cmd *__alloc_cmd(struct nullb_queue *nq) cmd = &nq->cmds[tag]; cmd->tag = tag; cmd->nq = nq; - if (irqmode == NULL_IRQ_TIMER) { + if (nq->dev->irqmode == NULL_IRQ_TIMER) { hrtimer_init(&cmd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); cmd->timer.function = null_cmd_timer_expired; @@ -229,19 +655,21 @@ static struct nullb_cmd *alloc_cmd(struct nullb_queue *nq, int can_wait) static void end_cmd(struct nullb_cmd *cmd) { struct request_queue *q = NULL; + int queue_mode = cmd->nq->dev->queue_mode; if (cmd->rq) q = cmd->rq->q; switch (queue_mode) { case NULL_Q_MQ: - blk_mq_end_request(cmd->rq, BLK_STS_OK); + blk_mq_end_request(cmd->rq, cmd->error); return; case NULL_Q_RQ: INIT_LIST_HEAD(&cmd->rq->queuelist); - blk_end_request_all(cmd->rq, BLK_STS_OK); + blk_end_request_all(cmd->rq, cmd->error); break; case NULL_Q_BIO: + cmd->bio->bi_status = cmd->error; bio_endio(cmd->bio); break; } @@ -267,25 +695,582 @@ static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer) static void null_cmd_end_timer(struct nullb_cmd *cmd) { - ktime_t kt = completion_nsec; + ktime_t kt = cmd->nq->dev->completion_nsec; hrtimer_start(&cmd->timer, kt, HRTIMER_MODE_REL); } static void null_softirq_done_fn(struct request *rq) { - if (queue_mode == NULL_Q_MQ) + struct nullb *nullb = rq->q->queuedata; + + if (nullb->dev->queue_mode == NULL_Q_MQ) end_cmd(blk_mq_rq_to_pdu(rq)); else end_cmd(rq->special); } -static inline void null_handle_cmd(struct nullb_cmd *cmd) +static struct nullb_page *null_alloc_page(gfp_t gfp_flags) +{ + struct nullb_page *t_page; + + t_page = kmalloc(sizeof(struct nullb_page), gfp_flags); + if (!t_page) + goto out; + + t_page->page = alloc_pages(gfp_flags, 0); + if (!t_page->page) + goto out_freepage; + + t_page->bitmap = 0; + return t_page; +out_freepage: + kfree(t_page); +out: + return NULL; +} + +static void null_free_page(struct nullb_page *t_page) +{ + __set_bit(NULLB_PAGE_FREE, &t_page->bitmap); + if (test_bit(NULLB_PAGE_LOCK, &t_page->bitmap)) + return; + __free_page(t_page->page); + kfree(t_page); +} + +static void null_free_sector(struct nullb *nullb, sector_t sector, + bool is_cache) +{ + unsigned int sector_bit; + u64 idx; + struct nullb_page *t_page, *ret; + struct radix_tree_root *root; + + root = is_cache ? &nullb->dev->cache : &nullb->dev->data; + idx = sector >> PAGE_SECTORS_SHIFT; + sector_bit = (sector & SECTOR_MASK); + + t_page = radix_tree_lookup(root, idx); + if (t_page) { + __clear_bit(sector_bit, &t_page->bitmap); + + if (!t_page->bitmap) { + ret = radix_tree_delete_item(root, idx, t_page); + WARN_ON(ret != t_page); + null_free_page(ret); + if (is_cache) + nullb->dev->curr_cache -= PAGE_SIZE; + } + } +} + +static struct nullb_page *null_radix_tree_insert(struct nullb *nullb, u64 idx, + struct nullb_page *t_page, bool is_cache) +{ + struct radix_tree_root *root; + + root = is_cache ? &nullb->dev->cache : &nullb->dev->data; + + if (radix_tree_insert(root, idx, t_page)) { + null_free_page(t_page); + t_page = radix_tree_lookup(root, idx); + WARN_ON(!t_page || t_page->page->index != idx); + } else if (is_cache) + nullb->dev->curr_cache += PAGE_SIZE; + + return t_page; +} + +static void null_free_device_storage(struct nullb_device *dev, bool is_cache) +{ + unsigned long pos = 0; + int nr_pages; + struct nullb_page *ret, *t_pages[FREE_BATCH]; + struct radix_tree_root *root; + + root = is_cache ? &dev->cache : &dev->data; + + do { + int i; + + nr_pages = radix_tree_gang_lookup(root, + (void **)t_pages, pos, FREE_BATCH); + + for (i = 0; i < nr_pages; i++) { + pos = t_pages[i]->page->index; + ret = radix_tree_delete_item(root, pos, t_pages[i]); + WARN_ON(ret != t_pages[i]); + null_free_page(ret); + } + + pos++; + } while (nr_pages == FREE_BATCH); + + if (is_cache) + dev->curr_cache = 0; +} + +static struct nullb_page *__null_lookup_page(struct nullb *nullb, + sector_t sector, bool for_write, bool is_cache) +{ + unsigned int sector_bit; + u64 idx; + struct nullb_page *t_page; + struct radix_tree_root *root; + + idx = sector >> PAGE_SECTORS_SHIFT; + sector_bit = (sector & SECTOR_MASK); + + root = is_cache ? &nullb->dev->cache : &nullb->dev->data; + t_page = radix_tree_lookup(root, idx); + WARN_ON(t_page && t_page->page->index != idx); + + if (t_page && (for_write || test_bit(sector_bit, &t_page->bitmap))) + return t_page; + + return NULL; +} + +static struct nullb_page *null_lookup_page(struct nullb *nullb, + sector_t sector, bool for_write, bool ignore_cache) +{ + struct nullb_page *page = NULL; + + if (!ignore_cache) + page = __null_lookup_page(nullb, sector, for_write, true); + if (page) + return page; + return __null_lookup_page(nullb, sector, for_write, false); +} + +static struct nullb_page *null_insert_page(struct nullb *nullb, + sector_t sector, bool ignore_cache) +{ + u64 idx; + struct nullb_page *t_page; + + t_page = null_lookup_page(nullb, sector, true, ignore_cache); + if (t_page) + return t_page; + + spin_unlock_irq(&nullb->lock); + + t_page = null_alloc_page(GFP_NOIO); + if (!t_page) + goto out_lock; + + if (radix_tree_preload(GFP_NOIO)) + goto out_freepage; + + spin_lock_irq(&nullb->lock); + idx = sector >> PAGE_SECTORS_SHIFT; + t_page->page->index = idx; + t_page = null_radix_tree_insert(nullb, idx, t_page, !ignore_cache); + radix_tree_preload_end(); + + return t_page; +out_freepage: + null_free_page(t_page); +out_lock: + spin_lock_irq(&nullb->lock); + return null_lookup_page(nullb, sector, true, ignore_cache); +} + +static int null_flush_cache_page(struct nullb *nullb, struct nullb_page *c_page) +{ + int i; + unsigned int offset; + u64 idx; + struct nullb_page *t_page, *ret; + void *dst, *src; + + idx = c_page->page->index; + + t_page = null_insert_page(nullb, idx << PAGE_SECTORS_SHIFT, true); + + __clear_bit(NULLB_PAGE_LOCK, &c_page->bitmap); + if (test_bit(NULLB_PAGE_FREE, &c_page->bitmap)) { + null_free_page(c_page); + if (t_page && t_page->bitmap == 0) { + ret = radix_tree_delete_item(&nullb->dev->data, + idx, t_page); + null_free_page(t_page); + } + return 0; + } + + if (!t_page) + return -ENOMEM; + + src = kmap_atomic(c_page->page); + dst = kmap_atomic(t_page->page); + + for (i = 0; i < PAGE_SECTORS; + i += (nullb->dev->blocksize >> SECTOR_SHIFT)) { + if (test_bit(i, &c_page->bitmap)) { + offset = (i << SECTOR_SHIFT); + memcpy(dst + offset, src + offset, + nullb->dev->blocksize); + __set_bit(i, &t_page->bitmap); + } + } + + kunmap_atomic(dst); + kunmap_atomic(src); + + ret = radix_tree_delete_item(&nullb->dev->cache, idx, c_page); + null_free_page(ret); + nullb->dev->curr_cache -= PAGE_SIZE; + + return 0; +} + +static int null_make_cache_space(struct nullb *nullb, unsigned long n) { + int i, err, nr_pages; + struct nullb_page *c_pages[FREE_BATCH]; + unsigned long flushed = 0, one_round; + +again: + if ((nullb->dev->cache_size * 1024 * 1024) > + nullb->dev->curr_cache + n || nullb->dev->curr_cache == 0) + return 0; + + nr_pages = radix_tree_gang_lookup(&nullb->dev->cache, + (void **)c_pages, nullb->cache_flush_pos, FREE_BATCH); + /* + * nullb_flush_cache_page could unlock before using the c_pages. To + * avoid race, we don't allow page free + */ + for (i = 0; i < nr_pages; i++) { + nullb->cache_flush_pos = c_pages[i]->page->index; + /* + * We found the page which is being flushed to disk by other + * threads + */ + if (test_bit(NULLB_PAGE_LOCK, &c_pages[i]->bitmap)) + c_pages[i] = NULL; + else + __set_bit(NULLB_PAGE_LOCK, &c_pages[i]->bitmap); + } + + one_round = 0; + for (i = 0; i < nr_pages; i++) { + if (c_pages[i] == NULL) + continue; + err = null_flush_cache_page(nullb, c_pages[i]); + if (err) + return err; + one_round++; + } + flushed += one_round << PAGE_SHIFT; + + if (n > flushed) { + if (nr_pages == 0) + nullb->cache_flush_pos = 0; + if (one_round == 0) { + /* give other threads a chance */ + spin_unlock_irq(&nullb->lock); + spin_lock_irq(&nullb->lock); + } + goto again; + } + return 0; +} + +static int copy_to_nullb(struct nullb *nullb, struct page *source, + unsigned int off, sector_t sector, size_t n, bool is_fua) +{ + size_t temp, count = 0; + unsigned int offset; + struct nullb_page *t_page; + void *dst, *src; + + while (count < n) { + temp = min_t(size_t, nullb->dev->blocksize, n - count); + + if (null_cache_active(nullb) && !is_fua) + null_make_cache_space(nullb, PAGE_SIZE); + + offset = (sector & SECTOR_MASK) << SECTOR_SHIFT; + t_page = null_insert_page(nullb, sector, + !null_cache_active(nullb) || is_fua); + if (!t_page) + return -ENOSPC; + + src = kmap_atomic(source); + dst = kmap_atomic(t_page->page); + memcpy(dst + offset, src + off + count, temp); + kunmap_atomic(dst); + kunmap_atomic(src); + + __set_bit(sector & SECTOR_MASK, &t_page->bitmap); + + if (is_fua) + null_free_sector(nullb, sector, true); + + count += temp; + sector += temp >> SECTOR_SHIFT; + } + return 0; +} + +static int copy_from_nullb(struct nullb *nullb, struct page *dest, + unsigned int off, sector_t sector, size_t n) +{ + size_t temp, count = 0; + unsigned int offset; + struct nullb_page *t_page; + void *dst, *src; + + while (count < n) { + temp = min_t(size_t, nullb->dev->blocksize, n - count); + + offset = (sector & SECTOR_MASK) << SECTOR_SHIFT; + t_page = null_lookup_page(nullb, sector, false, + !null_cache_active(nullb)); + + dst = kmap_atomic(dest); + if (!t_page) { + memset(dst + off + count, 0, temp); + goto next; + } + src = kmap_atomic(t_page->page); + memcpy(dst + off + count, src + offset, temp); + kunmap_atomic(src); +next: + kunmap_atomic(dst); + + count += temp; + sector += temp >> SECTOR_SHIFT; + } + return 0; +} + +static void null_handle_discard(struct nullb *nullb, sector_t sector, size_t n) +{ + size_t temp; + + spin_lock_irq(&nullb->lock); + while (n > 0) { + temp = min_t(size_t, n, nullb->dev->blocksize); + null_free_sector(nullb, sector, false); + if (null_cache_active(nullb)) + null_free_sector(nullb, sector, true); + sector += temp >> SECTOR_SHIFT; + n -= temp; + } + spin_unlock_irq(&nullb->lock); +} + +static int null_handle_flush(struct nullb *nullb) +{ + int err; + + if (!null_cache_active(nullb)) + return 0; + + spin_lock_irq(&nullb->lock); + while (true) { + err = null_make_cache_space(nullb, + nullb->dev->cache_size * 1024 * 1024); + if (err || nullb->dev->curr_cache == 0) + break; + } + + WARN_ON(!radix_tree_empty(&nullb->dev->cache)); + spin_unlock_irq(&nullb->lock); + return err; +} + +static int null_transfer(struct nullb *nullb, struct page *page, + unsigned int len, unsigned int off, bool is_write, sector_t sector, + bool is_fua) +{ + int err = 0; + + if (!is_write) { + err = copy_from_nullb(nullb, page, off, sector, len); + flush_dcache_page(page); + } else { + flush_dcache_page(page); + err = copy_to_nullb(nullb, page, off, sector, len, is_fua); + } + + return err; +} + +static int null_handle_rq(struct nullb_cmd *cmd) +{ + struct request *rq = cmd->rq; + struct nullb *nullb = cmd->nq->dev->nullb; + int err; + unsigned int len; + sector_t sector; + struct req_iterator iter; + struct bio_vec bvec; + + sector = blk_rq_pos(rq); + + if (req_op(rq) == REQ_OP_DISCARD) { + null_handle_discard(nullb, sector, blk_rq_bytes(rq)); + return 0; + } + + spin_lock_irq(&nullb->lock); + rq_for_each_segment(bvec, rq, iter) { + len = bvec.bv_len; + err = null_transfer(nullb, bvec.bv_page, len, bvec.bv_offset, + op_is_write(req_op(rq)), sector, + req_op(rq) & REQ_FUA); + if (err) { + spin_unlock_irq(&nullb->lock); + return err; + } + sector += len >> SECTOR_SHIFT; + } + spin_unlock_irq(&nullb->lock); + + return 0; +} + +static int null_handle_bio(struct nullb_cmd *cmd) +{ + struct bio *bio = cmd->bio; + struct nullb *nullb = cmd->nq->dev->nullb; + int err; + unsigned int len; + sector_t sector; + struct bio_vec bvec; + struct bvec_iter iter; + + sector = bio->bi_iter.bi_sector; + + if (bio_op(bio) == REQ_OP_DISCARD) { + null_handle_discard(nullb, sector, + bio_sectors(bio) << SECTOR_SHIFT); + return 0; + } + + spin_lock_irq(&nullb->lock); + bio_for_each_segment(bvec, bio, iter) { + len = bvec.bv_len; + err = null_transfer(nullb, bvec.bv_page, len, bvec.bv_offset, + op_is_write(bio_op(bio)), sector, + bio_op(bio) & REQ_FUA); + if (err) { + spin_unlock_irq(&nullb->lock); + return err; + } + sector += len >> SECTOR_SHIFT; + } + spin_unlock_irq(&nullb->lock); + return 0; +} + +static void null_stop_queue(struct nullb *nullb) +{ + struct request_queue *q = nullb->q; + + if (nullb->dev->queue_mode == NULL_Q_MQ) + blk_mq_stop_hw_queues(q); + else { + spin_lock_irq(q->queue_lock); + blk_stop_queue(q); + spin_unlock_irq(q->queue_lock); + } +} + +static void null_restart_queue_async(struct nullb *nullb) +{ + struct request_queue *q = nullb->q; + unsigned long flags; + + if (nullb->dev->queue_mode == NULL_Q_MQ) + blk_mq_start_stopped_hw_queues(q, true); + else { + spin_lock_irqsave(q->queue_lock, flags); + blk_start_queue_async(q); + spin_unlock_irqrestore(q->queue_lock, flags); + } +} + +static blk_status_t null_handle_cmd(struct nullb_cmd *cmd) +{ + struct nullb_device *dev = cmd->nq->dev; + struct nullb *nullb = dev->nullb; + int err = 0; + + if (test_bit(NULLB_DEV_FL_THROTTLED, &dev->flags)) { + struct request *rq = cmd->rq; + + if (!hrtimer_active(&nullb->bw_timer)) + hrtimer_restart(&nullb->bw_timer); + + if (atomic_long_sub_return(blk_rq_bytes(rq), + &nullb->cur_bytes) < 0) { + null_stop_queue(nullb); + /* race with timer */ + if (atomic_long_read(&nullb->cur_bytes) > 0) + null_restart_queue_async(nullb); + if (dev->queue_mode == NULL_Q_RQ) { + struct request_queue *q = nullb->q; + + spin_lock_irq(q->queue_lock); + rq->rq_flags |= RQF_DONTPREP; + blk_requeue_request(q, rq); + spin_unlock_irq(q->queue_lock); + return BLK_STS_OK; + } else + /* requeue request */ + return BLK_STS_RESOURCE; + } + } + + if (nullb->dev->badblocks.shift != -1) { + int bad_sectors; + sector_t sector, size, first_bad; + bool is_flush = true; + + if (dev->queue_mode == NULL_Q_BIO && + bio_op(cmd->bio) != REQ_OP_FLUSH) { + is_flush = false; + sector = cmd->bio->bi_iter.bi_sector; + size = bio_sectors(cmd->bio); + } + if (dev->queue_mode != NULL_Q_BIO && + req_op(cmd->rq) != REQ_OP_FLUSH) { + is_flush = false; + sector = blk_rq_pos(cmd->rq); + size = blk_rq_sectors(cmd->rq); + } + if (!is_flush && badblocks_check(&nullb->dev->badblocks, sector, + size, &first_bad, &bad_sectors)) { + cmd->error = BLK_STS_IOERR; + goto out; + } + } + + if (dev->memory_backed) { + if (dev->queue_mode == NULL_Q_BIO) { + if (bio_op(cmd->bio) == REQ_OP_FLUSH) + err = null_handle_flush(nullb); + else + err = null_handle_bio(cmd); + } else { + if (req_op(cmd->rq) == REQ_OP_FLUSH) + err = null_handle_flush(nullb); + else + err = null_handle_rq(cmd); + } + } + cmd->error = errno_to_blk_status(err); +out: /* Complete IO by inline, softirq or timer */ - switch (irqmode) { + switch (dev->irqmode) { case NULL_IRQ_SOFTIRQ: - switch (queue_mode) { + switch (dev->queue_mode) { case NULL_Q_MQ: blk_mq_complete_request(cmd->rq); break; @@ -307,6 +1292,34 @@ static inline void null_handle_cmd(struct nullb_cmd *cmd) null_cmd_end_timer(cmd); break; } + return BLK_STS_OK; +} + +static enum hrtimer_restart nullb_bwtimer_fn(struct hrtimer *timer) +{ + struct nullb *nullb = container_of(timer, struct nullb, bw_timer); + ktime_t timer_interval = ktime_set(0, TIMER_INTERVAL); + unsigned int mbps = nullb->dev->mbps; + + if (atomic_long_read(&nullb->cur_bytes) == mb_per_tick(mbps)) + return HRTIMER_NORESTART; + + atomic_long_set(&nullb->cur_bytes, mb_per_tick(mbps)); + null_restart_queue_async(nullb); + + hrtimer_forward_now(&nullb->bw_timer, timer_interval); + + return HRTIMER_RESTART; +} + +static void nullb_setup_bwtimer(struct nullb *nullb) +{ + ktime_t timer_interval = ktime_set(0, TIMER_INTERVAL); + + hrtimer_init(&nullb->bw_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + nullb->bw_timer.function = nullb_bwtimer_fn; + atomic_long_set(&nullb->cur_bytes, mb_per_tick(nullb->dev->mbps)); + hrtimer_start(&nullb->bw_timer, timer_interval, HRTIMER_MODE_REL); } static struct nullb_queue *nullb_to_queue(struct nullb *nullb) @@ -366,20 +1379,20 @@ static blk_status_t null_queue_rq(struct blk_mq_hw_ctx *hctx, const struct blk_mq_queue_data *bd) { struct nullb_cmd *cmd = blk_mq_rq_to_pdu(bd->rq); + struct nullb_queue *nq = hctx->driver_data; might_sleep_if(hctx->flags & BLK_MQ_F_BLOCKING); - if (irqmode == NULL_IRQ_TIMER) { + if (nq->dev->irqmode == NULL_IRQ_TIMER) { hrtimer_init(&cmd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); cmd->timer.function = null_cmd_timer_expired; } cmd->rq = bd->rq; - cmd->nq = hctx->driver_data; + cmd->nq = nq; blk_mq_start_request(bd->rq); - null_handle_cmd(cmd); - return BLK_STS_OK; + return null_handle_cmd(cmd); } static const struct blk_mq_ops null_mq_ops = { @@ -438,7 +1451,8 @@ static int null_lnvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd) static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id) { - sector_t size = gb * 1024 * 1024 * 1024ULL; + struct nullb *nullb = dev->q->queuedata; + sector_t size = (sector_t)nullb->dev->size * 1024 * 1024ULL; sector_t blksize; struct nvm_id_group *grp; @@ -460,7 +1474,7 @@ static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id) id->ppaf.ch_offset = 56; id->ppaf.ch_len = 8; - sector_div(size, bs); /* convert size to pages */ + sector_div(size, nullb->dev->blocksize); /* convert size to pages */ size >>= 8; /* concert size to pgs pr blk */ grp = &id->grp; grp->mtype = 0; @@ -474,8 +1488,8 @@ static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id) grp->num_blk = blksize; grp->num_pln = 1; - grp->fpg_sz = bs; - grp->csecs = bs; + grp->fpg_sz = nullb->dev->blocksize; + grp->csecs = nullb->dev->blocksize; grp->trdt = 25000; grp->trdm = 25000; grp->tprt = 500000; @@ -483,7 +1497,7 @@ static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id) grp->tbet = 1500000; grp->tbem = 1500000; grp->mpos = 0x010101; /* single plane rwe */ - grp->cpar = hw_queue_depth; + grp->cpar = nullb->dev->hw_queue_depth; return 0; } @@ -568,19 +1582,44 @@ static void null_nvm_unregister(struct nullb *nullb) {} static void null_del_dev(struct nullb *nullb) { + struct nullb_device *dev = nullb->dev; + + ida_simple_remove(&nullb_indexes, nullb->index); + list_del_init(&nullb->list); - if (use_lightnvm) + if (dev->use_lightnvm) null_nvm_unregister(nullb); else del_gendisk(nullb->disk); + + if (test_bit(NULLB_DEV_FL_THROTTLED, &nullb->dev->flags)) { + hrtimer_cancel(&nullb->bw_timer); + atomic_long_set(&nullb->cur_bytes, LONG_MAX); + null_restart_queue_async(nullb); + } + blk_cleanup_queue(nullb->q); - if (queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set) + if (dev->queue_mode == NULL_Q_MQ && + nullb->tag_set == &nullb->__tag_set) blk_mq_free_tag_set(nullb->tag_set); - if (!use_lightnvm) + if (!dev->use_lightnvm) put_disk(nullb->disk); cleanup_queues(nullb); + if (null_cache_active(nullb)) + null_free_device_storage(nullb->dev, true); kfree(nullb); + dev->nullb = NULL; +} + +static void null_config_discard(struct nullb *nullb) +{ + if (nullb->dev->discard == false) + return; + nullb->q->limits.discard_granularity = nullb->dev->blocksize; + nullb->q->limits.discard_alignment = nullb->dev->blocksize; + blk_queue_max_discard_sectors(nullb->q, UINT_MAX >> 9); + queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, nullb->q); } static int null_open(struct block_device *bdev, fmode_t mode) @@ -605,6 +1644,7 @@ static void null_init_queue(struct nullb *nullb, struct nullb_queue *nq) init_waitqueue_head(&nq->wait); nq->queue_depth = nullb->queue_depth; + nq->dev = nullb->dev; } static void null_init_queues(struct nullb *nullb) @@ -652,13 +1692,13 @@ static int setup_commands(struct nullb_queue *nq) static int setup_queues(struct nullb *nullb) { - nullb->queues = kzalloc(submit_queues * sizeof(struct nullb_queue), - GFP_KERNEL); + nullb->queues = kzalloc(nullb->dev->submit_queues * + sizeof(struct nullb_queue), GFP_KERNEL); if (!nullb->queues) return -ENOMEM; nullb->nr_queues = 0; - nullb->queue_depth = hw_queue_depth; + nullb->queue_depth = nullb->dev->hw_queue_depth; return 0; } @@ -668,7 +1708,7 @@ static int init_driver_queues(struct nullb *nullb) struct nullb_queue *nq; int i, ret = 0; - for (i = 0; i < submit_queues; i++) { + for (i = 0; i < nullb->dev->submit_queues; i++) { nq = &nullb->queues[i]; null_init_queue(nullb, nq); @@ -686,10 +1726,10 @@ static int null_gendisk_register(struct nullb *nullb) struct gendisk *disk; sector_t size; - disk = nullb->disk = alloc_disk_node(1, home_node); + disk = nullb->disk = alloc_disk_node(1, nullb->dev->home_node); if (!disk) return -ENOMEM; - size = gb * 1024 * 1024 * 1024ULL; + size = (sector_t)nullb->dev->size * 1024 * 1024ULL; set_capacity(disk, size >> 9); disk->flags |= GENHD_FL_EXT_DEVT | GENHD_FL_SUPPRESS_PARTITION_INFO; @@ -704,49 +1744,86 @@ static int null_gendisk_register(struct nullb *nullb) return 0; } -static int null_init_tag_set(struct blk_mq_tag_set *set) +static int null_init_tag_set(struct nullb *nullb, struct blk_mq_tag_set *set) { set->ops = &null_mq_ops; - set->nr_hw_queues = submit_queues; - set->queue_depth = hw_queue_depth; - set->numa_node = home_node; + set->nr_hw_queues = nullb ? nullb->dev->submit_queues : + g_submit_queues; + set->queue_depth = nullb ? nullb->dev->hw_queue_depth : + g_hw_queue_depth; + set->numa_node = nullb ? nullb->dev->home_node : g_home_node; set->cmd_size = sizeof(struct nullb_cmd); set->flags = BLK_MQ_F_SHOULD_MERGE; set->driver_data = NULL; - if (blocking) + if ((nullb && nullb->dev->blocking) || g_blocking) set->flags |= BLK_MQ_F_BLOCKING; return blk_mq_alloc_tag_set(set); } -static int null_add_dev(void) +static void null_validate_conf(struct nullb_device *dev) +{ + dev->blocksize = round_down(dev->blocksize, 512); + dev->blocksize = clamp_t(unsigned int, dev->blocksize, 512, 4096); + if (dev->use_lightnvm && dev->blocksize != 4096) + dev->blocksize = 4096; + + if (dev->use_lightnvm && dev->queue_mode != NULL_Q_MQ) + dev->queue_mode = NULL_Q_MQ; + + if (dev->queue_mode == NULL_Q_MQ && dev->use_per_node_hctx) { + if (dev->submit_queues != nr_online_nodes) + dev->submit_queues = nr_online_nodes; + } else if (dev->submit_queues > nr_cpu_ids) + dev->submit_queues = nr_cpu_ids; + else if (dev->submit_queues == 0) + dev->submit_queues = 1; + + dev->queue_mode = min_t(unsigned int, dev->queue_mode, NULL_Q_MQ); + dev->irqmode = min_t(unsigned int, dev->irqmode, NULL_IRQ_TIMER); + + /* Do memory allocation, so set blocking */ + if (dev->memory_backed) + dev->blocking = true; + else /* cache is meaningless */ + dev->cache_size = 0; + dev->cache_size = min_t(unsigned long, ULONG_MAX / 1024 / 1024, + dev->cache_size); + dev->mbps = min_t(unsigned int, 1024 * 40, dev->mbps); + /* can not stop a queue */ + if (dev->queue_mode == NULL_Q_BIO) + dev->mbps = 0; +} + +static int null_add_dev(struct nullb_device *dev) { struct nullb *nullb; int rv; - nullb = kzalloc_node(sizeof(*nullb), GFP_KERNEL, home_node); + null_validate_conf(dev); + + nullb = kzalloc_node(sizeof(*nullb), GFP_KERNEL, dev->home_node); if (!nullb) { rv = -ENOMEM; goto out; } + nullb->dev = dev; + dev->nullb = nullb; spin_lock_init(&nullb->lock); - if (queue_mode == NULL_Q_MQ && use_per_node_hctx) - submit_queues = nr_online_nodes; - rv = setup_queues(nullb); if (rv) goto out_free_nullb; - if (queue_mode == NULL_Q_MQ) { + if (dev->queue_mode == NULL_Q_MQ) { if (shared_tags) { nullb->tag_set = &tag_set; rv = 0; } else { nullb->tag_set = &nullb->__tag_set; - rv = null_init_tag_set(nullb->tag_set); + rv = null_init_tag_set(nullb, nullb->tag_set); } if (rv) @@ -758,8 +1835,8 @@ static int null_add_dev(void) goto out_cleanup_tags; } null_init_queues(nullb); - } else if (queue_mode == NULL_Q_BIO) { - nullb->q = blk_alloc_queue_node(GFP_KERNEL, home_node); + } else if (dev->queue_mode == NULL_Q_BIO) { + nullb->q = blk_alloc_queue_node(GFP_KERNEL, dev->home_node); if (!nullb->q) { rv = -ENOMEM; goto out_cleanup_queues; @@ -769,7 +1846,8 @@ static int null_add_dev(void) if (rv) goto out_cleanup_blk_queue; } else { - nullb->q = blk_init_queue_node(null_request_fn, &nullb->lock, home_node); + nullb->q = blk_init_queue_node(null_request_fn, &nullb->lock, + dev->home_node); if (!nullb->q) { rv = -ENOMEM; goto out_cleanup_queues; @@ -781,20 +1859,34 @@ static int null_add_dev(void) goto out_cleanup_blk_queue; } + if (dev->mbps) { + set_bit(NULLB_DEV_FL_THROTTLED, &dev->flags); + nullb_setup_bwtimer(nullb); + } + + if (dev->cache_size > 0) { + set_bit(NULLB_DEV_FL_CACHE, &nullb->dev->flags); + blk_queue_write_cache(nullb->q, true, true); + blk_queue_flush_queueable(nullb->q, true); + } + nullb->q->queuedata = nullb; queue_flag_set_unlocked(QUEUE_FLAG_NONROT, nullb->q); queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, nullb->q); mutex_lock(&lock); - nullb->index = nullb_indexes++; + nullb->index = ida_simple_get(&nullb_indexes, 0, 0, GFP_KERNEL); + dev->index = nullb->index; mutex_unlock(&lock); - blk_queue_logical_block_size(nullb->q, bs); - blk_queue_physical_block_size(nullb->q, bs); + blk_queue_logical_block_size(nullb->q, dev->blocksize); + blk_queue_physical_block_size(nullb->q, dev->blocksize); + + null_config_discard(nullb); sprintf(nullb->disk_name, "nullb%d", nullb->index); - if (use_lightnvm) + if (dev->use_lightnvm) rv = null_nvm_register(nullb); else rv = null_gendisk_register(nullb); @@ -810,7 +1902,7 @@ static int null_add_dev(void) out_cleanup_blk_queue: blk_cleanup_queue(nullb->q); out_cleanup_tags: - if (queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set) + if (dev->queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set) blk_mq_free_tag_set(nullb->tag_set); out_cleanup_queues: cleanup_queues(nullb); @@ -825,51 +1917,63 @@ static int __init null_init(void) int ret = 0; unsigned int i; struct nullb *nullb; + struct nullb_device *dev; + + /* check for nullb_page.bitmap */ + if (sizeof(unsigned long) * 8 - 2 < (PAGE_SIZE >> SECTOR_SHIFT)) + return -EINVAL; - if (bs > PAGE_SIZE) { + if (g_bs > PAGE_SIZE) { pr_warn("null_blk: invalid block size\n"); pr_warn("null_blk: defaults block size to %lu\n", PAGE_SIZE); - bs = PAGE_SIZE; + g_bs = PAGE_SIZE; } - if (use_lightnvm && bs != 4096) { + if (g_use_lightnvm && g_bs != 4096) { pr_warn("null_blk: LightNVM only supports 4k block size\n"); pr_warn("null_blk: defaults block size to 4k\n"); - bs = 4096; + g_bs = 4096; } - if (use_lightnvm && queue_mode != NULL_Q_MQ) { + if (g_use_lightnvm && g_queue_mode != NULL_Q_MQ) { pr_warn("null_blk: LightNVM only supported for blk-mq\n"); pr_warn("null_blk: defaults queue mode to blk-mq\n"); - queue_mode = NULL_Q_MQ; + g_queue_mode = NULL_Q_MQ; } - if (queue_mode == NULL_Q_MQ && use_per_node_hctx) { - if (submit_queues < nr_online_nodes) { - pr_warn("null_blk: submit_queues param is set to %u.", + if (g_queue_mode == NULL_Q_MQ && g_use_per_node_hctx) { + if (g_submit_queues != nr_online_nodes) { + pr_warn("null_blk: submit_queues param is set to %u.\n", nr_online_nodes); - submit_queues = nr_online_nodes; + g_submit_queues = nr_online_nodes; } - } else if (submit_queues > nr_cpu_ids) - submit_queues = nr_cpu_ids; - else if (!submit_queues) - submit_queues = 1; + } else if (g_submit_queues > nr_cpu_ids) + g_submit_queues = nr_cpu_ids; + else if (g_submit_queues <= 0) + g_submit_queues = 1; - if (queue_mode == NULL_Q_MQ && shared_tags) { - ret = null_init_tag_set(&tag_set); + if (g_queue_mode == NULL_Q_MQ && shared_tags) { + ret = null_init_tag_set(NULL, &tag_set); if (ret) return ret; } + config_group_init(&nullb_subsys.su_group); + mutex_init(&nullb_subsys.su_mutex); + + ret = configfs_register_subsystem(&nullb_subsys); + if (ret) + goto err_tagset; + mutex_init(&lock); null_major = register_blkdev(0, "nullb"); if (null_major < 0) { ret = null_major; - goto err_tagset; + goto err_conf; } - if (use_lightnvm) { + if (g_use_lightnvm) { ppa_cache = kmem_cache_create("ppa_cache", 64 * sizeof(u64), 0, 0, NULL); if (!ppa_cache) { @@ -880,9 +1984,14 @@ static int __init null_init(void) } for (i = 0; i < nr_devices; i++) { - ret = null_add_dev(); - if (ret) + dev = null_alloc_dev(); + if (!dev) + goto err_dev; + ret = null_add_dev(dev); + if (ret) { + null_free_dev(dev); goto err_dev; + } } pr_info("null: module loaded\n"); @@ -891,13 +2000,17 @@ static int __init null_init(void) err_dev: while (!list_empty(&nullb_list)) { nullb = list_entry(nullb_list.next, struct nullb, list); + dev = nullb->dev; null_del_dev(nullb); + null_free_dev(dev); } kmem_cache_destroy(ppa_cache); err_ppa: unregister_blkdev(null_major, "nullb"); +err_conf: + configfs_unregister_subsystem(&nullb_subsys); err_tagset: - if (queue_mode == NULL_Q_MQ && shared_tags) + if (g_queue_mode == NULL_Q_MQ && shared_tags) blk_mq_free_tag_set(&tag_set); return ret; } @@ -906,16 +2019,22 @@ static void __exit null_exit(void) { struct nullb *nullb; + configfs_unregister_subsystem(&nullb_subsys); + unregister_blkdev(null_major, "nullb"); mutex_lock(&lock); while (!list_empty(&nullb_list)) { + struct nullb_device *dev; + nullb = list_entry(nullb_list.next, struct nullb, list); + dev = nullb->dev; null_del_dev(nullb); + null_free_dev(dev); } mutex_unlock(&lock); - if (queue_mode == NULL_Q_MQ && shared_tags) + if (g_queue_mode == NULL_Q_MQ && shared_tags) blk_mq_free_tag_set(&tag_set); kmem_cache_destroy(ppa_cache); @@ -924,5 +2043,5 @@ static void __exit null_exit(void) module_init(null_init); module_exit(null_exit); -MODULE_AUTHOR("Jens Axboe <jaxboe@fusionio.com>"); +MODULE_AUTHOR("Jens Axboe <axboe@kernel.dk>"); MODULE_LICENSE("GPL"); diff --git a/drivers/block/pktcdvd.c b/drivers/block/pktcdvd.c index 6b8b097abbb9..67974796c350 100644 --- a/drivers/block/pktcdvd.c +++ b/drivers/block/pktcdvd.c @@ -1028,7 +1028,7 @@ static void pkt_gather_data(struct pktcdvd_device *pd, struct packet_data *pkt) bio = pkt->r_bios[f]; bio_reset(bio); bio->bi_iter.bi_sector = pkt->sector + f * (CD_FRAMESIZE >> 9); - bio->bi_bdev = pd->bdev; + bio_set_dev(bio, pd->bdev); bio->bi_end_io = pkt_end_io_read; bio->bi_private = pkt; @@ -1122,7 +1122,7 @@ static int pkt_start_recovery(struct packet_data *pkt) pkt->sector = new_sector; bio_reset(pkt->bio); - pkt->bio->bi_bdev = pd->bdev; + bio_set_set(pkt->bio, pd->bdev); bio_set_op_attrs(pkt->bio, REQ_OP_WRITE, 0); pkt->bio->bi_iter.bi_sector = new_sector; pkt->bio->bi_iter.bi_size = pkt->frames * CD_FRAMESIZE; @@ -1267,7 +1267,7 @@ static void pkt_start_write(struct pktcdvd_device *pd, struct packet_data *pkt) bio_reset(pkt->w_bio); pkt->w_bio->bi_iter.bi_sector = pkt->sector; - pkt->w_bio->bi_bdev = pd->bdev; + bio_set_dev(pkt->w_bio, pd->bdev); pkt->w_bio->bi_end_io = pkt_end_io_packet_write; pkt->w_bio->bi_private = pkt; @@ -2314,7 +2314,7 @@ static void pkt_make_request_read(struct pktcdvd_device *pd, struct bio *bio) psd->pd = pd; psd->bio = bio; - cloned_bio->bi_bdev = pd->bdev; + bio_set_dev(cloned_bio, pd->bdev); cloned_bio->bi_private = psd; cloned_bio->bi_end_io = pkt_end_io_read_cloned; pd->stats.secs_r += bio_sectors(bio); @@ -2415,8 +2415,7 @@ static blk_qc_t pkt_make_request(struct request_queue *q, struct bio *bio) pd = q->queuedata; if (!pd) { - pr_err("%s incorrect request queue\n", - bdevname(bio->bi_bdev, b)); + pr_err("%s incorrect request queue\n", bio_devname(bio, b)); goto end_io; } diff --git a/drivers/block/ps3vram.c b/drivers/block/ps3vram.c index e0e81cacd781..6a55959cbf78 100644 --- a/drivers/block/ps3vram.c +++ b/drivers/block/ps3vram.c @@ -409,10 +409,8 @@ static int ps3vram_cache_init(struct ps3_system_bus_device *dev) priv->cache.page_size = CACHE_PAGE_SIZE; priv->cache.tags = kzalloc(sizeof(struct ps3vram_tag) * CACHE_PAGE_COUNT, GFP_KERNEL); - if (priv->cache.tags == NULL) { - dev_err(&dev->core, "Could not allocate cache tags\n"); + if (!priv->cache.tags) return -ENOMEM; - } dev_info(&dev->core, "Created ram cache: %d entries, %d KiB each\n", CACHE_PAGE_COUNT, CACHE_PAGE_SIZE / 1024); @@ -743,7 +741,11 @@ static int ps3vram_probe(struct ps3_system_bus_device *dev) goto out_unmap_reports; } - ps3vram_cache_init(dev); + error = ps3vram_cache_init(dev); + if (error < 0) { + goto out_unmap_reports; + } + ps3vram_proc_init(dev); queue = blk_alloc_queue(GFP_KERNEL); diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index b008b6a98098..b640ad8a6d20 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -3435,7 +3435,7 @@ static void rbd_acquire_lock(struct work_struct *work) struct rbd_device *rbd_dev = container_of(to_delayed_work(work), struct rbd_device, lock_dwork); enum rbd_lock_state lock_state; - int ret; + int ret = 0; dout("%s rbd_dev %p\n", __func__, rbd_dev); again: diff --git a/drivers/block/rsxx/dev.c b/drivers/block/rsxx/dev.c index 7f4acebf4657..e397d3ee7308 100644 --- a/drivers/block/rsxx/dev.c +++ b/drivers/block/rsxx/dev.c @@ -112,7 +112,7 @@ static const struct block_device_operations rsxx_fops = { static void disk_stats_start(struct rsxx_cardinfo *card, struct bio *bio) { - generic_start_io_acct(bio_data_dir(bio), bio_sectors(bio), + generic_start_io_acct(card->queue, bio_data_dir(bio), bio_sectors(bio), &card->gendisk->part0); } @@ -120,8 +120,8 @@ static void disk_stats_complete(struct rsxx_cardinfo *card, struct bio *bio, unsigned long start_time) { - generic_end_io_acct(bio_data_dir(bio), &card->gendisk->part0, - start_time); + generic_end_io_acct(card->queue, bio_data_dir(bio), + &card->gendisk->part0, start_time); } static void bio_dma_done_cb(struct rsxx_cardinfo *card, diff --git a/drivers/block/skd_main.c b/drivers/block/skd_main.c index d0368682bd43..7cedb4295e9d 100644 --- a/drivers/block/skd_main.c +++ b/drivers/block/skd_main.c @@ -1,19 +1,12 @@ -/* Copyright 2012 STEC, Inc. +/* + * Driver for sTec s1120 PCIe SSDs. sTec was acquired in 2013 by HGST and HGST + * was acquired by Western Digital in 2012. + * + * Copyright 2012 sTec, Inc. + * Copyright (c) 2017 Western Digital Corporation or its affiliates. * - * This file is licensed under the terms of the 3-clause - * BSD License (http://opensource.org/licenses/BSD-3-Clause) - * or the GNU GPL-2.0 (http://www.gnu.org/licenses/gpl-2.0.html), - * at your option. Both licenses are also available in the LICENSE file - * distributed with this project. This file may not be copied, modified, - * or distributed except in accordance with those terms. - * Gordoni Waidhofer <gwaidhofer@stec-inc.com> - * Initial Driver Design! - * Thomas Swann <tswann@stec-inc.com> - * Interrupt handling. - * Ramprasad Chinthekindi <rchinthekindi@stec-inc.com> - * biomode implementation. - * Akhil Bhansali <abhansali@stec-inc.com> - * Added support for DISCARD / FLUSH and FUA. + * This file is part of the Linux kernel, and is made available under + * the terms of the GNU General Public License version 2. */ #include <linux/kernel.h> @@ -23,11 +16,11 @@ #include <linux/slab.h> #include <linux/spinlock.h> #include <linux/blkdev.h> +#include <linux/blk-mq.h> #include <linux/sched.h> #include <linux/interrupt.h> #include <linux/compiler.h> #include <linux/workqueue.h> -#include <linux/bitops.h> #include <linux/delay.h> #include <linux/time.h> #include <linux/hdreg.h> @@ -37,9 +30,9 @@ #include <linux/version.h> #include <linux/err.h> #include <linux/aer.h> -#include <linux/ctype.h> #include <linux/wait.h> -#include <linux/uio.h> +#include <linux/stringify.h> +#include <linux/slab_def.h> #include <scsi/scsi.h> #include <scsi/sg.h> #include <linux/io.h> @@ -51,19 +44,6 @@ static int skd_dbg_level; static int skd_isr_comp_limit = 4; -enum { - STEC_LINK_2_5GTS = 0, - STEC_LINK_5GTS = 1, - STEC_LINK_8GTS = 2, - STEC_LINK_UNKNOWN = 0xFF -}; - -enum { - SKD_FLUSH_INITIALIZER, - SKD_FLUSH_ZERO_SIZE_FIRST, - SKD_FLUSH_DATA_SECOND, -}; - #define SKD_ASSERT(expr) \ do { \ if (unlikely(!(expr))) { \ @@ -73,17 +53,11 @@ enum { } while (0) #define DRV_NAME "skd" -#define DRV_VERSION "2.2.1" -#define DRV_BUILD_ID "0260" #define PFX DRV_NAME ": " -#define DRV_BIN_VERSION 0x100 -#define DRV_VER_COMPL "2.2.1." DRV_BUILD_ID -MODULE_AUTHOR("bug-reports: support@stec-inc.com"); -MODULE_LICENSE("Dual BSD/GPL"); +MODULE_LICENSE("GPL"); -MODULE_DESCRIPTION("STEC s1120 PCIe SSD block driver (b" DRV_BUILD_ID ")"); -MODULE_VERSION(DRV_VERSION "-" DRV_BUILD_ID); +MODULE_DESCRIPTION("STEC s1120 PCIe SSD block driver"); #define PCI_VENDOR_ID_STEC 0x1B39 #define PCI_DEVICE_ID_S1120 0x0001 @@ -96,34 +70,32 @@ MODULE_VERSION(DRV_VERSION "-" DRV_BUILD_ID); #define SKD_PAUSE_TIMEOUT (5 * 1000) #define SKD_N_FITMSG_BYTES (512u) +#define SKD_MAX_REQ_PER_MSG 14 -#define SKD_N_SPECIAL_CONTEXT 32u #define SKD_N_SPECIAL_FITMSG_BYTES (128u) /* SG elements are 32 bytes, so we can make this 4096 and still be under the * 128KB limit. That allows 4096*4K = 16M xfer size */ #define SKD_N_SG_PER_REQ_DEFAULT 256u -#define SKD_N_SG_PER_SPECIAL 256u #define SKD_N_COMPLETION_ENTRY 256u #define SKD_N_READ_CAP_BYTES (8u) #define SKD_N_INTERNAL_BYTES (512u) +#define SKD_SKCOMP_SIZE \ + ((sizeof(struct fit_completion_entry_v1) + \ + sizeof(struct fit_comp_error_info)) * SKD_N_COMPLETION_ENTRY) + /* 5 bits of uniqifier, 0xF800 */ -#define SKD_ID_INCR (0x400) #define SKD_ID_TABLE_MASK (3u << 8u) #define SKD_ID_RW_REQUEST (0u << 8u) #define SKD_ID_INTERNAL (1u << 8u) -#define SKD_ID_SPECIAL_REQUEST (2u << 8u) #define SKD_ID_FIT_MSG (3u << 8u) #define SKD_ID_SLOT_MASK 0x00FFu #define SKD_ID_SLOT_AND_TABLE_MASK 0x03FFu -#define SKD_N_TIMEOUT_SLOT 4u -#define SKD_TIMEOUT_SLOT_MASK 3u - #define SKD_N_MAX_SECTORS 2048u #define SKD_MAX_RETRIES 2u @@ -141,7 +113,6 @@ enum skd_drvr_state { SKD_DRVR_STATE_ONLINE, SKD_DRVR_STATE_PAUSING, SKD_DRVR_STATE_PAUSED, - SKD_DRVR_STATE_DRAINING_TIMEOUT, SKD_DRVR_STATE_RESTARTING, SKD_DRVR_STATE_RESUMING, SKD_DRVR_STATE_STOPPING, @@ -158,7 +129,6 @@ enum skd_drvr_state { #define SKD_WAIT_BOOT_TIMO SKD_TIMER_SECONDS(90u) #define SKD_STARTING_TIMO SKD_TIMER_SECONDS(8u) #define SKD_RESTARTING_TIMO SKD_TIMER_MINUTES(4u) -#define SKD_DRAINING_TIMO SKD_TIMER_SECONDS(6u) #define SKD_BUSY_TIMO SKD_TIMER_MINUTES(20u) #define SKD_STARTED_BUSY_TIMO SKD_TIMER_SECONDS(60u) #define SKD_START_WAIT_SECONDS 90u @@ -169,12 +139,6 @@ enum skd_req_state { SKD_REQ_STATE_BUSY, SKD_REQ_STATE_COMPLETED, SKD_REQ_STATE_TIMEOUT, - SKD_REQ_STATE_ABORTED, -}; - -enum skd_fit_msg_state { - SKD_MSG_STATE_IDLE, - SKD_MSG_STATE_BUSY, }; enum skd_check_status_action { @@ -185,34 +149,29 @@ enum skd_check_status_action { SKD_CHECK_STATUS_BUSY_IMMINENT, }; -struct skd_fitmsg_context { - enum skd_fit_msg_state state; - - struct skd_fitmsg_context *next; +struct skd_msg_buf { + struct fit_msg_hdr fmh; + struct skd_scsi_request scsi[SKD_MAX_REQ_PER_MSG]; +}; +struct skd_fitmsg_context { u32 id; - u16 outstanding; u32 length; - u32 offset; - u8 *msg_buf; + struct skd_msg_buf *msg_buf; dma_addr_t mb_dma_address; }; struct skd_request_context { enum skd_req_state state; - struct skd_request_context *next; - u16 id; u32 fitmsg_id; - struct request *req; u8 flush_cmd; - u32 timeout_stamp; - u8 sg_data_dir; + enum dma_data_direction data_dir; struct scatterlist *sg; u32 n_sg; u32 sg_byte_count; @@ -224,38 +183,19 @@ struct skd_request_context { struct fit_comp_error_info err_info; + blk_status_t status; }; -#define SKD_DATA_DIR_HOST_TO_CARD 1 -#define SKD_DATA_DIR_CARD_TO_HOST 2 struct skd_special_context { struct skd_request_context req; - u8 orphaned; - void *data_buf; dma_addr_t db_dma_address; - u8 *msg_buf; + struct skd_msg_buf *msg_buf; dma_addr_t mb_dma_address; }; -struct skd_sg_io { - fmode_t mode; - void __user *argp; - - struct sg_io_hdr sg; - - u8 cdb[16]; - - u32 dxfer_len; - u32 iovcnt; - struct sg_iovec *iov; - struct sg_iovec no_iov_iov; - - struct skd_special_context *skspcl; -}; - typedef enum skd_irq_type { SKD_IRQ_LEGACY, SKD_IRQ_MSI, @@ -265,7 +205,7 @@ typedef enum skd_irq_type { #define SKD_MAX_BARS 2 struct skd_device { - volatile void __iomem *mem_map[SKD_MAX_BARS]; + void __iomem *mem_map[SKD_MAX_BARS]; resource_size_t mem_phys[SKD_MAX_BARS]; u32 mem_size[SKD_MAX_BARS]; @@ -276,21 +216,20 @@ struct skd_device { spinlock_t lock; struct gendisk *disk; + struct blk_mq_tag_set tag_set; struct request_queue *queue; + struct skd_fitmsg_context *skmsg; struct device *class_dev; int gendisk_on; int sync_done; - atomic_t device_count; u32 devno; u32 major; - char name[32]; char isr_name[30]; enum skd_drvr_state state; u32 drive_state; - u32 in_flight; u32 cur_max_queue_depth; u32 queue_low_water_mark; u32 dev_max_queue_depth; @@ -298,27 +237,20 @@ struct skd_device { u32 num_fitmsg_context; u32 num_req_context; - u32 timeout_slot[SKD_N_TIMEOUT_SLOT]; - u32 timeout_stamp; - struct skd_fitmsg_context *skmsg_free_list; struct skd_fitmsg_context *skmsg_table; - struct skd_request_context *skreq_free_list; - struct skd_request_context *skreq_table; - - struct skd_special_context *skspcl_free_list; - struct skd_special_context *skspcl_table; - struct skd_special_context internal_skspcl; u32 read_cap_blocksize; u32 read_cap_last_lba; int read_cap_is_valid; int inquiry_is_valid; u8 inq_serial_num[13]; /*12 chars plus null term */ - u8 id_str[80]; /* holds a composite name (pci + sernum) */ u8 skcomp_cycle; u32 skcomp_ix; + struct kmem_cache *msgbuf_cache; + struct kmem_cache *sglist_cache; + struct kmem_cache *databuf_cache; struct fit_completion_entry_v1 *skcomp_table; struct fit_comp_error_info *skerr_table; dma_addr_t cq_dma_address; @@ -329,7 +261,6 @@ struct skd_device { u32 timer_countdown; u32 timer_substate; - int n_special; int sgs_per_request; u32 last_mtd; @@ -343,7 +274,7 @@ struct skd_device { u32 timo_slot; - + struct work_struct start_queue; struct work_struct completion_worker; }; @@ -353,53 +284,32 @@ struct skd_device { static inline u32 skd_reg_read32(struct skd_device *skdev, u32 offset) { - u32 val; - - if (likely(skdev->dbg_level < 2)) - return readl(skdev->mem_map[1] + offset); - else { - barrier(); - val = readl(skdev->mem_map[1] + offset); - barrier(); - pr_debug("%s:%s:%d offset %x = %x\n", - skdev->name, __func__, __LINE__, offset, val); - return val; - } + u32 val = readl(skdev->mem_map[1] + offset); + if (unlikely(skdev->dbg_level >= 2)) + dev_dbg(&skdev->pdev->dev, "offset %x = %x\n", offset, val); + return val; } static inline void skd_reg_write32(struct skd_device *skdev, u32 val, u32 offset) { - if (likely(skdev->dbg_level < 2)) { - writel(val, skdev->mem_map[1] + offset); - barrier(); - } else { - barrier(); - writel(val, skdev->mem_map[1] + offset); - barrier(); - pr_debug("%s:%s:%d offset %x = %x\n", - skdev->name, __func__, __LINE__, offset, val); - } + writel(val, skdev->mem_map[1] + offset); + if (unlikely(skdev->dbg_level >= 2)) + dev_dbg(&skdev->pdev->dev, "offset %x = %x\n", offset, val); } static inline void skd_reg_write64(struct skd_device *skdev, u64 val, u32 offset) { - if (likely(skdev->dbg_level < 2)) { - writeq(val, skdev->mem_map[1] + offset); - barrier(); - } else { - barrier(); - writeq(val, skdev->mem_map[1] + offset); - barrier(); - pr_debug("%s:%s:%d offset %x = %016llx\n", - skdev->name, __func__, __LINE__, offset, val); - } + writeq(val, skdev->mem_map[1] + offset); + if (unlikely(skdev->dbg_level >= 2)) + dev_dbg(&skdev->pdev->dev, "offset %x = %016llx\n", offset, + val); } -#define SKD_IRQ_DEFAULT SKD_IRQ_MSI +#define SKD_IRQ_DEFAULT SKD_IRQ_MSIX static int skd_isr_type = SKD_IRQ_DEFAULT; module_param(skd_isr_type, int, 0444); @@ -412,7 +322,7 @@ static int skd_max_req_per_msg = SKD_MAX_REQ_PER_MSG_DEFAULT; module_param(skd_max_req_per_msg, int, 0444); MODULE_PARM_DESC(skd_max_req_per_msg, "Maximum SCSI requests packed in a single message." - " (1-14, default==1)"); + " (1-" __stringify(SKD_MAX_REQ_PER_MSG) ", default==1)"); #define SKD_MAX_QUEUE_DEPTH_DEFAULT 64 #define SKD_MAX_QUEUE_DEPTH_DEFAULT_STR "64" @@ -429,10 +339,10 @@ MODULE_PARM_DESC(skd_sgs_per_request, "Maximum SG elements per block request." " (1-4096, default==256)"); -static int skd_max_pass_thru = SKD_N_SPECIAL_CONTEXT; +static int skd_max_pass_thru = 1; module_param(skd_max_pass_thru, int, 0444); MODULE_PARM_DESC(skd_max_pass_thru, - "Maximum SCSI pass-thru at a time." " (1-50, default==32)"); + "Maximum SCSI pass-thru at a time. IGNORED"); module_param(skd_dbg_level, int, 0444); MODULE_PARM_DESC(skd_dbg_level, "s1120 debug level (0,1,2)"); @@ -449,9 +359,6 @@ static void skd_send_fitmsg(struct skd_device *skdev, struct skd_fitmsg_context *skmsg); static void skd_send_special_fitmsg(struct skd_device *skdev, struct skd_special_context *skspcl); -static void skd_request_fn(struct request_queue *rq); -static void skd_end_request(struct skd_device *skdev, - struct skd_request_context *skreq, blk_status_t status); static bool skd_preop_sg_list(struct skd_device *skdev, struct skd_request_context *skreq); static void skd_postop_sg_list(struct skd_device *skdev, @@ -460,19 +367,14 @@ static void skd_postop_sg_list(struct skd_device *skdev, static void skd_restart_device(struct skd_device *skdev); static int skd_quiesce_dev(struct skd_device *skdev); static int skd_unquiesce_dev(struct skd_device *skdev); -static void skd_release_special(struct skd_device *skdev, - struct skd_special_context *skspcl); static void skd_disable_interrupts(struct skd_device *skdev); static void skd_isr_fwstate(struct skd_device *skdev); -static void skd_recover_requests(struct skd_device *skdev, int requeue); +static void skd_recover_requests(struct skd_device *skdev); static void skd_soft_reset(struct skd_device *skdev); -static const char *skd_name(struct skd_device *skdev); const char *skd_drive_state_to_str(int state); const char *skd_skdev_state_to_str(enum skd_drvr_state state); static void skd_log_skdev(struct skd_device *skdev, const char *event); -static void skd_log_skmsg(struct skd_device *skdev, - struct skd_fitmsg_context *skmsg, const char *event); static void skd_log_skreq(struct skd_device *skdev, struct skd_request_context *skreq, const char *event); @@ -481,18 +383,20 @@ static void skd_log_skreq(struct skd_device *skdev, * READ/WRITE REQUESTS ***************************************************************************** */ -static void skd_fail_all_pending(struct skd_device *skdev) +static void skd_inc_in_flight(struct request *rq, void *data, bool reserved) { - struct request_queue *q = skdev->queue; - struct request *req; + int *count = data; - for (;; ) { - req = blk_peek_request(q); - if (req == NULL) - break; - blk_start_request(req); - __blk_end_request_all(req, BLK_STS_IOERR); - } + count++; +} + +static int skd_in_flight(struct skd_device *skdev) +{ + int count = 0; + + blk_mq_tagset_busy_iter(&skdev->tag_set, skd_inc_in_flight, &count); + + return count; } static void @@ -501,9 +405,9 @@ skd_prep_rw_cdb(struct skd_scsi_request *scsi_req, unsigned count) { if (data_dir == READ) - scsi_req->cdb[0] = 0x28; + scsi_req->cdb[0] = READ_10; else - scsi_req->cdb[0] = 0x2a; + scsi_req->cdb[0] = WRITE_10; scsi_req->cdb[1] = 0; scsi_req->cdb[2] = (lba & 0xff000000) >> 24; @@ -522,7 +426,7 @@ skd_prep_zerosize_flush_cdb(struct skd_scsi_request *scsi_req, { skreq->flush_cmd = 1; - scsi_req->cdb[0] = 0x35; + scsi_req->cdb[0] = SYNCHRONIZE_CACHE; scsi_req->cdb[1] = 0; scsi_req->cdb[2] = 0; scsi_req->cdb[3] = 0; @@ -534,307 +438,194 @@ skd_prep_zerosize_flush_cdb(struct skd_scsi_request *scsi_req, scsi_req->cdb[9] = 0; } -static void skd_request_fn_not_online(struct request_queue *q); - -static void skd_request_fn(struct request_queue *q) +/* + * Return true if and only if all pending requests should be failed. + */ +static bool skd_fail_all(struct request_queue *q) { struct skd_device *skdev = q->queuedata; - struct skd_fitmsg_context *skmsg = NULL; - struct fit_msg_hdr *fmh = NULL; - struct skd_request_context *skreq; - struct request *req = NULL; - struct skd_scsi_request *scsi_req; - unsigned long io_flags; - u32 lba; - u32 count; - int data_dir; - u32 be_lba; - u32 be_count; - u64 be_dmaa; - u64 cmdctxt; - u32 timo_slot; - void *cmd_ptr; - int flush, fua; - - if (skdev->state != SKD_DRVR_STATE_ONLINE) { - skd_request_fn_not_online(q); - return; - } - if (blk_queue_stopped(skdev->queue)) { - if (skdev->skmsg_free_list == NULL || - skdev->skreq_free_list == NULL || - skdev->in_flight >= skdev->queue_low_water_mark) - /* There is still some kind of shortage */ - return; - - queue_flag_clear(QUEUE_FLAG_STOPPED, skdev->queue); - } + SKD_ASSERT(skdev->state != SKD_DRVR_STATE_ONLINE); - /* - * Stop conditions: - * - There are no more native requests - * - There are already the maximum number of requests in progress - * - There are no more skd_request_context entries - * - There are no more FIT msg buffers + skd_log_skdev(skdev, "req_not_online"); + switch (skdev->state) { + case SKD_DRVR_STATE_PAUSING: + case SKD_DRVR_STATE_PAUSED: + case SKD_DRVR_STATE_STARTING: + case SKD_DRVR_STATE_RESTARTING: + case SKD_DRVR_STATE_WAIT_BOOT: + /* In case of starting, we haven't started the queue, + * so we can't get here... but requests are + * possibly hanging out waiting for us because we + * reported the dev/skd0 already. They'll wait + * forever if connect doesn't complete. + * What to do??? delay dev/skd0 ?? */ - for (;; ) { - - flush = fua = 0; - - req = blk_peek_request(q); - - /* Are there any native requests to start? */ - if (req == NULL) - break; - - lba = (u32)blk_rq_pos(req); - count = blk_rq_sectors(req); - data_dir = rq_data_dir(req); - io_flags = req->cmd_flags; - - if (req_op(req) == REQ_OP_FLUSH) - flush++; - - if (io_flags & REQ_FUA) - fua++; - - pr_debug("%s:%s:%d new req=%p lba=%u(0x%x) " - "count=%u(0x%x) dir=%d\n", - skdev->name, __func__, __LINE__, - req, lba, lba, count, count, data_dir); - - /* At this point we know there is a request */ + case SKD_DRVR_STATE_BUSY: + case SKD_DRVR_STATE_BUSY_IMMINENT: + case SKD_DRVR_STATE_BUSY_ERASE: + return false; - /* Are too many requets already in progress? */ - if (skdev->in_flight >= skdev->cur_max_queue_depth) { - pr_debug("%s:%s:%d qdepth %d, limit %d\n", - skdev->name, __func__, __LINE__, - skdev->in_flight, skdev->cur_max_queue_depth); - break; - } + case SKD_DRVR_STATE_BUSY_SANITIZE: + case SKD_DRVR_STATE_STOPPING: + case SKD_DRVR_STATE_SYNCING: + case SKD_DRVR_STATE_FAULT: + case SKD_DRVR_STATE_DISAPPEARED: + default: + return true; + } +} - /* Is a skd_request_context available? */ - skreq = skdev->skreq_free_list; - if (skreq == NULL) { - pr_debug("%s:%s:%d Out of req=%p\n", - skdev->name, __func__, __LINE__, q); - break; - } - SKD_ASSERT(skreq->state == SKD_REQ_STATE_IDLE); - SKD_ASSERT((skreq->id & SKD_ID_INCR) == 0); - - /* Now we check to see if we can get a fit msg */ - if (skmsg == NULL) { - if (skdev->skmsg_free_list == NULL) { - pr_debug("%s:%s:%d Out of msg\n", - skdev->name, __func__, __LINE__); - break; - } - } +static blk_status_t skd_mq_queue_rq(struct blk_mq_hw_ctx *hctx, + const struct blk_mq_queue_data *mqd) +{ + struct request *const req = mqd->rq; + struct request_queue *const q = req->q; + struct skd_device *skdev = q->queuedata; + struct skd_fitmsg_context *skmsg; + struct fit_msg_hdr *fmh; + const u32 tag = blk_mq_unique_tag(req); + struct skd_request_context *const skreq = blk_mq_rq_to_pdu(req); + struct skd_scsi_request *scsi_req; + unsigned long flags = 0; + const u32 lba = blk_rq_pos(req); + const u32 count = blk_rq_sectors(req); + const int data_dir = rq_data_dir(req); - skreq->flush_cmd = 0; - skreq->n_sg = 0; - skreq->sg_byte_count = 0; + if (unlikely(skdev->state != SKD_DRVR_STATE_ONLINE)) + return skd_fail_all(q) ? BLK_STS_IOERR : BLK_STS_RESOURCE; - /* - * OK to now dequeue request from q. - * - * At this point we are comitted to either start or reject - * the native request. Note that skd_request_context is - * available but is still at the head of the free list. - */ - blk_start_request(req); - skreq->req = req; - skreq->fitmsg_id = 0; - - /* Either a FIT msg is in progress or we have to start one. */ - if (skmsg == NULL) { - /* Are there any FIT msg buffers available? */ - skmsg = skdev->skmsg_free_list; - if (skmsg == NULL) { - pr_debug("%s:%s:%d Out of msg skdev=%p\n", - skdev->name, __func__, __LINE__, - skdev); - break; - } - SKD_ASSERT(skmsg->state == SKD_MSG_STATE_IDLE); - SKD_ASSERT((skmsg->id & SKD_ID_INCR) == 0); + blk_mq_start_request(req); - skdev->skmsg_free_list = skmsg->next; + WARN_ONCE(tag >= skd_max_queue_depth, "%#x > %#x (nr_requests = %lu)\n", + tag, skd_max_queue_depth, q->nr_requests); - skmsg->state = SKD_MSG_STATE_BUSY; - skmsg->id += SKD_ID_INCR; + SKD_ASSERT(skreq->state == SKD_REQ_STATE_IDLE); - /* Initialize the FIT msg header */ - fmh = (struct fit_msg_hdr *)skmsg->msg_buf; - memset(fmh, 0, sizeof(*fmh)); - fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT; - skmsg->length = sizeof(*fmh); - } + dev_dbg(&skdev->pdev->dev, + "new req=%p lba=%u(0x%x) count=%u(0x%x) dir=%d\n", req, lba, + lba, count, count, data_dir); - skreq->fitmsg_id = skmsg->id; + skreq->id = tag + SKD_ID_RW_REQUEST; + skreq->flush_cmd = 0; + skreq->n_sg = 0; + skreq->sg_byte_count = 0; - /* - * Note that a FIT msg may have just been started - * but contains no SoFIT requests yet. - */ + skreq->fitmsg_id = 0; - /* - * Transcode the request, checking as we go. The outcome of - * the transcoding is represented by the error variable. - */ - cmd_ptr = &skmsg->msg_buf[skmsg->length]; - memset(cmd_ptr, 0, 32); + skreq->data_dir = data_dir == READ ? DMA_FROM_DEVICE : DMA_TO_DEVICE; - be_lba = cpu_to_be32(lba); - be_count = cpu_to_be32(count); - be_dmaa = cpu_to_be64((u64)skreq->sksg_dma_address); - cmdctxt = skreq->id + SKD_ID_INCR; + if (req->bio && !skd_preop_sg_list(skdev, skreq)) { + dev_dbg(&skdev->pdev->dev, "error Out\n"); + skreq->status = BLK_STS_RESOURCE; + blk_mq_complete_request(req); + return BLK_STS_OK; + } - scsi_req = cmd_ptr; - scsi_req->hdr.tag = cmdctxt; - scsi_req->hdr.sg_list_dma_address = be_dmaa; + dma_sync_single_for_device(&skdev->pdev->dev, skreq->sksg_dma_address, + skreq->n_sg * + sizeof(struct fit_sg_descriptor), + DMA_TO_DEVICE); - if (data_dir == READ) - skreq->sg_data_dir = SKD_DATA_DIR_CARD_TO_HOST; - else - skreq->sg_data_dir = SKD_DATA_DIR_HOST_TO_CARD; + /* Either a FIT msg is in progress or we have to start one. */ + if (skd_max_req_per_msg == 1) { + skmsg = NULL; + } else { + spin_lock_irqsave(&skdev->lock, flags); + skmsg = skdev->skmsg; + } + if (!skmsg) { + skmsg = &skdev->skmsg_table[tag]; + skdev->skmsg = skmsg; - if (flush == SKD_FLUSH_ZERO_SIZE_FIRST) { - skd_prep_zerosize_flush_cdb(scsi_req, skreq); - SKD_ASSERT(skreq->flush_cmd == 1); + /* Initialize the FIT msg header */ + fmh = &skmsg->msg_buf->fmh; + memset(fmh, 0, sizeof(*fmh)); + fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT; + skmsg->length = sizeof(*fmh); + } else { + fmh = &skmsg->msg_buf->fmh; + } - } else { - skd_prep_rw_cdb(scsi_req, data_dir, lba, count); - } + skreq->fitmsg_id = skmsg->id; - if (fua) - scsi_req->cdb[1] |= SKD_FUA_NV; + scsi_req = &skmsg->msg_buf->scsi[fmh->num_protocol_cmds_coalesced]; + memset(scsi_req, 0, sizeof(*scsi_req)); - if (!req->bio) - goto skip_sg; + scsi_req->hdr.tag = skreq->id; + scsi_req->hdr.sg_list_dma_address = + cpu_to_be64(skreq->sksg_dma_address); - if (!skd_preop_sg_list(skdev, skreq)) { - /* - * Complete the native request with error. - * Note that the request context is still at the - * head of the free list, and that the SoFIT request - * was encoded into the FIT msg buffer but the FIT - * msg length has not been updated. In short, the - * only resource that has been allocated but might - * not be used is that the FIT msg could be empty. - */ - pr_debug("%s:%s:%d error Out\n", - skdev->name, __func__, __LINE__); - skd_end_request(skdev, skreq, BLK_STS_RESOURCE); - continue; - } + if (req_op(req) == REQ_OP_FLUSH) { + skd_prep_zerosize_flush_cdb(scsi_req, skreq); + SKD_ASSERT(skreq->flush_cmd == 1); + } else { + skd_prep_rw_cdb(scsi_req, data_dir, lba, count); + } -skip_sg: - scsi_req->hdr.sg_list_len_bytes = - cpu_to_be32(skreq->sg_byte_count); + if (req->cmd_flags & REQ_FUA) + scsi_req->cdb[1] |= SKD_FUA_NV; - /* Complete resource allocations. */ - skdev->skreq_free_list = skreq->next; - skreq->state = SKD_REQ_STATE_BUSY; - skreq->id += SKD_ID_INCR; + scsi_req->hdr.sg_list_len_bytes = cpu_to_be32(skreq->sg_byte_count); - skmsg->length += sizeof(struct skd_scsi_request); - fmh->num_protocol_cmds_coalesced++; + /* Complete resource allocations. */ + skreq->state = SKD_REQ_STATE_BUSY; - /* - * Update the active request counts. - * Capture the timeout timestamp. - */ - skreq->timeout_stamp = skdev->timeout_stamp; - timo_slot = skreq->timeout_stamp & SKD_TIMEOUT_SLOT_MASK; - skdev->timeout_slot[timo_slot]++; - skdev->in_flight++; - pr_debug("%s:%s:%d req=0x%x busy=%d\n", - skdev->name, __func__, __LINE__, - skreq->id, skdev->in_flight); + skmsg->length += sizeof(struct skd_scsi_request); + fmh->num_protocol_cmds_coalesced++; - /* - * If the FIT msg buffer is full send it. - */ - if (skmsg->length >= SKD_N_FITMSG_BYTES || - fmh->num_protocol_cmds_coalesced >= skd_max_req_per_msg) { - skd_send_fitmsg(skdev, skmsg); - skmsg = NULL; - fmh = NULL; - } - } + dev_dbg(&skdev->pdev->dev, "req=0x%x busy=%d\n", skreq->id, + skd_in_flight(skdev)); /* - * Is a FIT msg in progress? If it is empty put the buffer back - * on the free list. If it is non-empty send what we got. - * This minimizes latency when there are fewer requests than - * what fits in a FIT msg. + * If the FIT msg buffer is full send it. */ - if (skmsg != NULL) { - /* Bigger than just a FIT msg header? */ - if (skmsg->length > sizeof(struct fit_msg_hdr)) { - pr_debug("%s:%s:%d sending msg=%p, len %d\n", - skdev->name, __func__, __LINE__, - skmsg, skmsg->length); + if (skd_max_req_per_msg == 1) { + skd_send_fitmsg(skdev, skmsg); + } else { + if (mqd->last || + fmh->num_protocol_cmds_coalesced >= skd_max_req_per_msg) { skd_send_fitmsg(skdev, skmsg); - } else { - /* - * The FIT msg is empty. It means we got started - * on the msg, but the requests were rejected. - */ - skmsg->state = SKD_MSG_STATE_IDLE; - skmsg->id += SKD_ID_INCR; - skmsg->next = skdev->skmsg_free_list; - skdev->skmsg_free_list = skmsg; + skdev->skmsg = NULL; } - skmsg = NULL; - fmh = NULL; + spin_unlock_irqrestore(&skdev->lock, flags); } - /* - * If req is non-NULL it means there is something to do but - * we are out of a resource. - */ - if (req) - blk_stop_queue(skdev->queue); + return BLK_STS_OK; } -static void skd_end_request(struct skd_device *skdev, - struct skd_request_context *skreq, blk_status_t error) +static enum blk_eh_timer_return skd_timed_out(struct request *req, + bool reserved) { - if (unlikely(error)) { - struct request *req = skreq->req; - char *cmd = (rq_data_dir(req) == READ) ? "read" : "write"; - u32 lba = (u32)blk_rq_pos(req); - u32 count = blk_rq_sectors(req); - - pr_err("(%s): Error cmd=%s sect=%u count=%u id=0x%x\n", - skd_name(skdev), cmd, lba, count, skreq->id); - } else - pr_debug("%s:%s:%d id=0x%x error=%d\n", - skdev->name, __func__, __LINE__, skreq->id, error); + struct skd_device *skdev = req->q->queuedata; + + dev_err(&skdev->pdev->dev, "request with tag %#x timed out\n", + blk_mq_unique_tag(req)); - __blk_end_request_all(skreq->req, error); + return BLK_EH_RESET_TIMER; +} + +static void skd_complete_rq(struct request *req) +{ + struct skd_request_context *skreq = blk_mq_rq_to_pdu(req); + + blk_mq_end_request(req, skreq->status); } static bool skd_preop_sg_list(struct skd_device *skdev, struct skd_request_context *skreq) { - struct request *req = skreq->req; - int writing = skreq->sg_data_dir == SKD_DATA_DIR_HOST_TO_CARD; - int pci_dir = writing ? PCI_DMA_TODEVICE : PCI_DMA_FROMDEVICE; - struct scatterlist *sg = &skreq->sg[0]; + struct request *req = blk_mq_rq_from_pdu(skreq); + struct scatterlist *sgl = &skreq->sg[0], *sg; int n_sg; int i; skreq->sg_byte_count = 0; - /* SKD_ASSERT(skreq->sg_data_dir == SKD_DATA_DIR_HOST_TO_CARD || - skreq->sg_data_dir == SKD_DATA_DIR_CARD_TO_HOST); */ + WARN_ON_ONCE(skreq->data_dir != DMA_TO_DEVICE && + skreq->data_dir != DMA_FROM_DEVICE); - n_sg = blk_rq_map_sg(skdev->queue, req, sg); + n_sg = blk_rq_map_sg(skdev->queue, req, sgl); if (n_sg <= 0) return false; @@ -842,7 +633,7 @@ static bool skd_preop_sg_list(struct skd_device *skdev, * Map scatterlist to PCI bus addresses. * Note PCI might change the number of entries. */ - n_sg = pci_map_sg(skdev->pdev, sg, n_sg, pci_dir); + n_sg = pci_map_sg(skdev->pdev, sgl, n_sg, skreq->data_dir); if (n_sg <= 0) return false; @@ -850,10 +641,10 @@ static bool skd_preop_sg_list(struct skd_device *skdev, skreq->n_sg = n_sg; - for (i = 0; i < n_sg; i++) { + for_each_sg(sgl, sg, n_sg, i) { struct fit_sg_descriptor *sgd = &skreq->sksg_list[i]; - u32 cnt = sg_dma_len(&sg[i]); - uint64_t dma_addr = sg_dma_address(&sg[i]); + u32 cnt = sg_dma_len(sg); + uint64_t dma_addr = sg_dma_address(sg); sgd->control = FIT_SGD_CONTROL_NOT_LAST; sgd->byte_count = cnt; @@ -866,16 +657,16 @@ static bool skd_preop_sg_list(struct skd_device *skdev, skreq->sksg_list[n_sg - 1].control = FIT_SGD_CONTROL_LAST; if (unlikely(skdev->dbg_level > 1)) { - pr_debug("%s:%s:%d skreq=%x sksg_list=%p sksg_dma=%llx\n", - skdev->name, __func__, __LINE__, - skreq->id, skreq->sksg_list, skreq->sksg_dma_address); + dev_dbg(&skdev->pdev->dev, + "skreq=%x sksg_list=%p sksg_dma=%llx\n", + skreq->id, skreq->sksg_list, skreq->sksg_dma_address); for (i = 0; i < n_sg; i++) { struct fit_sg_descriptor *sgd = &skreq->sksg_list[i]; - pr_debug("%s:%s:%d sg[%d] count=%u ctrl=0x%x " - "addr=0x%llx next=0x%llx\n", - skdev->name, __func__, __LINE__, - i, sgd->byte_count, sgd->control, - sgd->host_side_addr, sgd->next_desc_ptr); + + dev_dbg(&skdev->pdev->dev, + " sg[%d] count=%u ctrl=0x%x addr=0x%llx next=0x%llx\n", + i, sgd->byte_count, sgd->control, + sgd->host_side_addr, sgd->next_desc_ptr); } } @@ -885,9 +676,6 @@ static bool skd_preop_sg_list(struct skd_device *skdev, static void skd_postop_sg_list(struct skd_device *skdev, struct skd_request_context *skreq) { - int writing = skreq->sg_data_dir == SKD_DATA_DIR_HOST_TO_CARD; - int pci_dir = writing ? PCI_DMA_TODEVICE : PCI_DMA_FROMDEVICE; - /* * restore the next ptr for next IO request so we * don't have to set it every time. @@ -895,51 +683,7 @@ static void skd_postop_sg_list(struct skd_device *skdev, skreq->sksg_list[skreq->n_sg - 1].next_desc_ptr = skreq->sksg_dma_address + ((skreq->n_sg) * sizeof(struct fit_sg_descriptor)); - pci_unmap_sg(skdev->pdev, &skreq->sg[0], skreq->n_sg, pci_dir); -} - -static void skd_request_fn_not_online(struct request_queue *q) -{ - struct skd_device *skdev = q->queuedata; - int error; - - SKD_ASSERT(skdev->state != SKD_DRVR_STATE_ONLINE); - - skd_log_skdev(skdev, "req_not_online"); - switch (skdev->state) { - case SKD_DRVR_STATE_PAUSING: - case SKD_DRVR_STATE_PAUSED: - case SKD_DRVR_STATE_STARTING: - case SKD_DRVR_STATE_RESTARTING: - case SKD_DRVR_STATE_WAIT_BOOT: - /* In case of starting, we haven't started the queue, - * so we can't get here... but requests are - * possibly hanging out waiting for us because we - * reported the dev/skd0 already. They'll wait - * forever if connect doesn't complete. - * What to do??? delay dev/skd0 ?? - */ - case SKD_DRVR_STATE_BUSY: - case SKD_DRVR_STATE_BUSY_IMMINENT: - case SKD_DRVR_STATE_BUSY_ERASE: - case SKD_DRVR_STATE_DRAINING_TIMEOUT: - return; - - case SKD_DRVR_STATE_BUSY_SANITIZE: - case SKD_DRVR_STATE_STOPPING: - case SKD_DRVR_STATE_SYNCING: - case SKD_DRVR_STATE_FAULT: - case SKD_DRVR_STATE_DISAPPEARED: - default: - error = -EIO; - break; - } - - /* If we get here, terminate all pending block requeusts - * with EIO and any scsi pass thru with appropriate sense - */ - - skd_fail_all_pending(skdev); + pci_unmap_sg(skdev->pdev, &skreq->sg[0], skreq->n_sg, skreq->data_dir); } /* @@ -950,12 +694,22 @@ static void skd_request_fn_not_online(struct request_queue *q) static void skd_timer_tick_not_online(struct skd_device *skdev); +static void skd_start_queue(struct work_struct *work) +{ + struct skd_device *skdev = container_of(work, typeof(*skdev), + start_queue); + + /* + * Although it is safe to call blk_start_queue() from interrupt + * context, blk_mq_start_hw_queues() must not be called from + * interrupt context. + */ + blk_mq_start_hw_queues(skdev->queue); +} + static void skd_timer_tick(ulong arg) { struct skd_device *skdev = (struct skd_device *)arg; - - u32 timo_slot; - u32 overdue_timestamp; unsigned long reqflags; u32 state; @@ -972,37 +726,9 @@ static void skd_timer_tick(ulong arg) if (state != skdev->drive_state) skd_isr_fwstate(skdev); - if (skdev->state != SKD_DRVR_STATE_ONLINE) { + if (skdev->state != SKD_DRVR_STATE_ONLINE) skd_timer_tick_not_online(skdev); - goto timer_func_out; - } - skdev->timeout_stamp++; - timo_slot = skdev->timeout_stamp & SKD_TIMEOUT_SLOT_MASK; - - /* - * All requests that happened during the previous use of - * this slot should be done by now. The previous use was - * over 7 seconds ago. - */ - if (skdev->timeout_slot[timo_slot] == 0) - goto timer_func_out; - - /* Something is overdue */ - overdue_timestamp = skdev->timeout_stamp - SKD_N_TIMEOUT_SLOT; - - pr_debug("%s:%s:%d found %d timeouts, draining busy=%d\n", - skdev->name, __func__, __LINE__, - skdev->timeout_slot[timo_slot], skdev->in_flight); - pr_err("(%s): Overdue IOs (%d), busy %d\n", - skd_name(skdev), skdev->timeout_slot[timo_slot], - skdev->in_flight); - skdev->timer_countdown = SKD_DRAINING_TIMO; - skdev->state = SKD_DRVR_STATE_DRAINING_TIMEOUT; - skdev->timo_slot = timo_slot; - blk_stop_queue(skdev->queue); - -timer_func_out: mod_timer(&skdev->timer, (jiffies + HZ)); spin_unlock_irqrestore(&skdev->lock, reqflags); @@ -1015,9 +741,9 @@ static void skd_timer_tick_not_online(struct skd_device *skdev) case SKD_DRVR_STATE_LOAD: break; case SKD_DRVR_STATE_BUSY_SANITIZE: - pr_debug("%s:%s:%d drive busy sanitize[%x], driver[%x]\n", - skdev->name, __func__, __LINE__, - skdev->drive_state, skdev->state); + dev_dbg(&skdev->pdev->dev, + "drive busy sanitize[%x], driver[%x]\n", + skdev->drive_state, skdev->state); /* If we've been in sanitize for 3 seconds, we figure we're not * going to get anymore completions, so recover requests now */ @@ -1025,22 +751,21 @@ static void skd_timer_tick_not_online(struct skd_device *skdev) skdev->timer_countdown--; return; } - skd_recover_requests(skdev, 0); + skd_recover_requests(skdev); break; case SKD_DRVR_STATE_BUSY: case SKD_DRVR_STATE_BUSY_IMMINENT: case SKD_DRVR_STATE_BUSY_ERASE: - pr_debug("%s:%s:%d busy[%x], countdown=%d\n", - skdev->name, __func__, __LINE__, - skdev->state, skdev->timer_countdown); + dev_dbg(&skdev->pdev->dev, "busy[%x], countdown=%d\n", + skdev->state, skdev->timer_countdown); if (skdev->timer_countdown > 0) { skdev->timer_countdown--; return; } - pr_debug("%s:%s:%d busy[%x], timedout=%d, restarting device.", - skdev->name, __func__, __LINE__, - skdev->state, skdev->timer_countdown); + dev_dbg(&skdev->pdev->dev, + "busy[%x], timedout=%d, restarting device.", + skdev->state, skdev->timer_countdown); skd_restart_device(skdev); break; @@ -1054,12 +779,12 @@ static void skd_timer_tick_not_online(struct skd_device *skdev) * revcover at some point. */ skdev->state = SKD_DRVR_STATE_FAULT; - pr_err("(%s): DriveFault Connect Timeout (%x)\n", - skd_name(skdev), skdev->drive_state); + dev_err(&skdev->pdev->dev, "DriveFault Connect Timeout (%x)\n", + skdev->drive_state); /*start the queue so we can respond with error to requests */ /* wakeup anyone waiting for startup complete */ - blk_start_queue(skdev->queue); + schedule_work(&skdev->start_queue); skdev->gendisk_on = -1; wake_up_interruptible(&skdev->waitq); break; @@ -1072,29 +797,6 @@ static void skd_timer_tick_not_online(struct skd_device *skdev) case SKD_DRVR_STATE_PAUSED: break; - case SKD_DRVR_STATE_DRAINING_TIMEOUT: - pr_debug("%s:%s:%d " - "draining busy [%d] tick[%d] qdb[%d] tmls[%d]\n", - skdev->name, __func__, __LINE__, - skdev->timo_slot, - skdev->timer_countdown, - skdev->in_flight, - skdev->timeout_slot[skdev->timo_slot]); - /* if the slot has cleared we can let the I/O continue */ - if (skdev->timeout_slot[skdev->timo_slot] == 0) { - pr_debug("%s:%s:%d Slot drained, starting queue.\n", - skdev->name, __func__, __LINE__); - skdev->state = SKD_DRVR_STATE_ONLINE; - blk_start_queue(skdev->queue); - return; - } - if (skdev->timer_countdown > 0) { - skdev->timer_countdown--; - return; - } - skd_restart_device(skdev); - break; - case SKD_DRVR_STATE_RESTARTING: if (skdev->timer_countdown > 0) { skdev->timer_countdown--; @@ -1103,8 +805,9 @@ static void skd_timer_tick_not_online(struct skd_device *skdev) /* For now, we fault the drive. Could attempt resets to * revcover at some point. */ skdev->state = SKD_DRVR_STATE_FAULT; - pr_err("(%s): DriveFault Reconnect Timeout (%x)\n", - skd_name(skdev), skdev->drive_state); + dev_err(&skdev->pdev->dev, + "DriveFault Reconnect Timeout (%x)\n", + skdev->drive_state); /* * Recovering does two things: @@ -1124,18 +827,18 @@ static void skd_timer_tick_not_online(struct skd_device *skdev) /* It never came out of soft reset. Try to * recover the requests and then let them * fail. This is to mitigate hung processes. */ - skd_recover_requests(skdev, 0); + skd_recover_requests(skdev); else { - pr_err("(%s): Disable BusMaster (%x)\n", - skd_name(skdev), skdev->drive_state); + dev_err(&skdev->pdev->dev, "Disable BusMaster (%x)\n", + skdev->drive_state); pci_disable_device(skdev->pdev); skd_disable_interrupts(skdev); - skd_recover_requests(skdev, 0); + skd_recover_requests(skdev); } /*start the queue so we can respond with error to requests */ /* wakeup anyone waiting for startup complete */ - blk_start_queue(skdev->queue); + schedule_work(&skdev->start_queue); skdev->gendisk_on = -1; wake_up_interruptible(&skdev->waitq); break; @@ -1154,13 +857,11 @@ static int skd_start_timer(struct skd_device *skdev) { int rc; - init_timer(&skdev->timer); setup_timer(&skdev->timer, skd_timer_tick, (ulong)skdev); rc = mod_timer(&skdev->timer, (jiffies + HZ)); if (rc) - pr_err("%s: failed to start timer %d\n", - __func__, rc); + dev_err(&skdev->pdev->dev, "failed to start timer %d\n", rc); return rc; } @@ -1171,634 +872,6 @@ static void skd_kill_timer(struct skd_device *skdev) /* ***************************************************************************** - * IOCTL - ***************************************************************************** - */ -static int skd_ioctl_sg_io(struct skd_device *skdev, - fmode_t mode, void __user *argp); -static int skd_sg_io_get_and_check_args(struct skd_device *skdev, - struct skd_sg_io *sksgio); -static int skd_sg_io_obtain_skspcl(struct skd_device *skdev, - struct skd_sg_io *sksgio); -static int skd_sg_io_prep_buffering(struct skd_device *skdev, - struct skd_sg_io *sksgio); -static int skd_sg_io_copy_buffer(struct skd_device *skdev, - struct skd_sg_io *sksgio, int dxfer_dir); -static int skd_sg_io_send_fitmsg(struct skd_device *skdev, - struct skd_sg_io *sksgio); -static int skd_sg_io_await(struct skd_device *skdev, struct skd_sg_io *sksgio); -static int skd_sg_io_release_skspcl(struct skd_device *skdev, - struct skd_sg_io *sksgio); -static int skd_sg_io_put_status(struct skd_device *skdev, - struct skd_sg_io *sksgio); - -static void skd_complete_special(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 - *skcomp, - volatile struct fit_comp_error_info *skerr, - struct skd_special_context *skspcl); - -static int skd_bdev_ioctl(struct block_device *bdev, fmode_t mode, - uint cmd_in, ulong arg) -{ - static const int sg_version_num = 30527; - int rc = 0, timeout; - struct gendisk *disk = bdev->bd_disk; - struct skd_device *skdev = disk->private_data; - int __user *p = (int __user *)arg; - - pr_debug("%s:%s:%d %s: CMD[%s] ioctl mode 0x%x, cmd 0x%x arg %0lx\n", - skdev->name, __func__, __LINE__, - disk->disk_name, current->comm, mode, cmd_in, arg); - - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - - switch (cmd_in) { - case SG_SET_TIMEOUT: - rc = get_user(timeout, p); - if (!rc) - disk->queue->sg_timeout = clock_t_to_jiffies(timeout); - break; - case SG_GET_TIMEOUT: - rc = jiffies_to_clock_t(disk->queue->sg_timeout); - break; - case SG_GET_VERSION_NUM: - rc = put_user(sg_version_num, p); - break; - case SG_IO: - rc = skd_ioctl_sg_io(skdev, mode, (void __user *)arg); - break; - - default: - rc = -ENOTTY; - break; - } - - pr_debug("%s:%s:%d %s: completion rc %d\n", - skdev->name, __func__, __LINE__, disk->disk_name, rc); - return rc; -} - -static int skd_ioctl_sg_io(struct skd_device *skdev, fmode_t mode, - void __user *argp) -{ - int rc; - struct skd_sg_io sksgio; - - memset(&sksgio, 0, sizeof(sksgio)); - sksgio.mode = mode; - sksgio.argp = argp; - sksgio.iov = &sksgio.no_iov_iov; - - switch (skdev->state) { - case SKD_DRVR_STATE_ONLINE: - case SKD_DRVR_STATE_BUSY_IMMINENT: - break; - - default: - pr_debug("%s:%s:%d drive not online\n", - skdev->name, __func__, __LINE__); - rc = -ENXIO; - goto out; - } - - rc = skd_sg_io_get_and_check_args(skdev, &sksgio); - if (rc) - goto out; - - rc = skd_sg_io_obtain_skspcl(skdev, &sksgio); - if (rc) - goto out; - - rc = skd_sg_io_prep_buffering(skdev, &sksgio); - if (rc) - goto out; - - rc = skd_sg_io_copy_buffer(skdev, &sksgio, SG_DXFER_TO_DEV); - if (rc) - goto out; - - rc = skd_sg_io_send_fitmsg(skdev, &sksgio); - if (rc) - goto out; - - rc = skd_sg_io_await(skdev, &sksgio); - if (rc) - goto out; - - rc = skd_sg_io_copy_buffer(skdev, &sksgio, SG_DXFER_FROM_DEV); - if (rc) - goto out; - - rc = skd_sg_io_put_status(skdev, &sksgio); - if (rc) - goto out; - - rc = 0; - -out: - skd_sg_io_release_skspcl(skdev, &sksgio); - - if (sksgio.iov != NULL && sksgio.iov != &sksgio.no_iov_iov) - kfree(sksgio.iov); - return rc; -} - -static int skd_sg_io_get_and_check_args(struct skd_device *skdev, - struct skd_sg_io *sksgio) -{ - struct sg_io_hdr *sgp = &sksgio->sg; - int i, acc; - - if (!access_ok(VERIFY_WRITE, sksgio->argp, sizeof(sg_io_hdr_t))) { - pr_debug("%s:%s:%d access sg failed %p\n", - skdev->name, __func__, __LINE__, sksgio->argp); - return -EFAULT; - } - - if (__copy_from_user(sgp, sksgio->argp, sizeof(sg_io_hdr_t))) { - pr_debug("%s:%s:%d copy_from_user sg failed %p\n", - skdev->name, __func__, __LINE__, sksgio->argp); - return -EFAULT; - } - - if (sgp->interface_id != SG_INTERFACE_ID_ORIG) { - pr_debug("%s:%s:%d interface_id invalid 0x%x\n", - skdev->name, __func__, __LINE__, sgp->interface_id); - return -EINVAL; - } - - if (sgp->cmd_len > sizeof(sksgio->cdb)) { - pr_debug("%s:%s:%d cmd_len invalid %d\n", - skdev->name, __func__, __LINE__, sgp->cmd_len); - return -EINVAL; - } - - if (sgp->iovec_count > 256) { - pr_debug("%s:%s:%d iovec_count invalid %d\n", - skdev->name, __func__, __LINE__, sgp->iovec_count); - return -EINVAL; - } - - if (sgp->dxfer_len > (PAGE_SIZE * SKD_N_SG_PER_SPECIAL)) { - pr_debug("%s:%s:%d dxfer_len invalid %d\n", - skdev->name, __func__, __LINE__, sgp->dxfer_len); - return -EINVAL; - } - - switch (sgp->dxfer_direction) { - case SG_DXFER_NONE: - acc = -1; - break; - - case SG_DXFER_TO_DEV: - acc = VERIFY_READ; - break; - - case SG_DXFER_FROM_DEV: - case SG_DXFER_TO_FROM_DEV: - acc = VERIFY_WRITE; - break; - - default: - pr_debug("%s:%s:%d dxfer_dir invalid %d\n", - skdev->name, __func__, __LINE__, sgp->dxfer_direction); - return -EINVAL; - } - - if (copy_from_user(sksgio->cdb, sgp->cmdp, sgp->cmd_len)) { - pr_debug("%s:%s:%d copy_from_user cmdp failed %p\n", - skdev->name, __func__, __LINE__, sgp->cmdp); - return -EFAULT; - } - - if (sgp->mx_sb_len != 0) { - if (!access_ok(VERIFY_WRITE, sgp->sbp, sgp->mx_sb_len)) { - pr_debug("%s:%s:%d access sbp failed %p\n", - skdev->name, __func__, __LINE__, sgp->sbp); - return -EFAULT; - } - } - - if (sgp->iovec_count == 0) { - sksgio->iov[0].iov_base = sgp->dxferp; - sksgio->iov[0].iov_len = sgp->dxfer_len; - sksgio->iovcnt = 1; - sksgio->dxfer_len = sgp->dxfer_len; - } else { - struct sg_iovec *iov; - uint nbytes = sizeof(*iov) * sgp->iovec_count; - size_t iov_data_len; - - iov = kmalloc(nbytes, GFP_KERNEL); - if (iov == NULL) { - pr_debug("%s:%s:%d alloc iovec failed %d\n", - skdev->name, __func__, __LINE__, - sgp->iovec_count); - return -ENOMEM; - } - sksgio->iov = iov; - sksgio->iovcnt = sgp->iovec_count; - - if (copy_from_user(iov, sgp->dxferp, nbytes)) { - pr_debug("%s:%s:%d copy_from_user iovec failed %p\n", - skdev->name, __func__, __LINE__, sgp->dxferp); - return -EFAULT; - } - - /* - * Sum up the vecs, making sure they don't overflow - */ - iov_data_len = 0; - for (i = 0; i < sgp->iovec_count; i++) { - if (iov_data_len + iov[i].iov_len < iov_data_len) - return -EINVAL; - iov_data_len += iov[i].iov_len; - } - - /* SG_IO howto says that the shorter of the two wins */ - if (sgp->dxfer_len < iov_data_len) { - sksgio->iovcnt = iov_shorten((struct iovec *)iov, - sgp->iovec_count, - sgp->dxfer_len); - sksgio->dxfer_len = sgp->dxfer_len; - } else - sksgio->dxfer_len = iov_data_len; - } - - if (sgp->dxfer_direction != SG_DXFER_NONE) { - struct sg_iovec *iov = sksgio->iov; - for (i = 0; i < sksgio->iovcnt; i++, iov++) { - if (!access_ok(acc, iov->iov_base, iov->iov_len)) { - pr_debug("%s:%s:%d access data failed %p/%d\n", - skdev->name, __func__, __LINE__, - iov->iov_base, (int)iov->iov_len); - return -EFAULT; - } - } - } - - return 0; -} - -static int skd_sg_io_obtain_skspcl(struct skd_device *skdev, - struct skd_sg_io *sksgio) -{ - struct skd_special_context *skspcl = NULL; - int rc; - - for (;;) { - ulong flags; - - spin_lock_irqsave(&skdev->lock, flags); - skspcl = skdev->skspcl_free_list; - if (skspcl != NULL) { - skdev->skspcl_free_list = - (struct skd_special_context *)skspcl->req.next; - skspcl->req.id += SKD_ID_INCR; - skspcl->req.state = SKD_REQ_STATE_SETUP; - skspcl->orphaned = 0; - skspcl->req.n_sg = 0; - } - spin_unlock_irqrestore(&skdev->lock, flags); - - if (skspcl != NULL) { - rc = 0; - break; - } - - pr_debug("%s:%s:%d blocking\n", - skdev->name, __func__, __LINE__); - - rc = wait_event_interruptible_timeout( - skdev->waitq, - (skdev->skspcl_free_list != NULL), - msecs_to_jiffies(sksgio->sg.timeout)); - - pr_debug("%s:%s:%d unblocking, rc=%d\n", - skdev->name, __func__, __LINE__, rc); - - if (rc <= 0) { - if (rc == 0) - rc = -ETIMEDOUT; - else - rc = -EINTR; - break; - } - /* - * If we get here rc > 0 meaning the timeout to - * wait_event_interruptible_timeout() had time left, hence the - * sought event -- non-empty free list -- happened. - * Retry the allocation. - */ - } - sksgio->skspcl = skspcl; - - return rc; -} - -static int skd_skreq_prep_buffering(struct skd_device *skdev, - struct skd_request_context *skreq, - u32 dxfer_len) -{ - u32 resid = dxfer_len; - - /* - * The DMA engine must have aligned addresses and byte counts. - */ - resid += (-resid) & 3; - skreq->sg_byte_count = resid; - - skreq->n_sg = 0; - - while (resid > 0) { - u32 nbytes = PAGE_SIZE; - u32 ix = skreq->n_sg; - struct scatterlist *sg = &skreq->sg[ix]; - struct fit_sg_descriptor *sksg = &skreq->sksg_list[ix]; - struct page *page; - - if (nbytes > resid) - nbytes = resid; - - page = alloc_page(GFP_KERNEL); - if (page == NULL) - return -ENOMEM; - - sg_set_page(sg, page, nbytes, 0); - - /* TODO: This should be going through a pci_???() - * routine to do proper mapping. */ - sksg->control = FIT_SGD_CONTROL_NOT_LAST; - sksg->byte_count = nbytes; - - sksg->host_side_addr = sg_phys(sg); - - sksg->dev_side_addr = 0; - sksg->next_desc_ptr = skreq->sksg_dma_address + - (ix + 1) * sizeof(*sksg); - - skreq->n_sg++; - resid -= nbytes; - } - - if (skreq->n_sg > 0) { - u32 ix = skreq->n_sg - 1; - struct fit_sg_descriptor *sksg = &skreq->sksg_list[ix]; - - sksg->control = FIT_SGD_CONTROL_LAST; - sksg->next_desc_ptr = 0; - } - - if (unlikely(skdev->dbg_level > 1)) { - u32 i; - - pr_debug("%s:%s:%d skreq=%x sksg_list=%p sksg_dma=%llx\n", - skdev->name, __func__, __LINE__, - skreq->id, skreq->sksg_list, skreq->sksg_dma_address); - for (i = 0; i < skreq->n_sg; i++) { - struct fit_sg_descriptor *sgd = &skreq->sksg_list[i]; - - pr_debug("%s:%s:%d sg[%d] count=%u ctrl=0x%x " - "addr=0x%llx next=0x%llx\n", - skdev->name, __func__, __LINE__, - i, sgd->byte_count, sgd->control, - sgd->host_side_addr, sgd->next_desc_ptr); - } - } - - return 0; -} - -static int skd_sg_io_prep_buffering(struct skd_device *skdev, - struct skd_sg_io *sksgio) -{ - struct skd_special_context *skspcl = sksgio->skspcl; - struct skd_request_context *skreq = &skspcl->req; - u32 dxfer_len = sksgio->dxfer_len; - int rc; - - rc = skd_skreq_prep_buffering(skdev, skreq, dxfer_len); - /* - * Eventually, errors or not, skd_release_special() is called - * to recover allocations including partial allocations. - */ - return rc; -} - -static int skd_sg_io_copy_buffer(struct skd_device *skdev, - struct skd_sg_io *sksgio, int dxfer_dir) -{ - struct skd_special_context *skspcl = sksgio->skspcl; - u32 iov_ix = 0; - struct sg_iovec curiov; - u32 sksg_ix = 0; - u8 *bufp = NULL; - u32 buf_len = 0; - u32 resid = sksgio->dxfer_len; - int rc; - - curiov.iov_len = 0; - curiov.iov_base = NULL; - - if (dxfer_dir != sksgio->sg.dxfer_direction) { - if (dxfer_dir != SG_DXFER_TO_DEV || - sksgio->sg.dxfer_direction != SG_DXFER_TO_FROM_DEV) - return 0; - } - - while (resid > 0) { - u32 nbytes = PAGE_SIZE; - - if (curiov.iov_len == 0) { - curiov = sksgio->iov[iov_ix++]; - continue; - } - - if (buf_len == 0) { - struct page *page; - page = sg_page(&skspcl->req.sg[sksg_ix++]); - bufp = page_address(page); - buf_len = PAGE_SIZE; - } - - nbytes = min_t(u32, nbytes, resid); - nbytes = min_t(u32, nbytes, curiov.iov_len); - nbytes = min_t(u32, nbytes, buf_len); - - if (dxfer_dir == SG_DXFER_TO_DEV) - rc = __copy_from_user(bufp, curiov.iov_base, nbytes); - else - rc = __copy_to_user(curiov.iov_base, bufp, nbytes); - - if (rc) - return -EFAULT; - - resid -= nbytes; - curiov.iov_len -= nbytes; - curiov.iov_base += nbytes; - buf_len -= nbytes; - } - - return 0; -} - -static int skd_sg_io_send_fitmsg(struct skd_device *skdev, - struct skd_sg_io *sksgio) -{ - struct skd_special_context *skspcl = sksgio->skspcl; - struct fit_msg_hdr *fmh = (struct fit_msg_hdr *)skspcl->msg_buf; - struct skd_scsi_request *scsi_req = (struct skd_scsi_request *)&fmh[1]; - - memset(skspcl->msg_buf, 0, SKD_N_SPECIAL_FITMSG_BYTES); - - /* Initialize the FIT msg header */ - fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT; - fmh->num_protocol_cmds_coalesced = 1; - - /* Initialize the SCSI request */ - if (sksgio->sg.dxfer_direction != SG_DXFER_NONE) - scsi_req->hdr.sg_list_dma_address = - cpu_to_be64(skspcl->req.sksg_dma_address); - scsi_req->hdr.tag = skspcl->req.id; - scsi_req->hdr.sg_list_len_bytes = - cpu_to_be32(skspcl->req.sg_byte_count); - memcpy(scsi_req->cdb, sksgio->cdb, sizeof(scsi_req->cdb)); - - skspcl->req.state = SKD_REQ_STATE_BUSY; - skd_send_special_fitmsg(skdev, skspcl); - - return 0; -} - -static int skd_sg_io_await(struct skd_device *skdev, struct skd_sg_io *sksgio) -{ - unsigned long flags; - int rc; - - rc = wait_event_interruptible_timeout(skdev->waitq, - (sksgio->skspcl->req.state != - SKD_REQ_STATE_BUSY), - msecs_to_jiffies(sksgio->sg. - timeout)); - - spin_lock_irqsave(&skdev->lock, flags); - - if (sksgio->skspcl->req.state == SKD_REQ_STATE_ABORTED) { - pr_debug("%s:%s:%d skspcl %p aborted\n", - skdev->name, __func__, __LINE__, sksgio->skspcl); - - /* Build check cond, sense and let command finish. */ - /* For a timeout, we must fabricate completion and sense - * data to complete the command */ - sksgio->skspcl->req.completion.status = - SAM_STAT_CHECK_CONDITION; - - memset(&sksgio->skspcl->req.err_info, 0, - sizeof(sksgio->skspcl->req.err_info)); - sksgio->skspcl->req.err_info.type = 0x70; - sksgio->skspcl->req.err_info.key = ABORTED_COMMAND; - sksgio->skspcl->req.err_info.code = 0x44; - sksgio->skspcl->req.err_info.qual = 0; - rc = 0; - } else if (sksgio->skspcl->req.state != SKD_REQ_STATE_BUSY) - /* No longer on the adapter. We finish. */ - rc = 0; - else { - /* Something's gone wrong. Still busy. Timeout or - * user interrupted (control-C). Mark as an orphan - * so it will be disposed when completed. */ - sksgio->skspcl->orphaned = 1; - sksgio->skspcl = NULL; - if (rc == 0) { - pr_debug("%s:%s:%d timed out %p (%u ms)\n", - skdev->name, __func__, __LINE__, - sksgio, sksgio->sg.timeout); - rc = -ETIMEDOUT; - } else { - pr_debug("%s:%s:%d cntlc %p\n", - skdev->name, __func__, __LINE__, sksgio); - rc = -EINTR; - } - } - - spin_unlock_irqrestore(&skdev->lock, flags); - - return rc; -} - -static int skd_sg_io_put_status(struct skd_device *skdev, - struct skd_sg_io *sksgio) -{ - struct sg_io_hdr *sgp = &sksgio->sg; - struct skd_special_context *skspcl = sksgio->skspcl; - int resid = 0; - - u32 nb = be32_to_cpu(skspcl->req.completion.num_returned_bytes); - - sgp->status = skspcl->req.completion.status; - resid = sksgio->dxfer_len - nb; - - sgp->masked_status = sgp->status & STATUS_MASK; - sgp->msg_status = 0; - sgp->host_status = 0; - sgp->driver_status = 0; - sgp->resid = resid; - if (sgp->masked_status || sgp->host_status || sgp->driver_status) - sgp->info |= SG_INFO_CHECK; - - pr_debug("%s:%s:%d status %x masked %x resid 0x%x\n", - skdev->name, __func__, __LINE__, - sgp->status, sgp->masked_status, sgp->resid); - - if (sgp->masked_status == SAM_STAT_CHECK_CONDITION) { - if (sgp->mx_sb_len > 0) { - struct fit_comp_error_info *ei = &skspcl->req.err_info; - u32 nbytes = sizeof(*ei); - - nbytes = min_t(u32, nbytes, sgp->mx_sb_len); - - sgp->sb_len_wr = nbytes; - - if (__copy_to_user(sgp->sbp, ei, nbytes)) { - pr_debug("%s:%s:%d copy_to_user sense failed %p\n", - skdev->name, __func__, __LINE__, - sgp->sbp); - return -EFAULT; - } - } - } - - if (__copy_to_user(sksgio->argp, sgp, sizeof(sg_io_hdr_t))) { - pr_debug("%s:%s:%d copy_to_user sg failed %p\n", - skdev->name, __func__, __LINE__, sksgio->argp); - return -EFAULT; - } - - return 0; -} - -static int skd_sg_io_release_skspcl(struct skd_device *skdev, - struct skd_sg_io *sksgio) -{ - struct skd_special_context *skspcl = sksgio->skspcl; - - if (skspcl != NULL) { - ulong flags; - - sksgio->skspcl = NULL; - - spin_lock_irqsave(&skdev->lock, flags); - skd_release_special(skdev, skspcl); - spin_unlock_irqrestore(&skdev->lock, flags); - } - - return 0; -} - -/* - ***************************************************************************** * INTERNAL REQUESTS -- generated by driver itself ***************************************************************************** */ @@ -1811,14 +884,15 @@ static int skd_format_internal_skspcl(struct skd_device *skdev) uint64_t dma_address; struct skd_scsi_request *scsi; - fmh = (struct fit_msg_hdr *)&skspcl->msg_buf[0]; + fmh = &skspcl->msg_buf->fmh; fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT; fmh->num_protocol_cmds_coalesced = 1; - scsi = (struct skd_scsi_request *)&skspcl->msg_buf[64]; + scsi = &skspcl->msg_buf->scsi[0]; memset(scsi, 0, sizeof(*scsi)); dma_address = skspcl->req.sksg_dma_address; scsi->hdr.sg_list_dma_address = cpu_to_be64(dma_address); + skspcl->req.n_sg = 1; sgd->control = FIT_SGD_CONTROL_LAST; sgd->byte_count = 0; sgd->host_side_addr = skspcl->db_dma_address; @@ -1846,11 +920,9 @@ static void skd_send_internal_skspcl(struct skd_device *skdev, */ return; - SKD_ASSERT((skspcl->req.id & SKD_ID_INCR) == 0); skspcl->req.state = SKD_REQ_STATE_BUSY; - skspcl->req.id += SKD_ID_INCR; - scsi = (struct skd_scsi_request *)&skspcl->msg_buf[64]; + scsi = &skspcl->msg_buf->scsi[0]; scsi->hdr.tag = skspcl->req.id; memset(scsi->cdb, 0, sizeof(scsi->cdb)); @@ -1940,32 +1012,35 @@ static void skd_log_check_status(struct skd_device *skdev, u8 status, u8 key, /* If the check condition is of special interest, log a message */ if ((status == SAM_STAT_CHECK_CONDITION) && (key == 0x02) && (code == 0x04) && (qual == 0x06)) { - pr_err("(%s): *** LOST_WRITE_DATA ERROR *** key/asc/" - "ascq/fruc %02x/%02x/%02x/%02x\n", - skd_name(skdev), key, code, qual, fruc); + dev_err(&skdev->pdev->dev, + "*** LOST_WRITE_DATA ERROR *** key/asc/ascq/fruc %02x/%02x/%02x/%02x\n", + key, code, qual, fruc); } } static void skd_complete_internal(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 - *skcomp, - volatile struct fit_comp_error_info *skerr, + struct fit_completion_entry_v1 *skcomp, + struct fit_comp_error_info *skerr, struct skd_special_context *skspcl) { u8 *buf = skspcl->data_buf; u8 status; int i; - struct skd_scsi_request *scsi = - (struct skd_scsi_request *)&skspcl->msg_buf[64]; + struct skd_scsi_request *scsi = &skspcl->msg_buf->scsi[0]; + + lockdep_assert_held(&skdev->lock); SKD_ASSERT(skspcl == &skdev->internal_skspcl); - pr_debug("%s:%s:%d complete internal %x\n", - skdev->name, __func__, __LINE__, scsi->cdb[0]); + dev_dbg(&skdev->pdev->dev, "complete internal %x\n", scsi->cdb[0]); + + dma_sync_single_for_cpu(&skdev->pdev->dev, + skspcl->db_dma_address, + skspcl->req.sksg_list[0].byte_count, + DMA_BIDIRECTIONAL); skspcl->req.completion = *skcomp; skspcl->req.state = SKD_REQ_STATE_IDLE; - skspcl->req.id += SKD_ID_INCR; status = skspcl->req.completion.status; @@ -1981,14 +1056,15 @@ static void skd_complete_internal(struct skd_device *skdev, skd_send_internal_skspcl(skdev, skspcl, WRITE_BUFFER); else { if (skdev->state == SKD_DRVR_STATE_STOPPING) { - pr_debug("%s:%s:%d TUR failed, don't send anymore state 0x%x\n", - skdev->name, __func__, __LINE__, - skdev->state); + dev_dbg(&skdev->pdev->dev, + "TUR failed, don't send anymore state 0x%x\n", + skdev->state); return; } - pr_debug("%s:%s:%d **** TUR failed, retry skerr\n", - skdev->name, __func__, __LINE__); - skd_send_internal_skspcl(skdev, skspcl, 0x00); + dev_dbg(&skdev->pdev->dev, + "**** TUR failed, retry skerr\n"); + skd_send_internal_skspcl(skdev, skspcl, + TEST_UNIT_READY); } break; @@ -1997,14 +1073,15 @@ static void skd_complete_internal(struct skd_device *skdev, skd_send_internal_skspcl(skdev, skspcl, READ_BUFFER); else { if (skdev->state == SKD_DRVR_STATE_STOPPING) { - pr_debug("%s:%s:%d write buffer failed, don't send anymore state 0x%x\n", - skdev->name, __func__, __LINE__, - skdev->state); + dev_dbg(&skdev->pdev->dev, + "write buffer failed, don't send anymore state 0x%x\n", + skdev->state); return; } - pr_debug("%s:%s:%d **** write buffer failed, retry skerr\n", - skdev->name, __func__, __LINE__); - skd_send_internal_skspcl(skdev, skspcl, 0x00); + dev_dbg(&skdev->pdev->dev, + "**** write buffer failed, retry skerr\n"); + skd_send_internal_skspcl(skdev, skspcl, + TEST_UNIT_READY); } break; @@ -2014,33 +1091,31 @@ static void skd_complete_internal(struct skd_device *skdev, skd_send_internal_skspcl(skdev, skspcl, READ_CAPACITY); else { - pr_err( - "(%s):*** W/R Buffer mismatch %d ***\n", - skd_name(skdev), skdev->connect_retries); + dev_err(&skdev->pdev->dev, + "*** W/R Buffer mismatch %d ***\n", + skdev->connect_retries); if (skdev->connect_retries < SKD_MAX_CONNECT_RETRIES) { skdev->connect_retries++; skd_soft_reset(skdev); } else { - pr_err( - "(%s): W/R Buffer Connect Error\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, + "W/R Buffer Connect Error\n"); return; } } } else { if (skdev->state == SKD_DRVR_STATE_STOPPING) { - pr_debug("%s:%s:%d " - "read buffer failed, don't send anymore state 0x%x\n", - skdev->name, __func__, __LINE__, - skdev->state); + dev_dbg(&skdev->pdev->dev, + "read buffer failed, don't send anymore state 0x%x\n", + skdev->state); return; } - pr_debug("%s:%s:%d " - "**** read buffer failed, retry skerr\n", - skdev->name, __func__, __LINE__); - skd_send_internal_skspcl(skdev, skspcl, 0x00); + dev_dbg(&skdev->pdev->dev, + "**** read buffer failed, retry skerr\n"); + skd_send_internal_skspcl(skdev, skspcl, + TEST_UNIT_READY); } break; @@ -2054,10 +1129,9 @@ static void skd_complete_internal(struct skd_device *skdev, (buf[4] << 24) | (buf[5] << 16) | (buf[6] << 8) | buf[7]; - pr_debug("%s:%s:%d last lba %d, bs %d\n", - skdev->name, __func__, __LINE__, - skdev->read_cap_last_lba, - skdev->read_cap_blocksize); + dev_dbg(&skdev->pdev->dev, "last lba %d, bs %d\n", + skdev->read_cap_last_lba, + skdev->read_cap_blocksize); set_capacity(skdev->disk, skdev->read_cap_last_lba + 1); @@ -2068,13 +1142,10 @@ static void skd_complete_internal(struct skd_device *skdev, (skerr->key == MEDIUM_ERROR)) { skdev->read_cap_last_lba = ~0; set_capacity(skdev->disk, skdev->read_cap_last_lba + 1); - pr_debug("%s:%s:%d " - "**** MEDIUM ERROR caused READCAP to fail, ignore failure and continue to inquiry\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "**** MEDIUM ERROR caused READCAP to fail, ignore failure and continue to inquiry\n"); skd_send_internal_skspcl(skdev, skspcl, INQUIRY); } else { - pr_debug("%s:%s:%d **** READCAP failed, retry TUR\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "**** READCAP failed, retry TUR\n"); skd_send_internal_skspcl(skdev, skspcl, TEST_UNIT_READY); } @@ -2091,8 +1162,7 @@ static void skd_complete_internal(struct skd_device *skdev, } if (skd_unquiesce_dev(skdev) < 0) - pr_debug("%s:%s:%d **** failed, to ONLINE device\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "**** failed, to ONLINE device\n"); /* connection is complete */ skdev->connect_retries = 0; break; @@ -2120,27 +1190,20 @@ static void skd_send_fitmsg(struct skd_device *skdev, struct skd_fitmsg_context *skmsg) { u64 qcmd; - struct fit_msg_hdr *fmh; - pr_debug("%s:%s:%d dma address 0x%llx, busy=%d\n", - skdev->name, __func__, __LINE__, - skmsg->mb_dma_address, skdev->in_flight); - pr_debug("%s:%s:%d msg_buf 0x%p, offset %x\n", - skdev->name, __func__, __LINE__, - skmsg->msg_buf, skmsg->offset); + dev_dbg(&skdev->pdev->dev, "dma address 0x%llx, busy=%d\n", + skmsg->mb_dma_address, skd_in_flight(skdev)); + dev_dbg(&skdev->pdev->dev, "msg_buf %p\n", skmsg->msg_buf); qcmd = skmsg->mb_dma_address; qcmd |= FIT_QCMD_QID_NORMAL; - fmh = (struct fit_msg_hdr *)skmsg->msg_buf; - skmsg->outstanding = fmh->num_protocol_cmds_coalesced; - if (unlikely(skdev->dbg_level > 1)) { u8 *bp = (u8 *)skmsg->msg_buf; int i; for (i = 0; i < skmsg->length; i += 8) { - pr_debug("%s:%s:%d msg[%2d] %8ph\n", - skdev->name, __func__, __LINE__, i, &bp[i]); + dev_dbg(&skdev->pdev->dev, "msg[%2d] %8ph\n", i, + &bp[i]); if (i == 0) i = 64 - 8; } @@ -2160,6 +1223,12 @@ static void skd_send_fitmsg(struct skd_device *skdev, */ qcmd |= FIT_QCMD_MSGSIZE_64; + dma_sync_single_for_device(&skdev->pdev->dev, skmsg->mb_dma_address, + skmsg->length, DMA_TO_DEVICE); + + /* Make sure skd_msg_buf is written before the doorbell is triggered. */ + smp_wmb(); + SKD_WRITEQ(skdev, qcmd, FIT_Q_COMMAND); } @@ -2168,30 +1237,31 @@ static void skd_send_special_fitmsg(struct skd_device *skdev, { u64 qcmd; + WARN_ON_ONCE(skspcl->req.n_sg != 1); + if (unlikely(skdev->dbg_level > 1)) { u8 *bp = (u8 *)skspcl->msg_buf; int i; for (i = 0; i < SKD_N_SPECIAL_FITMSG_BYTES; i += 8) { - pr_debug("%s:%s:%d spcl[%2d] %8ph\n", - skdev->name, __func__, __LINE__, i, &bp[i]); + dev_dbg(&skdev->pdev->dev, " spcl[%2d] %8ph\n", i, + &bp[i]); if (i == 0) i = 64 - 8; } - pr_debug("%s:%s:%d skspcl=%p id=%04x sksg_list=%p sksg_dma=%llx\n", - skdev->name, __func__, __LINE__, - skspcl, skspcl->req.id, skspcl->req.sksg_list, - skspcl->req.sksg_dma_address); + dev_dbg(&skdev->pdev->dev, + "skspcl=%p id=%04x sksg_list=%p sksg_dma=%llx\n", + skspcl, skspcl->req.id, skspcl->req.sksg_list, + skspcl->req.sksg_dma_address); for (i = 0; i < skspcl->req.n_sg; i++) { struct fit_sg_descriptor *sgd = &skspcl->req.sksg_list[i]; - pr_debug("%s:%s:%d sg[%d] count=%u ctrl=0x%x " - "addr=0x%llx next=0x%llx\n", - skdev->name, __func__, __LINE__, - i, sgd->byte_count, sgd->control, - sgd->host_side_addr, sgd->next_desc_ptr); + dev_dbg(&skdev->pdev->dev, + " sg[%d] count=%u ctrl=0x%x addr=0x%llx next=0x%llx\n", + i, sgd->byte_count, sgd->control, + sgd->host_side_addr, sgd->next_desc_ptr); } } @@ -2202,6 +1272,20 @@ static void skd_send_special_fitmsg(struct skd_device *skdev, qcmd = skspcl->mb_dma_address; qcmd |= FIT_QCMD_QID_NORMAL + FIT_QCMD_MSGSIZE_128; + dma_sync_single_for_device(&skdev->pdev->dev, skspcl->mb_dma_address, + SKD_N_SPECIAL_FITMSG_BYTES, DMA_TO_DEVICE); + dma_sync_single_for_device(&skdev->pdev->dev, + skspcl->req.sksg_dma_address, + 1 * sizeof(struct fit_sg_descriptor), + DMA_TO_DEVICE); + dma_sync_single_for_device(&skdev->pdev->dev, + skspcl->db_dma_address, + skspcl->req.sksg_list[0].byte_count, + DMA_BIDIRECTIONAL); + + /* Make sure skd_msg_buf is written before the doorbell is triggered. */ + smp_wmb(); + SKD_WRITEQ(skdev, qcmd, FIT_Q_COMMAND); } @@ -2212,8 +1296,8 @@ static void skd_send_special_fitmsg(struct skd_device *skdev, */ static void skd_complete_other(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 *skcomp, - volatile struct fit_comp_error_info *skerr); + struct fit_completion_entry_v1 *skcomp, + struct fit_comp_error_info *skerr); struct sns_info { u8 type; @@ -2262,21 +1346,20 @@ static struct sns_info skd_chkstat_table[] = { static enum skd_check_status_action skd_check_status(struct skd_device *skdev, - u8 cmp_status, volatile struct fit_comp_error_info *skerr) + u8 cmp_status, struct fit_comp_error_info *skerr) { - int i, n; + int i; - pr_err("(%s): key/asc/ascq/fruc %02x/%02x/%02x/%02x\n", - skd_name(skdev), skerr->key, skerr->code, skerr->qual, - skerr->fruc); + dev_err(&skdev->pdev->dev, "key/asc/ascq/fruc %02x/%02x/%02x/%02x\n", + skerr->key, skerr->code, skerr->qual, skerr->fruc); - pr_debug("%s:%s:%d stat: t=%02x stat=%02x k=%02x c=%02x q=%02x fruc=%02x\n", - skdev->name, __func__, __LINE__, skerr->type, cmp_status, - skerr->key, skerr->code, skerr->qual, skerr->fruc); + dev_dbg(&skdev->pdev->dev, + "stat: t=%02x stat=%02x k=%02x c=%02x q=%02x fruc=%02x\n", + skerr->type, cmp_status, skerr->key, skerr->code, skerr->qual, + skerr->fruc); /* Does the info match an entry in the good category? */ - n = sizeof(skd_chkstat_table) / sizeof(skd_chkstat_table[0]); - for (i = 0; i < n; i++) { + for (i = 0; i < ARRAY_SIZE(skd_chkstat_table); i++) { struct sns_info *sns = &skd_chkstat_table[i]; if (sns->mask & 0x10) @@ -2300,10 +1383,9 @@ skd_check_status(struct skd_device *skdev, continue; if (sns->action == SKD_CHECK_STATUS_REPORT_SMART_ALERT) { - pr_err("(%s): SMART Alert: sense key/asc/ascq " - "%02x/%02x/%02x\n", - skd_name(skdev), skerr->key, - skerr->code, skerr->qual); + dev_err(&skdev->pdev->dev, + "SMART Alert: sense key/asc/ascq %02x/%02x/%02x\n", + skerr->key, skerr->code, skerr->qual); } return sns->action; } @@ -2312,335 +1394,80 @@ skd_check_status(struct skd_device *skdev, * zero status means good */ if (cmp_status) { - pr_debug("%s:%s:%d status check: error\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "status check: error\n"); return SKD_CHECK_STATUS_REPORT_ERROR; } - pr_debug("%s:%s:%d status check good default\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "status check good default\n"); return SKD_CHECK_STATUS_REPORT_GOOD; } static void skd_resolve_req_exception(struct skd_device *skdev, - struct skd_request_context *skreq) + struct skd_request_context *skreq, + struct request *req) { u8 cmp_status = skreq->completion.status; switch (skd_check_status(skdev, cmp_status, &skreq->err_info)) { case SKD_CHECK_STATUS_REPORT_GOOD: case SKD_CHECK_STATUS_REPORT_SMART_ALERT: - skd_end_request(skdev, skreq, BLK_STS_OK); + skreq->status = BLK_STS_OK; + blk_mq_complete_request(req); break; case SKD_CHECK_STATUS_BUSY_IMMINENT: skd_log_skreq(skdev, skreq, "retry(busy)"); - blk_requeue_request(skdev->queue, skreq->req); - pr_info("(%s) drive BUSY imminent\n", skd_name(skdev)); + blk_requeue_request(skdev->queue, req); + dev_info(&skdev->pdev->dev, "drive BUSY imminent\n"); skdev->state = SKD_DRVR_STATE_BUSY_IMMINENT; skdev->timer_countdown = SKD_TIMER_MINUTES(20); skd_quiesce_dev(skdev); break; case SKD_CHECK_STATUS_REQUEUE_REQUEST: - if ((unsigned long) ++skreq->req->special < SKD_MAX_RETRIES) { + if ((unsigned long) ++req->special < SKD_MAX_RETRIES) { skd_log_skreq(skdev, skreq, "retry"); - blk_requeue_request(skdev->queue, skreq->req); + blk_requeue_request(skdev->queue, req); break; } - /* fall through to report error */ + /* fall through */ case SKD_CHECK_STATUS_REPORT_ERROR: default: - skd_end_request(skdev, skreq, BLK_STS_IOERR); + skreq->status = BLK_STS_IOERR; + blk_mq_complete_request(req); break; } } -/* assume spinlock is already held */ static void skd_release_skreq(struct skd_device *skdev, struct skd_request_context *skreq) { - u32 msg_slot; - struct skd_fitmsg_context *skmsg; - - u32 timo_slot; - - /* - * Reclaim the FIT msg buffer if this is - * the first of the requests it carried to - * be completed. The FIT msg buffer used to - * send this request cannot be reused until - * we are sure the s1120 card has copied - * it to its memory. The FIT msg might have - * contained several requests. As soon as - * any of them are completed we know that - * the entire FIT msg was transferred. - * Only the first completed request will - * match the FIT msg buffer id. The FIT - * msg buffer id is immediately updated. - * When subsequent requests complete the FIT - * msg buffer id won't match, so we know - * quite cheaply that it is already done. - */ - msg_slot = skreq->fitmsg_id & SKD_ID_SLOT_MASK; - SKD_ASSERT(msg_slot < skdev->num_fitmsg_context); - - skmsg = &skdev->skmsg_table[msg_slot]; - if (skmsg->id == skreq->fitmsg_id) { - SKD_ASSERT(skmsg->state == SKD_MSG_STATE_BUSY); - SKD_ASSERT(skmsg->outstanding > 0); - skmsg->outstanding--; - if (skmsg->outstanding == 0) { - skmsg->state = SKD_MSG_STATE_IDLE; - skmsg->id += SKD_ID_INCR; - skmsg->next = skdev->skmsg_free_list; - skdev->skmsg_free_list = skmsg; - } - } - - /* - * Decrease the number of active requests. - * Also decrements the count in the timeout slot. - */ - SKD_ASSERT(skdev->in_flight > 0); - skdev->in_flight -= 1; - - timo_slot = skreq->timeout_stamp & SKD_TIMEOUT_SLOT_MASK; - SKD_ASSERT(skdev->timeout_slot[timo_slot] > 0); - skdev->timeout_slot[timo_slot] -= 1; - - /* - * Reset backpointer - */ - skreq->req = NULL; - /* * Reclaim the skd_request_context */ skreq->state = SKD_REQ_STATE_IDLE; - skreq->id += SKD_ID_INCR; - skreq->next = skdev->skreq_free_list; - skdev->skreq_free_list = skreq; } -#define DRIVER_INQ_EVPD_PAGE_CODE 0xDA - -static void skd_do_inq_page_00(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 *skcomp, - volatile struct fit_comp_error_info *skerr, - uint8_t *cdb, uint8_t *buf) -{ - uint16_t insert_pt, max_bytes, drive_pages, drive_bytes, new_size; - - /* Caller requested "supported pages". The driver needs to insert - * its page. - */ - pr_debug("%s:%s:%d skd_do_driver_inquiry: modify supported pages.\n", - skdev->name, __func__, __LINE__); - - /* If the device rejected the request because the CDB was - * improperly formed, then just leave. - */ - if (skcomp->status == SAM_STAT_CHECK_CONDITION && - skerr->key == ILLEGAL_REQUEST && skerr->code == 0x24) - return; - - /* Get the amount of space the caller allocated */ - max_bytes = (cdb[3] << 8) | cdb[4]; - - /* Get the number of pages actually returned by the device */ - drive_pages = (buf[2] << 8) | buf[3]; - drive_bytes = drive_pages + 4; - new_size = drive_pages + 1; - - /* Supported pages must be in numerical order, so find where - * the driver page needs to be inserted into the list of - * pages returned by the device. - */ - for (insert_pt = 4; insert_pt < drive_bytes; insert_pt++) { - if (buf[insert_pt] == DRIVER_INQ_EVPD_PAGE_CODE) - return; /* Device using this page code. abort */ - else if (buf[insert_pt] > DRIVER_INQ_EVPD_PAGE_CODE) - break; - } - - if (insert_pt < max_bytes) { - uint16_t u; - - /* Shift everything up one byte to make room. */ - for (u = new_size + 3; u > insert_pt; u--) - buf[u] = buf[u - 1]; - buf[insert_pt] = DRIVER_INQ_EVPD_PAGE_CODE; - - /* SCSI byte order increment of num_returned_bytes by 1 */ - skcomp->num_returned_bytes = - be32_to_cpu(skcomp->num_returned_bytes) + 1; - skcomp->num_returned_bytes = - be32_to_cpu(skcomp->num_returned_bytes); - } - - /* update page length field to reflect the driver's page too */ - buf[2] = (uint8_t)((new_size >> 8) & 0xFF); - buf[3] = (uint8_t)((new_size >> 0) & 0xFF); -} - -static void skd_get_link_info(struct pci_dev *pdev, u8 *speed, u8 *width) -{ - int pcie_reg; - u16 pci_bus_speed; - u8 pci_lanes; - - pcie_reg = pci_find_capability(pdev, PCI_CAP_ID_EXP); - if (pcie_reg) { - u16 linksta; - pci_read_config_word(pdev, pcie_reg + PCI_EXP_LNKSTA, &linksta); - - pci_bus_speed = linksta & 0xF; - pci_lanes = (linksta & 0x3F0) >> 4; - } else { - *speed = STEC_LINK_UNKNOWN; - *width = 0xFF; - return; - } - - switch (pci_bus_speed) { - case 1: - *speed = STEC_LINK_2_5GTS; - break; - case 2: - *speed = STEC_LINK_5GTS; - break; - case 3: - *speed = STEC_LINK_8GTS; - break; - default: - *speed = STEC_LINK_UNKNOWN; - break; - } - - if (pci_lanes <= 0x20) - *width = pci_lanes; - else - *width = 0xFF; -} - -static void skd_do_inq_page_da(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 *skcomp, - volatile struct fit_comp_error_info *skerr, - uint8_t *cdb, uint8_t *buf) -{ - struct pci_dev *pdev = skdev->pdev; - unsigned max_bytes; - struct driver_inquiry_data inq; - u16 val; - - pr_debug("%s:%s:%d skd_do_driver_inquiry: return driver page\n", - skdev->name, __func__, __LINE__); - - memset(&inq, 0, sizeof(inq)); - - inq.page_code = DRIVER_INQ_EVPD_PAGE_CODE; - - skd_get_link_info(pdev, &inq.pcie_link_speed, &inq.pcie_link_lanes); - inq.pcie_bus_number = cpu_to_be16(pdev->bus->number); - inq.pcie_device_number = PCI_SLOT(pdev->devfn); - inq.pcie_function_number = PCI_FUNC(pdev->devfn); - - pci_read_config_word(pdev, PCI_VENDOR_ID, &val); - inq.pcie_vendor_id = cpu_to_be16(val); - - pci_read_config_word(pdev, PCI_DEVICE_ID, &val); - inq.pcie_device_id = cpu_to_be16(val); - - pci_read_config_word(pdev, PCI_SUBSYSTEM_VENDOR_ID, &val); - inq.pcie_subsystem_vendor_id = cpu_to_be16(val); - - pci_read_config_word(pdev, PCI_SUBSYSTEM_ID, &val); - inq.pcie_subsystem_device_id = cpu_to_be16(val); - - /* Driver version, fixed lenth, padded with spaces on the right */ - inq.driver_version_length = sizeof(inq.driver_version); - memset(&inq.driver_version, ' ', sizeof(inq.driver_version)); - memcpy(inq.driver_version, DRV_VER_COMPL, - min(sizeof(inq.driver_version), strlen(DRV_VER_COMPL))); - - inq.page_length = cpu_to_be16((sizeof(inq) - 4)); - - /* Clear the error set by the device */ - skcomp->status = SAM_STAT_GOOD; - memset((void *)skerr, 0, sizeof(*skerr)); - - /* copy response into output buffer */ - max_bytes = (cdb[3] << 8) | cdb[4]; - memcpy(buf, &inq, min_t(unsigned, max_bytes, sizeof(inq))); - - skcomp->num_returned_bytes = - be32_to_cpu(min_t(uint16_t, max_bytes, sizeof(inq))); -} - -static void skd_do_driver_inq(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 *skcomp, - volatile struct fit_comp_error_info *skerr, - uint8_t *cdb, uint8_t *buf) -{ - if (!buf) - return; - else if (cdb[0] != INQUIRY) - return; /* Not an INQUIRY */ - else if ((cdb[1] & 1) == 0) - return; /* EVPD not set */ - else if (cdb[2] == 0) - /* Need to add driver's page to supported pages list */ - skd_do_inq_page_00(skdev, skcomp, skerr, cdb, buf); - else if (cdb[2] == DRIVER_INQ_EVPD_PAGE_CODE) - /* Caller requested driver's page */ - skd_do_inq_page_da(skdev, skcomp, skerr, cdb, buf); -} - -static unsigned char *skd_sg_1st_page_ptr(struct scatterlist *sg) -{ - if (!sg) - return NULL; - if (!sg_page(sg)) - return NULL; - return sg_virt(sg); -} - -static void skd_process_scsi_inq(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 - *skcomp, - volatile struct fit_comp_error_info *skerr, - struct skd_special_context *skspcl) -{ - uint8_t *buf; - struct fit_msg_hdr *fmh = (struct fit_msg_hdr *)skspcl->msg_buf; - struct skd_scsi_request *scsi_req = (struct skd_scsi_request *)&fmh[1]; - - dma_sync_sg_for_cpu(skdev->class_dev, skspcl->req.sg, skspcl->req.n_sg, - skspcl->req.sg_data_dir); - buf = skd_sg_1st_page_ptr(skspcl->req.sg); - - if (buf) - skd_do_driver_inq(skdev, skcomp, skerr, scsi_req->cdb, buf); -} - - static int skd_isr_completion_posted(struct skd_device *skdev, int limit, int *enqueued) { - volatile struct fit_completion_entry_v1 *skcmp = NULL; - volatile struct fit_comp_error_info *skerr; + struct fit_completion_entry_v1 *skcmp; + struct fit_comp_error_info *skerr; u16 req_id; - u32 req_slot; + u32 tag; + u16 hwq = 0; + struct request *rq; struct skd_request_context *skreq; - u16 cmp_cntxt = 0; - u8 cmp_status = 0; - u8 cmp_cycle = 0; - u32 cmp_bytes = 0; + u16 cmp_cntxt; + u8 cmp_status; + u8 cmp_cycle; + u32 cmp_bytes; int rc = 0; int processed = 0; + lockdep_assert_held(&skdev->lock); + for (;; ) { SKD_ASSERT(skdev->skcomp_ix < SKD_N_COMPLETION_ENTRY); @@ -2652,16 +1479,14 @@ static int skd_isr_completion_posted(struct skd_device *skdev, skerr = &skdev->skerr_table[skdev->skcomp_ix]; - pr_debug("%s:%s:%d " - "cycle=%d ix=%d got cycle=%d cmdctxt=0x%x stat=%d " - "busy=%d rbytes=0x%x proto=%d\n", - skdev->name, __func__, __LINE__, skdev->skcomp_cycle, - skdev->skcomp_ix, cmp_cycle, cmp_cntxt, cmp_status, - skdev->in_flight, cmp_bytes, skdev->proto_ver); + dev_dbg(&skdev->pdev->dev, + "cycle=%d ix=%d got cycle=%d cmdctxt=0x%x stat=%d busy=%d rbytes=0x%x proto=%d\n", + skdev->skcomp_cycle, skdev->skcomp_ix, cmp_cycle, + cmp_cntxt, cmp_status, skd_in_flight(skdev), + cmp_bytes, skdev->proto_ver); if (cmp_cycle != skdev->skcomp_cycle) { - pr_debug("%s:%s:%d end of completions\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "end of completions\n"); break; } /* @@ -2680,49 +1505,38 @@ static int skd_isr_completion_posted(struct skd_device *skdev, * r/w request (see skd_start() above) or a special request. */ req_id = cmp_cntxt; - req_slot = req_id & SKD_ID_SLOT_AND_TABLE_MASK; + tag = req_id & SKD_ID_SLOT_AND_TABLE_MASK; /* Is this other than a r/w request? */ - if (req_slot >= skdev->num_req_context) { + if (tag >= skdev->num_req_context) { /* * This is not a completion for a r/w request. */ + WARN_ON_ONCE(blk_mq_tag_to_rq(skdev->tag_set.tags[hwq], + tag)); skd_complete_other(skdev, skcmp, skerr); continue; } - skreq = &skdev->skreq_table[req_slot]; + rq = blk_mq_tag_to_rq(skdev->tag_set.tags[hwq], tag); + if (WARN(!rq, "No request for tag %#x -> %#x\n", cmp_cntxt, + tag)) + continue; + skreq = blk_mq_rq_to_pdu(rq); /* * Make sure the request ID for the slot matches. */ if (skreq->id != req_id) { - pr_debug("%s:%s:%d mismatch comp_id=0x%x req_id=0x%x\n", - skdev->name, __func__, __LINE__, - req_id, skreq->id); - { - u16 new_id = cmp_cntxt; - pr_err("(%s): Completion mismatch " - "comp_id=0x%04x skreq=0x%04x new=0x%04x\n", - skd_name(skdev), req_id, - skreq->id, new_id); + dev_err(&skdev->pdev->dev, + "Completion mismatch comp_id=0x%04x skreq=0x%04x new=0x%04x\n", + req_id, skreq->id, cmp_cntxt); - continue; - } + continue; } SKD_ASSERT(skreq->state == SKD_REQ_STATE_BUSY); - if (skreq->state == SKD_REQ_STATE_ABORTED) { - pr_debug("%s:%s:%d reclaim req %p id=%04x\n", - skdev->name, __func__, __LINE__, - skreq, skreq->id); - /* a previously timed out command can - * now be cleaned up */ - skd_release_skreq(skdev, skreq); - continue; - } - skreq->completion = *skcmp; if (unlikely(cmp_status == SAM_STAT_CHECK_CONDITION)) { skreq->err_info = *skerr; @@ -2734,27 +1548,17 @@ static int skd_isr_completion_posted(struct skd_device *skdev, if (skreq->n_sg > 0) skd_postop_sg_list(skdev, skreq); - if (!skreq->req) { - pr_debug("%s:%s:%d NULL backptr skdreq %p, " - "req=0x%x req_id=0x%x\n", - skdev->name, __func__, __LINE__, - skreq, skreq->id, req_id); - } else { - /* - * Capture the outcome and post it back to the - * native request. - */ - if (likely(cmp_status == SAM_STAT_GOOD)) - skd_end_request(skdev, skreq, BLK_STS_OK); - else - skd_resolve_req_exception(skdev, skreq); - } + skd_release_skreq(skdev, skreq); /* - * Release the skreq, its FIT msg (if one), timeout slot, - * and queue depth. + * Capture the outcome and post it back to the native request. */ - skd_release_skreq(skdev, skreq); + if (likely(cmp_status == SAM_STAT_GOOD)) { + skreq->status = BLK_STS_OK; + blk_mq_complete_request(rq); + } else { + skd_resolve_req_exception(skdev, skreq, rq); + } /* skd_isr_comp_limit equal zero means no limit */ if (limit) { @@ -2765,8 +1569,8 @@ static int skd_isr_completion_posted(struct skd_device *skdev, } } - if ((skdev->state == SKD_DRVR_STATE_PAUSING) - && (skdev->in_flight) == 0) { + if (skdev->state == SKD_DRVR_STATE_PAUSING && + skd_in_flight(skdev) == 0) { skdev->state = SKD_DRVR_STATE_PAUSED; wake_up_interruptible(&skdev->waitq); } @@ -2775,21 +1579,22 @@ static int skd_isr_completion_posted(struct skd_device *skdev, } static void skd_complete_other(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 *skcomp, - volatile struct fit_comp_error_info *skerr) + struct fit_completion_entry_v1 *skcomp, + struct fit_comp_error_info *skerr) { u32 req_id = 0; u32 req_table; u32 req_slot; struct skd_special_context *skspcl; + lockdep_assert_held(&skdev->lock); + req_id = skcomp->tag; req_table = req_id & SKD_ID_TABLE_MASK; req_slot = req_id & SKD_ID_SLOT_MASK; - pr_debug("%s:%s:%d table=0x%x id=0x%x slot=%d\n", - skdev->name, __func__, __LINE__, - req_table, req_id, req_slot); + dev_dbg(&skdev->pdev->dev, "table=0x%x id=0x%x slot=%d\n", req_table, + req_id, req_slot); /* * Based on the request id, determine how to dispatch this completion. @@ -2799,28 +1604,12 @@ static void skd_complete_other(struct skd_device *skdev, switch (req_table) { case SKD_ID_RW_REQUEST: /* - * The caller, skd_completion_posted_isr() above, + * The caller, skd_isr_completion_posted() above, * handles r/w requests. The only way we get here * is if the req_slot is out of bounds. */ break; - case SKD_ID_SPECIAL_REQUEST: - /* - * Make sure the req_slot is in bounds and that the id - * matches. - */ - if (req_slot < skdev->n_special) { - skspcl = &skdev->skspcl_table[req_slot]; - if (skspcl->req.id == req_id && - skspcl->req.state == SKD_REQ_STATE_BUSY) { - skd_complete_special(skdev, - skcomp, skerr, skspcl); - return; - } - } - break; - case SKD_ID_INTERNAL: if (req_slot == 0) { skspcl = &skdev->internal_skspcl; @@ -2851,72 +1640,9 @@ static void skd_complete_other(struct skd_device *skdev, */ } -static void skd_complete_special(struct skd_device *skdev, - volatile struct fit_completion_entry_v1 - *skcomp, - volatile struct fit_comp_error_info *skerr, - struct skd_special_context *skspcl) -{ - pr_debug("%s:%s:%d completing special request %p\n", - skdev->name, __func__, __LINE__, skspcl); - if (skspcl->orphaned) { - /* Discard orphaned request */ - /* ?: Can this release directly or does it need - * to use a worker? */ - pr_debug("%s:%s:%d release orphaned %p\n", - skdev->name, __func__, __LINE__, skspcl); - skd_release_special(skdev, skspcl); - return; - } - - skd_process_scsi_inq(skdev, skcomp, skerr, skspcl); - - skspcl->req.state = SKD_REQ_STATE_COMPLETED; - skspcl->req.completion = *skcomp; - skspcl->req.err_info = *skerr; - - skd_log_check_status(skdev, skspcl->req.completion.status, skerr->key, - skerr->code, skerr->qual, skerr->fruc); - - wake_up_interruptible(&skdev->waitq); -} - -/* assume spinlock is already held */ -static void skd_release_special(struct skd_device *skdev, - struct skd_special_context *skspcl) -{ - int i, was_depleted; - - for (i = 0; i < skspcl->req.n_sg; i++) { - struct page *page = sg_page(&skspcl->req.sg[i]); - __free_page(page); - } - - was_depleted = (skdev->skspcl_free_list == NULL); - - skspcl->req.state = SKD_REQ_STATE_IDLE; - skspcl->req.id += SKD_ID_INCR; - skspcl->req.next = - (struct skd_request_context *)skdev->skspcl_free_list; - skdev->skspcl_free_list = (struct skd_special_context *)skspcl; - - if (was_depleted) { - pr_debug("%s:%s:%d skspcl was depleted\n", - skdev->name, __func__, __LINE__); - /* Free list was depleted. Their might be waiters. */ - wake_up_interruptible(&skdev->waitq); - } -} - static void skd_reset_skcomp(struct skd_device *skdev) { - u32 nbytes; - struct fit_completion_entry_v1 *skcomp; - - nbytes = sizeof(*skcomp) * SKD_N_COMPLETION_ENTRY; - nbytes += sizeof(struct fit_comp_error_info) * SKD_N_COMPLETION_ENTRY; - - memset(skdev->skcomp_table, 0, nbytes); + memset(skdev->skcomp_table, 0, SKD_SKCOMP_SIZE); skdev->skcomp_ix = 0; skdev->skcomp_cycle = 1; @@ -2941,7 +1667,7 @@ static void skd_completion_worker(struct work_struct *work) * process everything in compq */ skd_isr_completion_posted(skdev, 0, &flush_enqueued); - skd_request_fn(skdev->queue); + schedule_work(&skdev->start_queue); spin_unlock_irqrestore(&skdev->lock, flags); } @@ -2951,14 +1677,13 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev); static irqreturn_t skd_isr(int irq, void *ptr) { - struct skd_device *skdev; + struct skd_device *skdev = ptr; u32 intstat; u32 ack; int rc = 0; int deferred = 0; int flush_enqueued = 0; - skdev = (struct skd_device *)ptr; spin_lock(&skdev->lock); for (;; ) { @@ -2967,8 +1692,8 @@ skd_isr(int irq, void *ptr) ack = FIT_INT_DEF_MASK; ack &= intstat; - pr_debug("%s:%s:%d intstat=0x%x ack=0x%x\n", - skdev->name, __func__, __LINE__, intstat, ack); + dev_dbg(&skdev->pdev->dev, "intstat=0x%x ack=0x%x\n", intstat, + ack); /* As long as there is an int pending on device, keep * running loop. When none, get out, but if we've never @@ -3018,12 +1743,12 @@ skd_isr(int irq, void *ptr) } if (unlikely(flush_enqueued)) - skd_request_fn(skdev->queue); + schedule_work(&skdev->start_queue); if (deferred) schedule_work(&skdev->completion_worker); else if (!flush_enqueued) - skd_request_fn(skdev->queue); + schedule_work(&skdev->start_queue); spin_unlock(&skdev->lock); @@ -3033,13 +1758,13 @@ skd_isr(int irq, void *ptr) static void skd_drive_fault(struct skd_device *skdev) { skdev->state = SKD_DRVR_STATE_FAULT; - pr_err("(%s): Drive FAULT\n", skd_name(skdev)); + dev_err(&skdev->pdev->dev, "Drive FAULT\n"); } static void skd_drive_disappeared(struct skd_device *skdev) { skdev->state = SKD_DRVR_STATE_DISAPPEARED; - pr_err("(%s): Drive DISAPPEARED\n", skd_name(skdev)); + dev_err(&skdev->pdev->dev, "Drive DISAPPEARED\n"); } static void skd_isr_fwstate(struct skd_device *skdev) @@ -3052,10 +1777,9 @@ static void skd_isr_fwstate(struct skd_device *skdev) sense = SKD_READL(skdev, FIT_STATUS); state = sense & FIT_SR_DRIVE_STATE_MASK; - pr_err("(%s): s1120 state %s(%d)=>%s(%d)\n", - skd_name(skdev), - skd_drive_state_to_str(skdev->drive_state), skdev->drive_state, - skd_drive_state_to_str(state), state); + dev_err(&skdev->pdev->dev, "s1120 state %s(%d)=>%s(%d)\n", + skd_drive_state_to_str(skdev->drive_state), skdev->drive_state, + skd_drive_state_to_str(state), state); skdev->drive_state = state; @@ -3066,7 +1790,7 @@ static void skd_isr_fwstate(struct skd_device *skdev) break; } if (skdev->state == SKD_DRVR_STATE_RESTARTING) - skd_recover_requests(skdev, 0); + skd_recover_requests(skdev); if (skdev->state == SKD_DRVR_STATE_WAIT_BOOT) { skdev->timer_countdown = SKD_STARTING_TIMO; skdev->state = SKD_DRVR_STATE_STARTING; @@ -3087,11 +1811,11 @@ static void skd_isr_fwstate(struct skd_device *skdev) skdev->cur_max_queue_depth * 2 / 3 + 1; if (skdev->queue_low_water_mark < 1) skdev->queue_low_water_mark = 1; - pr_info( - "(%s): Queue depth limit=%d dev=%d lowat=%d\n", - skd_name(skdev), - skdev->cur_max_queue_depth, - skdev->dev_max_queue_depth, skdev->queue_low_water_mark); + dev_info(&skdev->pdev->dev, + "Queue depth limit=%d dev=%d lowat=%d\n", + skdev->cur_max_queue_depth, + skdev->dev_max_queue_depth, + skdev->queue_low_water_mark); skd_refresh_device_data(skdev); break; @@ -3107,7 +1831,7 @@ static void skd_isr_fwstate(struct skd_device *skdev) */ skdev->state = SKD_DRVR_STATE_BUSY_SANITIZE; skdev->timer_countdown = SKD_TIMER_SECONDS(3); - blk_start_queue(skdev->queue); + schedule_work(&skdev->start_queue); break; case FIT_SR_DRIVE_BUSY_ERASE: skdev->state = SKD_DRVR_STATE_BUSY_ERASE; @@ -3128,8 +1852,7 @@ static void skd_isr_fwstate(struct skd_device *skdev) } break; case FIT_SR_DRIVE_FW_BOOTING: - pr_debug("%s:%s:%d ISR FIT_SR_DRIVE_FW_BOOTING %s\n", - skdev->name, __func__, __LINE__, skdev->name); + dev_dbg(&skdev->pdev->dev, "ISR FIT_SR_DRIVE_FW_BOOTING\n"); skdev->state = SKD_DRVR_STATE_WAIT_BOOT; skdev->timer_countdown = SKD_WAIT_BOOT_TIMO; break; @@ -3141,17 +1864,17 @@ static void skd_isr_fwstate(struct skd_device *skdev) case FIT_SR_DRIVE_FAULT: skd_drive_fault(skdev); - skd_recover_requests(skdev, 0); - blk_start_queue(skdev->queue); + skd_recover_requests(skdev); + schedule_work(&skdev->start_queue); break; /* PCIe bus returned all Fs? */ case 0xFF: - pr_info("(%s): state=0x%x sense=0x%x\n", - skd_name(skdev), state, sense); + dev_info(&skdev->pdev->dev, "state=0x%x sense=0x%x\n", state, + sense); skd_drive_disappeared(skdev); - skd_recover_requests(skdev, 0); - blk_start_queue(skdev->queue); + skd_recover_requests(skdev); + schedule_work(&skdev->start_queue); break; default: /* @@ -3159,92 +1882,33 @@ static void skd_isr_fwstate(struct skd_device *skdev) */ break; } - pr_err("(%s): Driver state %s(%d)=>%s(%d)\n", - skd_name(skdev), - skd_skdev_state_to_str(prev_driver_state), prev_driver_state, - skd_skdev_state_to_str(skdev->state), skdev->state); + dev_err(&skdev->pdev->dev, "Driver state %s(%d)=>%s(%d)\n", + skd_skdev_state_to_str(prev_driver_state), prev_driver_state, + skd_skdev_state_to_str(skdev->state), skdev->state); } -static void skd_recover_requests(struct skd_device *skdev, int requeue) +static void skd_recover_request(struct request *req, void *data, bool reserved) { - int i; - - for (i = 0; i < skdev->num_req_context; i++) { - struct skd_request_context *skreq = &skdev->skreq_table[i]; - - if (skreq->state == SKD_REQ_STATE_BUSY) { - skd_log_skreq(skdev, skreq, "recover"); - - SKD_ASSERT((skreq->id & SKD_ID_INCR) != 0); - SKD_ASSERT(skreq->req != NULL); - - /* Release DMA resources for the request. */ - if (skreq->n_sg > 0) - skd_postop_sg_list(skdev, skreq); - - if (requeue && - (unsigned long) ++skreq->req->special < - SKD_MAX_RETRIES) - blk_requeue_request(skdev->queue, skreq->req); - else - skd_end_request(skdev, skreq, BLK_STS_IOERR); - - skreq->req = NULL; - - skreq->state = SKD_REQ_STATE_IDLE; - skreq->id += SKD_ID_INCR; - } - if (i > 0) - skreq[-1].next = skreq; - skreq->next = NULL; - } - skdev->skreq_free_list = skdev->skreq_table; + struct skd_device *const skdev = data; + struct skd_request_context *skreq = blk_mq_rq_to_pdu(req); - for (i = 0; i < skdev->num_fitmsg_context; i++) { - struct skd_fitmsg_context *skmsg = &skdev->skmsg_table[i]; + if (skreq->state != SKD_REQ_STATE_BUSY) + return; - if (skmsg->state == SKD_MSG_STATE_BUSY) { - skd_log_skmsg(skdev, skmsg, "salvaged"); - SKD_ASSERT((skmsg->id & SKD_ID_INCR) != 0); - skmsg->state = SKD_MSG_STATE_IDLE; - skmsg->id += SKD_ID_INCR; - } - if (i > 0) - skmsg[-1].next = skmsg; - skmsg->next = NULL; - } - skdev->skmsg_free_list = skdev->skmsg_table; + skd_log_skreq(skdev, skreq, "recover"); - for (i = 0; i < skdev->n_special; i++) { - struct skd_special_context *skspcl = &skdev->skspcl_table[i]; + /* Release DMA resources for the request. */ + if (skreq->n_sg > 0) + skd_postop_sg_list(skdev, skreq); - /* If orphaned, reclaim it because it has already been reported - * to the process as an error (it was just waiting for - * a completion that didn't come, and now it will never come) - * If busy, change to a state that will cause it to error - * out in the wait routine and let it do the normal - * reporting and reclaiming - */ - if (skspcl->req.state == SKD_REQ_STATE_BUSY) { - if (skspcl->orphaned) { - pr_debug("%s:%s:%d orphaned %p\n", - skdev->name, __func__, __LINE__, - skspcl); - skd_release_special(skdev, skspcl); - } else { - pr_debug("%s:%s:%d not orphaned %p\n", - skdev->name, __func__, __LINE__, - skspcl); - skspcl->req.state = SKD_REQ_STATE_ABORTED; - } - } - } - skdev->skspcl_free_list = skdev->skspcl_table; - - for (i = 0; i < SKD_N_TIMEOUT_SLOT; i++) - skdev->timeout_slot[i] = 0; + skreq->state = SKD_REQ_STATE_IDLE; + skreq->status = BLK_STS_IOERR; + blk_mq_complete_request(req); +} - skdev->in_flight = 0; +static void skd_recover_requests(struct skd_device *skdev) +{ + blk_mq_tagset_busy_iter(&skdev->tag_set, skd_recover_request, skdev); } static void skd_isr_msg_from_dev(struct skd_device *skdev) @@ -3255,8 +1919,8 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev) mfd = SKD_READL(skdev, FIT_MSG_FROM_DEVICE); - pr_debug("%s:%s:%d mfd=0x%x last_mtd=0x%x\n", - skdev->name, __func__, __LINE__, mfd, skdev->last_mtd); + dev_dbg(&skdev->pdev->dev, "mfd=0x%x last_mtd=0x%x\n", mfd, + skdev->last_mtd); /* ignore any mtd that is an ack for something we didn't send */ if (FIT_MXD_TYPE(mfd) != FIT_MXD_TYPE(skdev->last_mtd)) @@ -3267,13 +1931,10 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev) skdev->proto_ver = FIT_PROTOCOL_MAJOR_VER(mfd); if (skdev->proto_ver != FIT_PROTOCOL_VERSION_1) { - pr_err("(%s): protocol mismatch\n", - skdev->name); - pr_err("(%s): got=%d support=%d\n", - skdev->name, skdev->proto_ver, - FIT_PROTOCOL_VERSION_1); - pr_err("(%s): please upgrade driver\n", - skdev->name); + dev_err(&skdev->pdev->dev, "protocol mismatch\n"); + dev_err(&skdev->pdev->dev, " got=%d support=%d\n", + skdev->proto_ver, FIT_PROTOCOL_VERSION_1); + dev_err(&skdev->pdev->dev, " please upgrade driver\n"); skdev->state = SKD_DRVR_STATE_PROTOCOL_MISMATCH; skd_soft_reset(skdev); break; @@ -3327,9 +1988,8 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev) SKD_WRITEL(skdev, mtd, FIT_MSG_TO_DEVICE); skdev->last_mtd = mtd; - pr_err("(%s): Time sync driver=0x%x device=0x%x\n", - skd_name(skdev), - skdev->connect_time_stamp, skdev->drive_jiffies); + dev_err(&skdev->pdev->dev, "Time sync driver=0x%x device=0x%x\n", + skdev->connect_time_stamp, skdev->drive_jiffies); break; case FIT_MTD_ARM_QUEUE: @@ -3351,8 +2011,7 @@ static void skd_disable_interrupts(struct skd_device *skdev) sense = SKD_READL(skdev, FIT_CONTROL); sense &= ~FIT_CR_ENABLE_INTERRUPTS; SKD_WRITEL(skdev, sense, FIT_CONTROL); - pr_debug("%s:%s:%d sense 0x%x\n", - skdev->name, __func__, __LINE__, sense); + dev_dbg(&skdev->pdev->dev, "sense 0x%x\n", sense); /* Note that the 1s is written. A 1-bit means * disable, a 0 means enable. @@ -3371,13 +2030,11 @@ static void skd_enable_interrupts(struct skd_device *skdev) /* Note that the compliment of mask is written. A 1-bit means * disable, a 0 means enable. */ SKD_WRITEL(skdev, ~val, FIT_INT_MASK_HOST); - pr_debug("%s:%s:%d interrupt mask=0x%x\n", - skdev->name, __func__, __LINE__, ~val); + dev_dbg(&skdev->pdev->dev, "interrupt mask=0x%x\n", ~val); val = SKD_READL(skdev, FIT_CONTROL); val |= FIT_CR_ENABLE_INTERRUPTS; - pr_debug("%s:%s:%d control=0x%x\n", - skdev->name, __func__, __LINE__, val); + dev_dbg(&skdev->pdev->dev, "control=0x%x\n", val); SKD_WRITEL(skdev, val, FIT_CONTROL); } @@ -3393,8 +2050,7 @@ static void skd_soft_reset(struct skd_device *skdev) val = SKD_READL(skdev, FIT_CONTROL); val |= (FIT_CR_SOFT_RESET); - pr_debug("%s:%s:%d control=0x%x\n", - skdev->name, __func__, __LINE__, val); + dev_dbg(&skdev->pdev->dev, "control=0x%x\n", val); SKD_WRITEL(skdev, val, FIT_CONTROL); } @@ -3411,8 +2067,7 @@ static void skd_start_device(struct skd_device *skdev) sense = SKD_READL(skdev, FIT_STATUS); - pr_debug("%s:%s:%d initial status=0x%x\n", - skdev->name, __func__, __LINE__, sense); + dev_dbg(&skdev->pdev->dev, "initial status=0x%x\n", sense); state = sense & FIT_SR_DRIVE_STATE_MASK; skdev->drive_state = state; @@ -3425,25 +2080,23 @@ static void skd_start_device(struct skd_device *skdev) switch (skdev->drive_state) { case FIT_SR_DRIVE_OFFLINE: - pr_err("(%s): Drive offline...\n", skd_name(skdev)); + dev_err(&skdev->pdev->dev, "Drive offline...\n"); break; case FIT_SR_DRIVE_FW_BOOTING: - pr_debug("%s:%s:%d FIT_SR_DRIVE_FW_BOOTING %s\n", - skdev->name, __func__, __LINE__, skdev->name); + dev_dbg(&skdev->pdev->dev, "FIT_SR_DRIVE_FW_BOOTING\n"); skdev->state = SKD_DRVR_STATE_WAIT_BOOT; skdev->timer_countdown = SKD_WAIT_BOOT_TIMO; break; case FIT_SR_DRIVE_BUSY_SANITIZE: - pr_info("(%s): Start: BUSY_SANITIZE\n", - skd_name(skdev)); + dev_info(&skdev->pdev->dev, "Start: BUSY_SANITIZE\n"); skdev->state = SKD_DRVR_STATE_BUSY_SANITIZE; skdev->timer_countdown = SKD_STARTED_BUSY_TIMO; break; case FIT_SR_DRIVE_BUSY_ERASE: - pr_info("(%s): Start: BUSY_ERASE\n", skd_name(skdev)); + dev_info(&skdev->pdev->dev, "Start: BUSY_ERASE\n"); skdev->state = SKD_DRVR_STATE_BUSY_ERASE; skdev->timer_countdown = SKD_STARTED_BUSY_TIMO; break; @@ -3454,14 +2107,13 @@ static void skd_start_device(struct skd_device *skdev) break; case FIT_SR_DRIVE_BUSY: - pr_err("(%s): Drive Busy...\n", skd_name(skdev)); + dev_err(&skdev->pdev->dev, "Drive Busy...\n"); skdev->state = SKD_DRVR_STATE_BUSY; skdev->timer_countdown = SKD_STARTED_BUSY_TIMO; break; case FIT_SR_DRIVE_SOFT_RESET: - pr_err("(%s) drive soft reset in prog\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, "drive soft reset in prog\n"); break; case FIT_SR_DRIVE_FAULT: @@ -3471,9 +2123,8 @@ static void skd_start_device(struct skd_device *skdev) */ skd_drive_fault(skdev); /*start the queue so we can respond with error to requests */ - pr_debug("%s:%s:%d starting %s queue\n", - skdev->name, __func__, __LINE__, skdev->name); - blk_start_queue(skdev->queue); + dev_dbg(&skdev->pdev->dev, "starting queue\n"); + schedule_work(&skdev->start_queue); skdev->gendisk_on = -1; wake_up_interruptible(&skdev->waitq); break; @@ -3483,38 +2134,33 @@ static void skd_start_device(struct skd_device *skdev) * to the BAR1 addresses. */ skd_drive_disappeared(skdev); /*start the queue so we can respond with error to requests */ - pr_debug("%s:%s:%d starting %s queue to error-out reqs\n", - skdev->name, __func__, __LINE__, skdev->name); - blk_start_queue(skdev->queue); + dev_dbg(&skdev->pdev->dev, + "starting queue to error-out reqs\n"); + schedule_work(&skdev->start_queue); skdev->gendisk_on = -1; wake_up_interruptible(&skdev->waitq); break; default: - pr_err("(%s) Start: unknown state %x\n", - skd_name(skdev), skdev->drive_state); + dev_err(&skdev->pdev->dev, "Start: unknown state %x\n", + skdev->drive_state); break; } state = SKD_READL(skdev, FIT_CONTROL); - pr_debug("%s:%s:%d FIT Control Status=0x%x\n", - skdev->name, __func__, __LINE__, state); + dev_dbg(&skdev->pdev->dev, "FIT Control Status=0x%x\n", state); state = SKD_READL(skdev, FIT_INT_STATUS_HOST); - pr_debug("%s:%s:%d Intr Status=0x%x\n", - skdev->name, __func__, __LINE__, state); + dev_dbg(&skdev->pdev->dev, "Intr Status=0x%x\n", state); state = SKD_READL(skdev, FIT_INT_MASK_HOST); - pr_debug("%s:%s:%d Intr Mask=0x%x\n", - skdev->name, __func__, __LINE__, state); + dev_dbg(&skdev->pdev->dev, "Intr Mask=0x%x\n", state); state = SKD_READL(skdev, FIT_MSG_FROM_DEVICE); - pr_debug("%s:%s:%d Msg from Dev=0x%x\n", - skdev->name, __func__, __LINE__, state); + dev_dbg(&skdev->pdev->dev, "Msg from Dev=0x%x\n", state); state = SKD_READL(skdev, FIT_HW_VERSION); - pr_debug("%s:%s:%d HW version=0x%x\n", - skdev->name, __func__, __LINE__, state); + dev_dbg(&skdev->pdev->dev, "HW version=0x%x\n", state); spin_unlock_irqrestore(&skdev->lock, flags); } @@ -3529,14 +2175,12 @@ static void skd_stop_device(struct skd_device *skdev) spin_lock_irqsave(&skdev->lock, flags); if (skdev->state != SKD_DRVR_STATE_ONLINE) { - pr_err("(%s): skd_stop_device not online no sync\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, "%s not online no sync\n", __func__); goto stop_out; } if (skspcl->req.state != SKD_REQ_STATE_IDLE) { - pr_err("(%s): skd_stop_device no special\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, "%s no special\n", __func__); goto stop_out; } @@ -3554,16 +2198,13 @@ static void skd_stop_device(struct skd_device *skdev) switch (skdev->sync_done) { case 0: - pr_err("(%s): skd_stop_device no sync\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, "%s no sync\n", __func__); break; case 1: - pr_err("(%s): skd_stop_device sync done\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, "%s sync done\n", __func__); break; default: - pr_err("(%s): skd_stop_device sync error\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, "%s sync error\n", __func__); } stop_out: @@ -3593,8 +2234,8 @@ stop_out: } if (dev_state != FIT_SR_DRIVE_INIT) - pr_err("(%s): skd_stop_device state error 0x%02x\n", - skd_name(skdev), dev_state); + dev_err(&skdev->pdev->dev, "%s state error 0x%02x\n", __func__, + dev_state); } /* assume spinlock is held */ @@ -3607,8 +2248,7 @@ static void skd_restart_device(struct skd_device *skdev) state = SKD_READL(skdev, FIT_STATUS); - pr_debug("%s:%s:%d drive status=0x%x\n", - skdev->name, __func__, __LINE__, state); + dev_dbg(&skdev->pdev->dev, "drive status=0x%x\n", state); state &= FIT_SR_DRIVE_STATE_MASK; skdev->drive_state = state; @@ -3628,9 +2268,8 @@ static int skd_quiesce_dev(struct skd_device *skdev) switch (skdev->state) { case SKD_DRVR_STATE_BUSY: case SKD_DRVR_STATE_BUSY_IMMINENT: - pr_debug("%s:%s:%d stopping %s queue\n", - skdev->name, __func__, __LINE__, skdev->name); - blk_stop_queue(skdev->queue); + dev_dbg(&skdev->pdev->dev, "stopping queue\n"); + blk_mq_stop_hw_queues(skdev->queue); break; case SKD_DRVR_STATE_ONLINE: case SKD_DRVR_STATE_STOPPING: @@ -3642,8 +2281,8 @@ static int skd_quiesce_dev(struct skd_device *skdev) case SKD_DRVR_STATE_RESUMING: default: rc = -EINVAL; - pr_debug("%s:%s:%d state [%d] not implemented\n", - skdev->name, __func__, __LINE__, skdev->state); + dev_dbg(&skdev->pdev->dev, "state [%d] not implemented\n", + skdev->state); } return rc; } @@ -3655,8 +2294,7 @@ static int skd_unquiesce_dev(struct skd_device *skdev) skd_log_skdev(skdev, "unquiesce"); if (skdev->state == SKD_DRVR_STATE_ONLINE) { - pr_debug("%s:%s:%d **** device already ONLINE\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "**** device already ONLINE\n"); return 0; } if (skdev->drive_state != FIT_SR_DRIVE_ONLINE) { @@ -3669,8 +2307,7 @@ static int skd_unquiesce_dev(struct skd_device *skdev) * to become available. */ skdev->state = SKD_DRVR_STATE_BUSY; - pr_debug("%s:%s:%d drive BUSY state\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "drive BUSY state\n"); return 0; } @@ -3689,26 +2326,24 @@ static int skd_unquiesce_dev(struct skd_device *skdev) case SKD_DRVR_STATE_IDLE: case SKD_DRVR_STATE_LOAD: skdev->state = SKD_DRVR_STATE_ONLINE; - pr_err("(%s): Driver state %s(%d)=>%s(%d)\n", - skd_name(skdev), - skd_skdev_state_to_str(prev_driver_state), - prev_driver_state, skd_skdev_state_to_str(skdev->state), - skdev->state); - pr_debug("%s:%s:%d **** device ONLINE...starting block queue\n", - skdev->name, __func__, __LINE__); - pr_debug("%s:%s:%d starting %s queue\n", - skdev->name, __func__, __LINE__, skdev->name); - pr_info("(%s): STEC s1120 ONLINE\n", skd_name(skdev)); - blk_start_queue(skdev->queue); + dev_err(&skdev->pdev->dev, "Driver state %s(%d)=>%s(%d)\n", + skd_skdev_state_to_str(prev_driver_state), + prev_driver_state, skd_skdev_state_to_str(skdev->state), + skdev->state); + dev_dbg(&skdev->pdev->dev, + "**** device ONLINE...starting block queue\n"); + dev_dbg(&skdev->pdev->dev, "starting queue\n"); + dev_info(&skdev->pdev->dev, "STEC s1120 ONLINE\n"); + schedule_work(&skdev->start_queue); skdev->gendisk_on = 1; wake_up_interruptible(&skdev->waitq); break; case SKD_DRVR_STATE_DISAPPEARED: default: - pr_debug("%s:%s:%d **** driver state %d, not implemented \n", - skdev->name, __func__, __LINE__, - skdev->state); + dev_dbg(&skdev->pdev->dev, + "**** driver state %d, not implemented\n", + skdev->state); return -EBUSY; } return 0; @@ -3726,11 +2361,10 @@ static irqreturn_t skd_reserved_isr(int irq, void *skd_host_data) unsigned long flags; spin_lock_irqsave(&skdev->lock, flags); - pr_debug("%s:%s:%d MSIX = 0x%x\n", - skdev->name, __func__, __LINE__, - SKD_READL(skdev, FIT_INT_STATUS_HOST)); - pr_err("(%s): MSIX reserved irq %d = 0x%x\n", skd_name(skdev), - irq, SKD_READL(skdev, FIT_INT_STATUS_HOST)); + dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n", + SKD_READL(skdev, FIT_INT_STATUS_HOST)); + dev_err(&skdev->pdev->dev, "MSIX reserved irq %d = 0x%x\n", irq, + SKD_READL(skdev, FIT_INT_STATUS_HOST)); SKD_WRITEL(skdev, FIT_INT_RESERVED_MASK, FIT_INT_STATUS_HOST); spin_unlock_irqrestore(&skdev->lock, flags); return IRQ_HANDLED; @@ -3742,9 +2376,8 @@ static irqreturn_t skd_statec_isr(int irq, void *skd_host_data) unsigned long flags; spin_lock_irqsave(&skdev->lock, flags); - pr_debug("%s:%s:%d MSIX = 0x%x\n", - skdev->name, __func__, __LINE__, - SKD_READL(skdev, FIT_INT_STATUS_HOST)); + dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n", + SKD_READL(skdev, FIT_INT_STATUS_HOST)); SKD_WRITEL(skdev, FIT_ISH_FW_STATE_CHANGE, FIT_INT_STATUS_HOST); skd_isr_fwstate(skdev); spin_unlock_irqrestore(&skdev->lock, flags); @@ -3759,19 +2392,18 @@ static irqreturn_t skd_comp_q(int irq, void *skd_host_data) int deferred; spin_lock_irqsave(&skdev->lock, flags); - pr_debug("%s:%s:%d MSIX = 0x%x\n", - skdev->name, __func__, __LINE__, - SKD_READL(skdev, FIT_INT_STATUS_HOST)); + dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n", + SKD_READL(skdev, FIT_INT_STATUS_HOST)); SKD_WRITEL(skdev, FIT_ISH_COMPLETION_POSTED, FIT_INT_STATUS_HOST); deferred = skd_isr_completion_posted(skdev, skd_isr_comp_limit, &flush_enqueued); if (flush_enqueued) - skd_request_fn(skdev->queue); + schedule_work(&skdev->start_queue); if (deferred) schedule_work(&skdev->completion_worker); else if (!flush_enqueued) - skd_request_fn(skdev->queue); + schedule_work(&skdev->start_queue); spin_unlock_irqrestore(&skdev->lock, flags); @@ -3784,9 +2416,8 @@ static irqreturn_t skd_msg_isr(int irq, void *skd_host_data) unsigned long flags; spin_lock_irqsave(&skdev->lock, flags); - pr_debug("%s:%s:%d MSIX = 0x%x\n", - skdev->name, __func__, __LINE__, - SKD_READL(skdev, FIT_INT_STATUS_HOST)); + dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n", + SKD_READL(skdev, FIT_INT_STATUS_HOST)); SKD_WRITEL(skdev, FIT_ISH_MSG_FROM_DEV, FIT_INT_STATUS_HOST); skd_isr_msg_from_dev(skdev); spin_unlock_irqrestore(&skdev->lock, flags); @@ -3799,9 +2430,8 @@ static irqreturn_t skd_qfull_isr(int irq, void *skd_host_data) unsigned long flags; spin_lock_irqsave(&skdev->lock, flags); - pr_debug("%s:%s:%d MSIX = 0x%x\n", - skdev->name, __func__, __LINE__, - SKD_READL(skdev, FIT_INT_STATUS_HOST)); + dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n", + SKD_READL(skdev, FIT_INT_STATUS_HOST)); SKD_WRITEL(skdev, FIT_INT_QUEUE_FULL, FIT_INT_STATUS_HOST); spin_unlock_irqrestore(&skdev->lock, flags); return IRQ_HANDLED; @@ -3850,8 +2480,7 @@ static int skd_acquire_msix(struct skd_device *skdev) rc = pci_alloc_irq_vectors(pdev, SKD_MAX_MSIX_COUNT, SKD_MAX_MSIX_COUNT, PCI_IRQ_MSIX); if (rc < 0) { - pr_err("(%s): failed to enable MSI-X %d\n", - skd_name(skdev), rc); + dev_err(&skdev->pdev->dev, "failed to enable MSI-X %d\n", rc); goto out; } @@ -3859,8 +2488,7 @@ static int skd_acquire_msix(struct skd_device *skdev) sizeof(struct skd_msix_entry), GFP_KERNEL); if (!skdev->msix_entries) { rc = -ENOMEM; - pr_err("(%s): msix table allocation error\n", - skd_name(skdev)); + dev_err(&skdev->pdev->dev, "msix table allocation error\n"); goto out; } @@ -3877,16 +2505,15 @@ static int skd_acquire_msix(struct skd_device *skdev) msix_entries[i].handler, 0, qentry->isr_name, skdev); if (rc) { - pr_err("(%s): Unable to register(%d) MSI-X " - "handler %d: %s\n", - skd_name(skdev), rc, i, qentry->isr_name); + dev_err(&skdev->pdev->dev, + "Unable to register(%d) MSI-X handler %d: %s\n", + rc, i, qentry->isr_name); goto msix_out; } } - pr_debug("%s:%s:%d %s: <%s> msix %d irq(s) enabled\n", - skdev->name, __func__, __LINE__, - pci_name(pdev), skdev->name, SKD_MAX_MSIX_COUNT); + dev_dbg(&skdev->pdev->dev, "%d msix irq(s) enabled\n", + SKD_MAX_MSIX_COUNT); return 0; msix_out: @@ -3909,8 +2536,8 @@ static int skd_acquire_irq(struct skd_device *skdev) if (!rc) return 0; - pr_err("(%s): failed to enable MSI-X, re-trying with MSI %d\n", - skd_name(skdev), rc); + dev_err(&skdev->pdev->dev, + "failed to enable MSI-X, re-trying with MSI %d\n", rc); } snprintf(skdev->isr_name, sizeof(skdev->isr_name), "%s%d", DRV_NAME, @@ -3920,8 +2547,8 @@ static int skd_acquire_irq(struct skd_device *skdev) irq_flag |= PCI_IRQ_MSI; rc = pci_alloc_irq_vectors(pdev, 1, 1, irq_flag); if (rc < 0) { - pr_err("(%s): failed to allocate the MSI interrupt %d\n", - skd_name(skdev), rc); + dev_err(&skdev->pdev->dev, + "failed to allocate the MSI interrupt %d\n", rc); return rc; } @@ -3930,8 +2557,8 @@ static int skd_acquire_irq(struct skd_device *skdev) skdev->isr_name, skdev); if (rc) { pci_free_irq_vectors(pdev); - pr_err("(%s): failed to allocate interrupt %d\n", - skd_name(skdev), rc); + dev_err(&skdev->pdev->dev, "failed to allocate interrupt %d\n", + rc); return rc; } @@ -3965,20 +2592,45 @@ static void skd_release_irq(struct skd_device *skdev) ***************************************************************************** */ +static void *skd_alloc_dma(struct skd_device *skdev, struct kmem_cache *s, + dma_addr_t *dma_handle, gfp_t gfp, + enum dma_data_direction dir) +{ + struct device *dev = &skdev->pdev->dev; + void *buf; + + buf = kmem_cache_alloc(s, gfp); + if (!buf) + return NULL; + *dma_handle = dma_map_single(dev, buf, s->size, dir); + if (dma_mapping_error(dev, *dma_handle)) { + kfree(buf); + buf = NULL; + } + return buf; +} + +static void skd_free_dma(struct skd_device *skdev, struct kmem_cache *s, + void *vaddr, dma_addr_t dma_handle, + enum dma_data_direction dir) +{ + if (!vaddr) + return; + + dma_unmap_single(&skdev->pdev->dev, dma_handle, s->size, dir); + kmem_cache_free(s, vaddr); +} + static int skd_cons_skcomp(struct skd_device *skdev) { int rc = 0; struct fit_completion_entry_v1 *skcomp; - u32 nbytes; - nbytes = sizeof(*skcomp) * SKD_N_COMPLETION_ENTRY; - nbytes += sizeof(struct fit_comp_error_info) * SKD_N_COMPLETION_ENTRY; + dev_dbg(&skdev->pdev->dev, + "comp pci_alloc, total bytes %zd entries %d\n", + SKD_SKCOMP_SIZE, SKD_N_COMPLETION_ENTRY); - pr_debug("%s:%s:%d comp pci_alloc, total bytes %d entries %d\n", - skdev->name, __func__, __LINE__, - nbytes, SKD_N_COMPLETION_ENTRY); - - skcomp = pci_zalloc_consistent(skdev->pdev, nbytes, + skcomp = pci_zalloc_consistent(skdev->pdev, SKD_SKCOMP_SIZE, &skdev->cq_dma_address); if (skcomp == NULL) { @@ -4000,14 +2652,14 @@ static int skd_cons_skmsg(struct skd_device *skdev) int rc = 0; u32 i; - pr_debug("%s:%s:%d skmsg_table kzalloc, struct %lu, count %u total %lu\n", - skdev->name, __func__, __LINE__, - sizeof(struct skd_fitmsg_context), - skdev->num_fitmsg_context, - sizeof(struct skd_fitmsg_context) * skdev->num_fitmsg_context); + dev_dbg(&skdev->pdev->dev, + "skmsg_table kcalloc, struct %lu, count %u total %lu\n", + sizeof(struct skd_fitmsg_context), skdev->num_fitmsg_context, + sizeof(struct skd_fitmsg_context) * skdev->num_fitmsg_context); - skdev->skmsg_table = kzalloc(sizeof(struct skd_fitmsg_context) - *skdev->num_fitmsg_context, GFP_KERNEL); + skdev->skmsg_table = kcalloc(skdev->num_fitmsg_context, + sizeof(struct skd_fitmsg_context), + GFP_KERNEL); if (skdev->skmsg_table == NULL) { rc = -ENOMEM; goto err_out; @@ -4020,9 +2672,8 @@ static int skd_cons_skmsg(struct skd_device *skdev) skmsg->id = i + SKD_ID_FIT_MSG; - skmsg->state = SKD_MSG_STATE_IDLE; skmsg->msg_buf = pci_alloc_consistent(skdev->pdev, - SKD_N_FITMSG_BYTES + 64, + SKD_N_FITMSG_BYTES, &skmsg->mb_dma_address); if (skmsg->msg_buf == NULL) { @@ -4030,22 +2681,13 @@ static int skd_cons_skmsg(struct skd_device *skdev) goto err_out; } - skmsg->offset = (u32)((u64)skmsg->msg_buf & - (~FIT_QCMD_BASE_ADDRESS_MASK)); - skmsg->msg_buf += ~FIT_QCMD_BASE_ADDRESS_MASK; - skmsg->msg_buf = (u8 *)((u64)skmsg->msg_buf & - FIT_QCMD_BASE_ADDRESS_MASK); - skmsg->mb_dma_address += ~FIT_QCMD_BASE_ADDRESS_MASK; - skmsg->mb_dma_address &= FIT_QCMD_BASE_ADDRESS_MASK; + WARN(((uintptr_t)skmsg->msg_buf | skmsg->mb_dma_address) & + (FIT_QCMD_ALIGN - 1), + "not aligned: msg_buf %p mb_dma_address %#llx\n", + skmsg->msg_buf, skmsg->mb_dma_address); memset(skmsg->msg_buf, 0, SKD_N_FITMSG_BYTES); - - skmsg->next = &skmsg[1]; } - /* Free list is in order starting with the 0th entry. */ - skdev->skmsg_table[i - 1].next = NULL; - skdev->skmsg_free_list = skdev->skmsg_table; - err_out: return rc; } @@ -4055,18 +2697,14 @@ static struct fit_sg_descriptor *skd_cons_sg_list(struct skd_device *skdev, dma_addr_t *ret_dma_addr) { struct fit_sg_descriptor *sg_list; - u32 nbytes; - - nbytes = sizeof(*sg_list) * n_sg; - sg_list = pci_alloc_consistent(skdev->pdev, nbytes, ret_dma_addr); + sg_list = skd_alloc_dma(skdev, skdev->sglist_cache, ret_dma_addr, + GFP_DMA | __GFP_ZERO, DMA_TO_DEVICE); if (sg_list != NULL) { uint64_t dma_address = *ret_dma_addr; u32 i; - memset(sg_list, 0, nbytes); - for (i = 0; i < n_sg - 1; i++) { uint64_t ndp_off; ndp_off = (i + 1) * sizeof(struct fit_sg_descriptor); @@ -4079,153 +2717,63 @@ static struct fit_sg_descriptor *skd_cons_sg_list(struct skd_device *skdev, return sg_list; } -static int skd_cons_skreq(struct skd_device *skdev) +static void skd_free_sg_list(struct skd_device *skdev, + struct fit_sg_descriptor *sg_list, + dma_addr_t dma_addr) { - int rc = 0; - u32 i; - - pr_debug("%s:%s:%d skreq_table kzalloc, struct %lu, count %u total %lu\n", - skdev->name, __func__, __LINE__, - sizeof(struct skd_request_context), - skdev->num_req_context, - sizeof(struct skd_request_context) * skdev->num_req_context); - - skdev->skreq_table = kzalloc(sizeof(struct skd_request_context) - * skdev->num_req_context, GFP_KERNEL); - if (skdev->skreq_table == NULL) { - rc = -ENOMEM; - goto err_out; - } - - pr_debug("%s:%s:%d alloc sg_table sg_per_req %u scatlist %lu total %lu\n", - skdev->name, __func__, __LINE__, - skdev->sgs_per_request, sizeof(struct scatterlist), - skdev->sgs_per_request * sizeof(struct scatterlist)); - - for (i = 0; i < skdev->num_req_context; i++) { - struct skd_request_context *skreq; - - skreq = &skdev->skreq_table[i]; - - skreq->id = i + SKD_ID_RW_REQUEST; - skreq->state = SKD_REQ_STATE_IDLE; - - skreq->sg = kzalloc(sizeof(struct scatterlist) * - skdev->sgs_per_request, GFP_KERNEL); - if (skreq->sg == NULL) { - rc = -ENOMEM; - goto err_out; - } - sg_init_table(skreq->sg, skdev->sgs_per_request); - - skreq->sksg_list = skd_cons_sg_list(skdev, - skdev->sgs_per_request, - &skreq->sksg_dma_address); - - if (skreq->sksg_list == NULL) { - rc = -ENOMEM; - goto err_out; - } - - skreq->next = &skreq[1]; - } - - /* Free list is in order starting with the 0th entry. */ - skdev->skreq_table[i - 1].next = NULL; - skdev->skreq_free_list = skdev->skreq_table; + if (WARN_ON_ONCE(!sg_list)) + return; -err_out: - return rc; + skd_free_dma(skdev, skdev->sglist_cache, sg_list, dma_addr, + DMA_TO_DEVICE); } -static int skd_cons_skspcl(struct skd_device *skdev) +static int skd_init_request(struct blk_mq_tag_set *set, struct request *rq, + unsigned int hctx_idx, unsigned int numa_node) { - int rc = 0; - u32 i, nbytes; - - pr_debug("%s:%s:%d skspcl_table kzalloc, struct %lu, count %u total %lu\n", - skdev->name, __func__, __LINE__, - sizeof(struct skd_special_context), - skdev->n_special, - sizeof(struct skd_special_context) * skdev->n_special); - - skdev->skspcl_table = kzalloc(sizeof(struct skd_special_context) - * skdev->n_special, GFP_KERNEL); - if (skdev->skspcl_table == NULL) { - rc = -ENOMEM; - goto err_out; - } + struct skd_device *skdev = set->driver_data; + struct skd_request_context *skreq = blk_mq_rq_to_pdu(rq); - for (i = 0; i < skdev->n_special; i++) { - struct skd_special_context *skspcl; - - skspcl = &skdev->skspcl_table[i]; - - skspcl->req.id = i + SKD_ID_SPECIAL_REQUEST; - skspcl->req.state = SKD_REQ_STATE_IDLE; - - skspcl->req.next = &skspcl[1].req; - - nbytes = SKD_N_SPECIAL_FITMSG_BYTES; - - skspcl->msg_buf = - pci_zalloc_consistent(skdev->pdev, nbytes, - &skspcl->mb_dma_address); - if (skspcl->msg_buf == NULL) { - rc = -ENOMEM; - goto err_out; - } - - skspcl->req.sg = kzalloc(sizeof(struct scatterlist) * - SKD_N_SG_PER_SPECIAL, GFP_KERNEL); - if (skspcl->req.sg == NULL) { - rc = -ENOMEM; - goto err_out; - } - - skspcl->req.sksg_list = skd_cons_sg_list(skdev, - SKD_N_SG_PER_SPECIAL, - &skspcl->req. - sksg_dma_address); - if (skspcl->req.sksg_list == NULL) { - rc = -ENOMEM; - goto err_out; - } - } + skreq->state = SKD_REQ_STATE_IDLE; + skreq->sg = (void *)(skreq + 1); + sg_init_table(skreq->sg, skd_sgs_per_request); + skreq->sksg_list = skd_cons_sg_list(skdev, skd_sgs_per_request, + &skreq->sksg_dma_address); - /* Free list is in order starting with the 0th entry. */ - skdev->skspcl_table[i - 1].req.next = NULL; - skdev->skspcl_free_list = skdev->skspcl_table; + return skreq->sksg_list ? 0 : -ENOMEM; +} - return rc; +static void skd_exit_request(struct blk_mq_tag_set *set, struct request *rq, + unsigned int hctx_idx) +{ + struct skd_device *skdev = set->driver_data; + struct skd_request_context *skreq = blk_mq_rq_to_pdu(rq); -err_out: - return rc; + skd_free_sg_list(skdev, skreq->sksg_list, skreq->sksg_dma_address); } static int skd_cons_sksb(struct skd_device *skdev) { int rc = 0; struct skd_special_context *skspcl; - u32 nbytes; skspcl = &skdev->internal_skspcl; skspcl->req.id = 0 + SKD_ID_INTERNAL; skspcl->req.state = SKD_REQ_STATE_IDLE; - nbytes = SKD_N_INTERNAL_BYTES; - - skspcl->data_buf = pci_zalloc_consistent(skdev->pdev, nbytes, - &skspcl->db_dma_address); + skspcl->data_buf = skd_alloc_dma(skdev, skdev->databuf_cache, + &skspcl->db_dma_address, + GFP_DMA | __GFP_ZERO, + DMA_BIDIRECTIONAL); if (skspcl->data_buf == NULL) { rc = -ENOMEM; goto err_out; } - nbytes = SKD_N_SPECIAL_FITMSG_BYTES; - skspcl->msg_buf = pci_zalloc_consistent(skdev->pdev, nbytes, - &skspcl->mb_dma_address); + skspcl->msg_buf = skd_alloc_dma(skdev, skdev->msgbuf_cache, + &skspcl->mb_dma_address, + GFP_DMA | __GFP_ZERO, DMA_TO_DEVICE); if (skspcl->msg_buf == NULL) { rc = -ENOMEM; goto err_out; @@ -4247,6 +2795,14 @@ err_out: return rc; } +static const struct blk_mq_ops skd_mq_ops = { + .queue_rq = skd_mq_queue_rq, + .complete = skd_complete_rq, + .timeout = skd_timed_out, + .init_request = skd_init_request, + .exit_request = skd_exit_request, +}; + static int skd_cons_disk(struct skd_device *skdev) { int rc = 0; @@ -4268,31 +2824,46 @@ static int skd_cons_disk(struct skd_device *skdev) disk->fops = &skd_blockdev_ops; disk->private_data = skdev; - q = blk_init_queue(skd_request_fn, &skdev->lock); - if (!q) { - rc = -ENOMEM; + memset(&skdev->tag_set, 0, sizeof(skdev->tag_set)); + skdev->tag_set.ops = &skd_mq_ops; + skdev->tag_set.nr_hw_queues = 1; + skdev->tag_set.queue_depth = skd_max_queue_depth; + skdev->tag_set.cmd_size = sizeof(struct skd_request_context) + + skdev->sgs_per_request * sizeof(struct scatterlist); + skdev->tag_set.numa_node = NUMA_NO_NODE; + skdev->tag_set.flags = BLK_MQ_F_SHOULD_MERGE | + BLK_MQ_F_SG_MERGE | + BLK_ALLOC_POLICY_TO_MQ_FLAG(BLK_TAG_ALLOC_FIFO); + skdev->tag_set.driver_data = skdev; + rc = blk_mq_alloc_tag_set(&skdev->tag_set); + if (rc) + goto err_out; + q = blk_mq_init_queue(&skdev->tag_set); + if (IS_ERR(q)) { + blk_mq_free_tag_set(&skdev->tag_set); + rc = PTR_ERR(q); goto err_out; } - blk_queue_bounce_limit(q, BLK_BOUNCE_HIGH); + q->queuedata = skdev; skdev->queue = q; disk->queue = q; - q->queuedata = skdev; blk_queue_write_cache(q, true, true); blk_queue_max_segments(q, skdev->sgs_per_request); blk_queue_max_hw_sectors(q, SKD_N_MAX_SECTORS); - /* set sysfs ptimal_io_size to 8K */ + /* set optimal I/O size to 8KB */ blk_queue_io_opt(q, 8192); queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q); queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, q); + blk_queue_rq_timeout(q, 8 * HZ); + spin_lock_irqsave(&skdev->lock, flags); - pr_debug("%s:%s:%d stopping %s queue\n", - skdev->name, __func__, __LINE__, skdev->name); - blk_stop_queue(skdev->queue); + dev_dbg(&skdev->pdev->dev, "stopping queue\n"); + blk_mq_stop_hw_queues(skdev->queue); spin_unlock_irqrestore(&skdev->lock, flags); err_out: @@ -4306,13 +2877,13 @@ static struct skd_device *skd_construct(struct pci_dev *pdev) { struct skd_device *skdev; int blk_major = skd_major; + size_t size; int rc; skdev = kzalloc(sizeof(*skdev), GFP_KERNEL); if (!skdev) { - pr_err(PFX "(%s): memory alloc failure\n", - pci_name(pdev)); + dev_err(&pdev->dev, "memory alloc failure\n"); return NULL; } @@ -4320,60 +2891,71 @@ static struct skd_device *skd_construct(struct pci_dev *pdev) skdev->pdev = pdev; skdev->devno = skd_next_devno++; skdev->major = blk_major; - sprintf(skdev->name, DRV_NAME "%d", skdev->devno); skdev->dev_max_queue_depth = 0; skdev->num_req_context = skd_max_queue_depth; skdev->num_fitmsg_context = skd_max_queue_depth; - skdev->n_special = skd_max_pass_thru; skdev->cur_max_queue_depth = 1; skdev->queue_low_water_mark = 1; skdev->proto_ver = 99; skdev->sgs_per_request = skd_sgs_per_request; skdev->dbg_level = skd_dbg_level; - atomic_set(&skdev->device_count, 0); - spin_lock_init(&skdev->lock); + INIT_WORK(&skdev->start_queue, skd_start_queue); INIT_WORK(&skdev->completion_worker, skd_completion_worker); - pr_debug("%s:%s:%d skcomp\n", skdev->name, __func__, __LINE__); - rc = skd_cons_skcomp(skdev); - if (rc < 0) + size = max(SKD_N_FITMSG_BYTES, SKD_N_SPECIAL_FITMSG_BYTES); + skdev->msgbuf_cache = kmem_cache_create("skd-msgbuf", size, 0, + SLAB_HWCACHE_ALIGN, NULL); + if (!skdev->msgbuf_cache) goto err_out; - - pr_debug("%s:%s:%d skmsg\n", skdev->name, __func__, __LINE__); - rc = skd_cons_skmsg(skdev); - if (rc < 0) + WARN_ONCE(kmem_cache_size(skdev->msgbuf_cache) < size, + "skd-msgbuf: %d < %zd\n", + kmem_cache_size(skdev->msgbuf_cache), size); + size = skd_sgs_per_request * sizeof(struct fit_sg_descriptor); + skdev->sglist_cache = kmem_cache_create("skd-sglist", size, 0, + SLAB_HWCACHE_ALIGN, NULL); + if (!skdev->sglist_cache) + goto err_out; + WARN_ONCE(kmem_cache_size(skdev->sglist_cache) < size, + "skd-sglist: %d < %zd\n", + kmem_cache_size(skdev->sglist_cache), size); + size = SKD_N_INTERNAL_BYTES; + skdev->databuf_cache = kmem_cache_create("skd-databuf", size, 0, + SLAB_HWCACHE_ALIGN, NULL); + if (!skdev->databuf_cache) goto err_out; + WARN_ONCE(kmem_cache_size(skdev->databuf_cache) < size, + "skd-databuf: %d < %zd\n", + kmem_cache_size(skdev->databuf_cache), size); - pr_debug("%s:%s:%d skreq\n", skdev->name, __func__, __LINE__); - rc = skd_cons_skreq(skdev); + dev_dbg(&skdev->pdev->dev, "skcomp\n"); + rc = skd_cons_skcomp(skdev); if (rc < 0) goto err_out; - pr_debug("%s:%s:%d skspcl\n", skdev->name, __func__, __LINE__); - rc = skd_cons_skspcl(skdev); + dev_dbg(&skdev->pdev->dev, "skmsg\n"); + rc = skd_cons_skmsg(skdev); if (rc < 0) goto err_out; - pr_debug("%s:%s:%d sksb\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "sksb\n"); rc = skd_cons_sksb(skdev); if (rc < 0) goto err_out; - pr_debug("%s:%s:%d disk\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "disk\n"); rc = skd_cons_disk(skdev); if (rc < 0) goto err_out; - pr_debug("%s:%s:%d VICTORY\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "VICTORY\n"); return skdev; err_out: - pr_debug("%s:%s:%d construct failed\n", - skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "construct failed\n"); skd_destruct(skdev); return NULL; } @@ -4386,14 +2968,9 @@ err_out: static void skd_free_skcomp(struct skd_device *skdev) { - if (skdev->skcomp_table != NULL) { - u32 nbytes; - - nbytes = sizeof(skdev->skcomp_table[0]) * - SKD_N_COMPLETION_ENTRY; - pci_free_consistent(skdev->pdev, nbytes, + if (skdev->skcomp_table) + pci_free_consistent(skdev->pdev, SKD_SKCOMP_SIZE, skdev->skcomp_table, skdev->cq_dma_address); - } skdev->skcomp_table = NULL; skdev->cq_dma_address = 0; @@ -4412,8 +2989,6 @@ static void skd_free_skmsg(struct skd_device *skdev) skmsg = &skdev->skmsg_table[i]; if (skmsg->msg_buf != NULL) { - skmsg->msg_buf += skmsg->offset; - skmsg->mb_dma_address += skmsg->offset; pci_free_consistent(skdev->pdev, SKD_N_FITMSG_BYTES, skmsg->msg_buf, skmsg->mb_dma_address); @@ -4426,109 +3001,23 @@ static void skd_free_skmsg(struct skd_device *skdev) skdev->skmsg_table = NULL; } -static void skd_free_sg_list(struct skd_device *skdev, - struct fit_sg_descriptor *sg_list, - u32 n_sg, dma_addr_t dma_addr) -{ - if (sg_list != NULL) { - u32 nbytes; - - nbytes = sizeof(*sg_list) * n_sg; - - pci_free_consistent(skdev->pdev, nbytes, sg_list, dma_addr); - } -} - -static void skd_free_skreq(struct skd_device *skdev) -{ - u32 i; - - if (skdev->skreq_table == NULL) - return; - - for (i = 0; i < skdev->num_req_context; i++) { - struct skd_request_context *skreq; - - skreq = &skdev->skreq_table[i]; - - skd_free_sg_list(skdev, skreq->sksg_list, - skdev->sgs_per_request, - skreq->sksg_dma_address); - - skreq->sksg_list = NULL; - skreq->sksg_dma_address = 0; - - kfree(skreq->sg); - } - - kfree(skdev->skreq_table); - skdev->skreq_table = NULL; -} - -static void skd_free_skspcl(struct skd_device *skdev) -{ - u32 i; - u32 nbytes; - - if (skdev->skspcl_table == NULL) - return; - - for (i = 0; i < skdev->n_special; i++) { - struct skd_special_context *skspcl; - - skspcl = &skdev->skspcl_table[i]; - - if (skspcl->msg_buf != NULL) { - nbytes = SKD_N_SPECIAL_FITMSG_BYTES; - pci_free_consistent(skdev->pdev, nbytes, - skspcl->msg_buf, - skspcl->mb_dma_address); - } - - skspcl->msg_buf = NULL; - skspcl->mb_dma_address = 0; - - skd_free_sg_list(skdev, skspcl->req.sksg_list, - SKD_N_SG_PER_SPECIAL, - skspcl->req.sksg_dma_address); - - skspcl->req.sksg_list = NULL; - skspcl->req.sksg_dma_address = 0; - - kfree(skspcl->req.sg); - } - - kfree(skdev->skspcl_table); - skdev->skspcl_table = NULL; -} - static void skd_free_sksb(struct skd_device *skdev) { - struct skd_special_context *skspcl; - u32 nbytes; - - skspcl = &skdev->internal_skspcl; - - if (skspcl->data_buf != NULL) { - nbytes = SKD_N_INTERNAL_BYTES; + struct skd_special_context *skspcl = &skdev->internal_skspcl; - pci_free_consistent(skdev->pdev, nbytes, - skspcl->data_buf, skspcl->db_dma_address); - } + skd_free_dma(skdev, skdev->databuf_cache, skspcl->data_buf, + skspcl->db_dma_address, DMA_BIDIRECTIONAL); skspcl->data_buf = NULL; skspcl->db_dma_address = 0; - if (skspcl->msg_buf != NULL) { - nbytes = SKD_N_SPECIAL_FITMSG_BYTES; - pci_free_consistent(skdev->pdev, nbytes, - skspcl->msg_buf, skspcl->mb_dma_address); - } + skd_free_dma(skdev, skdev->msgbuf_cache, skspcl->msg_buf, + skspcl->mb_dma_address, DMA_TO_DEVICE); skspcl->msg_buf = NULL; skspcl->mb_dma_address = 0; - skd_free_sg_list(skdev, skspcl->req.sksg_list, 1, + skd_free_sg_list(skdev, skspcl->req.sksg_list, skspcl->req.sksg_dma_address); skspcl->req.sksg_list = NULL; @@ -4539,15 +3028,20 @@ static void skd_free_disk(struct skd_device *skdev) { struct gendisk *disk = skdev->disk; - if (disk != NULL) { - struct request_queue *q = disk->queue; + if (disk && (disk->flags & GENHD_FL_UP)) + del_gendisk(disk); - if (disk->flags & GENHD_FL_UP) - del_gendisk(disk); - if (q) - blk_cleanup_queue(q); - put_disk(disk); + if (skdev->queue) { + blk_cleanup_queue(skdev->queue); + skdev->queue = NULL; + if (disk) + disk->queue = NULL; } + + if (skdev->tag_set.tags) + blk_mq_free_tag_set(&skdev->tag_set); + + put_disk(disk); skdev->disk = NULL; } @@ -4556,26 +3050,25 @@ static void skd_destruct(struct skd_device *skdev) if (skdev == NULL) return; + cancel_work_sync(&skdev->start_queue); - pr_debug("%s:%s:%d disk\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "disk\n"); skd_free_disk(skdev); - pr_debug("%s:%s:%d sksb\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "sksb\n"); skd_free_sksb(skdev); - pr_debug("%s:%s:%d skspcl\n", skdev->name, __func__, __LINE__); - skd_free_skspcl(skdev); - - pr_debug("%s:%s:%d skreq\n", skdev->name, __func__, __LINE__); - skd_free_skreq(skdev); - - pr_debug("%s:%s:%d skmsg\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "skmsg\n"); skd_free_skmsg(skdev); - pr_debug("%s:%s:%d skcomp\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "skcomp\n"); skd_free_skcomp(skdev); - pr_debug("%s:%s:%d skdev\n", skdev->name, __func__, __LINE__); + kmem_cache_destroy(skdev->databuf_cache); + kmem_cache_destroy(skdev->sglist_cache); + kmem_cache_destroy(skdev->msgbuf_cache); + + dev_dbg(&skdev->pdev->dev, "skdev\n"); kfree(skdev); } @@ -4592,9 +3085,8 @@ static int skd_bdev_getgeo(struct block_device *bdev, struct hd_geometry *geo) skdev = bdev->bd_disk->private_data; - pr_debug("%s:%s:%d %s: CMD[%s] getgeo device\n", - skdev->name, __func__, __LINE__, - bdev->bd_disk->disk_name, current->comm); + dev_dbg(&skdev->pdev->dev, "%s: CMD[%s] getgeo device\n", + bdev->bd_disk->disk_name, current->comm); if (skdev->read_cap_is_valid) { capacity = get_capacity(skdev->disk); @@ -4609,18 +3101,16 @@ static int skd_bdev_getgeo(struct block_device *bdev, struct hd_geometry *geo) static int skd_bdev_attach(struct device *parent, struct skd_device *skdev) { - pr_debug("%s:%s:%d add_disk\n", skdev->name, __func__, __LINE__); + dev_dbg(&skdev->pdev->dev, "add_disk\n"); device_add_disk(parent, skdev->disk); return 0; } static const struct block_device_operations skd_blockdev_ops = { .owner = THIS_MODULE, - .ioctl = skd_bdev_ioctl, .getgeo = skd_bdev_getgeo, }; - /* ***************************************************************************** * PCIe DRIVER GLUE @@ -4671,10 +3161,8 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) char pci_str[32]; struct skd_device *skdev; - pr_info("STEC s1120 Driver(%s) version %s-b%s\n", - DRV_NAME, DRV_VERSION, DRV_BUILD_ID); - pr_info("(skd?:??:[%s]): vendor=%04X device=%04x\n", - pci_name(pdev), pdev->vendor, pdev->device); + dev_dbg(&pdev->dev, "vendor=%04X device=%04x\n", pdev->vendor, + pdev->device); rc = pci_enable_device(pdev); if (rc) @@ -4685,16 +3173,13 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(64)); if (!rc) { if (pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64))) { - - pr_err("(%s): consistent DMA mask error %d\n", - pci_name(pdev), rc); + dev_err(&pdev->dev, "consistent DMA mask error %d\n", + rc); } } else { - (rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32))); + rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) { - - pr_err("(%s): DMA mask error %d\n", - pci_name(pdev), rc); + dev_err(&pdev->dev, "DMA mask error %d\n", rc); goto err_out_regions; } } @@ -4714,19 +3199,17 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) } skd_pci_info(skdev, pci_str); - pr_info("(%s): %s 64bit\n", skd_name(skdev), pci_str); + dev_info(&pdev->dev, "%s 64bit\n", pci_str); pci_set_master(pdev); rc = pci_enable_pcie_error_reporting(pdev); if (rc) { - pr_err( - "(%s): bad enable of PCIe error reporting rc=%d\n", - skd_name(skdev), rc); + dev_err(&pdev->dev, + "bad enable of PCIe error reporting rc=%d\n", rc); skdev->pcie_error_reporting_is_enabled = 0; } else skdev->pcie_error_reporting_is_enabled = 1; - pci_set_drvdata(pdev, skdev); for (i = 0; i < SKD_MAX_BARS; i++) { @@ -4735,21 +3218,19 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) skdev->mem_map[i] = ioremap(skdev->mem_phys[i], skdev->mem_size[i]); if (!skdev->mem_map[i]) { - pr_err("(%s): Unable to map adapter memory!\n", - skd_name(skdev)); + dev_err(&pdev->dev, + "Unable to map adapter memory!\n"); rc = -ENODEV; goto err_out_iounmap; } - pr_debug("%s:%s:%d mem_map=%p, phyd=%016llx, size=%d\n", - skdev->name, __func__, __LINE__, - skdev->mem_map[i], - (uint64_t)skdev->mem_phys[i], skdev->mem_size[i]); + dev_dbg(&pdev->dev, "mem_map=%p, phyd=%016llx, size=%d\n", + skdev->mem_map[i], (uint64_t)skdev->mem_phys[i], + skdev->mem_size[i]); } rc = skd_acquire_irq(skdev); if (rc) { - pr_err("(%s): interrupt resource error %d\n", - skd_name(skdev), rc); + dev_err(&pdev->dev, "interrupt resource error %d\n", rc); goto err_out_iounmap; } @@ -4771,29 +3252,14 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) } else { /* we timed out, something is wrong with the device, don't add the disk structure */ - pr_err( - "(%s): error: waiting for s1120 timed out %d!\n", - skd_name(skdev), rc); + dev_err(&pdev->dev, "error: waiting for s1120 timed out %d!\n", + rc); /* in case of no error; we timeout with ENXIO */ if (!rc) rc = -ENXIO; goto err_out_timer; } - -#ifdef SKD_VMK_POLL_HANDLER - if (skdev->irq_type == SKD_IRQ_MSIX) { - /* MSIX completion handler is being used for coredump */ - vmklnx_scsi_register_poll_handler(skdev->scsi_host, - skdev->msix_entries[5].vector, - skd_comp_q, skdev); - } else { - vmklnx_scsi_register_poll_handler(skdev->scsi_host, - skdev->pdev->irq, skd_isr, - skdev); - } -#endif /* SKD_VMK_POLL_HANDLER */ - return rc; err_out_timer: @@ -4826,7 +3292,7 @@ static void skd_pci_remove(struct pci_dev *pdev) skdev = pci_get_drvdata(pdev); if (!skdev) { - pr_err("%s: no device data for PCI\n", pci_name(pdev)); + dev_err(&pdev->dev, "no device data for PCI\n"); return; } skd_stop_device(skdev); @@ -4834,7 +3300,7 @@ static void skd_pci_remove(struct pci_dev *pdev) for (i = 0; i < SKD_MAX_BARS; i++) if (skdev->mem_map[i]) - iounmap((u32 *)skdev->mem_map[i]); + iounmap(skdev->mem_map[i]); if (skdev->pcie_error_reporting_is_enabled) pci_disable_pcie_error_reporting(pdev); @@ -4855,7 +3321,7 @@ static int skd_pci_suspend(struct pci_dev *pdev, pm_message_t state) skdev = pci_get_drvdata(pdev); if (!skdev) { - pr_err("%s: no device data for PCI\n", pci_name(pdev)); + dev_err(&pdev->dev, "no device data for PCI\n"); return -EIO; } @@ -4865,7 +3331,7 @@ static int skd_pci_suspend(struct pci_dev *pdev, pm_message_t state) for (i = 0; i < SKD_MAX_BARS; i++) if (skdev->mem_map[i]) - iounmap((u32 *)skdev->mem_map[i]); + iounmap(skdev->mem_map[i]); if (skdev->pcie_error_reporting_is_enabled) pci_disable_pcie_error_reporting(pdev); @@ -4885,7 +3351,7 @@ static int skd_pci_resume(struct pci_dev *pdev) skdev = pci_get_drvdata(pdev); if (!skdev) { - pr_err("%s: no device data for PCI\n", pci_name(pdev)); + dev_err(&pdev->dev, "no device data for PCI\n"); return -1; } @@ -4903,15 +3369,14 @@ static int skd_pci_resume(struct pci_dev *pdev) if (!rc) { if (pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64))) { - pr_err("(%s): consistent DMA mask error %d\n", - pci_name(pdev), rc); + dev_err(&pdev->dev, "consistent DMA mask error %d\n", + rc); } } else { rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); if (rc) { - pr_err("(%s): DMA mask error %d\n", - pci_name(pdev), rc); + dev_err(&pdev->dev, "DMA mask error %d\n", rc); goto err_out_regions; } } @@ -4919,8 +3384,8 @@ static int skd_pci_resume(struct pci_dev *pdev) pci_set_master(pdev); rc = pci_enable_pcie_error_reporting(pdev); if (rc) { - pr_err("(%s): bad enable of PCIe error reporting rc=%d\n", - skdev->name, rc); + dev_err(&pdev->dev, + "bad enable of PCIe error reporting rc=%d\n", rc); skdev->pcie_error_reporting_is_enabled = 0; } else skdev->pcie_error_reporting_is_enabled = 1; @@ -4932,21 +3397,17 @@ static int skd_pci_resume(struct pci_dev *pdev) skdev->mem_map[i] = ioremap(skdev->mem_phys[i], skdev->mem_size[i]); if (!skdev->mem_map[i]) { - pr_err("(%s): Unable to map adapter memory!\n", - skd_name(skdev)); + dev_err(&pdev->dev, "Unable to map adapter memory!\n"); rc = -ENODEV; goto err_out_iounmap; } - pr_debug("%s:%s:%d mem_map=%p, phyd=%016llx, size=%d\n", - skdev->name, __func__, __LINE__, - skdev->mem_map[i], - (uint64_t)skdev->mem_phys[i], skdev->mem_size[i]); + dev_dbg(&pdev->dev, "mem_map=%p, phyd=%016llx, size=%d\n", + skdev->mem_map[i], (uint64_t)skdev->mem_phys[i], + skdev->mem_size[i]); } rc = skd_acquire_irq(skdev); if (rc) { - - pr_err("(%s): interrupt resource error %d\n", - pci_name(pdev), rc); + dev_err(&pdev->dev, "interrupt resource error %d\n", rc); goto err_out_iounmap; } @@ -4984,15 +3445,15 @@ static void skd_pci_shutdown(struct pci_dev *pdev) { struct skd_device *skdev; - pr_err("skd_pci_shutdown called\n"); + dev_err(&pdev->dev, "%s called\n", __func__); skdev = pci_get_drvdata(pdev); if (!skdev) { - pr_err("%s: no device data for PCI\n", pci_name(pdev)); + dev_err(&pdev->dev, "no device data for PCI\n"); return; } - pr_err("%s: calling stop\n", skd_name(skdev)); + dev_err(&pdev->dev, "calling stop\n"); skd_stop_device(skdev); } @@ -5012,21 +3473,6 @@ static struct pci_driver skd_driver = { ***************************************************************************** */ -static const char *skd_name(struct skd_device *skdev) -{ - memset(skdev->id_str, 0, sizeof(skdev->id_str)); - - if (skdev->inquiry_is_valid) - snprintf(skdev->id_str, sizeof(skdev->id_str), "%s:%s:[%s]", - skdev->name, skdev->inq_serial_num, - pci_name(skdev->pdev)); - else - snprintf(skdev->id_str, sizeof(skdev->id_str), "%s:??:[%s]", - skdev->name, pci_name(skdev->pdev)); - - return skdev->id_str; -} - const char *skd_drive_state_to_str(int state) { switch (state) { @@ -5078,8 +3524,6 @@ const char *skd_skdev_state_to_str(enum skd_drvr_state state) return "PAUSING"; case SKD_DRVR_STATE_PAUSED: return "PAUSED"; - case SKD_DRVR_STATE_DRAINING_TIMEOUT: - return "DRAINING_TIMEOUT"; case SKD_DRVR_STATE_RESTARTING: return "RESTARTING"; case SKD_DRVR_STATE_RESUMING: @@ -5106,18 +3550,6 @@ const char *skd_skdev_state_to_str(enum skd_drvr_state state) } } -static const char *skd_skmsg_state_to_str(enum skd_fit_msg_state state) -{ - switch (state) { - case SKD_MSG_STATE_IDLE: - return "IDLE"; - case SKD_MSG_STATE_BUSY: - return "BUSY"; - default: - return "???"; - } -} - static const char *skd_skreq_state_to_str(enum skd_req_state state) { switch (state) { @@ -5131,8 +3563,6 @@ static const char *skd_skreq_state_to_str(enum skd_req_state state) return "COMPLETED"; case SKD_REQ_STATE_TIMEOUT: return "TIMEOUT"; - case SKD_REQ_STATE_ABORTED: - return "ABORTED"; default: return "???"; } @@ -5140,58 +3570,34 @@ static const char *skd_skreq_state_to_str(enum skd_req_state state) static void skd_log_skdev(struct skd_device *skdev, const char *event) { - pr_debug("%s:%s:%d (%s) skdev=%p event='%s'\n", - skdev->name, __func__, __LINE__, skdev->name, skdev, event); - pr_debug("%s:%s:%d drive_state=%s(%d) driver_state=%s(%d)\n", - skdev->name, __func__, __LINE__, - skd_drive_state_to_str(skdev->drive_state), skdev->drive_state, - skd_skdev_state_to_str(skdev->state), skdev->state); - pr_debug("%s:%s:%d busy=%d limit=%d dev=%d lowat=%d\n", - skdev->name, __func__, __LINE__, - skdev->in_flight, skdev->cur_max_queue_depth, - skdev->dev_max_queue_depth, skdev->queue_low_water_mark); - pr_debug("%s:%s:%d timestamp=0x%x cycle=%d cycle_ix=%d\n", - skdev->name, __func__, __LINE__, - skdev->timeout_stamp, skdev->skcomp_cycle, skdev->skcomp_ix); -} - -static void skd_log_skmsg(struct skd_device *skdev, - struct skd_fitmsg_context *skmsg, const char *event) -{ - pr_debug("%s:%s:%d (%s) skmsg=%p event='%s'\n", - skdev->name, __func__, __LINE__, skdev->name, skmsg, event); - pr_debug("%s:%s:%d state=%s(%d) id=0x%04x length=%d\n", - skdev->name, __func__, __LINE__, - skd_skmsg_state_to_str(skmsg->state), skmsg->state, - skmsg->id, skmsg->length); + dev_dbg(&skdev->pdev->dev, "skdev=%p event='%s'\n", skdev, event); + dev_dbg(&skdev->pdev->dev, " drive_state=%s(%d) driver_state=%s(%d)\n", + skd_drive_state_to_str(skdev->drive_state), skdev->drive_state, + skd_skdev_state_to_str(skdev->state), skdev->state); + dev_dbg(&skdev->pdev->dev, " busy=%d limit=%d dev=%d lowat=%d\n", + skd_in_flight(skdev), skdev->cur_max_queue_depth, + skdev->dev_max_queue_depth, skdev->queue_low_water_mark); + dev_dbg(&skdev->pdev->dev, " cycle=%d cycle_ix=%d\n", + skdev->skcomp_cycle, skdev->skcomp_ix); } static void skd_log_skreq(struct skd_device *skdev, struct skd_request_context *skreq, const char *event) { - pr_debug("%s:%s:%d (%s) skreq=%p event='%s'\n", - skdev->name, __func__, __LINE__, skdev->name, skreq, event); - pr_debug("%s:%s:%d state=%s(%d) id=0x%04x fitmsg=0x%04x\n", - skdev->name, __func__, __LINE__, - skd_skreq_state_to_str(skreq->state), skreq->state, - skreq->id, skreq->fitmsg_id); - pr_debug("%s:%s:%d timo=0x%x sg_dir=%d n_sg=%d\n", - skdev->name, __func__, __LINE__, - skreq->timeout_stamp, skreq->sg_data_dir, skreq->n_sg); - - if (skreq->req != NULL) { - struct request *req = skreq->req; - u32 lba = (u32)blk_rq_pos(req); - u32 count = blk_rq_sectors(req); - - pr_debug("%s:%s:%d " - "req=%p lba=%u(0x%x) count=%u(0x%x) dir=%d\n", - skdev->name, __func__, __LINE__, - req, lba, lba, count, count, - (int)rq_data_dir(req)); - } else - pr_debug("%s:%s:%d req=NULL\n", - skdev->name, __func__, __LINE__); + struct request *req = blk_mq_rq_from_pdu(skreq); + u32 lba = blk_rq_pos(req); + u32 count = blk_rq_sectors(req); + + dev_dbg(&skdev->pdev->dev, "skreq=%p event='%s'\n", skreq, event); + dev_dbg(&skdev->pdev->dev, " state=%s(%d) id=0x%04x fitmsg=0x%04x\n", + skd_skreq_state_to_str(skreq->state), skreq->state, skreq->id, + skreq->fitmsg_id); + dev_dbg(&skdev->pdev->dev, " sg_dir=%d n_sg=%d\n", + skreq->data_dir, skreq->n_sg); + + dev_dbg(&skdev->pdev->dev, + "req=%p lba=%u(0x%x) count=%u(0x%x) dir=%d\n", req, lba, lba, + count, count, (int)rq_data_dir(req)); } /* @@ -5202,7 +3608,14 @@ static void skd_log_skreq(struct skd_device *skdev, static int __init skd_init(void) { - pr_info(PFX " v%s-b%s loaded\n", DRV_VERSION, DRV_BUILD_ID); + BUILD_BUG_ON(sizeof(struct fit_completion_entry_v1) != 8); + BUILD_BUG_ON(sizeof(struct fit_comp_error_info) != 32); + BUILD_BUG_ON(sizeof(struct skd_command_header) != 16); + BUILD_BUG_ON(sizeof(struct skd_scsi_request) != 32); + BUILD_BUG_ON(sizeof(struct driver_inquiry_data) != 44); + BUILD_BUG_ON(offsetof(struct skd_msg_buf, fmh) != 0); + BUILD_BUG_ON(offsetof(struct skd_msg_buf, scsi) != 64); + BUILD_BUG_ON(sizeof(struct skd_msg_buf) != SKD_N_FITMSG_BYTES); switch (skd_isr_type) { case SKD_IRQ_LEGACY: @@ -5222,7 +3635,8 @@ static int __init skd_init(void) skd_max_queue_depth = SKD_MAX_QUEUE_DEPTH_DEFAULT; } - if (skd_max_req_per_msg < 1 || skd_max_req_per_msg > 14) { + if (skd_max_req_per_msg < 1 || + skd_max_req_per_msg > SKD_MAX_REQ_PER_MSG) { pr_err(PFX "skd_max_req_per_msg %d invalid, re-set to %d\n", skd_max_req_per_msg, SKD_MAX_REQ_PER_MSG_DEFAULT); skd_max_req_per_msg = SKD_MAX_REQ_PER_MSG_DEFAULT; @@ -5246,19 +3660,11 @@ static int __init skd_init(void) skd_isr_comp_limit = 0; } - if (skd_max_pass_thru < 1 || skd_max_pass_thru > 50) { - pr_err(PFX "skd_max_pass_thru %d invalid, re-set to %d\n", - skd_max_pass_thru, SKD_N_SPECIAL_CONTEXT); - skd_max_pass_thru = SKD_N_SPECIAL_CONTEXT; - } - return pci_register_driver(&skd_driver); } static void __exit skd_exit(void) { - pr_info(PFX " v%s-b%s unloading\n", DRV_VERSION, DRV_BUILD_ID); - pci_unregister_driver(&skd_driver); if (skd_major) diff --git a/drivers/block/skd_s1120.h b/drivers/block/skd_s1120.h index 61c757ff0161..de35f47e953c 100644 --- a/drivers/block/skd_s1120.h +++ b/drivers/block/skd_s1120.h @@ -1,19 +1,15 @@ -/* Copyright 2012 STEC, Inc. +/* + * Copyright 2012 STEC, Inc. + * Copyright (c) 2017 Western Digital Corporation or its affiliates. * - * This file is licensed under the terms of the 3-clause - * BSD License (http://opensource.org/licenses/BSD-3-Clause) - * or the GNU GPL-2.0 (http://www.gnu.org/licenses/gpl-2.0.html), - * at your option. Both licenses are also available in the LICENSE file - * distributed with this project. This file may not be copied, modified, - * or distributed except in accordance with those terms. + * This file is part of the Linux kernel, and is made available under + * the terms of the GNU General Public License version 2. */ #ifndef SKD_S1120_H #define SKD_S1120_H -#pragma pack(push, s1120_h, 1) - /* * Q-channel, 64-bit r/w */ @@ -30,7 +26,7 @@ #define FIT_QCMD_MSGSIZE_128 (0x1 << 4) #define FIT_QCMD_MSGSIZE_256 (0x2 << 4) #define FIT_QCMD_MSGSIZE_512 (0x3 << 4) -#define FIT_QCMD_BASE_ADDRESS_MASK (0xFFFFFFFFFFFFFFC0ull) +#define FIT_QCMD_ALIGN L1_CACHE_BYTES /* * Control, 32-bit r/w @@ -250,7 +246,7 @@ struct fit_msg_hdr { * 20-23 of the FIT_MTD_FITFW_INIT response. */ struct fit_completion_entry_v1 { - uint32_t num_returned_bytes; + __be32 num_returned_bytes; uint16_t tag; uint8_t status; /* SCSI status */ uint8_t cycle; @@ -278,7 +274,7 @@ struct fit_comp_error_info { uint16_t sks_low; /* 10: Sense Key Specific (LSW) */ uint16_t reserved3; /* 12: Part of additional sense bytes (unused) */ uint16_t uec; /* 14: Additional Sense Bytes */ - uint64_t per; /* 16: Additional Sense Bytes */ + uint64_t per __packed; /* 16: Additional Sense Bytes */ uint8_t reserved4[2]; /* 1E: Additional Sense Bytes (unused) */ }; @@ -292,11 +288,11 @@ struct fit_comp_error_info { * Version one has the last 32 bits sg_list_len_bytes; */ struct skd_command_header { - uint64_t sg_list_dma_address; + __be64 sg_list_dma_address; uint16_t tag; uint8_t attribute; uint8_t add_cdb_len; /* In 32 bit words */ - uint32_t sg_list_len_bytes; + __be32 sg_list_len_bytes; }; struct skd_scsi_request { @@ -309,22 +305,20 @@ struct driver_inquiry_data { uint8_t peripheral_device_type:5; uint8_t qualifier:3; uint8_t page_code; - uint16_t page_length; - uint16_t pcie_bus_number; + __be16 page_length; + __be16 pcie_bus_number; uint8_t pcie_device_number; uint8_t pcie_function_number; uint8_t pcie_link_speed; uint8_t pcie_link_lanes; - uint16_t pcie_vendor_id; - uint16_t pcie_device_id; - uint16_t pcie_subsystem_vendor_id; - uint16_t pcie_subsystem_device_id; + __be16 pcie_vendor_id; + __be16 pcie_device_id; + __be16 pcie_subsystem_vendor_id; + __be16 pcie_subsystem_device_id; uint8_t reserved1[2]; uint8_t reserved2[3]; uint8_t driver_version_length; uint8_t driver_version[0x14]; }; -#pragma pack(pop, s1120_h) - #endif /* SKD_S1120_H */ diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c index 1498b899a593..34e17ee799be 100644 --- a/drivers/block/virtio_blk.c +++ b/drivers/block/virtio_blk.c @@ -265,7 +265,7 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx, } spin_lock_irqsave(&vblk->vqs[qid].lock, flags); - if (req_op(req) == REQ_OP_SCSI_IN || req_op(req) == REQ_OP_SCSI_OUT) + if (blk_rq_is_scsi(req)) err = virtblk_add_req_scsi(vblk->vqs[qid].vq, vbr, vbr->sg, num); else err = virtblk_add_req(vblk->vqs[qid].vq, vbr, vbr->sg, num); @@ -381,6 +381,7 @@ static void virtblk_config_changed_work(struct work_struct *work) struct request_queue *q = vblk->disk->queue; char cap_str_2[10], cap_str_10[10]; char *envp[] = { "RESIZE=1", NULL }; + unsigned long long nblocks; u64 capacity; /* Host must always specify the capacity. */ @@ -393,16 +394,19 @@ static void virtblk_config_changed_work(struct work_struct *work) capacity = (sector_t)-1; } - string_get_size(capacity, queue_logical_block_size(q), + nblocks = DIV_ROUND_UP_ULL(capacity, queue_logical_block_size(q) >> 9); + + string_get_size(nblocks, queue_logical_block_size(q), STRING_UNITS_2, cap_str_2, sizeof(cap_str_2)); - string_get_size(capacity, queue_logical_block_size(q), + string_get_size(nblocks, queue_logical_block_size(q), STRING_UNITS_10, cap_str_10, sizeof(cap_str_10)); dev_notice(&vdev->dev, - "new size: %llu %d-byte logical blocks (%s/%s)\n", - (unsigned long long)capacity, - queue_logical_block_size(q), - cap_str_10, cap_str_2); + "new size: %llu %d-byte logical blocks (%s/%s)\n", + nblocks, + queue_logical_block_size(q), + cap_str_10, + cap_str_2); set_capacity(vblk->disk, capacity); revalidate_disk(vblk->disk); diff --git a/drivers/block/xen-blkback/blkback.c b/drivers/block/xen-blkback/blkback.c index fe7cd58c43d0..987d665e82de 100644 --- a/drivers/block/xen-blkback/blkback.c +++ b/drivers/block/xen-blkback/blkback.c @@ -705,9 +705,9 @@ static unsigned int xen_blkbk_unmap_prepare( GNTMAP_host_map, pages[i]->handle); pages[i]->handle = BLKBACK_INVALID_HANDLE; invcount++; - } + } - return invcount; + return invcount; } static void xen_blkbk_unmap_and_respond_callback(int result, struct gntab_unmap_queue_data *data) @@ -1251,6 +1251,7 @@ static int dispatch_rw_block_io(struct xen_blkif_ring *ring, break; case BLKIF_OP_WRITE_BARRIER: drain = true; + /* fall through */ case BLKIF_OP_FLUSH_DISKCACHE: ring->st_f_req++; operation = REQ_OP_WRITE; @@ -1362,7 +1363,7 @@ static int dispatch_rw_block_io(struct xen_blkif_ring *ring, goto fail_put_bio; biolist[nbio++] = bio; - bio->bi_bdev = preq.bdev; + bio_set_dev(bio, preq.bdev); bio->bi_private = pending_req; bio->bi_end_io = end_block_io_op; bio->bi_iter.bi_sector = preq.sector_number; @@ -1381,7 +1382,7 @@ static int dispatch_rw_block_io(struct xen_blkif_ring *ring, goto fail_put_bio; biolist[nbio++] = bio; - bio->bi_bdev = preq.bdev; + bio_set_dev(bio, preq.bdev); bio->bi_private = pending_req; bio->bi_end_io = end_block_io_op; bio_set_op_attrs(bio, operation, operation_flags); diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-blkback/xenbus.c index 792da683e70d..21c1be1eb226 100644 --- a/drivers/block/xen-blkback/xenbus.c +++ b/drivers/block/xen-blkback/xenbus.c @@ -244,6 +244,7 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif) { struct pending_req *req, *n; unsigned int j, r; + bool busy = false; for (r = 0; r < blkif->nr_rings; r++) { struct xen_blkif_ring *ring = &blkif->rings[r]; @@ -261,8 +262,10 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif) * don't have any discard_io or other_io requests. So, checking * for inflight IO is enough. */ - if (atomic_read(&ring->inflight) > 0) - return -EBUSY; + if (atomic_read(&ring->inflight) > 0) { + busy = true; + continue; + } if (ring->irq) { unbind_from_irqhandler(ring->irq, ring); @@ -300,6 +303,9 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif) WARN_ON(i != (XEN_BLKIF_REQS_PER_PAGE * blkif->nr_ring_pages)); ring->active = false; } + if (busy) + return -EBUSY; + blkif->nr_ring_pages = 0; /* * blkif->rings was allocated in connect_ring, so we should free it in @@ -810,7 +816,8 @@ static void frontend_changed(struct xenbus_device *dev, xenbus_switch_state(dev, XenbusStateClosed); if (xenbus_dev_is_online(dev)) break; - /* fall through if not online */ + /* fall through */ + /* if not online */ case XenbusStateUnknown: /* implies xen_blkif_disconnect() via xen_blkbk_remove() */ device_unregister(&dev->dev); diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c index 98e34e4c62b8..891265acb10e 100644 --- a/drivers/block/xen-blkfront.c +++ b/drivers/block/xen-blkfront.c @@ -2075,9 +2075,9 @@ static int blkfront_resume(struct xenbus_device *dev) /* * Get the bios in the request so we can re-queue them. */ - if (req_op(shadow[i].request) == REQ_OP_FLUSH || - req_op(shadow[i].request) == REQ_OP_DISCARD || - req_op(shadow[i].request) == REQ_OP_SECURE_ERASE || + if (req_op(shadow[j].request) == REQ_OP_FLUSH || + req_op(shadow[j].request) == REQ_OP_DISCARD || + req_op(shadow[j].request) == REQ_OP_SECURE_ERASE || shadow[j].request->cmd_flags & REQ_FUA) { /* * Flush operations don't contain bios, so @@ -2456,7 +2456,7 @@ static void blkback_changed(struct xenbus_device *dev, case XenbusStateClosed: if (dev->state == XenbusStateClosed) break; - /* Missed the backend's Closing state -- fallthrough */ + /* fall through */ case XenbusStateClosing: if (info) blkfront_closing(info); diff --git a/drivers/block/zram/Kconfig b/drivers/block/zram/Kconfig index b8ecba6dcd3b..7cd4a8ec3c8f 100644 --- a/drivers/block/zram/Kconfig +++ b/drivers/block/zram/Kconfig @@ -13,3 +13,15 @@ config ZRAM disks and maybe many more. See zram.txt for more information. + +config ZRAM_WRITEBACK + bool "Write back incompressible page to backing device" + depends on ZRAM + default n + help + With incompressible page, there is no memory saving to keep it + in memory. Instead, write it out to backing device. + For this feature, admin should set up backing device via + /sys/block/zramX/backing_dev. + + See zram.txt for more infomration. diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c index 3b1b6340ba13..2981c27d3aae 100644 --- a/drivers/block/zram/zram_drv.c +++ b/drivers/block/zram/zram_drv.c @@ -175,20 +175,11 @@ static inline void update_used_max(struct zram *zram, } while (old_max != cur_max); } -static inline void zram_fill_page(char *ptr, unsigned long len, +static inline void zram_fill_page(void *ptr, unsigned long len, unsigned long value) { - int i; - unsigned long *page = (unsigned long *)ptr; - WARN_ON_ONCE(!IS_ALIGNED(len, sizeof(unsigned long))); - - if (likely(value == 0)) { - memset(ptr, 0, len); - } else { - for (i = 0; i < len / sizeof(*page); i++) - page[i] = value; - } + memset_l(ptr, value, len / sizeof(unsigned long)); } static bool page_same_filled(void *ptr, unsigned long *element) @@ -270,6 +261,349 @@ static ssize_t mem_used_max_store(struct device *dev, return len; } +#ifdef CONFIG_ZRAM_WRITEBACK +static bool zram_wb_enabled(struct zram *zram) +{ + return zram->backing_dev; +} + +static void reset_bdev(struct zram *zram) +{ + struct block_device *bdev; + + if (!zram_wb_enabled(zram)) + return; + + bdev = zram->bdev; + if (zram->old_block_size) + set_blocksize(bdev, zram->old_block_size); + blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL); + /* hope filp_close flush all of IO */ + filp_close(zram->backing_dev, NULL); + zram->backing_dev = NULL; + zram->old_block_size = 0; + zram->bdev = NULL; + + kvfree(zram->bitmap); + zram->bitmap = NULL; +} + +static ssize_t backing_dev_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct zram *zram = dev_to_zram(dev); + struct file *file = zram->backing_dev; + char *p; + ssize_t ret; + + down_read(&zram->init_lock); + if (!zram_wb_enabled(zram)) { + memcpy(buf, "none\n", 5); + up_read(&zram->init_lock); + return 5; + } + + p = file_path(file, buf, PAGE_SIZE - 1); + if (IS_ERR(p)) { + ret = PTR_ERR(p); + goto out; + } + + ret = strlen(p); + memmove(buf, p, ret); + buf[ret++] = '\n'; +out: + up_read(&zram->init_lock); + return ret; +} + +static ssize_t backing_dev_store(struct device *dev, + struct device_attribute *attr, const char *buf, size_t len) +{ + char *file_name; + struct file *backing_dev = NULL; + struct inode *inode; + struct address_space *mapping; + unsigned int bitmap_sz, old_block_size = 0; + unsigned long nr_pages, *bitmap = NULL; + struct block_device *bdev = NULL; + int err; + struct zram *zram = dev_to_zram(dev); + + file_name = kmalloc(PATH_MAX, GFP_KERNEL); + if (!file_name) + return -ENOMEM; + + down_write(&zram->init_lock); + if (init_done(zram)) { + pr_info("Can't setup backing device for initialized device\n"); + err = -EBUSY; + goto out; + } + + strlcpy(file_name, buf, len); + + backing_dev = filp_open(file_name, O_RDWR|O_LARGEFILE, 0); + if (IS_ERR(backing_dev)) { + err = PTR_ERR(backing_dev); + backing_dev = NULL; + goto out; + } + + mapping = backing_dev->f_mapping; + inode = mapping->host; + + /* Support only block device in this moment */ + if (!S_ISBLK(inode->i_mode)) { + err = -ENOTBLK; + goto out; + } + + bdev = bdgrab(I_BDEV(inode)); + err = blkdev_get(bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL, zram); + if (err < 0) + goto out; + + nr_pages = i_size_read(inode) >> PAGE_SHIFT; + bitmap_sz = BITS_TO_LONGS(nr_pages) * sizeof(long); + bitmap = kvzalloc(bitmap_sz, GFP_KERNEL); + if (!bitmap) { + err = -ENOMEM; + goto out; + } + + old_block_size = block_size(bdev); + err = set_blocksize(bdev, PAGE_SIZE); + if (err) + goto out; + + reset_bdev(zram); + spin_lock_init(&zram->bitmap_lock); + + zram->old_block_size = old_block_size; + zram->bdev = bdev; + zram->backing_dev = backing_dev; + zram->bitmap = bitmap; + zram->nr_pages = nr_pages; + up_write(&zram->init_lock); + + pr_info("setup backing device %s\n", file_name); + kfree(file_name); + + return len; +out: + if (bitmap) + kvfree(bitmap); + + if (bdev) + blkdev_put(bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL); + + if (backing_dev) + filp_close(backing_dev, NULL); + + up_write(&zram->init_lock); + + kfree(file_name); + + return err; +} + +static unsigned long get_entry_bdev(struct zram *zram) +{ + unsigned long entry; + + spin_lock(&zram->bitmap_lock); + /* skip 0 bit to confuse zram.handle = 0 */ + entry = find_next_zero_bit(zram->bitmap, zram->nr_pages, 1); + if (entry == zram->nr_pages) { + spin_unlock(&zram->bitmap_lock); + return 0; + } + + set_bit(entry, zram->bitmap); + spin_unlock(&zram->bitmap_lock); + + return entry; +} + +static void put_entry_bdev(struct zram *zram, unsigned long entry) +{ + int was_set; + + spin_lock(&zram->bitmap_lock); + was_set = test_and_clear_bit(entry, zram->bitmap); + spin_unlock(&zram->bitmap_lock); + WARN_ON_ONCE(!was_set); +} + +void zram_page_end_io(struct bio *bio) +{ + struct page *page = bio->bi_io_vec[0].bv_page; + + page_endio(page, op_is_write(bio_op(bio)), + blk_status_to_errno(bio->bi_status)); + bio_put(bio); +} + +/* + * Returns 1 if the submission is successful. + */ +static int read_from_bdev_async(struct zram *zram, struct bio_vec *bvec, + unsigned long entry, struct bio *parent) +{ + struct bio *bio; + + bio = bio_alloc(GFP_ATOMIC, 1); + if (!bio) + return -ENOMEM; + + bio->bi_iter.bi_sector = entry * (PAGE_SIZE >> 9); + bio_set_dev(bio, zram->bdev); + if (!bio_add_page(bio, bvec->bv_page, bvec->bv_len, bvec->bv_offset)) { + bio_put(bio); + return -EIO; + } + + if (!parent) { + bio->bi_opf = REQ_OP_READ; + bio->bi_end_io = zram_page_end_io; + } else { + bio->bi_opf = parent->bi_opf; + bio_chain(bio, parent); + } + + submit_bio(bio); + return 1; +} + +struct zram_work { + struct work_struct work; + struct zram *zram; + unsigned long entry; + struct bio *bio; +}; + +#if PAGE_SIZE != 4096 +static void zram_sync_read(struct work_struct *work) +{ + struct bio_vec bvec; + struct zram_work *zw = container_of(work, struct zram_work, work); + struct zram *zram = zw->zram; + unsigned long entry = zw->entry; + struct bio *bio = zw->bio; + + read_from_bdev_async(zram, &bvec, entry, bio); +} + +/* + * Block layer want one ->make_request_fn to be active at a time + * so if we use chained IO with parent IO in same context, + * it's a deadlock. To avoid, it, it uses worker thread context. + */ +static int read_from_bdev_sync(struct zram *zram, struct bio_vec *bvec, + unsigned long entry, struct bio *bio) +{ + struct zram_work work; + + work.zram = zram; + work.entry = entry; + work.bio = bio; + + INIT_WORK_ONSTACK(&work.work, zram_sync_read); + queue_work(system_unbound_wq, &work.work); + flush_work(&work.work); + destroy_work_on_stack(&work.work); + + return 1; +} +#else +static int read_from_bdev_sync(struct zram *zram, struct bio_vec *bvec, + unsigned long entry, struct bio *bio) +{ + WARN_ON(1); + return -EIO; +} +#endif + +static int read_from_bdev(struct zram *zram, struct bio_vec *bvec, + unsigned long entry, struct bio *parent, bool sync) +{ + if (sync) + return read_from_bdev_sync(zram, bvec, entry, parent); + else + return read_from_bdev_async(zram, bvec, entry, parent); +} + +static int write_to_bdev(struct zram *zram, struct bio_vec *bvec, + u32 index, struct bio *parent, + unsigned long *pentry) +{ + struct bio *bio; + unsigned long entry; + + bio = bio_alloc(GFP_ATOMIC, 1); + if (!bio) + return -ENOMEM; + + entry = get_entry_bdev(zram); + if (!entry) { + bio_put(bio); + return -ENOSPC; + } + + bio->bi_iter.bi_sector = entry * (PAGE_SIZE >> 9); + bio_set_dev(bio, zram->bdev); + if (!bio_add_page(bio, bvec->bv_page, bvec->bv_len, + bvec->bv_offset)) { + bio_put(bio); + put_entry_bdev(zram, entry); + return -EIO; + } + + if (!parent) { + bio->bi_opf = REQ_OP_WRITE | REQ_SYNC; + bio->bi_end_io = zram_page_end_io; + } else { + bio->bi_opf = parent->bi_opf; + bio_chain(bio, parent); + } + + submit_bio(bio); + *pentry = entry; + + return 0; +} + +static void zram_wb_clear(struct zram *zram, u32 index) +{ + unsigned long entry; + + zram_clear_flag(zram, index, ZRAM_WB); + entry = zram_get_element(zram, index); + zram_set_element(zram, index, 0); + put_entry_bdev(zram, entry); +} + +#else +static bool zram_wb_enabled(struct zram *zram) { return false; } +static inline void reset_bdev(struct zram *zram) {}; +static int write_to_bdev(struct zram *zram, struct bio_vec *bvec, + u32 index, struct bio *parent, + unsigned long *pentry) + +{ + return -EIO; +} + +static int read_from_bdev(struct zram *zram, struct bio_vec *bvec, + unsigned long entry, struct bio *parent, bool sync) +{ + return -EIO; +} +static void zram_wb_clear(struct zram *zram, u32 index) {} +#endif + + /* * We switched to per-cpu streams and this attr is not needed anymore. * However, we will keep it around for some time, because: @@ -453,30 +787,6 @@ static bool zram_same_page_read(struct zram *zram, u32 index, return false; } -static bool zram_same_page_write(struct zram *zram, u32 index, - struct page *page) -{ - unsigned long element; - void *mem = kmap_atomic(page); - - if (page_same_filled(mem, &element)) { - kunmap_atomic(mem); - /* Free memory associated with this sector now. */ - zram_slot_lock(zram, index); - zram_free_page(zram, index); - zram_set_flag(zram, index, ZRAM_SAME); - zram_set_element(zram, index, element); - zram_slot_unlock(zram, index); - - atomic64_inc(&zram->stats.same_pages); - atomic64_inc(&zram->stats.pages_stored); - return true; - } - kunmap_atomic(mem); - - return false; -} - static void zram_meta_free(struct zram *zram, u64 disksize) { size_t num_pages = disksize >> PAGE_SHIFT; @@ -515,7 +825,13 @@ static bool zram_meta_alloc(struct zram *zram, u64 disksize) */ static void zram_free_page(struct zram *zram, size_t index) { - unsigned long handle = zram_get_handle(zram, index); + unsigned long handle; + + if (zram_wb_enabled(zram) && zram_test_flag(zram, index, ZRAM_WB)) { + zram_wb_clear(zram, index); + atomic64_dec(&zram->stats.pages_stored); + return; + } /* * No memory is allocated for same element filled pages. @@ -529,6 +845,7 @@ static void zram_free_page(struct zram *zram, size_t index) return; } + handle = zram_get_handle(zram, index); if (!handle) return; @@ -542,13 +859,31 @@ static void zram_free_page(struct zram *zram, size_t index) zram_set_obj_size(zram, index, 0); } -static int zram_decompress_page(struct zram *zram, struct page *page, u32 index) +static int __zram_bvec_read(struct zram *zram, struct page *page, u32 index, + struct bio *bio, bool partial_io) { int ret; unsigned long handle; unsigned int size; void *src, *dst; + if (zram_wb_enabled(zram)) { + zram_slot_lock(zram, index); + if (zram_test_flag(zram, index, ZRAM_WB)) { + struct bio_vec bvec; + + zram_slot_unlock(zram, index); + + bvec.bv_page = page; + bvec.bv_len = PAGE_SIZE; + bvec.bv_offset = 0; + return read_from_bdev(zram, &bvec, + zram_get_element(zram, index), + bio, partial_io); + } + zram_slot_unlock(zram, index); + } + if (zram_same_page_read(zram, index, page, 0, PAGE_SIZE)) return 0; @@ -581,7 +916,7 @@ static int zram_decompress_page(struct zram *zram, struct page *page, u32 index) } static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec, - u32 index, int offset) + u32 index, int offset, struct bio *bio) { int ret; struct page *page; @@ -594,7 +929,7 @@ static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec, return -ENOMEM; } - ret = zram_decompress_page(zram, page, index); + ret = __zram_bvec_read(zram, page, index, bio, is_partial_io(bvec)); if (unlikely(ret)) goto out; @@ -613,30 +948,57 @@ out: return ret; } -static int zram_compress(struct zram *zram, struct zcomp_strm **zstrm, - struct page *page, - unsigned long *out_handle, unsigned int *out_comp_len) +static int __zram_bvec_write(struct zram *zram, struct bio_vec *bvec, + u32 index, struct bio *bio) { - int ret; - unsigned int comp_len; - void *src; + int ret = 0; unsigned long alloced_pages; unsigned long handle = 0; + unsigned int comp_len = 0; + void *src, *dst, *mem; + struct zcomp_strm *zstrm; + struct page *page = bvec->bv_page; + unsigned long element = 0; + enum zram_pageflags flags = 0; + bool allow_wb = true; + + mem = kmap_atomic(page); + if (page_same_filled(mem, &element)) { + kunmap_atomic(mem); + /* Free memory associated with this sector now. */ + flags = ZRAM_SAME; + atomic64_inc(&zram->stats.same_pages); + goto out; + } + kunmap_atomic(mem); compress_again: + zstrm = zcomp_stream_get(zram->comp); src = kmap_atomic(page); - ret = zcomp_compress(*zstrm, src, &comp_len); + ret = zcomp_compress(zstrm, src, &comp_len); kunmap_atomic(src); if (unlikely(ret)) { + zcomp_stream_put(zram->comp); pr_err("Compression failed! err=%d\n", ret); - if (handle) - zs_free(zram->mem_pool, handle); + zs_free(zram->mem_pool, handle); return ret; } - if (unlikely(comp_len > max_zpage_size)) + if (unlikely(comp_len > max_zpage_size)) { + if (zram_wb_enabled(zram) && allow_wb) { + zcomp_stream_put(zram->comp); + ret = write_to_bdev(zram, bvec, index, bio, &element); + if (!ret) { + flags = ZRAM_WB; + ret = 1; + goto out; + } + allow_wb = false; + goto compress_again; + } comp_len = PAGE_SIZE; + } /* * handle allocation has 2 paths: @@ -663,7 +1025,6 @@ compress_again: handle = zs_malloc(zram->mem_pool, comp_len, GFP_NOIO | __GFP_HIGHMEM | __GFP_MOVABLE); - *zstrm = zcomp_stream_get(zram->comp); if (handle) goto compress_again; return -ENOMEM; @@ -673,34 +1034,11 @@ compress_again: update_used_max(zram, alloced_pages); if (zram->limit_pages && alloced_pages > zram->limit_pages) { + zcomp_stream_put(zram->comp); zs_free(zram->mem_pool, handle); return -ENOMEM; } - *out_handle = handle; - *out_comp_len = comp_len; - return 0; -} - -static int __zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index) -{ - int ret; - unsigned long handle; - unsigned int comp_len; - void *src, *dst; - struct zcomp_strm *zstrm; - struct page *page = bvec->bv_page; - - if (zram_same_page_write(zram, index, page)) - return 0; - - zstrm = zcomp_stream_get(zram->comp); - ret = zram_compress(zram, &zstrm, page, &handle, &comp_len); - if (ret) { - zcomp_stream_put(zram->comp); - return ret; - } - dst = zs_map_object(zram->mem_pool, handle, ZS_MM_WO); src = zstrm->buffer; @@ -712,25 +1050,31 @@ static int __zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index) zcomp_stream_put(zram->comp); zs_unmap_object(zram->mem_pool, handle); - + atomic64_add(comp_len, &zram->stats.compr_data_size); +out: /* * Free memory associated with this sector * before overwriting unused sectors. */ zram_slot_lock(zram, index); zram_free_page(zram, index); - zram_set_handle(zram, index, handle); - zram_set_obj_size(zram, index, comp_len); + + if (flags) { + zram_set_flag(zram, index, flags); + zram_set_element(zram, index, element); + } else { + zram_set_handle(zram, index, handle); + zram_set_obj_size(zram, index, comp_len); + } zram_slot_unlock(zram, index); /* Update stats */ - atomic64_add(comp_len, &zram->stats.compr_data_size); atomic64_inc(&zram->stats.pages_stored); - return 0; + return ret; } static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, - u32 index, int offset) + u32 index, int offset, struct bio *bio) { int ret; struct page *page = NULL; @@ -748,7 +1092,7 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, if (!page) return -ENOMEM; - ret = zram_decompress_page(zram, page, index); + ret = __zram_bvec_read(zram, page, index, bio, true); if (ret) goto out; @@ -763,7 +1107,7 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, vec.bv_offset = 0; } - ret = __zram_bvec_write(zram, &vec, index); + ret = __zram_bvec_write(zram, &vec, index, bio); out: if (is_partial_io(bvec)) __free_page(page); @@ -808,28 +1152,34 @@ static void zram_bio_discard(struct zram *zram, u32 index, } } +/* + * Returns errno if it has some problem. Otherwise return 0 or 1. + * Returns 0 if IO request was done synchronously + * Returns 1 if IO request was successfully submitted. + */ static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index, - int offset, bool is_write) + int offset, bool is_write, struct bio *bio) { unsigned long start_time = jiffies; int rw_acct = is_write ? REQ_OP_WRITE : REQ_OP_READ; + struct request_queue *q = zram->disk->queue; int ret; - generic_start_io_acct(rw_acct, bvec->bv_len >> SECTOR_SHIFT, + generic_start_io_acct(q, rw_acct, bvec->bv_len >> SECTOR_SHIFT, &zram->disk->part0); if (!is_write) { atomic64_inc(&zram->stats.num_reads); - ret = zram_bvec_read(zram, bvec, index, offset); + ret = zram_bvec_read(zram, bvec, index, offset, bio); flush_dcache_page(bvec->bv_page); } else { atomic64_inc(&zram->stats.num_writes); - ret = zram_bvec_write(zram, bvec, index, offset); + ret = zram_bvec_write(zram, bvec, index, offset, bio); } - generic_end_io_acct(rw_acct, &zram->disk->part0, start_time); + generic_end_io_acct(q, rw_acct, &zram->disk->part0, start_time); - if (unlikely(ret)) { + if (unlikely(ret < 0)) { if (!is_write) atomic64_inc(&zram->stats.failed_reads); else @@ -868,7 +1218,7 @@ static void __zram_make_request(struct zram *zram, struct bio *bio) bv.bv_len = min_t(unsigned int, PAGE_SIZE - offset, unwritten); if (zram_bvec_rw(zram, &bv, index, offset, - op_is_write(bio_op(bio))) < 0) + op_is_write(bio_op(bio)), bio) < 0) goto out; bv.bv_offset += bv.bv_len; @@ -922,16 +1272,18 @@ static void zram_slot_free_notify(struct block_device *bdev, static int zram_rw_page(struct block_device *bdev, sector_t sector, struct page *page, bool is_write) { - int offset, err = -EIO; + int offset, ret; u32 index; struct zram *zram; struct bio_vec bv; + if (PageTransHuge(page)) + return -ENOTSUPP; zram = bdev->bd_disk->private_data; if (!valid_io_request(zram, sector, PAGE_SIZE)) { atomic64_inc(&zram->stats.invalid_io); - err = -EINVAL; + ret = -EINVAL; goto out; } @@ -942,7 +1294,7 @@ static int zram_rw_page(struct block_device *bdev, sector_t sector, bv.bv_len = PAGE_SIZE; bv.bv_offset = 0; - err = zram_bvec_rw(zram, &bv, index, offset, is_write); + ret = zram_bvec_rw(zram, &bv, index, offset, is_write, NULL); out: /* * If I/O fails, just return error(ie, non-zero) without @@ -952,9 +1304,20 @@ out: * bio->bi_end_io does things to handle the error * (e.g., SetPageError, set_page_dirty and extra works). */ - if (err == 0) + if (unlikely(ret < 0)) + return ret; + + switch (ret) { + case 0: page_endio(page, is_write, 0); - return err; + break; + case 1: + ret = 0; + break; + default: + WARN_ON(1); + } + return ret; } static void zram_reset_device(struct zram *zram) @@ -983,6 +1346,7 @@ static void zram_reset_device(struct zram *zram) zram_meta_free(zram, disksize); memset(&zram->stats, 0, sizeof(zram->stats)); zcomp_destroy(comp); + reset_bdev(zram); } static ssize_t disksize_store(struct device *dev, @@ -1108,6 +1472,9 @@ static DEVICE_ATTR_WO(mem_limit); static DEVICE_ATTR_WO(mem_used_max); static DEVICE_ATTR_RW(max_comp_streams); static DEVICE_ATTR_RW(comp_algorithm); +#ifdef CONFIG_ZRAM_WRITEBACK +static DEVICE_ATTR_RW(backing_dev); +#endif static struct attribute *zram_disk_attrs[] = { &dev_attr_disksize.attr, @@ -1118,6 +1485,9 @@ static struct attribute *zram_disk_attrs[] = { &dev_attr_mem_used_max.attr, &dev_attr_max_comp_streams.attr, &dev_attr_comp_algorithm.attr, +#ifdef CONFIG_ZRAM_WRITEBACK + &dev_attr_backing_dev.attr, +#endif &dev_attr_io_stat.attr, &dev_attr_mm_stat.attr, &dev_attr_debug_stat.attr, diff --git a/drivers/block/zram/zram_drv.h b/drivers/block/zram/zram_drv.h index e34e44d02e3e..31762db861e3 100644 --- a/drivers/block/zram/zram_drv.h +++ b/drivers/block/zram/zram_drv.h @@ -60,9 +60,10 @@ static const size_t max_zpage_size = PAGE_SIZE / 4 * 3; /* Flags for zram pages (table[page_no].value) */ enum zram_pageflags { - /* Page consists entirely of zeros */ + /* Page consists the same element */ ZRAM_SAME = ZRAM_FLAG_SHIFT, ZRAM_ACCESS, /* page is now accessed */ + ZRAM_WB, /* page is stored on backing_device */ __NR_ZRAM_PAGEFLAGS, }; @@ -115,5 +116,13 @@ struct zram { * zram is claimed so open request will be failed */ bool claim; /* Protected by bdev->bd_mutex */ +#ifdef CONFIG_ZRAM_WRITEBACK + struct file *backing_dev; + struct block_device *bdev; + unsigned int old_block_size; + unsigned long *bitmap; + unsigned long nr_pages; + spinlock_t bitmap_lock; +#endif }; #endif |