diff options
author | Thomas Gleixner <tglx@mtd.linutronix.de> | 2005-11-06 15:36:37 +0100 |
---|---|---|
committer | Thomas Gleixner <tglx@mtd.linutronix.de> | 2005-11-06 15:36:37 +0100 |
commit | 2fc2991175bf77395e6b15fe6b2304d3bf72da40 (patch) | |
tree | b0ff38c09240e7c00e1577d447ebe89143d752dc /arch/sparc64 | |
parent | [MTD] mtdchar: Return EINVAL for bad seeks instead of fixing up to valid byte (diff) | |
parent | [PATCH] nvidiafb: Geforce 7800 series support added (diff) | |
download | linux-dev-2fc2991175bf77395e6b15fe6b2304d3bf72da40.tar.xz linux-dev-2fc2991175bf77395e6b15fe6b2304d3bf72da40.zip |
Merge branch 'master' of /home/tglx/work/mtd/git/linux-2.6.git/
Diffstat (limited to 'arch/sparc64')
76 files changed, 2782 insertions, 4354 deletions
diff --git a/arch/sparc64/Kconfig b/arch/sparc64/Kconfig index 9afd28e2c4d5..1e9d8638a28a 100644 --- a/arch/sparc64/Kconfig +++ b/arch/sparc64/Kconfig @@ -5,6 +5,16 @@ mainmenu "Linux/UltraSPARC Kernel Configuration" +config SPARC64 + bool + default y + help + SPARC is a family of RISC microprocessors designed and marketed by + Sun Microsystems, incorporated. This port covers the newer 64-bit + UltraSPARC. The UltraLinux project maintains both the SPARC32 and + SPARC64 ports; its web page is available at + <http://www.ultralinux.org/>. + config 64BIT def_bool y @@ -16,6 +26,10 @@ config TIME_INTERPOLATION bool default y +config ARCH_MAY_HAVE_PC_FDC + bool + default y + choice prompt "Kernel page size" default SPARC64_PAGE_SIZE_8KB @@ -71,75 +85,6 @@ config SYSVIPC_COMPAT menu "General machine setup" -config BBC_I2C - tristate "UltraSPARC-III bootbus i2c controller driver" - depends on PCI - help - The BBC devices on the UltraSPARC III have two I2C controllers. The - first I2C controller connects mainly to configuration PROMs (NVRAM, - CPU configuration, DIMM types, etc.). The second I2C controller - connects to environmental control devices such as fans and - temperature sensors. The second controller also connects to the - smartcard reader, if present. Say Y to enable support for these. - -config VT - bool "Virtual terminal" if EMBEDDED - select INPUT - default y - ---help--- - If you say Y here, you will get support for terminal devices with - display and keyboard devices. These are called "virtual" because you - can run several virtual terminals (also called virtual consoles) on - one physical terminal. This is rather useful, for example one - virtual terminal can collect system messages and warnings, another - one can be used for a text-mode user session, and a third could run - an X session, all in parallel. Switching between virtual terminals - is done with certain key combinations, usually Alt-<function key>. - - The setterm command ("man setterm") can be used to change the - properties (such as colors or beeping) of a virtual terminal. The - man page console_codes(4) ("man console_codes") contains the special - character sequences that can be used to change those properties - directly. The fonts used on virtual terminals can be changed with - the setfont ("man setfont") command and the key bindings are defined - with the loadkeys ("man loadkeys") command. - - You need at least one virtual terminal device in order to make use - of your keyboard and monitor. Therefore, only people configuring an - embedded system would want to say N here in order to save some - memory; the only way to log into such a system is then via a serial - or network connection. - - If unsure, say Y, or else you won't be able to do much with your new - shiny Linux system :-) - -config VT_CONSOLE - bool "Support for console on virtual terminal" if EMBEDDED - depends on VT - default y - ---help--- - The system console is the device which receives all kernel messages - and warnings and which allows logins in single user mode. If you - answer Y here, a virtual terminal (the device used to interact with - a physical terminal) can be used as system console. This is the most - common mode of operations, so you should say Y here unless you want - the kernel messages be output only to a serial port (in which case - you should say Y to "Console on serial port", below). - - If you do say Y here, by default the currently visible virtual - terminal (/dev/tty0) will be used as system console. You can change - that with a kernel command line option such as "console=tty3" which - would use the third virtual terminal as system console. (Try "man - bootparam" or see the documentation of your boot loader (lilo or - loadlin) about how to pass options to the kernel at boot time.) - - If unsure, say Y. - -config HW_CONSOLE - bool - depends on VT - default y - config SMP bool "Symmetric multi-processing support" ---help--- @@ -205,17 +150,6 @@ config US2E_FREQ If in doubt, say N. -# Identify this as a Sparc64 build -config SPARC64 - bool - default y - help - SPARC is a family of RISC microprocessors designed and marketed by - Sun Microsystems, incorporated. This port covers the newer 64-bit - UltraSPARC. The UltraLinux project maintains both the SPARC32 and - SPARC64 ports; its web page is available at - <http://www.ultralinux.org/>. - # Global things across all Sun machines. config RWSEM_GENERIC_SPINLOCK bool @@ -246,6 +180,10 @@ config HUGETLB_PAGE_SIZE_64K endchoice +endmenu + +source "mm/Kconfig" + config GENERIC_ISA_DMA bool default y @@ -344,33 +282,6 @@ config PCI_DOMAINS bool default PCI -config RTC - tristate - depends on PCI - default y - ---help--- - If you say Y here and create a character special file /dev/rtc with - major number 10 and minor number 135 using mknod ("man mknod"), you - will get access to the real time clock (or hardware clock) built - into your computer. - - Every PC has such a clock built in. It can be used to generate - signals from as low as 1Hz up to 8192Hz, and can also be used - as a 24 hour alarm. It reports status information via the file - /proc/driver/rtc and its behaviour is set by various ioctls on - /dev/rtc. - - If you run Linux on a multiprocessor machine and said Y to - "Symmetric Multi Processing" above, you should say Y here to read - and set the RTC in an SMP compatible fashion. - - If you think you have a use for such a device (such as periodic data - sampling), then say Y here, and read <file:Documentation/rtc.txt> - for details. - - To compile this driver as a module, choose M here: the - module will be called rtc. - source "drivers/pci/Kconfig" config SUN_OPENPROMFS @@ -414,6 +325,8 @@ config BINFMT_AOUT32 If you want to run SunOS binaries (see SunOS binary emulation below) or other a.out binaries, say Y. If unsure, say N. +menu "Executable file formats" + source "fs/Kconfig.binfmt" config SUNOS_EMUL @@ -436,74 +349,7 @@ config SOLARIS_EMUL To compile this code as a module, choose M here: the module will be called solaris. -source "drivers/parport/Kconfig" - -config PRINTER - tristate "Parallel printer support" - depends on PARPORT - ---help--- - If you intend to attach a printer to the parallel port of your Linux - box (as opposed to using a serial printer; if the connector at the - printer has 9 or 25 holes ["female"], then it's serial), say Y. - Also read the Printing-HOWTO, available from - <http://www.tldp.org/docs.html#howto>. - - It is possible to share one parallel port among several devices - (e.g. printer and ZIP drive) and it is safe to compile the - corresponding drivers into the kernel. - To compile this driver as a module, choose M here and read - <file:Documentation/parport.txt>. The module will be called lp. - - If you have several parallel ports, you can specify which ports to - use with the "lp" kernel command line option. (Try "man bootparam" - or see the documentation of your boot loader (lilo or loadlin) about - how to pass options to the kernel at boot time.) The syntax of the - "lp" command line option can be found in <file:drivers/char/lp.c>. - - If you have more than 8 printers, you need to increase the LP_NO - macro in lp.c and the PARPORT_MAX macro in parport.h. - -config PPDEV - tristate "Support for user-space parallel port device drivers" - depends on PARPORT - ---help--- - Saying Y to this adds support for /dev/parport device nodes. This - is needed for programs that want portable access to the parallel - port, for instance deviceid (which displays Plug-and-Play device - IDs). - - This is the parallel port equivalent of SCSI generic support (sg). - It is safe to say N to this -- it is not needed for normal printing - or parallel port CD-ROM/disk support. - - To compile this driver as a module, choose M here: the - module will be called ppdev. - - If unsure, say N. - -config ENVCTRL - tristate "SUNW, envctrl support" - depends on PCI - help - Kernel support for temperature and fan monitoring on Sun SME - machines. - - To compile this driver as a module, choose M here: the - module will be called envctrl. - -config DISPLAY7SEG - tristate "7-Segment Display support" - depends on PCI - ---help--- - This is the driver for the 7-segment display and LED present on - Sun Microsystems CompactPCI models CP1400 and CP1500. - - To compile this driver as a module, choose M here: the - module will be called display7seg. - - If you do not have a CompactPCI model CP1400 or CP1500, or - another UltraSPARC-IIi-cEngine boardset with a 7-segment display, - you should say N to this option. +endmenu config CMDLINE_BOOL bool "Default bootloader kernel arguments" @@ -521,148 +367,16 @@ config CMDLINE NOTE: This option WILL override the PROM bootargs setting! -source "mm/Kconfig" - -endmenu - source "net/Kconfig" -source "drivers/base/Kconfig" - -source "drivers/video/Kconfig" - -source "drivers/serial/Kconfig" +source "drivers/Kconfig" source "drivers/sbus/char/Kconfig" -source "drivers/mtd/Kconfig" - -source "drivers/block/Kconfig" - -source "drivers/ide/Kconfig" - -source "drivers/scsi/Kconfig" - source "drivers/fc4/Kconfig" -source "drivers/md/Kconfig" - -if PCI -source "drivers/message/fusion/Kconfig" -endif - -source "drivers/ieee1394/Kconfig" - -source "drivers/net/Kconfig" - -source "drivers/isdn/Kconfig" - -source "drivers/telephony/Kconfig" - -# This one must be before the filesystem configs. -DaveM - -menu "Unix98 PTY support" - -config UNIX98_PTYS - bool "Unix98 PTY support" - ---help--- - A pseudo terminal (PTY) is a software device consisting of two - halves: a master and a slave. The slave device behaves identical to - a physical terminal; the master device is used by a process to - read data from and write data to the slave, thereby emulating a - terminal. Typical programs for the master side are telnet servers - and xterms. - - Linux has traditionally used the BSD-like names /dev/ptyxx for - masters and /dev/ttyxx for slaves of pseudo terminals. This scheme - has a number of problems. The GNU C library glibc 2.1 and later, - however, supports the Unix98 naming standard: in order to acquire a - pseudo terminal, a process opens /dev/ptmx; the number of the pseudo - terminal is then made available to the process and the pseudo - terminal slave can be accessed as /dev/pts/<number>. What was - traditionally /dev/ttyp2 will then be /dev/pts/2, for example. - - The entries in /dev/pts/ are created on the fly by a virtual - file system; therefore, if you say Y here you should say Y to - "/dev/pts file system for Unix98 PTYs" as well. - - If you want to say Y here, you need to have the C library glibc 2.1 - or later (equal to libc-6.1, check with "ls -l /lib/libc.so.*"). - Read the instructions in <file:Documentation/Changes> pertaining to - pseudo terminals. It's safe to say N. - -config UNIX98_PTY_COUNT - int "Maximum number of Unix98 PTYs in use (0-2048)" - depends on UNIX98_PTYS - default "256" - help - The maximum number of Unix98 PTYs that can be used at any one time. - The default is 256, and should be enough for desktop systems. Server - machines which support incoming telnet/rlogin/ssh connections and/or - serve several X terminals may want to increase this: every incoming - connection and every xterm uses up one PTY. - - When not in use, each additional set of 256 PTYs occupy - approximately 8 KB of kernel memory on 32-bit architectures. - -endmenu - -menu "XFree86 DRI support" - -config DRM - bool "Direct Rendering Manager (XFree86 DRI support)" - help - Kernel-level support for the Direct Rendering Infrastructure (DRI) - introduced in XFree86 4.0. If you say Y here, you need to select - the module that's right for your graphics card from the list below. - These modules provide support for synchronization, security, and - DMA transfers. Please see <http://dri.sourceforge.net/> for more - details. You should also select and configure AGP - (/dev/agpgart) support. - -config DRM_FFB - tristate "Creator/Creator3D" - depends on DRM && BROKEN - help - Choose this option if you have one of Sun's Creator3D-based graphics - and frame buffer cards. Product page at - <http://www.sun.com/desktop/products/Graphics/creator3d.html>. - -config DRM_TDFX - tristate "3dfx Banshee/Voodoo3+" - depends on DRM - help - Choose this option if you have a 3dfx Banshee or Voodoo3 (or later), - graphics card. If M is selected, the module will be called tdfx. - -config DRM_R128 - tristate "ATI Rage 128" - depends on DRM - help - Choose this option if you have an ATI Rage 128 graphics card. If M - is selected, the module will be called r128. AGP support for - this card is strongly suggested (unless you have a PCI version). - -endmenu - -source "drivers/input/Kconfig" - -source "drivers/i2c/Kconfig" - -source "drivers/hwmon/Kconfig" - source "fs/Kconfig" -source "drivers/media/Kconfig" - -source "sound/Kconfig" - -source "drivers/usb/Kconfig" - -source "drivers/infiniband/Kconfig" - -source "drivers/char/watchdog/Kconfig" - source "arch/sparc64/oprofile/Kconfig" source "arch/sparc64/Kconfig.debug" diff --git a/arch/sparc64/Kconfig.debug b/arch/sparc64/Kconfig.debug index cd8d39fb954d..fa06ea04837b 100644 --- a/arch/sparc64/Kconfig.debug +++ b/arch/sparc64/Kconfig.debug @@ -33,13 +33,13 @@ config DEBUG_BOOTMEM depends on DEBUG_KERNEL bool "Debug BOOTMEM initialization" -# We have a custom atomic_dec_and_lock() implementation but it's not -# compatible with spinlock debugging so we need to fall back on -# the generic version in that case. -config HAVE_DEC_LOCK - bool - depends on SMP && !DEBUG_SPINLOCK - default y +config DEBUG_PAGEALLOC + bool "Page alloc debugging" + depends on DEBUG_KERNEL && !SOFTWARE_SUSPEND + help + Unmap pages from the kernel linear mapping after free_pages(). + This results in a large slowdown, but helps to find certain types + of memory corruptions. config MCOUNT bool diff --git a/arch/sparc64/kernel/Makefile b/arch/sparc64/kernel/Makefile index 093281bdf85f..6f00ab8b9d23 100644 --- a/arch/sparc64/kernel/Makefile +++ b/arch/sparc64/kernel/Makefile @@ -8,7 +8,7 @@ EXTRA_CFLAGS := -Werror extra-y := head.o init_task.o vmlinux.lds obj-y := process.o setup.o cpu.o idprom.o \ - traps.o devices.o auxio.o \ + traps.o devices.o auxio.o una_asm.o \ irq.o ptrace.o time.o sys_sparc.o signal.o \ unaligned.o central.o pci.o starfire.o semaphore.o \ power.o sbus.o iommu_common.o sparc64_ksyms.o chmc.o diff --git a/arch/sparc64/kernel/asm-offsets.c b/arch/sparc64/kernel/asm-offsets.c new file mode 100644 index 000000000000..9e263112a6e2 --- /dev/null +++ b/arch/sparc64/kernel/asm-offsets.c @@ -0,0 +1 @@ +/* Dummy asm-offsets.c file. Required by kbuild and ready to be used - hint! */ diff --git a/arch/sparc64/kernel/binfmt_aout32.c b/arch/sparc64/kernel/binfmt_aout32.c index b2854ef221d0..edf52d06b280 100644 --- a/arch/sparc64/kernel/binfmt_aout32.c +++ b/arch/sparc64/kernel/binfmt_aout32.c @@ -241,7 +241,6 @@ static int load_aout32_binary(struct linux_binprm * bprm, struct pt_regs * regs) current->mm->brk = ex.a_bss + (current->mm->start_brk = N_BSSADDR(ex)); - set_mm_counter(current->mm, rss, 0); current->mm->mmap = NULL; compute_creds(bprm); current->flags &= ~PF_FORKNOEXEC; diff --git a/arch/sparc64/kernel/cpu.c b/arch/sparc64/kernel/cpu.c index 48756958116b..77ef5df4e5a7 100644 --- a/arch/sparc64/kernel/cpu.c +++ b/arch/sparc64/kernel/cpu.c @@ -39,6 +39,8 @@ struct cpu_fp_info linux_sparc_fpu[] = { { 0x3e, 0x15, 0, "UltraSparc III+ integrated FPU"}, { 0x3e, 0x16, 0, "UltraSparc IIIi integrated FPU"}, { 0x3e, 0x18, 0, "UltraSparc IV integrated FPU"}, + { 0x3e, 0x19, 0, "UltraSparc IV+ integrated FPU"}, + { 0x3e, 0x22, 0, "UltraSparc IIIi+ integrated FPU"}, }; #define NSPARCFPU (sizeof(linux_sparc_fpu)/sizeof(struct cpu_fp_info)) @@ -53,6 +55,8 @@ struct cpu_iu_info linux_sparc_chips[] = { { 0x3e, 0x15, "TI UltraSparc III+ (Cheetah+)"}, { 0x3e, 0x16, "TI UltraSparc IIIi (Jalapeno)"}, { 0x3e, 0x18, "TI UltraSparc IV (Jaguar)"}, + { 0x3e, 0x19, "TI UltraSparc IV+ (Panther)"}, + { 0x3e, 0x22, "TI UltraSparc IIIi+ (Serrano)"}, }; #define NSPARCCHIPS (sizeof(linux_sparc_chips)/sizeof(struct cpu_iu_info)) diff --git a/arch/sparc64/kernel/devices.c b/arch/sparc64/kernel/devices.c index d710274e516b..df9a1ca8fd77 100644 --- a/arch/sparc64/kernel/devices.c +++ b/arch/sparc64/kernel/devices.c @@ -135,6 +135,28 @@ void __init device_scan(void) cpu_data(0).clock_tick = prom_getintdefault(cpu_node, "clock-frequency", 0); + cpu_data(0).dcache_size = prom_getintdefault(cpu_node, + "dcache-size", + 16 * 1024); + cpu_data(0).dcache_line_size = + prom_getintdefault(cpu_node, "dcache-line-size", 32); + cpu_data(0).icache_size = prom_getintdefault(cpu_node, + "icache-size", + 16 * 1024); + cpu_data(0).icache_line_size = + prom_getintdefault(cpu_node, "icache-line-size", 32); + cpu_data(0).ecache_size = prom_getintdefault(cpu_node, + "ecache-size", + 4 * 1024 * 1024); + cpu_data(0).ecache_line_size = + prom_getintdefault(cpu_node, "ecache-line-size", 64); + printk("CPU[0]: Caches " + "D[sz(%d):line_sz(%d)] " + "I[sz(%d):line_sz(%d)] " + "E[sz(%d):line_sz(%d)]\n", + cpu_data(0).dcache_size, cpu_data(0).dcache_line_size, + cpu_data(0).icache_size, cpu_data(0).icache_line_size, + cpu_data(0).ecache_size, cpu_data(0).ecache_line_size); } #endif diff --git a/arch/sparc64/kernel/dtlb_backend.S b/arch/sparc64/kernel/dtlb_backend.S index 538522848ad4..acc889a7f9c1 100644 --- a/arch/sparc64/kernel/dtlb_backend.S +++ b/arch/sparc64/kernel/dtlb_backend.S @@ -9,17 +9,7 @@ #include <asm/pgtable.h> #include <asm/mmu.h> -#if PAGE_SHIFT == 13 -#define SZ_BITS _PAGE_SZ8K -#elif PAGE_SHIFT == 16 -#define SZ_BITS _PAGE_SZ64K -#elif PAGE_SHIFT == 19 -#define SZ_BITS _PAGE_SZ512K -#elif PAGE_SHIFT == 22 -#define SZ_BITS _PAGE_SZ4MB -#endif - -#define VALID_SZ_BITS (_PAGE_VALID | SZ_BITS) +#define VALID_SZ_BITS (_PAGE_VALID | _PAGE_SZBITS) #define VPTE_BITS (_PAGE_CP | _PAGE_CV | _PAGE_P ) #define VPTE_SHIFT (PAGE_SHIFT - 3) @@ -163,7 +153,6 @@ sparc64_vpte_continue: stxa %g4, [%g1 + %g1] ASI_DMMU ! Restore previous TAG_ACCESS retry ! Load PTE once again -#undef SZ_BITS #undef VALID_SZ_BITS #undef VPTE_SHIFT #undef VPTE_BITS diff --git a/arch/sparc64/kernel/dtlb_base.S b/arch/sparc64/kernel/dtlb_base.S index ded2fed23fcc..6528786840c0 100644 --- a/arch/sparc64/kernel/dtlb_base.S +++ b/arch/sparc64/kernel/dtlb_base.S @@ -53,39 +53,36 @@ * be guaranteed to be 0 ... mmu_context.h does guarantee this * by only using 10 bits in the hwcontext value. */ -#define CREATE_VPTE_OFFSET1(r1, r2) +#define CREATE_VPTE_OFFSET1(r1, r2) nop #define CREATE_VPTE_OFFSET2(r1, r2) \ srax r1, 10, r2 -#define CREATE_VPTE_NOP nop #else #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, PAGE_SHIFT, r2 #define CREATE_VPTE_OFFSET2(r1, r2) \ sllx r2, 3, r2 -#define CREATE_VPTE_NOP #endif /* DTLB ** ICACHE line 1: Quick user TLB misses */ + mov TLB_SFSR, %g1 ldxa [%g1 + %g1] ASI_DMMU, %g4 ! Get TAG_ACCESS andcc %g4, TAG_CONTEXT_BITS, %g0 ! From Nucleus? from_tl1_trap: rdpr %tl, %g5 ! For TL==3 test CREATE_VPTE_OFFSET1(%g4, %g6) ! Create VPTE offset - be,pn %xcc, 3f ! Yep, special processing + be,pn %xcc, kvmap ! Yep, special processing CREATE_VPTE_OFFSET2(%g4, %g6) ! Create VPTE offset cmp %g5, 4 ! Last trap level? - be,pn %xcc, longpath ! Yep, cannot risk VPTE miss - nop ! delay slot /* DTLB ** ICACHE line 2: User finish + quick kernel TLB misses */ + be,pn %xcc, longpath ! Yep, cannot risk VPTE miss + nop ! delay slot ldxa [%g3 + %g6] ASI_S, %g5 ! Load VPTE 1: brgez,pn %g5, longpath ! Invalid, branch out nop ! Delay-slot 9: stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB retry ! Trap return -3: brlz,pt %g4, 9b ! Kernel virtual map? - xor %g2, %g4, %g5 ! Finish bit twiddles - ba,a,pt %xcc, kvmap ! Yep, go check for obp/vmalloc + nop /* DTLB ** ICACHE line 3: winfixups+real_faults */ longpath: @@ -106,8 +103,7 @@ longpath: nop nop nop - CREATE_VPTE_NOP + nop #undef CREATE_VPTE_OFFSET1 #undef CREATE_VPTE_OFFSET2 -#undef CREATE_VPTE_NOP diff --git a/arch/sparc64/kernel/dtlb_prot.S b/arch/sparc64/kernel/dtlb_prot.S index d848bb7374bb..e0a920162604 100644 --- a/arch/sparc64/kernel/dtlb_prot.S +++ b/arch/sparc64/kernel/dtlb_prot.S @@ -14,14 +14,14 @@ */ /* PROT ** ICACHE line 1: User DTLB protection trap */ - stxa %g0, [%g1] ASI_DMMU ! Clear SFSR FaultValid bit - membar #Sync ! Synchronize ASI stores - rdpr %pstate, %g5 ! Move into alternate globals + mov TLB_SFSR, %g1 + stxa %g0, [%g1] ASI_DMMU ! Clear FaultValid bit + membar #Sync ! Synchronize stores + rdpr %pstate, %g5 ! Move into alt-globals wrpr %g5, PSTATE_AG|PSTATE_MG, %pstate - rdpr %tl, %g1 ! Need to do a winfixup? + rdpr %tl, %g1 ! Need a winfixup? cmp %g1, 1 ! Trap level >1? - mov TLB_TAG_ACCESS, %g4 ! Prepare reload of vaddr - nop + mov TLB_TAG_ACCESS, %g4 ! For reload of vaddr /* PROT ** ICACHE line 2: More real fault processing */ bgu,pn %xcc, winfix_trampoline ! Yes, perform winfixup diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S index 88332f00094a..11a848402fb1 100644 --- a/arch/sparc64/kernel/entry.S +++ b/arch/sparc64/kernel/entry.S @@ -21,6 +21,7 @@ #include <asm/visasm.h> #include <asm/estate.h> #include <asm/auxio.h> +#include <asm/sfafsr.h> #define curptr g6 @@ -29,164 +30,10 @@ .text .align 32 - .globl sparc64_vpte_patchme1 - .globl sparc64_vpte_patchme2 -/* - * On a second level vpte miss, check whether the original fault is to the OBP - * range (note that this is only possible for instruction miss, data misses to - * obp range do not use vpte). If so, go back directly to the faulting address. - * This is because we want to read the tpc, otherwise we have no way of knowing - * the 8k aligned faulting address if we are using >8k kernel pagesize. This - * also ensures no vpte range addresses are dropped into tlb while obp is - * executing (see inherit_locked_prom_mappings() rant). - */ -sparc64_vpte_nucleus: - /* Load 0xf0000000, which is LOW_OBP_ADDRESS. */ - mov 0xf, %g5 - sllx %g5, 28, %g5 - - /* Is addr >= LOW_OBP_ADDRESS? */ - cmp %g4, %g5 - blu,pn %xcc, sparc64_vpte_patchme1 - mov 0x1, %g5 - - /* Load 0x100000000, which is HI_OBP_ADDRESS. */ - sllx %g5, 32, %g5 - - /* Is addr < HI_OBP_ADDRESS? */ - cmp %g4, %g5 - blu,pn %xcc, obp_iaddr_patch - nop - - /* These two instructions are patched by paginig_init(). */ -sparc64_vpte_patchme1: - sethi %hi(0), %g5 -sparc64_vpte_patchme2: - or %g5, %lo(0), %g5 - - /* With kernel PGD in %g5, branch back into dtlb_backend. */ - ba,pt %xcc, sparc64_kpte_continue - andn %g1, 0x3, %g1 /* Finish PMD offset adjustment. */ - -vpte_noent: - /* Restore previous TAG_ACCESS, %g5 is zero, and we will - * skip over the trap instruction so that the top level - * TLB miss handler will thing this %g5 value is just an - * invalid PTE, thus branching to full fault processing. - */ - mov TLB_SFSR, %g1 - stxa %g4, [%g1 + %g1] ASI_DMMU - done - - .globl obp_iaddr_patch -obp_iaddr_patch: - /* These two instructions patched by inherit_prom_mappings(). */ - sethi %hi(0), %g5 - or %g5, %lo(0), %g5 - - /* Behave as if we are at TL0. */ - wrpr %g0, 1, %tl - rdpr %tpc, %g4 /* Find original faulting iaddr */ - srlx %g4, 13, %g4 /* Throw out context bits */ - sllx %g4, 13, %g4 /* g4 has vpn + ctx0 now */ - - /* Restore previous TAG_ACCESS. */ - mov TLB_SFSR, %g1 - stxa %g4, [%g1 + %g1] ASI_IMMU - - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop - - /* TLB load and return from trap. */ - stxa %g5, [%g0] ASI_ITLB_DATA_IN - retry - - .globl obp_daddr_patch -obp_daddr_patch: - /* These two instructions patched by inherit_prom_mappings(). */ - sethi %hi(0), %g5 - or %g5, %lo(0), %g5 - - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop - - /* TLB load and return from trap. */ - stxa %g5, [%g0] ASI_DTLB_DATA_IN - retry - -/* - * On a first level data miss, check whether this is to the OBP range (note - * that such accesses can be made by prom, as well as by kernel using - * prom_getproperty on "address"), and if so, do not use vpte access ... - * rather, use information saved during inherit_prom_mappings() using 8k - * pagesize. - */ -kvmap: - /* Load 0xf0000000, which is LOW_OBP_ADDRESS. */ - mov 0xf, %g5 - sllx %g5, 28, %g5 - - /* Is addr >= LOW_OBP_ADDRESS? */ - cmp %g4, %g5 - blu,pn %xcc, vmalloc_addr - mov 0x1, %g5 - - /* Load 0x100000000, which is HI_OBP_ADDRESS. */ - sllx %g5, 32, %g5 - - /* Is addr < HI_OBP_ADDRESS? */ - cmp %g4, %g5 - blu,pn %xcc, obp_daddr_patch - nop - -vmalloc_addr: - /* If we get here, a vmalloc addr accessed, load kernel VPTE. */ - ldxa [%g3 + %g6] ASI_N, %g5 - brgez,pn %g5, longpath - nop - - /* PTE is valid, load into TLB and return from trap. */ - stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB - retry - /* This is trivial with the new code... */ .globl do_fpdis do_fpdis: - sethi %hi(TSTATE_PEF), %g4 ! IEU0 + sethi %hi(TSTATE_PEF), %g4 rdpr %tstate, %g5 andcc %g5, %g4, %g0 be,pt %xcc, 1f @@ -203,18 +50,18 @@ do_fpdis: add %g0, %g0, %g0 ba,a,pt %xcc, rtrap_clr_l6 -1: ldub [%g6 + TI_FPSAVED], %g5 ! Load Group - wr %g0, FPRS_FEF, %fprs ! LSU Group+4bubbles - andcc %g5, FPRS_FEF, %g0 ! IEU1 Group - be,a,pt %icc, 1f ! CTI - clr %g7 ! IEU0 - ldx [%g6 + TI_GSR], %g7 ! Load Group -1: andcc %g5, FPRS_DL, %g0 ! IEU1 - bne,pn %icc, 2f ! CTI - fzero %f0 ! FPA - andcc %g5, FPRS_DU, %g0 ! IEU1 Group - bne,pn %icc, 1f ! CTI - fzero %f2 ! FPA +1: ldub [%g6 + TI_FPSAVED], %g5 + wr %g0, FPRS_FEF, %fprs + andcc %g5, FPRS_FEF, %g0 + be,a,pt %icc, 1f + clr %g7 + ldx [%g6 + TI_GSR], %g7 +1: andcc %g5, FPRS_DL, %g0 + bne,pn %icc, 2f + fzero %f0 + andcc %g5, FPRS_DU, %g0 + bne,pn %icc, 1f + fzero %f2 faddd %f0, %f2, %f4 fmuld %f0, %f2, %f6 faddd %f0, %f2, %f8 @@ -250,15 +97,17 @@ do_fpdis: faddd %f0, %f2, %f4 fmuld %f0, %f2, %f6 ldxa [%g3] ASI_DMMU, %g5 -cplus_fptrap_insn_1: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync add %g6, TI_FPREGS + 0xc0, %g2 faddd %f0, %f2, %f8 fmuld %f0, %f2, %f10 - ldda [%g1] ASI_BLK_S, %f32 ! grrr, where is ASI_BLK_NUCLEUS 8-( + membar #Sync + ldda [%g1] ASI_BLK_S, %f32 ldda [%g2] ASI_BLK_S, %f48 + membar #Sync faddd %f0, %f2, %f12 fmuld %f0, %f2, %f14 faddd %f0, %f2, %f16 @@ -269,7 +118,6 @@ cplus_fptrap_insn_1: fmuld %f0, %f2, %f26 faddd %f0, %f2, %f28 fmuld %f0, %f2, %f30 - membar #Sync b,pt %xcc, fpdis_exit nop 2: andcc %g5, FPRS_DU, %g0 @@ -279,15 +127,17 @@ cplus_fptrap_insn_1: fzero %f34 ldxa [%g3] ASI_DMMU, %g5 add %g6, TI_FPREGS, %g1 -cplus_fptrap_insn_2: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync add %g6, TI_FPREGS + 0x40, %g2 faddd %f32, %f34, %f36 fmuld %f32, %f34, %f38 - ldda [%g1] ASI_BLK_S, %f0 ! grrr, where is ASI_BLK_NUCLEUS 8-( + membar #Sync + ldda [%g1] ASI_BLK_S, %f0 ldda [%g2] ASI_BLK_S, %f16 + membar #Sync faddd %f32, %f34, %f40 fmuld %f32, %f34, %f42 faddd %f32, %f34, %f44 @@ -300,18 +150,18 @@ cplus_fptrap_insn_2: fmuld %f32, %f34, %f58 faddd %f32, %f34, %f60 fmuld %f32, %f34, %f62 - membar #Sync ba,pt %xcc, fpdis_exit nop 3: mov SECONDARY_CONTEXT, %g3 add %g6, TI_FPREGS, %g1 ldxa [%g3] ASI_DMMU, %g5 -cplus_fptrap_insn_3: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync mov 0x40, %g2 - ldda [%g1] ASI_BLK_S, %f0 ! grrr, where is ASI_BLK_NUCLEUS 8-( + membar #Sync + ldda [%g1] ASI_BLK_S, %f0 ldda [%g1 + %g2] ASI_BLK_S, %f16 add %g1, 0x80, %g1 ldda [%g1] ASI_BLK_S, %f32 @@ -472,8 +322,8 @@ do_fptrap_after_fsr: stx %g3, [%g6 + TI_GSR] mov SECONDARY_CONTEXT, %g3 ldxa [%g3] ASI_DMMU, %g5 -cplus_fptrap_insn_4: - sethi %hi(0), %g2 + sethi %hi(sparc64_kern_sec_context), %g2 + ldx [%g2 + %lo(sparc64_kern_sec_context)], %g2 stxa %g2, [%g3] ASI_DMMU membar #Sync add %g6, TI_FPREGS, %g2 @@ -494,45 +344,17 @@ cplus_fptrap_insn_4: ba,pt %xcc, etrap wr %g0, 0, %fprs -cplus_fptrap_1: - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - - .globl cheetah_plus_patch_fpdis -cheetah_plus_patch_fpdis: - /* We configure the dTLB512_0 for 4MB pages and the - * dTLB512_1 for 8K pages when in context zero. - */ - sethi %hi(cplus_fptrap_1), %o0 - lduw [%o0 + %lo(cplus_fptrap_1)], %o1 - - set cplus_fptrap_insn_1, %o2 - stw %o1, [%o2] - flush %o2 - set cplus_fptrap_insn_2, %o2 - stw %o1, [%o2] - flush %o2 - set cplus_fptrap_insn_3, %o2 - stw %o1, [%o2] - flush %o2 - set cplus_fptrap_insn_4, %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop - /* The registers for cross calls will be: * * DATA 0: [low 32-bits] Address of function to call, jmp to this * [high 32-bits] MMU Context Argument 0, place in %g5 - * DATA 1: Address Argument 1, place in %g6 + * DATA 1: Address Argument 1, place in %g1 * DATA 2: Address Argument 2, place in %g7 * * With this method we can do most of the cross-call tlb/cache * flushing very quickly. * - * Current CPU's IRQ worklist table is locked into %g1, - * don't touch. + * Current CPU's IRQ worklist table is locked into %g6, don't touch. */ .text .align 32 @@ -690,14 +512,159 @@ netbsd_syscall: retl nop - /* These next few routines must be sure to clear the - * SFSR FaultValid bit so that the fast tlb data protection - * handler does not flush the wrong context and lock up the - * box. + /* We need to carefully read the error status, ACK + * the errors, prevent recursive traps, and pass the + * information on to C code for logging. + * + * We pass the AFAR in as-is, and we encode the status + * information as described in asm-sparc64/sfafsr.h + */ + .globl __spitfire_access_error +__spitfire_access_error: + /* Disable ESTATE error reporting so that we do not + * take recursive traps and RED state the processor. + */ + stxa %g0, [%g0] ASI_ESTATE_ERROR_EN + membar #Sync + + mov UDBE_UE, %g1 + ldxa [%g0] ASI_AFSR, %g4 ! Get AFSR + + /* __spitfire_cee_trap branches here with AFSR in %g4 and + * UDBE_CE in %g1. It only clears ESTATE_ERR_CE in the + * ESTATE Error Enable register. + */ +__spitfire_cee_trap_continue: + ldxa [%g0] ASI_AFAR, %g5 ! Get AFAR + + rdpr %tt, %g3 + and %g3, 0x1ff, %g3 ! Paranoia + sllx %g3, SFSTAT_TRAP_TYPE_SHIFT, %g3 + or %g4, %g3, %g4 + rdpr %tl, %g3 + cmp %g3, 1 + mov 1, %g3 + bleu %xcc, 1f + sllx %g3, SFSTAT_TL_GT_ONE_SHIFT, %g3 + + or %g4, %g3, %g4 + + /* Read in the UDB error register state, clearing the + * sticky error bits as-needed. We only clear them if + * the UE bit is set. Likewise, __spitfire_cee_trap + * below will only do so if the CE bit is set. + * + * NOTE: UltraSparc-I/II have high and low UDB error + * registers, corresponding to the two UDB units + * present on those chips. UltraSparc-IIi only + * has a single UDB, called "SDB" in the manual. + * For IIi the upper UDB register always reads + * as zero so for our purposes things will just + * work with the checks below. + */ +1: ldxa [%g0] ASI_UDBH_ERROR_R, %g3 + and %g3, 0x3ff, %g7 ! Paranoia + sllx %g7, SFSTAT_UDBH_SHIFT, %g7 + or %g4, %g7, %g4 + andcc %g3, %g1, %g3 ! UDBE_UE or UDBE_CE + be,pn %xcc, 1f + nop + stxa %g3, [%g0] ASI_UDB_ERROR_W + membar #Sync + +1: mov 0x18, %g3 + ldxa [%g3] ASI_UDBL_ERROR_R, %g3 + and %g3, 0x3ff, %g7 ! Paranoia + sllx %g7, SFSTAT_UDBL_SHIFT, %g7 + or %g4, %g7, %g4 + andcc %g3, %g1, %g3 ! UDBE_UE or UDBE_CE + be,pn %xcc, 1f + nop + mov 0x18, %g7 + stxa %g3, [%g7] ASI_UDB_ERROR_W + membar #Sync + +1: /* Ok, now that we've latched the error state, + * clear the sticky bits in the AFSR. */ - .globl __do_data_access_exception - .globl __do_data_access_exception_tl1 -__do_data_access_exception_tl1: + stxa %g4, [%g0] ASI_AFSR + membar #Sync + + rdpr %tl, %g2 + cmp %g2, 1 + rdpr %pil, %g2 + bleu,pt %xcc, 1f + wrpr %g0, 15, %pil + + ba,pt %xcc, etraptl1 + rd %pc, %g7 + + ba,pt %xcc, 2f + nop + +1: ba,pt %xcc, etrap_irq + rd %pc, %g7 + +2: mov %l4, %o1 + mov %l5, %o2 + call spitfire_access_error + add %sp, PTREGS_OFF, %o0 + ba,pt %xcc, rtrap + clr %l6 + + /* This is the trap handler entry point for ECC correctable + * errors. They are corrected, but we listen for the trap + * so that the event can be logged. + * + * Disrupting errors are either: + * 1) single-bit ECC errors during UDB reads to system + * memory + * 2) data parity errors during write-back events + * + * As far as I can make out from the manual, the CEE trap + * is only for correctable errors during memory read + * accesses by the front-end of the processor. + * + * The code below is only for trap level 1 CEE events, + * as it is the only situation where we can safely record + * and log. For trap level >1 we just clear the CE bit + * in the AFSR and return. + * + * This is just like __spiftire_access_error above, but it + * specifically handles correctable errors. If an + * uncorrectable error is indicated in the AFSR we + * will branch directly above to __spitfire_access_error + * to handle it instead. Uncorrectable therefore takes + * priority over correctable, and the error logging + * C code will notice this case by inspecting the + * trap type. + */ + .globl __spitfire_cee_trap +__spitfire_cee_trap: + ldxa [%g0] ASI_AFSR, %g4 ! Get AFSR + mov 1, %g3 + sllx %g3, SFAFSR_UE_SHIFT, %g3 + andcc %g4, %g3, %g0 ! Check for UE + bne,pn %xcc, __spitfire_access_error + nop + + /* Ok, in this case we only have a correctable error. + * Indicate we only wish to capture that state in register + * %g1, and we only disable CE error reporting unlike UE + * handling which disables all errors. + */ + ldxa [%g0] ASI_ESTATE_ERROR_EN, %g3 + andn %g3, ESTATE_ERR_CE, %g3 + stxa %g3, [%g0] ASI_ESTATE_ERROR_EN + membar #Sync + + /* Preserve AFSR in %g4, indicate UDB state to capture in %g1 */ + ba,pt %xcc, __spitfire_cee_trap_continue + mov UDBE_CE, %g1 + + .globl __spitfire_data_access_exception + .globl __spitfire_data_access_exception_tl1 +__spitfire_data_access_exception_tl1: rdpr %pstate, %g4 wrpr %g4, PSTATE_MG|PSTATE_AG, %pstate mov TLB_SFSR, %g3 @@ -706,9 +673,25 @@ __do_data_access_exception_tl1: ldxa [%g5] ASI_DMMU, %g5 ! Get SFAR stxa %g0, [%g3] ASI_DMMU ! Clear SFSR.FaultValid bit membar #Sync + rdpr %tt, %g3 + cmp %g3, 0x80 ! first win spill/fill trap + blu,pn %xcc, 1f + cmp %g3, 0xff ! last win spill/fill trap + bgu,pn %xcc, 1f + nop ba,pt %xcc, winfix_dax rdpr %tpc, %g3 -__do_data_access_exception: +1: sethi %hi(109f), %g7 + ba,pt %xcc, etraptl1 +109: or %g7, %lo(109b), %g7 + mov %l4, %o1 + mov %l5, %o2 + call spitfire_data_access_exception_tl1 + add %sp, PTREGS_OFF, %o0 + ba,pt %xcc, rtrap + clr %l6 + +__spitfire_data_access_exception: rdpr %pstate, %g4 wrpr %g4, PSTATE_MG|PSTATE_AG, %pstate mov TLB_SFSR, %g3 @@ -722,20 +705,19 @@ __do_data_access_exception: 109: or %g7, %lo(109b), %g7 mov %l4, %o1 mov %l5, %o2 - call data_access_exception + call spitfire_data_access_exception add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap clr %l6 - .globl __do_instruction_access_exception - .globl __do_instruction_access_exception_tl1 -__do_instruction_access_exception_tl1: + .globl __spitfire_insn_access_exception + .globl __spitfire_insn_access_exception_tl1 +__spitfire_insn_access_exception_tl1: rdpr %pstate, %g4 wrpr %g4, PSTATE_MG|PSTATE_AG, %pstate mov TLB_SFSR, %g3 - mov DMMU_SFAR, %g5 - ldxa [%g3] ASI_DMMU, %g4 ! Get SFSR - ldxa [%g5] ASI_DMMU, %g5 ! Get SFAR + ldxa [%g3] ASI_IMMU, %g4 ! Get SFSR + rdpr %tpc, %g5 ! IMMU has no SFAR, use TPC stxa %g0, [%g3] ASI_IMMU ! Clear FaultValid bit membar #Sync sethi %hi(109f), %g7 @@ -743,18 +725,17 @@ __do_instruction_access_exception_tl1: 109: or %g7, %lo(109b), %g7 mov %l4, %o1 mov %l5, %o2 - call instruction_access_exception_tl1 + call spitfire_insn_access_exception_tl1 add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap clr %l6 -__do_instruction_access_exception: +__spitfire_insn_access_exception: rdpr %pstate, %g4 wrpr %g4, PSTATE_MG|PSTATE_AG, %pstate mov TLB_SFSR, %g3 - mov DMMU_SFAR, %g5 - ldxa [%g3] ASI_DMMU, %g4 ! Get SFSR - ldxa [%g5] ASI_DMMU, %g5 ! Get SFAR + ldxa [%g3] ASI_IMMU, %g4 ! Get SFSR + rdpr %tpc, %g5 ! IMMU has no SFAR, use TPC stxa %g0, [%g3] ASI_IMMU ! Clear FaultValid bit membar #Sync sethi %hi(109f), %g7 @@ -762,235 +743,11 @@ __do_instruction_access_exception: 109: or %g7, %lo(109b), %g7 mov %l4, %o1 mov %l5, %o2 - call instruction_access_exception + call spitfire_insn_access_exception add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap clr %l6 - /* This is the trap handler entry point for ECC correctable - * errors. They are corrected, but we listen for the trap - * so that the event can be logged. - * - * Disrupting errors are either: - * 1) single-bit ECC errors during UDB reads to system - * memory - * 2) data parity errors during write-back events - * - * As far as I can make out from the manual, the CEE trap - * is only for correctable errors during memory read - * accesses by the front-end of the processor. - * - * The code below is only for trap level 1 CEE events, - * as it is the only situation where we can safely record - * and log. For trap level >1 we just clear the CE bit - * in the AFSR and return. - */ - - /* Our trap handling infrastructure allows us to preserve - * two 64-bit values during etrap for arguments to - * subsequent C code. Therefore we encode the information - * as follows: - * - * value 1) Full 64-bits of AFAR - * value 2) Low 33-bits of AFSR, then bits 33-->42 - * are UDBL error status and bits 43-->52 - * are UDBH error status - */ - .align 64 - .globl cee_trap -cee_trap: - ldxa [%g0] ASI_AFSR, %g1 ! Read AFSR - ldxa [%g0] ASI_AFAR, %g2 ! Read AFAR - sllx %g1, 31, %g1 ! Clear reserved bits - srlx %g1, 31, %g1 ! in AFSR - - /* NOTE: UltraSparc-I/II have high and low UDB error - * registers, corresponding to the two UDB units - * present on those chips. UltraSparc-IIi only - * has a single UDB, called "SDB" in the manual. - * For IIi the upper UDB register always reads - * as zero so for our purposes things will just - * work with the checks below. - */ - ldxa [%g0] ASI_UDBL_ERROR_R, %g3 ! Read UDB-Low error status - andcc %g3, (1 << 8), %g4 ! Check CE bit - sllx %g3, (64 - 10), %g3 ! Clear reserved bits - srlx %g3, (64 - 10), %g3 ! in UDB-Low error status - - sllx %g3, (33 + 0), %g3 ! Shift up to encoding area - or %g1, %g3, %g1 ! Or it in - be,pn %xcc, 1f ! Branch if CE bit was clear - nop - stxa %g4, [%g0] ASI_UDB_ERROR_W ! Clear CE sticky bit in UDBL - membar #Sync ! Synchronize ASI stores -1: mov 0x18, %g5 ! Addr of UDB-High error status - ldxa [%g5] ASI_UDBH_ERROR_R, %g3 ! Read it - - andcc %g3, (1 << 8), %g4 ! Check CE bit - sllx %g3, (64 - 10), %g3 ! Clear reserved bits - srlx %g3, (64 - 10), %g3 ! in UDB-High error status - sllx %g3, (33 + 10), %g3 ! Shift up to encoding area - or %g1, %g3, %g1 ! Or it in - be,pn %xcc, 1f ! Branch if CE bit was clear - nop - nop - - stxa %g4, [%g5] ASI_UDB_ERROR_W ! Clear CE sticky bit in UDBH - membar #Sync ! Synchronize ASI stores -1: mov 1, %g5 ! AFSR CE bit is - sllx %g5, 20, %g5 ! bit 20 - stxa %g5, [%g0] ASI_AFSR ! Clear CE sticky bit in AFSR - membar #Sync ! Synchronize ASI stores - sllx %g2, (64 - 41), %g2 ! Clear reserved bits - srlx %g2, (64 - 41), %g2 ! in latched AFAR - - andn %g2, 0x0f, %g2 ! Finish resv bit clearing - mov %g1, %g4 ! Move AFSR+UDB* into save reg - mov %g2, %g5 ! Move AFAR into save reg - rdpr %pil, %g2 - wrpr %g0, 15, %pil - ba,pt %xcc, etrap_irq - rd %pc, %g7 - mov %l4, %o0 - - mov %l5, %o1 - call cee_log - add %sp, PTREGS_OFF, %o2 - ba,a,pt %xcc, rtrap_irq - - /* Capture I/D/E-cache state into per-cpu error scoreboard. - * - * %g1: (TL>=0) ? 1 : 0 - * %g2: scratch - * %g3: scratch - * %g4: AFSR - * %g5: AFAR - * %g6: current thread ptr - * %g7: scratch - */ -#define CHEETAH_LOG_ERROR \ - /* Put "TL1" software bit into AFSR. */ \ - and %g1, 0x1, %g1; \ - sllx %g1, 63, %g2; \ - or %g4, %g2, %g4; \ - /* Get log entry pointer for this cpu at this trap level. */ \ - BRANCH_IF_JALAPENO(g2,g3,50f) \ - ldxa [%g0] ASI_SAFARI_CONFIG, %g2; \ - srlx %g2, 17, %g2; \ - ba,pt %xcc, 60f; \ - and %g2, 0x3ff, %g2; \ -50: ldxa [%g0] ASI_JBUS_CONFIG, %g2; \ - srlx %g2, 17, %g2; \ - and %g2, 0x1f, %g2; \ -60: sllx %g2, 9, %g2; \ - sethi %hi(cheetah_error_log), %g3; \ - ldx [%g3 + %lo(cheetah_error_log)], %g3; \ - brz,pn %g3, 80f; \ - nop; \ - add %g3, %g2, %g3; \ - sllx %g1, 8, %g1; \ - add %g3, %g1, %g1; \ - /* %g1 holds pointer to the top of the logging scoreboard */ \ - ldx [%g1 + 0x0], %g7; \ - cmp %g7, -1; \ - bne,pn %xcc, 80f; \ - nop; \ - stx %g4, [%g1 + 0x0]; \ - stx %g5, [%g1 + 0x8]; \ - add %g1, 0x10, %g1; \ - /* %g1 now points to D-cache logging area */ \ - set 0x3ff8, %g2; /* DC_addr mask */ \ - and %g5, %g2, %g2; /* DC_addr bits of AFAR */ \ - srlx %g5, 12, %g3; \ - or %g3, 1, %g3; /* PHYS tag + valid */ \ -10: ldxa [%g2] ASI_DCACHE_TAG, %g7; \ - cmp %g3, %g7; /* TAG match? */ \ - bne,pt %xcc, 13f; \ - nop; \ - /* Yep, what we want, capture state. */ \ - stx %g2, [%g1 + 0x20]; \ - stx %g7, [%g1 + 0x28]; \ - /* A membar Sync is required before and after utag access. */ \ - membar #Sync; \ - ldxa [%g2] ASI_DCACHE_UTAG, %g7; \ - membar #Sync; \ - stx %g7, [%g1 + 0x30]; \ - ldxa [%g2] ASI_DCACHE_SNOOP_TAG, %g7; \ - stx %g7, [%g1 + 0x38]; \ - clr %g3; \ -12: ldxa [%g2 + %g3] ASI_DCACHE_DATA, %g7; \ - stx %g7, [%g1]; \ - add %g3, (1 << 5), %g3; \ - cmp %g3, (4 << 5); \ - bl,pt %xcc, 12b; \ - add %g1, 0x8, %g1; \ - ba,pt %xcc, 20f; \ - add %g1, 0x20, %g1; \ -13: sethi %hi(1 << 14), %g7; \ - add %g2, %g7, %g2; \ - srlx %g2, 14, %g7; \ - cmp %g7, 4; \ - bl,pt %xcc, 10b; \ - nop; \ - add %g1, 0x40, %g1; \ -20: /* %g1 now points to I-cache logging area */ \ - set 0x1fe0, %g2; /* IC_addr mask */ \ - and %g5, %g2, %g2; /* IC_addr bits of AFAR */ \ - sllx %g2, 1, %g2; /* IC_addr[13:6]==VA[12:5] */ \ - srlx %g5, (13 - 8), %g3; /* Make PTAG */ \ - andn %g3, 0xff, %g3; /* Mask off undefined bits */ \ -21: ldxa [%g2] ASI_IC_TAG, %g7; \ - andn %g7, 0xff, %g7; \ - cmp %g3, %g7; \ - bne,pt %xcc, 23f; \ - nop; \ - /* Yep, what we want, capture state. */ \ - stx %g2, [%g1 + 0x40]; \ - stx %g7, [%g1 + 0x48]; \ - add %g2, (1 << 3), %g2; \ - ldxa [%g2] ASI_IC_TAG, %g7; \ - add %g2, (1 << 3), %g2; \ - stx %g7, [%g1 + 0x50]; \ - ldxa [%g2] ASI_IC_TAG, %g7; \ - add %g2, (1 << 3), %g2; \ - stx %g7, [%g1 + 0x60]; \ - ldxa [%g2] ASI_IC_TAG, %g7; \ - stx %g7, [%g1 + 0x68]; \ - sub %g2, (3 << 3), %g2; \ - ldxa [%g2] ASI_IC_STAG, %g7; \ - stx %g7, [%g1 + 0x58]; \ - clr %g3; \ - srlx %g2, 2, %g2; \ -22: ldxa [%g2 + %g3] ASI_IC_INSTR, %g7; \ - stx %g7, [%g1]; \ - add %g3, (1 << 3), %g3; \ - cmp %g3, (8 << 3); \ - bl,pt %xcc, 22b; \ - add %g1, 0x8, %g1; \ - ba,pt %xcc, 30f; \ - add %g1, 0x30, %g1; \ -23: sethi %hi(1 << 14), %g7; \ - add %g2, %g7, %g2; \ - srlx %g2, 14, %g7; \ - cmp %g7, 4; \ - bl,pt %xcc, 21b; \ - nop; \ - add %g1, 0x70, %g1; \ -30: /* %g1 now points to E-cache logging area */ \ - andn %g5, (32 - 1), %g2; /* E-cache subblock */ \ - stx %g2, [%g1 + 0x20]; \ - ldxa [%g2] ASI_EC_TAG_DATA, %g7; \ - stx %g7, [%g1 + 0x28]; \ - ldxa [%g2] ASI_EC_R, %g0; \ - clr %g3; \ -31: ldxa [%g3] ASI_EC_DATA, %g7; \ - stx %g7, [%g1 + %g3]; \ - add %g3, 0x8, %g3; \ - cmp %g3, 0x20; \ - bl,pt %xcc, 31b; \ - nop; \ -80: /* DONE */ - /* These get patched into the trap table at boot time * once we know we have a cheetah processor. */ @@ -1071,13 +828,14 @@ cheetah_plus_dcpe_trap_vector: nop do_cheetah_plus_data_parity: - ba,pt %xcc, etrap + rdpr %pil, %g2 + wrpr %g0, 15, %pil + ba,pt %xcc, etrap_irq rd %pc, %g7 mov 0x0, %o0 call cheetah_plus_parity_error add %sp, PTREGS_OFF, %o1 - ba,pt %xcc, rtrap - clr %l6 + ba,a,pt %xcc, rtrap_irq cheetah_plus_dcpe_trap_vector_tl1: membar #Sync @@ -1101,13 +859,14 @@ cheetah_plus_icpe_trap_vector: nop do_cheetah_plus_insn_parity: - ba,pt %xcc, etrap + rdpr %pil, %g2 + wrpr %g0, 15, %pil + ba,pt %xcc, etrap_irq rd %pc, %g7 mov 0x1, %o0 call cheetah_plus_parity_error add %sp, PTREGS_OFF, %o1 - ba,pt %xcc, rtrap - clr %l6 + ba,a,pt %xcc, rtrap_irq cheetah_plus_icpe_trap_vector_tl1: membar #Sync @@ -1140,6 +899,10 @@ do_dcpe_tl1: nop wrpr %g1, %tl ! Restore original trap level do_dcpe_tl1_nonfatal: /* Ok we may use interrupt globals safely. */ + sethi %hi(dcache_parity_tl1_occurred), %g2 + lduw [%g2 + %lo(dcache_parity_tl1_occurred)], %g1 + add %g1, 1, %g1 + stw %g1, [%g2 + %lo(dcache_parity_tl1_occurred)] /* Reset D-cache parity */ sethi %hi(1 << 16), %g1 ! D-cache size mov (1 << 5), %g2 ! D-cache line size @@ -1186,6 +949,10 @@ do_icpe_tl1: nop wrpr %g1, %tl ! Restore original trap level do_icpe_tl1_nonfatal: /* Ok we may use interrupt globals safely. */ + sethi %hi(icache_parity_tl1_occurred), %g2 + lduw [%g2 + %lo(icache_parity_tl1_occurred)], %g1 + add %g1, 1, %g1 + stw %g1, [%g2 + %lo(icache_parity_tl1_occurred)] /* Flush I-cache */ sethi %hi(1 << 15), %g1 ! I-cache size mov (1 << 5), %g2 ! I-cache line size @@ -1227,6 +994,170 @@ dcpe_icpe_tl1_common: membar #Sync retry + /* Capture I/D/E-cache state into per-cpu error scoreboard. + * + * %g1: (TL>=0) ? 1 : 0 + * %g2: scratch + * %g3: scratch + * %g4: AFSR + * %g5: AFAR + * %g6: current thread ptr + * %g7: scratch + */ +__cheetah_log_error: + /* Put "TL1" software bit into AFSR. */ + and %g1, 0x1, %g1 + sllx %g1, 63, %g2 + or %g4, %g2, %g4 + + /* Get log entry pointer for this cpu at this trap level. */ + BRANCH_IF_JALAPENO(g2,g3,50f) + ldxa [%g0] ASI_SAFARI_CONFIG, %g2 + srlx %g2, 17, %g2 + ba,pt %xcc, 60f + and %g2, 0x3ff, %g2 + +50: ldxa [%g0] ASI_JBUS_CONFIG, %g2 + srlx %g2, 17, %g2 + and %g2, 0x1f, %g2 + +60: sllx %g2, 9, %g2 + sethi %hi(cheetah_error_log), %g3 + ldx [%g3 + %lo(cheetah_error_log)], %g3 + brz,pn %g3, 80f + nop + + add %g3, %g2, %g3 + sllx %g1, 8, %g1 + add %g3, %g1, %g1 + + /* %g1 holds pointer to the top of the logging scoreboard */ + ldx [%g1 + 0x0], %g7 + cmp %g7, -1 + bne,pn %xcc, 80f + nop + + stx %g4, [%g1 + 0x0] + stx %g5, [%g1 + 0x8] + add %g1, 0x10, %g1 + + /* %g1 now points to D-cache logging area */ + set 0x3ff8, %g2 /* DC_addr mask */ + and %g5, %g2, %g2 /* DC_addr bits of AFAR */ + srlx %g5, 12, %g3 + or %g3, 1, %g3 /* PHYS tag + valid */ + +10: ldxa [%g2] ASI_DCACHE_TAG, %g7 + cmp %g3, %g7 /* TAG match? */ + bne,pt %xcc, 13f + nop + + /* Yep, what we want, capture state. */ + stx %g2, [%g1 + 0x20] + stx %g7, [%g1 + 0x28] + + /* A membar Sync is required before and after utag access. */ + membar #Sync + ldxa [%g2] ASI_DCACHE_UTAG, %g7 + membar #Sync + stx %g7, [%g1 + 0x30] + ldxa [%g2] ASI_DCACHE_SNOOP_TAG, %g7 + stx %g7, [%g1 + 0x38] + clr %g3 + +12: ldxa [%g2 + %g3] ASI_DCACHE_DATA, %g7 + stx %g7, [%g1] + add %g3, (1 << 5), %g3 + cmp %g3, (4 << 5) + bl,pt %xcc, 12b + add %g1, 0x8, %g1 + + ba,pt %xcc, 20f + add %g1, 0x20, %g1 + +13: sethi %hi(1 << 14), %g7 + add %g2, %g7, %g2 + srlx %g2, 14, %g7 + cmp %g7, 4 + bl,pt %xcc, 10b + nop + + add %g1, 0x40, %g1 + + /* %g1 now points to I-cache logging area */ +20: set 0x1fe0, %g2 /* IC_addr mask */ + and %g5, %g2, %g2 /* IC_addr bits of AFAR */ + sllx %g2, 1, %g2 /* IC_addr[13:6]==VA[12:5] */ + srlx %g5, (13 - 8), %g3 /* Make PTAG */ + andn %g3, 0xff, %g3 /* Mask off undefined bits */ + +21: ldxa [%g2] ASI_IC_TAG, %g7 + andn %g7, 0xff, %g7 + cmp %g3, %g7 + bne,pt %xcc, 23f + nop + + /* Yep, what we want, capture state. */ + stx %g2, [%g1 + 0x40] + stx %g7, [%g1 + 0x48] + add %g2, (1 << 3), %g2 + ldxa [%g2] ASI_IC_TAG, %g7 + add %g2, (1 << 3), %g2 + stx %g7, [%g1 + 0x50] + ldxa [%g2] ASI_IC_TAG, %g7 + add %g2, (1 << 3), %g2 + stx %g7, [%g1 + 0x60] + ldxa [%g2] ASI_IC_TAG, %g7 + stx %g7, [%g1 + 0x68] + sub %g2, (3 << 3), %g2 + ldxa [%g2] ASI_IC_STAG, %g7 + stx %g7, [%g1 + 0x58] + clr %g3 + srlx %g2, 2, %g2 + +22: ldxa [%g2 + %g3] ASI_IC_INSTR, %g7 + stx %g7, [%g1] + add %g3, (1 << 3), %g3 + cmp %g3, (8 << 3) + bl,pt %xcc, 22b + add %g1, 0x8, %g1 + + ba,pt %xcc, 30f + add %g1, 0x30, %g1 + +23: sethi %hi(1 << 14), %g7 + add %g2, %g7, %g2 + srlx %g2, 14, %g7 + cmp %g7, 4 + bl,pt %xcc, 21b + nop + + add %g1, 0x70, %g1 + + /* %g1 now points to E-cache logging area */ +30: andn %g5, (32 - 1), %g2 + stx %g2, [%g1 + 0x20] + ldxa [%g2] ASI_EC_TAG_DATA, %g7 + stx %g7, [%g1 + 0x28] + ldxa [%g2] ASI_EC_R, %g0 + clr %g3 + +31: ldxa [%g3] ASI_EC_DATA, %g7 + stx %g7, [%g1 + %g3] + add %g3, 0x8, %g3 + cmp %g3, 0x20 + + bl,pt %xcc, 31b + nop +80: + rdpr %tt, %g2 + cmp %g2, 0x70 + be c_fast_ecc + cmp %g2, 0x63 + be c_cee + nop + ba,pt %xcc, c_deferred + /* Cheetah FECC trap handling, we get here from tl{0,1}_fecc * in the trap table. That code has done a memory barrier * and has disabled both the I-cache and D-cache in the DCU @@ -1252,8 +1183,10 @@ cheetah_fast_ecc: stxa %g4, [%g0] ASI_AFSR membar #Sync - CHEETAH_LOG_ERROR + ba,pt %xcc, __cheetah_log_error + nop +c_fast_ecc: rdpr %pil, %g2 wrpr %g0, 15, %pil ba,pt %xcc, etrap_irq @@ -1278,8 +1211,10 @@ cheetah_cee: stxa %g4, [%g0] ASI_AFSR membar #Sync - CHEETAH_LOG_ERROR + ba,pt %xcc, __cheetah_log_error + nop +c_cee: rdpr %pil, %g2 wrpr %g0, 15, %pil ba,pt %xcc, etrap_irq @@ -1304,8 +1239,10 @@ cheetah_deferred_trap: stxa %g4, [%g0] ASI_AFSR membar #Sync - CHEETAH_LOG_ERROR + ba,pt %xcc, __cheetah_log_error + nop +c_deferred: rdpr %pil, %g2 wrpr %g0, 15, %pil ba,pt %xcc, etrap_irq diff --git a/arch/sparc64/kernel/etrap.S b/arch/sparc64/kernel/etrap.S index 50d2af1d98ae..0d8eba21111b 100644 --- a/arch/sparc64/kernel/etrap.S +++ b/arch/sparc64/kernel/etrap.S @@ -68,12 +68,8 @@ etrap_irq: wrpr %g3, 0, %otherwin wrpr %g2, 0, %wstate -cplus_etrap_insn_1: - sethi %hi(0), %g3 - sllx %g3, 32, %g3 -cplus_etrap_insn_2: - sethi %hi(0), %g2 - or %g3, %g2, %g3 + sethi %hi(sparc64_kern_pri_context), %g2 + ldx [%g2 + %lo(sparc64_kern_pri_context)], %g3 stxa %g3, [%l4] ASI_DMMU flush %l6 wr %g0, ASI_AIUS, %asi @@ -215,12 +211,8 @@ scetrap: rdpr %pil, %g2 mov PRIMARY_CONTEXT, %l4 wrpr %g3, 0, %otherwin wrpr %g2, 0, %wstate -cplus_etrap_insn_3: - sethi %hi(0), %g3 - sllx %g3, 32, %g3 -cplus_etrap_insn_4: - sethi %hi(0), %g2 - or %g3, %g2, %g3 + sethi %hi(sparc64_kern_pri_context), %g2 + ldx [%g2 + %lo(sparc64_kern_pri_context)], %g3 stxa %g3, [%l4] ASI_DMMU flush %l6 @@ -264,38 +256,3 @@ cplus_etrap_insn_4: #undef TASK_REGOFF #undef ETRAP_PSTATE1 - -cplus_einsn_1: - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %g3 -cplus_einsn_2: - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - - .globl cheetah_plus_patch_etrap -cheetah_plus_patch_etrap: - /* We configure the dTLB512_0 for 4MB pages and the - * dTLB512_1 for 8K pages when in context zero. - */ - sethi %hi(cplus_einsn_1), %o0 - sethi %hi(cplus_etrap_insn_1), %o2 - lduw [%o0 + %lo(cplus_einsn_1)], %o1 - or %o2, %lo(cplus_etrap_insn_1), %o2 - stw %o1, [%o2] - flush %o2 - sethi %hi(cplus_etrap_insn_3), %o2 - or %o2, %lo(cplus_etrap_insn_3), %o2 - stw %o1, [%o2] - flush %o2 - - sethi %hi(cplus_einsn_2), %o0 - sethi %hi(cplus_etrap_insn_2), %o2 - lduw [%o0 + %lo(cplus_einsn_2)], %o1 - or %o2, %lo(cplus_etrap_insn_2), %o2 - stw %o1, [%o2] - flush %o2 - sethi %hi(cplus_etrap_insn_4), %o2 - or %o2, %lo(cplus_etrap_insn_4), %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 8104a56ca2d8..b49dcd4504b0 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -28,19 +28,14 @@ #include <asm/mmu.h> /* This section from from _start to sparc64_boot_end should fit into - * 0x0000.0000.0040.4000 to 0x0000.0000.0040.8000 and will be sharing space - * with bootup_user_stack, which is from 0x0000.0000.0040.4000 to - * 0x0000.0000.0040.6000 and empty_bad_page, which is from - * 0x0000.0000.0040.6000 to 0x0000.0000.0040.8000. + * 0x0000000000404000 to 0x0000000000408000. */ - .text .globl start, _start, stext, _stext _start: start: _stext: stext: -bootup_user_stack: ! 0x0000000000404000 b sparc64_boot flushw /* Flush register file. */ @@ -80,15 +75,169 @@ sparc_ramdisk_image64: .xword 0 .word _end - /* We must be careful, 32-bit OpenBOOT will get confused if it - * tries to save away a register window to a 64-bit kernel - * stack address. Flush all windows, disable interrupts, - * remap if necessary, jump onto kernel trap table, then kernel - * stack, or else we die. + /* PROM cif handler code address is in %o4. */ +sparc64_boot: +1: rd %pc, %g7 + set 1b, %g1 + cmp %g1, %g7 + be,pn %xcc, sparc64_boot_after_remap + mov %o4, %l7 + + /* We need to remap the kernel. Use position independant + * code to remap us to KERNBASE. * - * PROM entry point is on %o4 + * SILO can invoke us with 32-bit address masking enabled, + * so make sure that's clear. */ -sparc64_boot: + rdpr %pstate, %g1 + andn %g1, PSTATE_AM, %g1 + wrpr %g1, 0x0, %pstate + ba,a,pt %xcc, 1f + + .globl prom_finddev_name, prom_chosen_path + .globl prom_getprop_name, prom_mmu_name + .globl prom_callmethod_name, prom_translate_name + .globl prom_map_name, prom_unmap_name, prom_mmu_ihandle_cache + .globl prom_boot_mapped_pc, prom_boot_mapping_mode + .globl prom_boot_mapping_phys_high, prom_boot_mapping_phys_low +prom_finddev_name: + .asciz "finddevice" +prom_chosen_path: + .asciz "/chosen" +prom_getprop_name: + .asciz "getprop" +prom_mmu_name: + .asciz "mmu" +prom_callmethod_name: + .asciz "call-method" +prom_translate_name: + .asciz "translate" +prom_map_name: + .asciz "map" +prom_unmap_name: + .asciz "unmap" + .align 4 +prom_mmu_ihandle_cache: + .word 0 +prom_boot_mapped_pc: + .word 0 +prom_boot_mapping_mode: + .word 0 + .align 8 +prom_boot_mapping_phys_high: + .xword 0 +prom_boot_mapping_phys_low: + .xword 0 +1: + rd %pc, %l0 + mov (1b - prom_finddev_name), %l1 + mov (1b - prom_chosen_path), %l2 + mov (1b - prom_boot_mapped_pc), %l3 + sub %l0, %l1, %l1 + sub %l0, %l2, %l2 + sub %l0, %l3, %l3 + stw %l0, [%l3] + sub %sp, (192 + 128), %sp + + /* chosen_node = prom_finddevice("/chosen") */ + stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "finddevice" + mov 1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 1 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1 + stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1, "/chosen" + stx %g0, [%sp + 2047 + 128 + 0x20] ! ret1 + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + ldx [%sp + 2047 + 128 + 0x20], %l4 ! chosen device node + + mov (1b - prom_getprop_name), %l1 + mov (1b - prom_mmu_name), %l2 + mov (1b - prom_mmu_ihandle_cache), %l5 + sub %l0, %l1, %l1 + sub %l0, %l2, %l2 + sub %l0, %l5, %l5 + + /* prom_mmu_ihandle_cache = prom_getint(chosen_node, "mmu") */ + stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "getprop" + mov 4, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 4 + mov 1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1 + stx %l4, [%sp + 2047 + 128 + 0x18] ! arg1, chosen_node + stx %l2, [%sp + 2047 + 128 + 0x20] ! arg2, "mmu" + stx %l5, [%sp + 2047 + 128 + 0x28] ! arg3, &prom_mmu_ihandle_cache + mov 4, %l3 + stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4, sizeof(arg3) + stx %g0, [%sp + 2047 + 128 + 0x38] ! ret1 + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + mov (1b - prom_callmethod_name), %l1 + mov (1b - prom_translate_name), %l2 + sub %l0, %l1, %l1 + sub %l0, %l2, %l2 + lduw [%l5], %l5 ! prom_mmu_ihandle_cache + + stx %l1, [%sp + 2047 + 128 + 0x00] ! service, "call-method" + mov 3, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 3 + mov 5, %l3 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 5 + stx %l2, [%sp + 2047 + 128 + 0x18] ! arg1: "translate" + stx %l5, [%sp + 2047 + 128 + 0x20] ! arg2: prom_mmu_ihandle_cache + /* PAGE align */ + srlx %l0, 13, %l3 + sllx %l3, 13, %l3 + stx %l3, [%sp + 2047 + 128 + 0x28] ! arg3: vaddr, our PC + stx %g0, [%sp + 2047 + 128 + 0x30] ! res1 + stx %g0, [%sp + 2047 + 128 + 0x38] ! res2 + stx %g0, [%sp + 2047 + 128 + 0x40] ! res3 + stx %g0, [%sp + 2047 + 128 + 0x48] ! res4 + stx %g0, [%sp + 2047 + 128 + 0x50] ! res5 + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + ldx [%sp + 2047 + 128 + 0x40], %l1 ! translation mode + mov (1b - prom_boot_mapping_mode), %l4 + sub %l0, %l4, %l4 + stw %l1, [%l4] + mov (1b - prom_boot_mapping_phys_high), %l4 + sub %l0, %l4, %l4 + ldx [%sp + 2047 + 128 + 0x48], %l2 ! physaddr high + stx %l2, [%l4 + 0x0] + ldx [%sp + 2047 + 128 + 0x50], %l3 ! physaddr low + /* 4MB align */ + srlx %l3, 22, %l3 + sllx %l3, 22, %l3 + stx %l3, [%l4 + 0x8] + + /* Leave service as-is, "call-method" */ + mov 7, %l3 + stx %l3, [%sp + 2047 + 128 + 0x08] ! num_args, 7 + mov 1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x10] ! num_rets, 1 + mov (1b - prom_map_name), %l3 + sub %l0, %l3, %l3 + stx %l3, [%sp + 2047 + 128 + 0x18] ! arg1: "map" + /* Leave arg2 as-is, prom_mmu_ihandle_cache */ + mov -1, %l3 + stx %l3, [%sp + 2047 + 128 + 0x28] ! arg3: mode (-1 default) + sethi %hi(8 * 1024 * 1024), %l3 + stx %l3, [%sp + 2047 + 128 + 0x30] ! arg4: size (8MB) + sethi %hi(KERNBASE), %l3 + stx %l3, [%sp + 2047 + 128 + 0x38] ! arg5: vaddr (KERNBASE) + stx %g0, [%sp + 2047 + 128 + 0x40] ! arg6: empty + mov (1b - prom_boot_mapping_phys_low), %l3 + sub %l0, %l3, %l3 + ldx [%l3], %l3 + stx %l3, [%sp + 2047 + 128 + 0x48] ! arg7: phys addr + call %l7 + add %sp, (2047 + 128), %o0 ! argument array + + add %sp, (192 + 128), %sp + +sparc64_boot_after_remap: BRANCH_IF_CHEETAH_BASE(g1,g7,cheetah_boot) BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,cheetah_plus_boot) ba,pt %xcc, spitfire_boot @@ -125,185 +274,7 @@ cheetah_generic_boot: stxa %g0, [%g3] ASI_IMMU membar #Sync - wrpr %g0, (PSTATE_PRIV|PSTATE_PEF|PSTATE_IE), %pstate - wr %g0, 0, %fprs - - /* Just like for Spitfire, we probe itlb-2 for a mapping which - * matches our current %pc. We take the physical address in - * that mapping and use it to make our own. - */ - - /* %g5 holds the tlb data */ - sethi %uhi(_PAGE_VALID | _PAGE_SZ4MB), %g5 - sllx %g5, 32, %g5 - or %g5, (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W | _PAGE_G), %g5 - - /* Put PADDR tlb data mask into %g3. */ - sethi %uhi(_PAGE_PADDR), %g3 - or %g3, %ulo(_PAGE_PADDR), %g3 - sllx %g3, 32, %g3 - sethi %hi(_PAGE_PADDR), %g7 - or %g7, %lo(_PAGE_PADDR), %g7 - or %g3, %g7, %g3 - - set 2 << 16, %l0 /* TLB entry walker. */ - set 0x1fff, %l2 /* Page mask. */ - rd %pc, %l3 - andn %l3, %l2, %g2 /* vaddr comparator */ - -1: ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g2 - be,pn %xcc, cheetah_got_tlbentry - nop - and %l0, (127 << 3), %g1 - cmp %g1, (127 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* Search the small TLB. OBP never maps us like that but - * newer SILO can. - */ - clr %l0 - -1: ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g2 - be,pn %xcc, cheetah_got_tlbentry - nop - cmp %l0, (15 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* BUG() if we get here... */ - ta 0x5 - -cheetah_got_tlbentry: - ldxa [%l0] ASI_ITLB_DATA_ACCESS, %g0 - ldxa [%l0] ASI_ITLB_DATA_ACCESS, %g1 - membar #Sync - and %g1, %g3, %g1 - set 0x5fff, %l0 - andn %g1, %l0, %g1 - or %g5, %g1, %g5 - - /* Clear out any KERNBASE area entries. */ - set 2 << 16, %l0 - sethi %hi(KERNBASE), %g3 - sethi %hi(KERNBASE<<1), %g7 - mov TLB_TAG_ACCESS, %l7 - - /* First, check ITLB */ -1: ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_IMMU - membar #Sync - stxa %g0, [%l0] ASI_ITLB_DATA_ACCESS - membar #Sync - -2: and %l0, (127 << 3), %g1 - cmp %g1, (127 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* Next, check DTLB */ - set 2 << 16, %l0 -1: ldxa [%l0] ASI_DTLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_DMMU - membar #Sync - stxa %g0, [%l0] ASI_DTLB_DATA_ACCESS - membar #Sync - -2: and %l0, (511 << 3), %g1 - cmp %g1, (511 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* On Cheetah+, have to check second DTLB. */ - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,l0,2f) - ba,pt %xcc, 9f - nop - -2: set 3 << 16, %l0 -1: ldxa [%l0] ASI_DTLB_TAG_READ, %g1 - membar #Sync - andn %g1, %l2, %g1 - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_DMMU - membar #Sync - stxa %g0, [%l0] ASI_DTLB_DATA_ACCESS - membar #Sync - -2: and %l0, (511 << 3), %g1 - cmp %g1, (511 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - -9: - - /* Now lock the TTE we created into ITLB-0 and DTLB-0, - * entry 15 (and maybe 14 too). - */ - sethi %hi(KERNBASE), %g3 - set (0 << 16) | (15 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU - membar #Sync - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - stxa %g3, [%l7] ASI_IMMU - membar #Sync - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - flush %g3 - membar #Sync - sethi %hi(_end), %g3 /* Check for bigkernel case */ - or %g3, %lo(_end), %g3 - srl %g3, 23, %g3 /* Check if _end > 8M */ - brz,pt %g3, 1f - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ - sethi %hi(0x400000), %g3 - or %g3, %lo(0x400000), %g3 - add %g5, %g3, %g5 /* New tte data */ - andn %g5, (_PAGE_G), %g5 - sethi %hi(KERNBASE+0x400000), %g3 - or %g3, %lo(KERNBASE+0x400000), %g3 - set (0 << 16) | (14 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU - membar #Sync - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - stxa %g3, [%l7] ASI_IMMU - membar #Sync - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - flush %g3 - membar #Sync - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ - ba,pt %xcc, 1f - nop - -1: set sun4u_init, %g2 - jmpl %g2 + %g0, %g0 - nop + ba,a,pt %xcc, jump_to_sun4u_init spitfire_boot: /* Typically PROM has already enabled both MMU's and both on-chip @@ -313,6 +284,7 @@ spitfire_boot: stxa %g1, [%g0] ASI_LSU_CONTROL membar #Sync +jump_to_sun4u_init: /* * Make sure we are in privileged mode, have address masking, * using the ordinary globals and have enabled floating @@ -324,151 +296,6 @@ spitfire_boot: wrpr %g0, (PSTATE_PRIV|PSTATE_PEF|PSTATE_IE), %pstate wr %g0, 0, %fprs -spitfire_create_mappings: - /* %g5 holds the tlb data */ - sethi %uhi(_PAGE_VALID | _PAGE_SZ4MB), %g5 - sllx %g5, 32, %g5 - or %g5, (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W | _PAGE_G), %g5 - - /* Base of physical memory cannot reliably be assumed to be - * at 0x0! Figure out where it happens to be. -DaveM - */ - - /* Put PADDR tlb data mask into %g3. */ - sethi %uhi(_PAGE_PADDR_SF), %g3 - or %g3, %ulo(_PAGE_PADDR_SF), %g3 - sllx %g3, 32, %g3 - sethi %hi(_PAGE_PADDR_SF), %g7 - or %g7, %lo(_PAGE_PADDR_SF), %g7 - or %g3, %g7, %g3 - - /* Walk through entire ITLB, looking for entry which maps - * our %pc currently, stick PADDR from there into %g5 tlb data. - */ - clr %l0 /* TLB entry walker. */ - set 0x1fff, %l2 /* Page mask. */ - rd %pc, %l3 - andn %l3, %l2, %g2 /* vaddr comparator */ -1: - /* Yes, the nops seem to be necessary for now, don't ask me why. -DaveM */ - ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - nop - nop - nop - andn %g1, %l2, %g1 /* Get vaddr */ - cmp %g1, %g2 - be,a,pn %xcc, spitfire_got_tlbentry - ldxa [%l0] ASI_ITLB_DATA_ACCESS, %g1 - cmp %l0, (63 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - /* BUG() if we get here... */ - ta 0x5 - -spitfire_got_tlbentry: - /* Nops here again, perhaps Cheetah/Blackbird are better behaved... */ - nop - nop - nop - and %g1, %g3, %g1 /* Mask to just get paddr bits. */ - set 0x5fff, %l3 /* Mask offset to get phys base. */ - andn %g1, %l3, %g1 - - /* NOTE: We hold on to %g1 paddr base as we need it below to lock - * NOTE: the PROM cif code into the TLB. - */ - - or %g5, %g1, %g5 /* Or it into TAG being built. */ - - clr %l0 /* TLB entry walker. */ - sethi %hi(KERNBASE), %g3 /* 4M lower limit */ - sethi %hi(KERNBASE<<1), %g7 /* 8M upper limit */ - mov TLB_TAG_ACCESS, %l7 -1: - /* Yes, the nops seem to be necessary for now, don't ask me why. -DaveM */ - ldxa [%l0] ASI_ITLB_TAG_READ, %g1 - nop - nop - nop - andn %g1, %l2, %g1 /* Get vaddr */ - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_IMMU - stxa %g0, [%l0] ASI_ITLB_DATA_ACCESS - membar #Sync -2: - cmp %l0, (63 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - nop; nop; nop - - clr %l0 /* TLB entry walker. */ -1: - /* Yes, the nops seem to be necessary for now, don't ask me why. -DaveM */ - ldxa [%l0] ASI_DTLB_TAG_READ, %g1 - nop - nop - nop - andn %g1, %l2, %g1 /* Get vaddr */ - cmp %g1, %g3 - blu,pn %xcc, 2f - cmp %g1, %g7 - bgeu,pn %xcc, 2f - nop - stxa %g0, [%l7] ASI_DMMU - stxa %g0, [%l0] ASI_DTLB_DATA_ACCESS - membar #Sync -2: - cmp %l0, (63 << 3) - blu,pt %xcc, 1b - add %l0, (1 << 3), %l0 - - nop; nop; nop - - - /* PROM never puts any TLB entries into the MMU with the lock bit - * set. So we gladly use tlb entry 63 for KERNBASE. And maybe 62 too. - */ - - sethi %hi(KERNBASE), %g3 - mov (63 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU /* KERNBASE into TLB TAG */ - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS /* TTE into TLB DATA */ - membar #Sync - stxa %g3, [%l7] ASI_IMMU /* KERNBASE into TLB TAG */ - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS /* TTE into TLB DATA */ - membar #Sync - flush %g3 - membar #Sync - sethi %hi(_end), %g3 /* Check for bigkernel case */ - or %g3, %lo(_end), %g3 - srl %g3, 23, %g3 /* Check if _end > 8M */ - brz,pt %g3, 2f - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ - sethi %hi(0x400000), %g3 - or %g3, %lo(0x400000), %g3 - add %g5, %g3, %g5 /* New tte data */ - andn %g5, (_PAGE_G), %g5 - sethi %hi(KERNBASE+0x400000), %g3 - or %g3, %lo(KERNBASE+0x400000), %g3 - mov (62 << 3), %g7 - stxa %g3, [%l7] ASI_DMMU - stxa %g5, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - stxa %g3, [%l7] ASI_IMMU - stxa %g5, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - flush %g3 - membar #Sync - sethi %hi(KERNBASE), %g3 /* Restore for fixup code below */ -2: ba,pt %xcc, 1f - nop -1: set sun4u_init, %g2 jmpl %g2 + %g0, %g0 nop @@ -483,38 +310,12 @@ sun4u_init: stxa %g0, [%g7] ASI_DMMU membar #Sync - /* We are now safely (we hope) in Nucleus context (0), rewrite - * the KERNBASE TTE's so they no longer have the global bit set. - * Don't forget to setup TAG_ACCESS first 8-) - */ - mov TLB_TAG_ACCESS, %g2 - stxa %g3, [%g2] ASI_IMMU - stxa %g3, [%g2] ASI_DMMU - membar #Sync - BRANCH_IF_ANY_CHEETAH(g1,g7,cheetah_tlb_fixup) ba,pt %xcc, spitfire_tlb_fixup nop cheetah_tlb_fixup: - set (0 << 16) | (15 << 3), %g7 - ldxa [%g7] ASI_ITLB_DATA_ACCESS, %g0 - ldxa [%g7] ASI_ITLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - - ldxa [%g7] ASI_DTLB_DATA_ACCESS, %g0 - ldxa [%g7] ASI_DTLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - - /* Kill instruction prefetch queues. */ - flush %g3 - membar #Sync - mov 2, %g2 /* Set TLB type to cheetah+. */ BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,1f) @@ -523,26 +324,11 @@ cheetah_tlb_fixup: 1: sethi %hi(tlb_type), %g1 stw %g2, [%g1 + %lo(tlb_type)] - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g1,g7,1f) - ba,pt %xcc, 2f - nop - -1: /* Patch context register writes to support nucleus page - * size correctly. - */ - call cheetah_plus_patch_etrap - nop - call cheetah_plus_patch_rtrap - nop - call cheetah_plus_patch_fpdis - nop - call cheetah_plus_patch_winfixup - nop - - -2: /* Patch copy/page operations to cheetah optimized versions. */ + /* Patch copy/page operations to cheetah optimized versions. */ call cheetah_patch_copyops nop + call cheetah_patch_copy_page + nop call cheetah_patch_cachetlbops nop @@ -550,21 +336,6 @@ cheetah_tlb_fixup: nop spitfire_tlb_fixup: - mov (63 << 3), %g7 - ldxa [%g7] ASI_ITLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_ITLB_DATA_ACCESS - membar #Sync - - ldxa [%g7] ASI_DTLB_DATA_ACCESS, %g1 - andn %g1, (_PAGE_G), %g1 - stxa %g1, [%g7] ASI_DTLB_DATA_ACCESS - membar #Sync - - /* Kill instruction prefetch queues. */ - flush %g3 - membar #Sync - /* Set TLB type to spitfire. */ mov 0, %g2 sethi %hi(tlb_type), %g1 @@ -577,24 +348,6 @@ tlb_fixup_done: mov %sp, %l6 mov %o4, %l7 -#if 0 /* We don't do it like this anymore, but for historical hack value - * I leave this snippet here to show how crazy we can be sometimes. 8-) - */ - - /* Setup "Linux Current Register", thanks Sun 8-) */ - wr %g0, 0x1, %pcr - - /* Blackbird errata workaround. See commentary in - * smp.c:smp_percpu_timer_interrupt() for more - * information. - */ - ba,pt %xcc, 99f - nop - .align 64 -99: wr %g6, %g0, %pic - rd %pic, %g0 -#endif - wr %g0, ASI_P, %asi mov 1, %g1 sllx %g1, THREAD_SHIFT, %g1 @@ -628,32 +381,78 @@ tlb_fixup_done: nop /* Not reached... */ -/* IMPORTANT NOTE: Whenever making changes here, check - * trampoline.S as well. -jj */ - .globl setup_tba -setup_tba: /* i0 = is_starfire */ - save %sp, -160, %sp + /* This is meant to allow the sharing of this code between + * boot processor invocation (via setup_tba() below) and + * secondary processor startup (via trampoline.S). The + * former does use this code, the latter does not yet due + * to some complexities. That should be fixed up at some + * point. + * + * There used to be enormous complexity wrt. transferring + * over from the firwmare's trap table to the Linux kernel's. + * For example, there was a chicken & egg problem wrt. building + * the OBP page tables, yet needing to be on the Linux kernel + * trap table (to translate PAGE_OFFSET addresses) in order to + * do that. + * + * We now handle OBP tlb misses differently, via linear lookups + * into the prom_trans[] array. So that specific problem no + * longer exists. Yet, unfortunately there are still some issues + * preventing trampoline.S from using this code... ho hum. + */ + .globl setup_trap_table +setup_trap_table: + save %sp, -192, %sp - rdpr %tba, %g7 - sethi %hi(prom_tba), %o1 - or %o1, %lo(prom_tba), %o1 - stx %g7, [%o1] + /* Force interrupts to be disabled. */ + rdpr %pstate, %o1 + andn %o1, PSTATE_IE, %o1 + wrpr %o1, 0x0, %pstate + wrpr %g0, 15, %pil + + /* Make the firmware call to jump over to the Linux trap table. */ + call prom_set_trap_table + sethi %hi(sparc64_ttable_tl0), %o0 + + /* Start using proper page size encodings in ctx register. */ + sethi %hi(sparc64_kern_pri_context), %g3 + ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2 + mov PRIMARY_CONTEXT, %g1 + stxa %g2, [%g1] ASI_DMMU + membar #Sync + + /* The Linux trap handlers expect various trap global registers + * to be setup with some fixed values. So here we set these + * up very carefully. These globals are: + * + * Alternate Globals (PSTATE_AG): + * + * %g6 --> current_thread_info() + * + * MMU Globals (PSTATE_MG): + * + * %g1 --> TLB_SFSR + * %g2 --> ((_PAGE_VALID | _PAGE_SZ4MB | + * _PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_W) + * ^ 0xfffff80000000000) + * (this %g2 value is used for computing the PAGE_OFFSET kernel + * TLB entries quickly, the virtual address of the fault XOR'd + * with this %g2 value is the PTE to load into the TLB) + * %g3 --> VPTE_BASE_CHEETAH or VPTE_BASE_SPITFIRE + * + * Interrupt Globals (PSTATE_IG, setup by init_irqwork_curcpu()): + * + * %g6 --> __irq_work[smp_processor_id()] + */ - /* Setup "Linux" globals 8-) */ rdpr %pstate, %o1 mov %g6, %o2 - wrpr %o1, (PSTATE_AG|PSTATE_IE), %pstate - sethi %hi(sparc64_ttable_tl0), %g1 - wrpr %g1, %tba + wrpr %o1, PSTATE_AG, %pstate mov %o2, %g6 - /* Set up MMU globals */ - wrpr %o1, (PSTATE_MG|PSTATE_IE), %pstate - - /* Set fixed globals used by dTLB miss handler. */ #define KERN_HIGHBITS ((_PAGE_VALID|_PAGE_SZ4MB)^0xfffff80000000000) #define KERN_LOWBITS (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_W) - + wrpr %o1, PSTATE_MG, %pstate mov TSB_REG, %g1 stxa %g0, [%g1] ASI_DMMU membar #Sync @@ -665,17 +464,17 @@ setup_tba: /* i0 = is_starfire */ sllx %g2, 32, %g2 or %g2, KERN_LOWBITS, %g2 - BRANCH_IF_ANY_CHEETAH(g3,g7,cheetah_vpte_base) - ba,pt %xcc, spitfire_vpte_base + BRANCH_IF_ANY_CHEETAH(g3,g7,8f) + ba,pt %xcc, 9f nop -cheetah_vpte_base: +8: sethi %uhi(VPTE_BASE_CHEETAH), %g3 or %g3, %ulo(VPTE_BASE_CHEETAH), %g3 ba,pt %xcc, 2f sllx %g3, 32, %g3 -spitfire_vpte_base: +9: sethi %uhi(VPTE_BASE_SPITFIRE), %g3 or %g3, %ulo(VPTE_BASE_SPITFIRE), %g3 sllx %g3, 32, %g3 @@ -701,48 +500,55 @@ spitfire_vpte_base: sllx %o2, 32, %o2 wr %o2, %asr25 - /* Ok, we're done setting up all the state our trap mechanims needs, - * now get back into normal globals and let the PROM know what is up. - */ 2: wrpr %g0, %g0, %wstate - wrpr %o1, PSTATE_IE, %pstate + wrpr %o1, 0x0, %pstate call init_irqwork_curcpu nop - call prom_set_trap_table - sethi %hi(sparc64_ttable_tl0), %o0 - - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g2,g3,1f) - ba,pt %xcc, 2f - nop - -1: /* Start using proper page size encodings in ctx register. */ - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %g3 - mov PRIMARY_CONTEXT, %g1 - sllx %g3, 32, %g3 - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - or %g3, %g2, %g3 - stxa %g3, [%g1] ASI_DMMU - membar #Sync - -2: + /* Now we can turn interrupts back on. */ rdpr %pstate, %o1 or %o1, PSTATE_IE, %o1 wrpr %o1, 0, %pstate + wrpr %g0, 0x0, %pil + + ret + restore + + .globl setup_tba +setup_tba: /* i0 = is_starfire */ + save %sp, -192, %sp + + /* The boot processor is the only cpu which invokes this + * routine, the other cpus set things up via trampoline.S. + * So save the OBP trap table address here. + */ + rdpr %tba, %g7 + sethi %hi(prom_tba), %o1 + or %o1, %lo(prom_tba), %o1 + stx %g7, [%o1] + + call setup_trap_table + nop ret restore +sparc64_boot_end: + +#include "systbls.S" +#include "ktlb.S" +#include "etrap.S" +#include "rtrap.S" +#include "winfixup.S" +#include "entry.S" /* - * The following skips make sure the trap table in ttable.S is aligned + * The following skip makes sure the trap table in ttable.S is aligned * on a 32K boundary as required by the v9 specs for TBA register. */ -sparc64_boot_end: - .skip 0x2000 + _start - sparc64_boot_end -bootup_user_stack_end: - .skip 0x2000 +1: + .skip 0x4000 + _start - 1b #ifdef CONFIG_SBUS /* This is just a hack to fool make depend config.h discovering @@ -754,20 +560,6 @@ bootup_user_stack_end: ! 0x0000000000408000 #include "ttable.S" -#include "systbls.S" - - .align 1024 - .globl swapper_pg_dir -swapper_pg_dir: - .word 0 - -#include "etrap.S" -#include "rtrap.S" -#include "winfixup.S" -#include "entry.S" - - /* This is just anal retentiveness on my part... */ - .align 16384 .data .align 8 @@ -775,8 +567,11 @@ swapper_pg_dir: prom_tba: .xword 0 tlb_type: .word 0 /* Must NOT end up in BSS */ .section ".fixup",#alloc,#execinstr - .globl __ret_efault + + .globl __ret_efault, __retl_efault __ret_efault: ret restore %g0, -EFAULT, %o0 - +__retl_efault: + retl + mov -EFAULT, %o0 diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 43fc3173d480..e6a00325075a 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -475,9 +475,6 @@ IOCTL_TABLE_START #include <linux/compat_ioctl.h> #define DECLARES #include "compat_ioctl.c" -COMPATIBLE_IOCTL(TIOCSTART) -COMPATIBLE_IOCTL(TIOCSTOP) -COMPATIBLE_IOCTL(TIOCSLTC) COMPATIBLE_IOCTL(FBIOGTYPE) COMPATIBLE_IOCTL(FBIOSATTR) COMPATIBLE_IOCTL(FBIOGATTR) diff --git a/arch/sparc64/kernel/irq.c b/arch/sparc64/kernel/irq.c index c9b69167632a..233526ba3abe 100644 --- a/arch/sparc64/kernel/irq.c +++ b/arch/sparc64/kernel/irq.c @@ -27,6 +27,7 @@ #include <asm/atomic.h> #include <asm/system.h> #include <asm/irq.h> +#include <asm/io.h> #include <asm/sbus.h> #include <asm/iommu.h> #include <asm/upa.h> diff --git a/arch/sparc64/kernel/itlb_base.S b/arch/sparc64/kernel/itlb_base.S index b5e32dfa4fbc..4951ff8f6877 100644 --- a/arch/sparc64/kernel/itlb_base.S +++ b/arch/sparc64/kernel/itlb_base.S @@ -15,14 +15,12 @@ */ #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, 10, r2 -#define CREATE_VPTE_OFFSET2(r1, r2) -#define CREATE_VPTE_NOP nop +#define CREATE_VPTE_OFFSET2(r1, r2) nop #else /* PAGE_SHIFT */ #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, PAGE_SHIFT, r2 #define CREATE_VPTE_OFFSET2(r1, r2) \ sllx r2, 3, r2 -#define CREATE_VPTE_NOP #endif /* PAGE_SHIFT */ @@ -36,6 +34,7 @@ */ /* ITLB ** ICACHE line 1: Quick user TLB misses */ + mov TLB_SFSR, %g1 ldxa [%g1 + %g1] ASI_IMMU, %g4 ! Get TAG_ACCESS CREATE_VPTE_OFFSET1(%g4, %g6) ! Create VPTE offset CREATE_VPTE_OFFSET2(%g4, %g6) ! Create VPTE offset @@ -43,41 +42,38 @@ 1: brgez,pn %g5, 3f ! Not valid, branch out sethi %hi(_PAGE_EXEC), %g4 ! Delay-slot andcc %g5, %g4, %g0 ! Executable? + +/* ITLB ** ICACHE line 2: Real faults */ be,pn %xcc, 3f ! Nope, branch. nop ! Delay-slot 2: stxa %g5, [%g0] ASI_ITLB_DATA_IN ! Load PTE into TLB retry ! Trap return -3: rdpr %pstate, %g4 ! Move into alternate globals - -/* ITLB ** ICACHE line 2: Real faults */ +3: rdpr %pstate, %g4 ! Move into alt-globals wrpr %g4, PSTATE_AG|PSTATE_MG, %pstate rdpr %tpc, %g5 ! And load faulting VA mov FAULT_CODE_ITLB, %g4 ! It was read from ITLB -sparc64_realfault_common: ! Called by TL0 dtlb_miss too + +/* ITLB ** ICACHE line 3: Finish faults */ +sparc64_realfault_common: ! Called by dtlb_miss stb %g4, [%g6 + TI_FAULT_CODE] stx %g5, [%g6 + TI_FAULT_ADDR] ba,pt %xcc, etrap ! Save state 1: rd %pc, %g7 ! ... - nop - -/* ITLB ** ICACHE line 3: Finish faults + window fixups */ call do_sparc64_fault ! Call fault handler add %sp, PTREGS_OFF, %o0! Compute pt_regs arg ba,pt %xcc, rtrap_clr_l6 ! Restore cpu state nop + +/* ITLB ** ICACHE line 4: Window fixups */ winfix_trampoline: rdpr %tpc, %g3 ! Prepare winfixup TNPC - or %g3, 0x7c, %g3 ! Compute offset to branch + or %g3, 0x7c, %g3 ! Compute branch offset wrpr %g3, %tnpc ! Write it into TNPC done ! Do it to it - -/* ITLB ** ICACHE line 4: Unused... */ nop nop nop nop - CREATE_VPTE_NOP #undef CREATE_VPTE_OFFSET1 #undef CREATE_VPTE_OFFSET2 -#undef CREATE_VPTE_NOP diff --git a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c index bbf11f85dab1..0d66d07c8c6e 100644 --- a/arch/sparc64/kernel/kprobes.c +++ b/arch/sparc64/kernel/kprobes.c @@ -8,6 +8,7 @@ #include <linux/kprobes.h> #include <asm/kdebug.h> #include <asm/signal.h> +#include <asm/cacheflush.h> /* We do not have hardware single-stepping on sparc64. * So we implement software single-stepping with breakpoint @@ -37,31 +38,31 @@ * - Mark that we are no longer actively in a kprobe. */ -int arch_prepare_kprobe(struct kprobe *p) +int __kprobes arch_prepare_kprobe(struct kprobe *p) { return 0; } -void arch_copy_kprobe(struct kprobe *p) +void __kprobes arch_copy_kprobe(struct kprobe *p) { p->ainsn.insn[0] = *p->addr; p->ainsn.insn[1] = BREAKPOINT_INSTRUCTION_2; p->opcode = *p->addr; } -void arch_arm_kprobe(struct kprobe *p) +void __kprobes arch_arm_kprobe(struct kprobe *p) { *p->addr = BREAKPOINT_INSTRUCTION; flushi(p->addr); } -void arch_disarm_kprobe(struct kprobe *p) +void __kprobes arch_disarm_kprobe(struct kprobe *p) { *p->addr = p->opcode; flushi(p->addr); } -void arch_remove_kprobe(struct kprobe *p) +void __kprobes arch_remove_kprobe(struct kprobe *p) { } @@ -111,7 +112,7 @@ static inline void prepare_singlestep(struct kprobe *p, struct pt_regs *regs) } } -static int kprobe_handler(struct pt_regs *regs) +static int __kprobes kprobe_handler(struct pt_regs *regs) { struct kprobe *p; void *addr = (void *) regs->tpc; @@ -191,8 +192,9 @@ no_kprobe: * The original INSN location was REAL_PC, it actually * executed at PC and produced destination address NPC. */ -static unsigned long relbranch_fixup(u32 insn, unsigned long real_pc, - unsigned long pc, unsigned long npc) +static unsigned long __kprobes relbranch_fixup(u32 insn, unsigned long real_pc, + unsigned long pc, + unsigned long npc) { /* Branch not taken, no mods necessary. */ if (npc == pc + 0x4UL) @@ -217,7 +219,8 @@ static unsigned long relbranch_fixup(u32 insn, unsigned long real_pc, /* If INSN is an instruction which writes it's PC location * into a destination register, fix that up. */ -static void retpc_fixup(struct pt_regs *regs, u32 insn, unsigned long real_pc) +static void __kprobes retpc_fixup(struct pt_regs *regs, u32 insn, + unsigned long real_pc) { unsigned long *slot = NULL; @@ -257,7 +260,7 @@ static void retpc_fixup(struct pt_regs *regs, u32 insn, unsigned long real_pc) * This function prepares to return from the post-single-step * breakpoint trap. */ -static void resume_execution(struct kprobe *p, struct pt_regs *regs) +static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs) { u32 insn = p->ainsn.insn[0]; @@ -315,8 +318,8 @@ static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr) /* * Wrapper routine to for handling exceptions. */ -int kprobe_exceptions_notify(struct notifier_block *self, unsigned long val, - void *data) +int __kprobes kprobe_exceptions_notify(struct notifier_block *self, + unsigned long val, void *data) { struct die_args *args = (struct die_args *)data; switch (val) { @@ -344,7 +347,8 @@ int kprobe_exceptions_notify(struct notifier_block *self, unsigned long val, return NOTIFY_DONE; } -asmlinkage void kprobe_trap(unsigned long trap_level, struct pt_regs *regs) +asmlinkage void __kprobes kprobe_trap(unsigned long trap_level, + struct pt_regs *regs) { BUG_ON(trap_level != 0x170 && trap_level != 0x171); @@ -368,7 +372,7 @@ static struct pt_regs jprobe_saved_regs; static struct pt_regs *jprobe_saved_regs_location; static struct sparc_stackf jprobe_saved_stack; -int setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) +int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) { struct jprobe *jp = container_of(p, struct jprobe, kp); @@ -390,7 +394,7 @@ int setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs) return 1; } -void jprobe_return(void) +void __kprobes jprobe_return(void) { preempt_enable_no_resched(); __asm__ __volatile__( @@ -403,7 +407,7 @@ extern void jprobe_return_trap_instruction(void); extern void __show_regs(struct pt_regs * regs); -int longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) +int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) { u32 *addr = (u32 *) regs->tpc; diff --git a/arch/sparc64/kernel/ktlb.S b/arch/sparc64/kernel/ktlb.S new file mode 100644 index 000000000000..d9244d3c9f73 --- /dev/null +++ b/arch/sparc64/kernel/ktlb.S @@ -0,0 +1,194 @@ +/* arch/sparc64/kernel/ktlb.S: Kernel mapping TLB miss handling. + * + * Copyright (C) 1995, 1997, 2005 David S. Miller <davem@davemloft.net> + * Copyright (C) 1996 Eddie C. Dost (ecd@brainaid.de) + * Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx) + * Copyright (C) 1996,98,99 Jakub Jelinek (jj@sunsite.mff.cuni.cz) +*/ + +#include <linux/config.h> +#include <asm/head.h> +#include <asm/asi.h> +#include <asm/page.h> +#include <asm/pgtable.h> + + .text + .align 32 + +/* + * On a second level vpte miss, check whether the original fault is to the OBP + * range (note that this is only possible for instruction miss, data misses to + * obp range do not use vpte). If so, go back directly to the faulting address. + * This is because we want to read the tpc, otherwise we have no way of knowing + * the 8k aligned faulting address if we are using >8k kernel pagesize. This + * also ensures no vpte range addresses are dropped into tlb while obp is + * executing (see inherit_locked_prom_mappings() rant). + */ +sparc64_vpte_nucleus: + /* Note that kvmap below has verified that the address is + * in the range MODULES_VADDR --> VMALLOC_END already. So + * here we need only check if it is an OBP address or not. + */ + sethi %hi(LOW_OBP_ADDRESS), %g5 + cmp %g4, %g5 + blu,pn %xcc, kern_vpte + mov 0x1, %g5 + sllx %g5, 32, %g5 + cmp %g4, %g5 + blu,pn %xcc, vpte_insn_obp + nop + + /* These two instructions are patched by paginig_init(). */ +kern_vpte: + sethi %hi(swapper_pgd_zero), %g5 + lduw [%g5 + %lo(swapper_pgd_zero)], %g5 + + /* With kernel PGD in %g5, branch back into dtlb_backend. */ + ba,pt %xcc, sparc64_kpte_continue + andn %g1, 0x3, %g1 /* Finish PMD offset adjustment. */ + +vpte_noent: + /* Restore previous TAG_ACCESS, %g5 is zero, and we will + * skip over the trap instruction so that the top level + * TLB miss handler will thing this %g5 value is just an + * invalid PTE, thus branching to full fault processing. + */ + mov TLB_SFSR, %g1 + stxa %g4, [%g1 + %g1] ASI_DMMU + done + +vpte_insn_obp: + /* Behave as if we are at TL0. */ + wrpr %g0, 1, %tl + rdpr %tpc, %g4 /* Find original faulting iaddr */ + srlx %g4, 13, %g4 /* Throw out context bits */ + sllx %g4, 13, %g4 /* g4 has vpn + ctx0 now */ + + /* Restore previous TAG_ACCESS. */ + mov TLB_SFSR, %g1 + stxa %g4, [%g1 + %g1] ASI_IMMU + + sethi %hi(prom_trans), %g5 + or %g5, %lo(prom_trans), %g5 + +1: ldx [%g5 + 0x00], %g6 ! base + brz,a,pn %g6, longpath ! no more entries, fail + mov TLB_SFSR, %g1 ! and restore %g1 + ldx [%g5 + 0x08], %g1 ! len + add %g6, %g1, %g1 ! end + cmp %g6, %g4 + bgu,pt %xcc, 2f + cmp %g4, %g1 + bgeu,pt %xcc, 2f + ldx [%g5 + 0x10], %g1 ! PTE + + /* TLB load, restore %g1, and return from trap. */ + sub %g4, %g6, %g6 + add %g1, %g6, %g5 + mov TLB_SFSR, %g1 + stxa %g5, [%g0] ASI_ITLB_DATA_IN + retry + +2: ba,pt %xcc, 1b + add %g5, (3 * 8), %g5 ! next entry + +kvmap_do_obp: + sethi %hi(prom_trans), %g5 + or %g5, %lo(prom_trans), %g5 + srlx %g4, 13, %g4 + sllx %g4, 13, %g4 + +1: ldx [%g5 + 0x00], %g6 ! base + brz,a,pn %g6, longpath ! no more entries, fail + mov TLB_SFSR, %g1 ! and restore %g1 + ldx [%g5 + 0x08], %g1 ! len + add %g6, %g1, %g1 ! end + cmp %g6, %g4 + bgu,pt %xcc, 2f + cmp %g4, %g1 + bgeu,pt %xcc, 2f + ldx [%g5 + 0x10], %g1 ! PTE + + /* TLB load, restore %g1, and return from trap. */ + sub %g4, %g6, %g6 + add %g1, %g6, %g5 + mov TLB_SFSR, %g1 + stxa %g5, [%g0] ASI_DTLB_DATA_IN + retry + +2: ba,pt %xcc, 1b + add %g5, (3 * 8), %g5 ! next entry + +/* + * On a first level data miss, check whether this is to the OBP range (note + * that such accesses can be made by prom, as well as by kernel using + * prom_getproperty on "address"), and if so, do not use vpte access ... + * rather, use information saved during inherit_prom_mappings() using 8k + * pagesize. + */ + .align 32 +kvmap: + brgez,pn %g4, kvmap_nonlinear + nop + +#ifdef CONFIG_DEBUG_PAGEALLOC + .globl kvmap_linear_patch +kvmap_linear_patch: +#endif + ba,pt %xcc, kvmap_load + xor %g2, %g4, %g5 + +#ifdef CONFIG_DEBUG_PAGEALLOC + sethi %hi(swapper_pg_dir), %g5 + or %g5, %lo(swapper_pg_dir), %g5 + sllx %g4, 64 - (PGDIR_SHIFT + PGDIR_BITS), %g6 + srlx %g6, 64 - PAGE_SHIFT, %g6 + andn %g6, 0x3, %g6 + lduw [%g5 + %g6], %g5 + brz,pn %g5, longpath + sllx %g4, 64 - (PMD_SHIFT + PMD_BITS), %g6 + srlx %g6, 64 - PAGE_SHIFT, %g6 + sllx %g5, 11, %g5 + andn %g6, 0x3, %g6 + lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brz,pn %g5, longpath + sllx %g4, 64 - PMD_SHIFT, %g6 + srlx %g6, 64 - PAGE_SHIFT, %g6 + sllx %g5, 11, %g5 + andn %g6, 0x7, %g6 + ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 + brz,pn %g5, longpath + nop + ba,a,pt %xcc, kvmap_load +#endif + +kvmap_nonlinear: + sethi %hi(MODULES_VADDR), %g5 + cmp %g4, %g5 + blu,pn %xcc, longpath + mov (VMALLOC_END >> 24), %g5 + sllx %g5, 24, %g5 + cmp %g4, %g5 + bgeu,pn %xcc, longpath + nop + +kvmap_check_obp: + sethi %hi(LOW_OBP_ADDRESS), %g5 + cmp %g4, %g5 + blu,pn %xcc, kvmap_vmalloc_addr + mov 0x1, %g5 + sllx %g5, 32, %g5 + cmp %g4, %g5 + blu,pn %xcc, kvmap_do_obp + nop + +kvmap_vmalloc_addr: + /* If we get here, a vmalloc addr was accessed, load kernel VPTE. */ + ldxa [%g3 + %g6] ASI_N, %g5 + brgez,pn %g5, longpath + nop + +kvmap_load: + /* PTE is valid, load into TLB and return from trap. */ + stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB + retry diff --git a/arch/sparc64/kernel/pci.c b/arch/sparc64/kernel/pci.c index bba140d98b1b..2ff7c32ab0ce 100644 --- a/arch/sparc64/kernel/pci.c +++ b/arch/sparc64/kernel/pci.c @@ -359,134 +359,17 @@ void pcibios_fixup_bus(struct pci_bus *pbus) pbus->resource[1] = &pbm->mem_space; } -int pci_claim_resource(struct pci_dev *pdev, int resource) +struct resource *pcibios_select_root(struct pci_dev *pdev, struct resource *r) { struct pci_pbm_info *pbm = pdev->bus->sysdata; - struct resource *res = &pdev->resource[resource]; - struct resource *root; - - if (!pbm) - return -EINVAL; + struct resource *root = NULL; - if (res->flags & IORESOURCE_IO) + if (r->flags & IORESOURCE_IO) root = &pbm->io_space; - else + if (r->flags & IORESOURCE_MEM) root = &pbm->mem_space; - pbm->parent->resource_adjust(pdev, res, root); - - return request_resource(root, res); -} - -/* - * Given the PCI bus a device resides on, try to - * find an acceptable resource allocation for a - * specific device resource.. - */ -static int pci_assign_bus_resource(const struct pci_bus *bus, - struct pci_dev *dev, - struct resource *res, - unsigned long size, - unsigned long min, - int resno) -{ - unsigned int type_mask; - int i; - - type_mask = IORESOURCE_IO | IORESOURCE_MEM; - for (i = 0 ; i < 4; i++) { - struct resource *r = bus->resource[i]; - if (!r) - continue; - - /* type_mask must match */ - if ((res->flags ^ r->flags) & type_mask) - continue; - - /* Ok, try it out.. */ - if (allocate_resource(r, res, size, min, -1, size, NULL, NULL) < 0) - continue; - - /* PCI config space updated by caller. */ - return 0; - } - return -EBUSY; -} - -int pci_assign_resource(struct pci_dev *pdev, int resource) -{ - struct pcidev_cookie *pcp = pdev->sysdata; - struct pci_pbm_info *pbm = pcp->pbm; - struct resource *res = &pdev->resource[resource]; - unsigned long min, size; - int err; - - if (res->flags & IORESOURCE_IO) - min = pbm->io_space.start + 0x400UL; - else - min = pbm->mem_space.start; - - size = res->end - res->start + 1; - - err = pci_assign_bus_resource(pdev->bus, pdev, res, size, min, resource); - - if (err < 0) { - printk("PCI: Failed to allocate resource %d for %s\n", - resource, pci_name(pdev)); - } else { - /* Update PCI config space. */ - pbm->parent->base_address_update(pdev, resource); - } - - return err; -} - -/* Sort resources by alignment */ -void pdev_sort_resources(struct pci_dev *dev, struct resource_list *head) -{ - int i; - - for (i = 0; i < PCI_NUM_RESOURCES; i++) { - struct resource *r; - struct resource_list *list, *tmp; - unsigned long r_align; - - r = &dev->resource[i]; - r_align = r->end - r->start; - - if (!(r->flags) || r->parent) - continue; - if (!r_align) { - printk(KERN_WARNING "PCI: Ignore bogus resource %d " - "[%lx:%lx] of %s\n", - i, r->start, r->end, pci_name(dev)); - continue; - } - r_align = (i < PCI_BRIDGE_RESOURCES) ? r_align + 1 : r->start; - for (list = head; ; list = list->next) { - unsigned long align = 0; - struct resource_list *ln = list->next; - int idx; - - if (ln) { - idx = ln->res - &ln->dev->resource[0]; - align = (idx < PCI_BRIDGE_RESOURCES) ? - ln->res->end - ln->res->start + 1 : - ln->res->start; - } - if (r_align > align) { - tmp = kmalloc(sizeof(*tmp), GFP_KERNEL); - if (!tmp) - panic("pdev_sort_resources(): " - "kmalloc() failed!\n"); - tmp->next = ln; - tmp->res = r; - tmp->dev = dev; - list->next = tmp; - break; - } - } - } + return root; } void pcibios_update_irq(struct pci_dev *pdev, int irq) @@ -540,6 +423,7 @@ void pcibios_bus_to_resource(struct pci_dev *pdev, struct resource *res, pbm->parent->resource_adjust(pdev, res, root); } +EXPORT_SYMBOL(pcibios_bus_to_resource); char * __init pcibios_setup(char *str) { @@ -735,8 +619,7 @@ static void __pci_mmap_set_flags(struct pci_dev *dev, struct vm_area_struct *vma static void __pci_mmap_set_pgprot(struct pci_dev *dev, struct vm_area_struct *vma, enum pci_mmap_state mmap_state) { - /* Our io_remap_page_range/io_remap_pfn_range takes care of this, - do nothing. */ + /* Our io_remap_pfn_range takes care of this, do nothing. */ } /* Perform the actual remap of the pages for a PCI device mapping, as appropriate diff --git a/arch/sparc64/kernel/pci_iommu.c b/arch/sparc64/kernel/pci_iommu.c index 2803bc7c2c79..a11910be1013 100644 --- a/arch/sparc64/kernel/pci_iommu.c +++ b/arch/sparc64/kernel/pci_iommu.c @@ -49,12 +49,6 @@ static void __iommu_flushall(struct pci_iommu *iommu) /* Ensure completion of previous PIO writes. */ (void) pci_iommu_read(iommu->write_complete_reg); - - /* Now update everyone's flush point. */ - for (entry = 0; entry < PBM_NCLUSTERS; entry++) { - iommu->alloc_info[entry].flush = - iommu->alloc_info[entry].next; - } } #define IOPTE_CONSISTENT(CTX) \ @@ -80,120 +74,117 @@ static void inline iopte_make_dummy(struct pci_iommu *iommu, iopte_t *iopte) iopte_val(*iopte) = val; } -void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize) +/* Based largely upon the ppc64 iommu allocator. */ +static long pci_arena_alloc(struct pci_iommu *iommu, unsigned long npages) { - int i; - - tsbsize /= sizeof(iopte_t); - - for (i = 0; i < tsbsize; i++) - iopte_make_dummy(iommu, &iommu->page_table[i]); -} - -static iopte_t *alloc_streaming_cluster(struct pci_iommu *iommu, unsigned long npages) -{ - iopte_t *iopte, *limit, *first; - unsigned long cnum, ent, flush_point; - - cnum = 0; - while ((1UL << cnum) < npages) - cnum++; - iopte = (iommu->page_table + - (cnum << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); - - if (cnum == 0) - limit = (iommu->page_table + - iommu->lowest_consistent_map); - else - limit = (iopte + - (1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); - - iopte += ((ent = iommu->alloc_info[cnum].next) << cnum); - flush_point = iommu->alloc_info[cnum].flush; - - first = iopte; - for (;;) { - if (IOPTE_IS_DUMMY(iommu, iopte)) { - if ((iopte + (1 << cnum)) >= limit) - ent = 0; - else - ent = ent + 1; - iommu->alloc_info[cnum].next = ent; - if (ent == flush_point) - __iommu_flushall(iommu); - break; + struct pci_iommu_arena *arena = &iommu->arena; + unsigned long n, i, start, end, limit; + int pass; + + limit = arena->limit; + start = arena->hint; + pass = 0; + +again: + n = find_next_zero_bit(arena->map, limit, start); + end = n + npages; + if (unlikely(end >= limit)) { + if (likely(pass < 1)) { + limit = start; + start = 0; + __iommu_flushall(iommu); + pass++; + goto again; + } else { + /* Scanned the whole thing, give up. */ + return -1; } - iopte += (1 << cnum); - ent++; - if (iopte >= limit) { - iopte = (iommu->page_table + - (cnum << - (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); - ent = 0; + } + + for (i = n; i < end; i++) { + if (test_bit(i, arena->map)) { + start = i + 1; + goto again; } - if (ent == flush_point) - __iommu_flushall(iommu); - if (iopte == first) - goto bad; } - /* I've got your streaming cluster right here buddy boy... */ - return iopte; + for (i = n; i < end; i++) + __set_bit(i, arena->map); -bad: - printk(KERN_EMERG "pci_iommu: alloc_streaming_cluster of npages(%ld) failed!\n", - npages); - return NULL; + arena->hint = end; + + return n; } -static void free_streaming_cluster(struct pci_iommu *iommu, dma_addr_t base, - unsigned long npages, unsigned long ctx) +static void pci_arena_free(struct pci_iommu_arena *arena, unsigned long base, unsigned long npages) { - unsigned long cnum, ent; + unsigned long i; - cnum = 0; - while ((1UL << cnum) < npages) - cnum++; + for (i = base; i < (base + npages); i++) + __clear_bit(i, arena->map); +} - ent = (base << (32 - IO_PAGE_SHIFT + PBM_LOGCLUSTERS - iommu->page_table_sz_bits)) - >> (32 + PBM_LOGCLUSTERS + cnum - iommu->page_table_sz_bits); +void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize, u32 dma_offset, u32 dma_addr_mask) +{ + unsigned long i, tsbbase, order, sz, num_tsb_entries; + + num_tsb_entries = tsbsize / sizeof(iopte_t); + + /* Setup initial software IOMMU state. */ + spin_lock_init(&iommu->lock); + iommu->ctx_lowest_free = 1; + iommu->page_table_map_base = dma_offset; + iommu->dma_addr_mask = dma_addr_mask; + + /* Allocate and initialize the free area map. */ + sz = num_tsb_entries / 8; + sz = (sz + 7UL) & ~7UL; + iommu->arena.map = kmalloc(sz, GFP_KERNEL); + if (!iommu->arena.map) { + prom_printf("PCI_IOMMU: Error, kmalloc(arena.map) failed.\n"); + prom_halt(); + } + memset(iommu->arena.map, 0, sz); + iommu->arena.limit = num_tsb_entries; - /* If the global flush might not have caught this entry, - * adjust the flush point such that we will flush before - * ever trying to reuse it. + /* Allocate and initialize the dummy page which we + * set inactive IO PTEs to point to. */ -#define between(X,Y,Z) (((Z) - (Y)) >= ((X) - (Y))) - if (between(ent, iommu->alloc_info[cnum].next, iommu->alloc_info[cnum].flush)) - iommu->alloc_info[cnum].flush = ent; -#undef between + iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); + if (!iommu->dummy_page) { + prom_printf("PCI_IOMMU: Error, gfp(dummy_page) failed.\n"); + prom_halt(); + } + memset((void *)iommu->dummy_page, 0, PAGE_SIZE); + iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); + + /* Now allocate and setup the IOMMU page table itself. */ + order = get_order(tsbsize); + tsbbase = __get_free_pages(GFP_KERNEL, order); + if (!tsbbase) { + prom_printf("PCI_IOMMU: Error, gfp(tsb) failed.\n"); + prom_halt(); + } + iommu->page_table = (iopte_t *)tsbbase; + + for (i = 0; i < num_tsb_entries; i++) + iopte_make_dummy(iommu, &iommu->page_table[i]); } -/* We allocate consistent mappings from the end of cluster zero. */ -static iopte_t *alloc_consistent_cluster(struct pci_iommu *iommu, unsigned long npages) +static inline iopte_t *alloc_npages(struct pci_iommu *iommu, unsigned long npages) { - iopte_t *iopte; + long entry; - iopte = iommu->page_table + (1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS)); - while (iopte > iommu->page_table) { - iopte--; - if (IOPTE_IS_DUMMY(iommu, iopte)) { - unsigned long tmp = npages; + entry = pci_arena_alloc(iommu, npages); + if (unlikely(entry < 0)) + return NULL; - while (--tmp) { - iopte--; - if (!IOPTE_IS_DUMMY(iommu, iopte)) - break; - } - if (tmp == 0) { - u32 entry = (iopte - iommu->page_table); + return iommu->page_table + entry; +} - if (entry < iommu->lowest_consistent_map) - iommu->lowest_consistent_map = entry; - return iopte; - } - } - } - return NULL; +static inline void free_npages(struct pci_iommu *iommu, dma_addr_t base, unsigned long npages) +{ + pci_arena_free(&iommu->arena, base >> IO_PAGE_SHIFT, npages); } static int iommu_alloc_ctx(struct pci_iommu *iommu) @@ -233,7 +224,7 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad struct pcidev_cookie *pcp; struct pci_iommu *iommu; iopte_t *iopte; - unsigned long flags, order, first_page, ctx; + unsigned long flags, order, first_page; void *ret; int npages; @@ -251,9 +242,10 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad iommu = pcp->pbm->iommu; spin_lock_irqsave(&iommu->lock, flags); - iopte = alloc_consistent_cluster(iommu, size >> IO_PAGE_SHIFT); - if (iopte == NULL) { - spin_unlock_irqrestore(&iommu->lock, flags); + iopte = alloc_npages(iommu, size >> IO_PAGE_SHIFT); + spin_unlock_irqrestore(&iommu->lock, flags); + + if (unlikely(iopte == NULL)) { free_pages(first_page, order); return NULL; } @@ -262,31 +254,15 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad ((iopte - iommu->page_table) << IO_PAGE_SHIFT)); ret = (void *) first_page; npages = size >> IO_PAGE_SHIFT; - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = iommu_alloc_ctx(iommu); first_page = __pa(first_page); while (npages--) { - iopte_val(*iopte) = (IOPTE_CONSISTENT(ctx) | + iopte_val(*iopte) = (IOPTE_CONSISTENT(0UL) | IOPTE_WRITE | (first_page & IOPTE_PAGE)); iopte++; first_page += IO_PAGE_SIZE; } - { - int i; - u32 daddr = *dma_addrp; - - npages = size >> IO_PAGE_SHIFT; - for (i = 0; i < npages; i++) { - pci_iommu_write(iommu->iommu_flush, daddr); - daddr += IO_PAGE_SIZE; - } - } - - spin_unlock_irqrestore(&iommu->lock, flags); - return ret; } @@ -296,7 +272,7 @@ void pci_free_consistent(struct pci_dev *pdev, size_t size, void *cpu, dma_addr_ struct pcidev_cookie *pcp; struct pci_iommu *iommu; iopte_t *iopte; - unsigned long flags, order, npages, i, ctx; + unsigned long flags, order, npages; npages = IO_PAGE_ALIGN(size) >> IO_PAGE_SHIFT; pcp = pdev->sysdata; @@ -306,46 +282,7 @@ void pci_free_consistent(struct pci_dev *pdev, size_t size, void *cpu, dma_addr_ spin_lock_irqsave(&iommu->lock, flags); - if ((iopte - iommu->page_table) == - iommu->lowest_consistent_map) { - iopte_t *walk = iopte + npages; - iopte_t *limit; - - limit = (iommu->page_table + - (1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS))); - while (walk < limit) { - if (!IOPTE_IS_DUMMY(iommu, walk)) - break; - walk++; - } - iommu->lowest_consistent_map = - (walk - iommu->page_table); - } - - /* Data for consistent mappings cannot enter the streaming - * buffers, so we only need to update the TSB. We flush - * the IOMMU here as well to prevent conflicts with the - * streaming mapping deferred tlb flush scheme. - */ - - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = (iopte_val(*iopte) & IOPTE_CONTEXT) >> 47UL; - - for (i = 0; i < npages; i++, iopte++) - iopte_make_dummy(iommu, iopte); - - if (iommu->iommu_ctxflush) { - pci_iommu_write(iommu->iommu_ctxflush, ctx); - } else { - for (i = 0; i < npages; i++) { - u32 daddr = dvma + (i << IO_PAGE_SHIFT); - - pci_iommu_write(iommu->iommu_flush, daddr); - } - } - - iommu_free_ctx(iommu, ctx); + free_npages(iommu, dvma, npages); spin_unlock_irqrestore(&iommu->lock, flags); @@ -372,25 +309,27 @@ dma_addr_t pci_map_single(struct pci_dev *pdev, void *ptr, size_t sz, int direct iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) + goto bad_no_ctx; oaddr = (unsigned long)ptr; npages = IO_PAGE_ALIGN(oaddr + sz) - (oaddr & IO_PAGE_MASK); npages >>= IO_PAGE_SHIFT; spin_lock_irqsave(&iommu->lock, flags); + base = alloc_npages(iommu, npages); + ctx = 0; + if (iommu->iommu_ctxflush) + ctx = iommu_alloc_ctx(iommu); + spin_unlock_irqrestore(&iommu->lock, flags); - base = alloc_streaming_cluster(iommu, npages); - if (base == NULL) + if (unlikely(!base)) goto bad; + bus_addr = (iommu->page_table_map_base + ((base - iommu->page_table) << IO_PAGE_SHIFT)); ret = bus_addr | (oaddr & ~IO_PAGE_MASK); base_paddr = __pa(oaddr & IO_PAGE_MASK); - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = iommu_alloc_ctx(iommu); if (strbuf->strbuf_enabled) iopte_protection = IOPTE_STREAMING(ctx); else @@ -401,12 +340,13 @@ dma_addr_t pci_map_single(struct pci_dev *pdev, void *ptr, size_t sz, int direct for (i = 0; i < npages; i++, base++, base_paddr += IO_PAGE_SIZE) iopte_val(*base) = iopte_protection | base_paddr; - spin_unlock_irqrestore(&iommu->lock, flags); - return ret; bad: - spin_unlock_irqrestore(&iommu->lock, flags); + iommu_free_ctx(iommu, ctx); +bad_no_ctx: + if (printk_ratelimit()) + WARN_ON(1); return PCI_DMA_ERROR_CODE; } @@ -466,7 +406,7 @@ do_flush_sync: if (!limit) break; udelay(1); - membar("#LoadLoad"); + rmb(); } if (!limit) printk(KERN_WARNING "pci_strbuf_flush: flushflag timeout " @@ -481,10 +421,13 @@ void pci_unmap_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, int struct pci_iommu *iommu; struct pci_strbuf *strbuf; iopte_t *base; - unsigned long flags, npages, ctx; + unsigned long flags, npages, ctx, i; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) { + if (printk_ratelimit()) + WARN_ON(1); + return; + } pcp = pdev->sysdata; iommu = pcp->pbm->iommu; @@ -510,13 +453,14 @@ void pci_unmap_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, int /* Step 1: Kick data out of streaming buffers if necessary. */ if (strbuf->strbuf_enabled) - pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); + pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, + npages, direction); - /* Step 2: Clear out first TSB entry. */ - iopte_make_dummy(iommu, base); + /* Step 2: Clear out TSB entries. */ + for (i = 0; i < npages; i++) + iopte_make_dummy(iommu, base + i); - free_streaming_cluster(iommu, bus_addr - iommu->page_table_map_base, - npages, ctx); + free_npages(iommu, bus_addr - iommu->page_table_map_base, npages); iommu_free_ctx(iommu, ctx); @@ -621,6 +565,8 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int pci_map_single(pdev, (page_address(sglist->page) + sglist->offset), sglist->length, direction); + if (unlikely(sglist->dma_address == PCI_DMA_ERROR_CODE)) + return 0; sglist->dma_length = sglist->length; return 1; } @@ -629,21 +575,29 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) + goto bad_no_ctx; /* Step 1: Prepare scatter list. */ npages = prepare_sg(sglist, nelems); - /* Step 2: Allocate a cluster. */ + /* Step 2: Allocate a cluster and context, if necessary. */ spin_lock_irqsave(&iommu->lock, flags); - base = alloc_streaming_cluster(iommu, npages); + base = alloc_npages(iommu, npages); + ctx = 0; + if (iommu->iommu_ctxflush) + ctx = iommu_alloc_ctx(iommu); + + spin_unlock_irqrestore(&iommu->lock, flags); + if (base == NULL) goto bad; - dma_base = iommu->page_table_map_base + ((base - iommu->page_table) << IO_PAGE_SHIFT); + + dma_base = iommu->page_table_map_base + + ((base - iommu->page_table) << IO_PAGE_SHIFT); /* Step 3: Normalize DMA addresses. */ used = nelems; @@ -656,30 +610,28 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int } used = nelems - used; - /* Step 4: Choose a context if necessary. */ - ctx = 0; - if (iommu->iommu_ctxflush) - ctx = iommu_alloc_ctx(iommu); - - /* Step 5: Create the mappings. */ + /* Step 4: Create the mappings. */ if (strbuf->strbuf_enabled) iopte_protection = IOPTE_STREAMING(ctx); else iopte_protection = IOPTE_CONSISTENT(ctx); if (direction != PCI_DMA_TODEVICE) iopte_protection |= IOPTE_WRITE; - fill_sg (base, sglist, used, nelems, iopte_protection); + + fill_sg(base, sglist, used, nelems, iopte_protection); + #ifdef VERIFY_SG verify_sglist(sglist, nelems, base, npages); #endif - spin_unlock_irqrestore(&iommu->lock, flags); - return used; bad: - spin_unlock_irqrestore(&iommu->lock, flags); - return PCI_DMA_ERROR_CODE; + iommu_free_ctx(iommu, ctx); +bad_no_ctx: + if (printk_ratelimit()) + WARN_ON(1); + return 0; } /* Unmap a set of streaming mode DMA translations. */ @@ -692,8 +644,10 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, unsigned long flags, ctx, i, npages; u32 bus_addr; - if (direction == PCI_DMA_NONE) - BUG(); + if (unlikely(direction == PCI_DMA_NONE)) { + if (printk_ratelimit()) + WARN_ON(1); + } pcp = pdev->sysdata; iommu = pcp->pbm->iommu; @@ -705,7 +659,8 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, if (sglist[i].dma_length == 0) break; i--; - npages = (IO_PAGE_ALIGN(sglist[i].dma_address + sglist[i].dma_length) - bus_addr) >> IO_PAGE_SHIFT; + npages = (IO_PAGE_ALIGN(sglist[i].dma_address + sglist[i].dma_length) - + bus_addr) >> IO_PAGE_SHIFT; base = iommu->page_table + ((bus_addr - iommu->page_table_map_base) >> IO_PAGE_SHIFT); @@ -726,11 +681,11 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, if (strbuf->strbuf_enabled) pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); - /* Step 2: Clear out first TSB entry. */ - iopte_make_dummy(iommu, base); + /* Step 2: Clear out the TSB entries. */ + for (i = 0; i < npages; i++) + iopte_make_dummy(iommu, base + i); - free_streaming_cluster(iommu, bus_addr - iommu->page_table_map_base, - npages, ctx); + free_npages(iommu, bus_addr - iommu->page_table_map_base, npages); iommu_free_ctx(iommu, ctx); diff --git a/arch/sparc64/kernel/pci_psycho.c b/arch/sparc64/kernel/pci_psycho.c index 91ab466d6c66..c03ed5f49d31 100644 --- a/arch/sparc64/kernel/pci_psycho.c +++ b/arch/sparc64/kernel/pci_psycho.c @@ -307,7 +307,7 @@ static unsigned char psycho_pil_table[] = { /*0x32*/15, /* Power Management */ }; -static int __init psycho_ino_to_pil(struct pci_dev *pdev, unsigned int ino) +static int psycho_ino_to_pil(struct pci_dev *pdev, unsigned int ino) { int ret; @@ -344,9 +344,9 @@ static int __init psycho_ino_to_pil(struct pci_dev *pdev, unsigned int ino) return ret; } -static unsigned int __init psycho_irq_build(struct pci_pbm_info *pbm, - struct pci_dev *pdev, - unsigned int ino) +static unsigned int psycho_irq_build(struct pci_pbm_info *pbm, + struct pci_dev *pdev, + unsigned int ino) { struct ino_bucket *bucket; unsigned long imap, iclr; @@ -1024,7 +1024,7 @@ static irqreturn_t psycho_pcierr_intr(int irq, void *dev_id, struct pt_regs *reg #define PSYCHO_CE_INO 0x2f #define PSYCHO_PCIERR_A_INO 0x30 #define PSYCHO_PCIERR_B_INO 0x31 -static void __init psycho_register_error_handlers(struct pci_controller_info *p) +static void psycho_register_error_handlers(struct pci_controller_info *p) { struct pci_pbm_info *pbm = &p->pbm_A; /* arbitrary */ unsigned long base = p->pbm_A.controller_regs; @@ -1091,15 +1091,15 @@ static void __init psycho_register_error_handlers(struct pci_controller_info *p) } /* PSYCHO boot time probing and initialization. */ -static void __init psycho_resource_adjust(struct pci_dev *pdev, - struct resource *res, - struct resource *root) +static void psycho_resource_adjust(struct pci_dev *pdev, + struct resource *res, + struct resource *root) { res->start += root->start; res->end += root->start; } -static void __init psycho_base_address_update(struct pci_dev *pdev, int resource) +static void psycho_base_address_update(struct pci_dev *pdev, int resource) { struct pcidev_cookie *pcp = pdev->sysdata; struct pci_pbm_info *pbm = pcp->pbm; @@ -1144,7 +1144,7 @@ static void __init psycho_base_address_update(struct pci_dev *pdev, int resource pci_write_config_dword(pdev, where + 4, 0); } -static void __init pbm_config_busmastering(struct pci_pbm_info *pbm) +static void pbm_config_busmastering(struct pci_pbm_info *pbm) { u8 *addr; @@ -1161,8 +1161,8 @@ static void __init pbm_config_busmastering(struct pci_pbm_info *pbm) pci_config_write8(addr, 64); } -static void __init pbm_scan_bus(struct pci_controller_info *p, - struct pci_pbm_info *pbm) +static void pbm_scan_bus(struct pci_controller_info *p, + struct pci_pbm_info *pbm) { struct pcidev_cookie *cookie = kmalloc(sizeof(*cookie), GFP_KERNEL); @@ -1189,7 +1189,7 @@ static void __init pbm_scan_bus(struct pci_controller_info *p, pci_setup_busmastering(pbm, pbm->pci_bus); } -static void __init psycho_scan_bus(struct pci_controller_info *p) +static void psycho_scan_bus(struct pci_controller_info *p) { pbm_config_busmastering(&p->pbm_B); p->pbm_B.is_66mhz_capable = 0; @@ -1204,16 +1204,12 @@ static void __init psycho_scan_bus(struct pci_controller_info *p) psycho_register_error_handlers(p); } -static void __init psycho_iommu_init(struct pci_controller_info *p) +static void psycho_iommu_init(struct pci_controller_info *p) { struct pci_iommu *iommu = p->pbm_A.iommu; - unsigned long tsbbase, i; + unsigned long i; u64 control; - /* Setup initial software IOMMU state. */ - spin_lock_init(&iommu->lock); - iommu->ctx_lowest_free = 1; - /* Register addresses. */ iommu->iommu_control = p->pbm_A.controller_regs + PSYCHO_IOMMU_CONTROL; iommu->iommu_tsbbase = p->pbm_A.controller_regs + PSYCHO_IOMMU_TSBBASE; @@ -1240,40 +1236,10 @@ static void __init psycho_iommu_init(struct pci_controller_info *p) /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ + pci_iommu_table_init(iommu, IO_TSB_SIZE, 0xc0000000, 0xffffffff); - iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); - if (!iommu->dummy_page) { - prom_printf("PSYCHO_IOMMU: Error, gfp(dummy_page) failed.\n"); - prom_halt(); - } - memset((void *)iommu->dummy_page, 0, PAGE_SIZE); - iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); - - /* Using assumed page size 8K with 128K entries we need 1MB iommu page - * table (128K ioptes * 8 bytes per iopte). This is - * page order 7 on UltraSparc. - */ - tsbbase = __get_free_pages(GFP_KERNEL, get_order(IO_TSB_SIZE)); - if (!tsbbase) { - prom_printf("PSYCHO_IOMMU: Error, gfp(tsb) failed.\n"); - prom_halt(); - } - iommu->page_table = (iopte_t *)tsbbase; - iommu->page_table_sz_bits = 17; - iommu->page_table_map_base = 0xc0000000; - iommu->dma_addr_mask = 0xffffffff; - pci_iommu_table_init(iommu, IO_TSB_SIZE); - - /* We start with no consistent mappings. */ - iommu->lowest_consistent_map = - 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); - - for (i = 0; i < PBM_NCLUSTERS; i++) { - iommu->alloc_info[i].flush = 0; - iommu->alloc_info[i].next = 0; - } - - psycho_write(p->pbm_A.controller_regs + PSYCHO_IOMMU_TSBBASE, __pa(tsbbase)); + psycho_write(p->pbm_A.controller_regs + PSYCHO_IOMMU_TSBBASE, + __pa(iommu->page_table)); control = psycho_read(p->pbm_A.controller_regs + PSYCHO_IOMMU_CONTROL); control &= ~(PSYCHO_IOMMU_CTRL_TSBSZ | PSYCHO_IOMMU_CTRL_TBWSZ); @@ -1281,7 +1247,7 @@ static void __init psycho_iommu_init(struct pci_controller_info *p) psycho_write(p->pbm_A.controller_regs + PSYCHO_IOMMU_CONTROL, control); /* If necessary, hook us up for starfire IRQ translations. */ - if(this_is_starfire) + if (this_is_starfire) p->starfire_cookie = starfire_hookup(p->pbm_A.portid); else p->starfire_cookie = NULL; @@ -1327,8 +1293,8 @@ static void psycho_controller_hwinit(struct pci_controller_info *p) psycho_write(p->pbm_A.controller_regs + PSYCHO_PCIB_DIAG, tmp); } -static void __init pbm_register_toplevel_resources(struct pci_controller_info *p, - struct pci_pbm_info *pbm) +static void pbm_register_toplevel_resources(struct pci_controller_info *p, + struct pci_pbm_info *pbm) { char *name = pbm->name; @@ -1481,7 +1447,7 @@ static void psycho_pbm_init(struct pci_controller_info *p, #define PSYCHO_CONFIGSPACE 0x001000000UL -void __init psycho_init(int node, char *model_name) +void psycho_init(int node, char *model_name) { struct linux_prom64_registers pr_regs[3]; struct pci_controller_info *p; diff --git a/arch/sparc64/kernel/pci_sabre.c b/arch/sparc64/kernel/pci_sabre.c index 52bf3431a422..da8e1364194f 100644 --- a/arch/sparc64/kernel/pci_sabre.c +++ b/arch/sparc64/kernel/pci_sabre.c @@ -554,7 +554,7 @@ static unsigned char sabre_pil_table[] = { /*0x32*/15, /* Power Management */ }; -static int __init sabre_ino_to_pil(struct pci_dev *pdev, unsigned int ino) +static int sabre_ino_to_pil(struct pci_dev *pdev, unsigned int ino) { int ret; @@ -612,9 +612,9 @@ static void sabre_wsync_handler(struct ino_bucket *bucket, void *_arg1, void *_a sabre_read(sync_reg); } -static unsigned int __init sabre_irq_build(struct pci_pbm_info *pbm, - struct pci_dev *pdev, - unsigned int ino) +static unsigned int sabre_irq_build(struct pci_pbm_info *pbm, + struct pci_dev *pdev, + unsigned int ino) { struct ino_bucket *bucket; unsigned long imap, iclr; @@ -1009,7 +1009,7 @@ static irqreturn_t sabre_pcierr_intr(int irq, void *dev_id, struct pt_regs *regs #define SABRE_UE_INO 0x2e #define SABRE_CE_INO 0x2f #define SABRE_PCIERR_INO 0x30 -static void __init sabre_register_error_handlers(struct pci_controller_info *p) +static void sabre_register_error_handlers(struct pci_controller_info *p) { struct pci_pbm_info *pbm = &p->pbm_A; /* arbitrary */ unsigned long base = pbm->controller_regs; @@ -1056,9 +1056,9 @@ static void __init sabre_register_error_handlers(struct pci_controller_info *p) sabre_write(base + SABRE_PCICTRL, tmp); } -static void __init sabre_resource_adjust(struct pci_dev *pdev, - struct resource *res, - struct resource *root) +static void sabre_resource_adjust(struct pci_dev *pdev, + struct resource *res, + struct resource *root) { struct pci_pbm_info *pbm = pdev->bus->sysdata; unsigned long base; @@ -1072,7 +1072,7 @@ static void __init sabre_resource_adjust(struct pci_dev *pdev, res->end += base; } -static void __init sabre_base_address_update(struct pci_dev *pdev, int resource) +static void sabre_base_address_update(struct pci_dev *pdev, int resource) { struct pcidev_cookie *pcp = pdev->sysdata; struct pci_pbm_info *pbm = pcp->pbm; @@ -1118,7 +1118,7 @@ static void __init sabre_base_address_update(struct pci_dev *pdev, int resource) pci_write_config_dword(pdev, where + 4, 0); } -static void __init apb_init(struct pci_controller_info *p, struct pci_bus *sabre_bus) +static void apb_init(struct pci_controller_info *p, struct pci_bus *sabre_bus) { struct pci_dev *pdev; @@ -1181,7 +1181,7 @@ static struct pcidev_cookie *alloc_bridge_cookie(struct pci_pbm_info *pbm) return cookie; } -static void __init sabre_scan_bus(struct pci_controller_info *p) +static void sabre_scan_bus(struct pci_controller_info *p) { static int once; struct pci_bus *sabre_bus, *pbus; @@ -1262,18 +1262,14 @@ static void __init sabre_scan_bus(struct pci_controller_info *p) sabre_register_error_handlers(p); } -static void __init sabre_iommu_init(struct pci_controller_info *p, - int tsbsize, unsigned long dvma_offset, - u32 dma_mask) +static void sabre_iommu_init(struct pci_controller_info *p, + int tsbsize, unsigned long dvma_offset, + u32 dma_mask) { struct pci_iommu *iommu = p->pbm_A.iommu; - unsigned long tsbbase, i, order; + unsigned long i; u64 control; - /* Setup initial software IOMMU state. */ - spin_lock_init(&iommu->lock); - iommu->ctx_lowest_free = 1; - /* Register addresses. */ iommu->iommu_control = p->pbm_A.controller_regs + SABRE_IOMMU_CONTROL; iommu->iommu_tsbbase = p->pbm_A.controller_regs + SABRE_IOMMU_TSBBASE; @@ -1295,26 +1291,10 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ + pci_iommu_table_init(iommu, tsbsize * 1024 * 8, dvma_offset, dma_mask); - iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); - if (!iommu->dummy_page) { - prom_printf("PSYCHO_IOMMU: Error, gfp(dummy_page) failed.\n"); - prom_halt(); - } - memset((void *)iommu->dummy_page, 0, PAGE_SIZE); - iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); - - tsbbase = __get_free_pages(GFP_KERNEL, order = get_order(tsbsize * 1024 * 8)); - if (!tsbbase) { - prom_printf("SABRE_IOMMU: Error, gfp(tsb) failed.\n"); - prom_halt(); - } - iommu->page_table = (iopte_t *)tsbbase; - iommu->page_table_map_base = dvma_offset; - iommu->dma_addr_mask = dma_mask; - pci_iommu_table_init(iommu, PAGE_SIZE << order); - - sabre_write(p->pbm_A.controller_regs + SABRE_IOMMU_TSBBASE, __pa(tsbbase)); + sabre_write(p->pbm_A.controller_regs + SABRE_IOMMU_TSBBASE, + __pa(iommu->page_table)); control = sabre_read(p->pbm_A.controller_regs + SABRE_IOMMU_CONTROL); control &= ~(SABRE_IOMMUCTRL_TSBSZ | SABRE_IOMMUCTRL_TBWSZ); @@ -1322,11 +1302,9 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, switch(tsbsize) { case 64: control |= SABRE_IOMMU_TSBSZ_64K; - iommu->page_table_sz_bits = 16; break; case 128: control |= SABRE_IOMMU_TSBSZ_128K; - iommu->page_table_sz_bits = 17; break; default: prom_printf("iommu_init: Illegal TSB size %d\n", tsbsize); @@ -1334,19 +1312,10 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, break; } sabre_write(p->pbm_A.controller_regs + SABRE_IOMMU_CONTROL, control); - - /* We start with no consistent mappings. */ - iommu->lowest_consistent_map = - 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); - - for (i = 0; i < PBM_NCLUSTERS; i++) { - iommu->alloc_info[i].flush = 0; - iommu->alloc_info[i].next = 0; - } } -static void __init pbm_register_toplevel_resources(struct pci_controller_info *p, - struct pci_pbm_info *pbm) +static void pbm_register_toplevel_resources(struct pci_controller_info *p, + struct pci_pbm_info *pbm) { char *name = pbm->name; unsigned long ibase = p->pbm_A.controller_regs + SABRE_IOSPACE; @@ -1415,7 +1384,7 @@ static void __init pbm_register_toplevel_resources(struct pci_controller_info *p &pbm->mem_space); } -static void __init sabre_pbm_init(struct pci_controller_info *p, int sabre_node, u32 dma_begin) +static void sabre_pbm_init(struct pci_controller_info *p, int sabre_node, u32 dma_begin) { struct pci_pbm_info *pbm; char namebuf[128]; @@ -1552,7 +1521,7 @@ static void __init sabre_pbm_init(struct pci_controller_info *p, int sabre_node, } } -void __init sabre_init(int pnode, char *model_name) +void sabre_init(int pnode, char *model_name) { struct linux_prom64_registers pr_regs[2]; struct pci_controller_info *p; diff --git a/arch/sparc64/kernel/pci_schizo.c b/arch/sparc64/kernel/pci_schizo.c index 6a182bb66281..d8c4e0919b4e 100644 --- a/arch/sparc64/kernel/pci_schizo.c +++ b/arch/sparc64/kernel/pci_schizo.c @@ -285,7 +285,7 @@ static unsigned char schizo_pil_table[] = { /*0x3f*/0, /* Reserved for NewLink */ }; -static int __init schizo_ino_to_pil(struct pci_dev *pdev, unsigned int ino) +static int schizo_ino_to_pil(struct pci_dev *pdev, unsigned int ino) { int ret; @@ -330,7 +330,7 @@ static int __init schizo_ino_to_pil(struct pci_dev *pdev, unsigned int ino) static void tomatillo_wsync_handler(struct ino_bucket *bucket, void *_arg1, void *_arg2) { unsigned long sync_reg = (unsigned long) _arg2; - u64 mask = 1 << (__irq_ino(__irq(bucket)) & IMAP_INO); + u64 mask = 1UL << (__irq_ino(__irq(bucket)) & IMAP_INO); u64 val; int limit; @@ -1221,7 +1221,7 @@ static irqreturn_t schizo_safarierr_intr(int irq, void *dev_id, struct pt_regs * * PCI bus units of the same Tomatillo. I still have not really * figured this out... */ -static void __init tomatillo_register_error_handlers(struct pci_controller_info *p) +static void tomatillo_register_error_handlers(struct pci_controller_info *p) { struct pci_pbm_info *pbm; unsigned int irq; @@ -1359,7 +1359,7 @@ static void __init tomatillo_register_error_handlers(struct pci_controller_info (SCHIZO_SAFIRQCTRL_EN | (BUS_ERROR_UNMAP))); } -static void __init schizo_register_error_handlers(struct pci_controller_info *p) +static void schizo_register_error_handlers(struct pci_controller_info *p) { struct pci_pbm_info *pbm; unsigned int irq; @@ -1505,7 +1505,7 @@ static void __init schizo_register_error_handlers(struct pci_controller_info *p) (SCHIZO_SAFIRQCTRL_EN | (BUS_ERROR_UNMAP))); } -static void __init pbm_config_busmastering(struct pci_pbm_info *pbm) +static void pbm_config_busmastering(struct pci_pbm_info *pbm) { u8 *addr; @@ -1522,8 +1522,8 @@ static void __init pbm_config_busmastering(struct pci_pbm_info *pbm) pci_config_write8(addr, 64); } -static void __init pbm_scan_bus(struct pci_controller_info *p, - struct pci_pbm_info *pbm) +static void pbm_scan_bus(struct pci_controller_info *p, + struct pci_pbm_info *pbm) { struct pcidev_cookie *cookie = kmalloc(sizeof(*cookie), GFP_KERNEL); @@ -1550,8 +1550,8 @@ static void __init pbm_scan_bus(struct pci_controller_info *p, pci_setup_busmastering(pbm, pbm->pci_bus); } -static void __init __schizo_scan_bus(struct pci_controller_info *p, - int chip_type) +static void __schizo_scan_bus(struct pci_controller_info *p, + int chip_type) { if (!p->pbm_B.prom_node || !p->pbm_A.prom_node) { printk("PCI: Only one PCI bus module of controller found.\n"); @@ -1577,17 +1577,17 @@ static void __init __schizo_scan_bus(struct pci_controller_info *p, schizo_register_error_handlers(p); } -static void __init schizo_scan_bus(struct pci_controller_info *p) +static void schizo_scan_bus(struct pci_controller_info *p) { __schizo_scan_bus(p, PBM_CHIP_TYPE_SCHIZO); } -static void __init tomatillo_scan_bus(struct pci_controller_info *p) +static void tomatillo_scan_bus(struct pci_controller_info *p) { __schizo_scan_bus(p, PBM_CHIP_TYPE_TOMATILLO); } -static void __init schizo_base_address_update(struct pci_dev *pdev, int resource) +static void schizo_base_address_update(struct pci_dev *pdev, int resource) { struct pcidev_cookie *pcp = pdev->sysdata; struct pci_pbm_info *pbm = pcp->pbm; @@ -1632,9 +1632,9 @@ static void __init schizo_base_address_update(struct pci_dev *pdev, int resource pci_write_config_dword(pdev, where + 4, 0); } -static void __init schizo_resource_adjust(struct pci_dev *pdev, - struct resource *res, - struct resource *root) +static void schizo_resource_adjust(struct pci_dev *pdev, + struct resource *res, + struct resource *root) { res->start += root->start; res->end += root->start; @@ -1702,8 +1702,8 @@ static void schizo_determine_mem_io_space(struct pci_pbm_info *pbm) pbm->mem_space.start); } -static void __init pbm_register_toplevel_resources(struct pci_controller_info *p, - struct pci_pbm_info *pbm) +static void pbm_register_toplevel_resources(struct pci_controller_info *p, + struct pci_pbm_info *pbm) { pbm->io_space.name = pbm->mem_space.name = pbm->name; @@ -1765,7 +1765,7 @@ static void schizo_pbm_strbuf_init(struct pci_pbm_info *pbm) static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) { struct pci_iommu *iommu = pbm->iommu; - unsigned long tsbbase, i, tagbase, database, order; + unsigned long i, tagbase, database; u32 vdma[2], dma_mask; u64 control; int err, tsbsize; @@ -1800,10 +1800,6 @@ static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) prom_halt(); }; - /* Setup initial software IOMMU state. */ - spin_lock_init(&iommu->lock); - iommu->ctx_lowest_free = 1; - /* Register addresses, SCHIZO has iommu ctx flushing. */ iommu->iommu_control = pbm->pbm_regs + SCHIZO_IOMMU_CONTROL; iommu->iommu_tsbbase = pbm->pbm_regs + SCHIZO_IOMMU_TSBBASE; @@ -1832,56 +1828,9 @@ static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) /* Leave diag mode enabled for full-flushing done * in pci_iommu.c */ + pci_iommu_table_init(iommu, tsbsize * 8 * 1024, vdma[0], dma_mask); - iommu->dummy_page = __get_free_pages(GFP_KERNEL, 0); - if (!iommu->dummy_page) { - prom_printf("PSYCHO_IOMMU: Error, gfp(dummy_page) failed.\n"); - prom_halt(); - } - memset((void *)iommu->dummy_page, 0, PAGE_SIZE); - iommu->dummy_page_pa = (unsigned long) __pa(iommu->dummy_page); - - /* Using assumed page size 8K with 128K entries we need 1MB iommu page - * table (128K ioptes * 8 bytes per iopte). This is - * page order 7 on UltraSparc. - */ - order = get_order(tsbsize * 8 * 1024); - tsbbase = __get_free_pages(GFP_KERNEL, order); - if (!tsbbase) { - prom_printf("%s: Error, gfp(tsb) failed.\n", pbm->name); - prom_halt(); - } - - iommu->page_table = (iopte_t *)tsbbase; - iommu->page_table_map_base = vdma[0]; - iommu->dma_addr_mask = dma_mask; - pci_iommu_table_init(iommu, PAGE_SIZE << order); - - switch (tsbsize) { - case 64: - iommu->page_table_sz_bits = 16; - break; - - case 128: - iommu->page_table_sz_bits = 17; - break; - - default: - prom_printf("iommu_init: Illegal TSB size %d\n", tsbsize); - prom_halt(); - break; - }; - - /* We start with no consistent mappings. */ - iommu->lowest_consistent_map = - 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); - - for (i = 0; i < PBM_NCLUSTERS; i++) { - iommu->alloc_info[i].flush = 0; - iommu->alloc_info[i].next = 0; - } - - schizo_write(iommu->iommu_tsbbase, __pa(tsbbase)); + schizo_write(iommu->iommu_tsbbase, __pa(iommu->page_table)); control = schizo_read(iommu->iommu_control); control &= ~(SCHIZO_IOMMU_CTRL_TSBSZ | SCHIZO_IOMMU_CTRL_TBWSZ); @@ -1932,7 +1881,7 @@ static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) #define TOMATILLO_PCI_IOC_TDIAG (0x2250UL) #define TOMATILLO_PCI_IOC_DDIAG (0x2290UL) -static void __init schizo_pbm_hw_init(struct pci_pbm_info *pbm) +static void schizo_pbm_hw_init(struct pci_pbm_info *pbm) { u64 tmp; @@ -1986,9 +1935,9 @@ static void __init schizo_pbm_hw_init(struct pci_pbm_info *pbm) } } -static void __init schizo_pbm_init(struct pci_controller_info *p, - int prom_node, u32 portid, - int chip_type) +static void schizo_pbm_init(struct pci_controller_info *p, + int prom_node, u32 portid, + int chip_type) { struct linux_prom64_registers pr_regs[4]; unsigned int busrange[2]; @@ -2145,7 +2094,7 @@ static inline int portid_compare(u32 x, u32 y, int chip_type) return (x == y); } -static void __init __schizo_init(int node, char *model_name, int chip_type) +static void __schizo_init(int node, char *model_name, int chip_type) { struct pci_controller_info *p; struct pci_iommu *iommu; @@ -2213,17 +2162,17 @@ static void __init __schizo_init(int node, char *model_name, int chip_type) schizo_pbm_init(p, node, portid, chip_type); } -void __init schizo_init(int node, char *model_name) +void schizo_init(int node, char *model_name) { __schizo_init(node, model_name, PBM_CHIP_TYPE_SCHIZO); } -void __init schizo_plus_init(int node, char *model_name) +void schizo_plus_init(int node, char *model_name) { __schizo_init(node, model_name, PBM_CHIP_TYPE_SCHIZO_PLUS); } -void __init tomatillo_init(int node, char *model_name) +void tomatillo_init(int node, char *model_name) { __schizo_init(node, model_name, PBM_CHIP_TYPE_TOMATILLO); } diff --git a/arch/sparc64/kernel/power.c b/arch/sparc64/kernel/power.c index 946cee0257ea..9e8362ea3104 100644 --- a/arch/sparc64/kernel/power.c +++ b/arch/sparc64/kernel/power.c @@ -17,6 +17,7 @@ #include <asm/system.h> #include <asm/ebus.h> +#include <asm/isa.h> #include <asm/auxio.h> #include <linux/unistd.h> @@ -100,46 +101,83 @@ again: return 0; } -static int __init has_button_interrupt(struct linux_ebus_device *edev) +static int __init has_button_interrupt(unsigned int irq, int prom_node) { - if (edev->irqs[0] == PCI_IRQ_NONE) + if (irq == PCI_IRQ_NONE) return 0; - if (!prom_node_has_property(edev->prom_node, "button")) + if (!prom_node_has_property(prom_node, "button")) return 0; return 1; } -void __init power_init(void) +static int __init power_probe_ebus(struct resource **resp, unsigned int *irq_p, int *prom_node_p) { struct linux_ebus *ebus; struct linux_ebus_device *edev; + + for_each_ebus(ebus) { + for_each_ebusdev(edev, ebus) { + if (!strcmp(edev->prom_name, "power")) { + *resp = &edev->resource[0]; + *irq_p = edev->irqs[0]; + *prom_node_p = edev->prom_node; + return 0; + } + } + } + return -ENODEV; +} + +static int __init power_probe_isa(struct resource **resp, unsigned int *irq_p, int *prom_node_p) +{ + struct sparc_isa_bridge *isa_bus; + struct sparc_isa_device *isa_dev; + + for_each_isa(isa_bus) { + for_each_isadev(isa_dev, isa_bus) { + if (!strcmp(isa_dev->prom_name, "power")) { + *resp = &isa_dev->resource; + *irq_p = isa_dev->irq; + *prom_node_p = isa_dev->prom_node; + return 0; + } + } + } + return -ENODEV; +} + +void __init power_init(void) +{ + struct resource *res = NULL; + unsigned int irq; + int prom_node; static int invoked; if (invoked) return; invoked = 1; - for_each_ebus(ebus) { - for_each_ebusdev(edev, ebus) { - if (!strcmp(edev->prom_name, "power")) - goto found; - } - } + if (!power_probe_ebus(&res, &irq, &prom_node)) + goto found; + + if (!power_probe_isa(&res, &irq, &prom_node)) + goto found; + return; found: - power_reg = ioremap(edev->resource[0].start, 0x4); + power_reg = ioremap(res->start, 0x4); printk("power: Control reg at %p ... ", power_reg); poweroff_method = machine_halt; /* able to use the standard halt */ - if (has_button_interrupt(edev)) { + if (has_button_interrupt(irq, prom_node)) { if (kernel_thread(powerd, NULL, CLONE_FS) < 0) { printk("Failed to start power daemon.\n"); return; } printk("powerd running.\n"); - if (request_irq(edev->irqs[0], + if (request_irq(irq, power_handler, SA_SHIRQ, "power", NULL) < 0) printk("power: Error, cannot register IRQ handler.\n"); } else { diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c index 07424b075938..7d10b0397091 100644 --- a/arch/sparc64/kernel/process.c +++ b/arch/sparc64/kernel/process.c @@ -103,7 +103,7 @@ void cpu_idle(void) * other cpus see our increasing idleness for the buddy * redistribution algorithm. -DaveM */ - membar("#StoreStore | #StoreLoad"); + membar_storeload_storestore(); } } @@ -607,11 +607,6 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp, struct thread_info *t = p->thread_info; char *child_trap_frame; -#ifdef CONFIG_DEBUG_SPINLOCK - p->thread.smp_lock_count = 0; - p->thread.smp_lock_pc = 0; -#endif - /* Calculate offset to stack_frame & pt_regs */ child_trap_frame = ((char *)t) + (THREAD_SIZE - (TRACEREG_SZ+STACKFRAME_SZ)); memcpy(child_trap_frame, (((struct sparc_stackf *)regs)-1), (TRACEREG_SZ+STACKFRAME_SZ)); diff --git a/arch/sparc64/kernel/ptrace.c b/arch/sparc64/kernel/ptrace.c index 23ad839d113f..774ecbb8a031 100644 --- a/arch/sparc64/kernel/ptrace.c +++ b/arch/sparc64/kernel/ptrace.c @@ -30,6 +30,8 @@ #include <asm/psrcompat.h> #include <asm/visasm.h> #include <asm/spitfire.h> +#include <asm/page.h> +#include <asm/cpudata.h> /* Returning from ptrace is a bit tricky because the syscall return * low level code assumes any value returned which is negative and @@ -128,20 +130,24 @@ void flush_ptrace_access(struct vm_area_struct *vma, struct page *page, * is mapped to in the user's address space, we can skip the * D-cache flush. */ - if ((uaddr ^ kaddr) & (1UL << 13)) { + if ((uaddr ^ (unsigned long) kaddr) & (1UL << 13)) { unsigned long start = __pa(kaddr); unsigned long end = start + len; + unsigned long dcache_line_size; + + dcache_line_size = local_cpu_data().dcache_line_size; if (tlb_type == spitfire) { - for (; start < end; start += 32) - spitfire_put_dcache_tag(va & 0x3fe0, 0x0); + for (; start < end; start += dcache_line_size) + spitfire_put_dcache_tag(start & 0x3fe0, 0x0); } else { - for (; start < end; start += 32) + start &= ~(dcache_line_size - 1); + for (; start < end; start += dcache_line_size) __asm__ __volatile__( "stxa %%g0, [%0] %1\n\t" "membar #Sync" : /* no outputs */ - : "r" (va), + : "r" (start), "i" (ASI_DCACHE_INVALIDATE)); } } @@ -149,8 +155,11 @@ void flush_ptrace_access(struct vm_area_struct *vma, struct page *page, if (write && tlb_type == spitfire) { unsigned long start = (unsigned long) kaddr; unsigned long end = start + len; + unsigned long icache_line_size; + + icache_line_size = local_cpu_data().icache_line_size; - for (; start < end; start += 32) + for (; start < end; start += icache_line_size) flushi(start); } } diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S index 0696ed4b9d64..090dcca00d2a 100644 --- a/arch/sparc64/kernel/rtrap.S +++ b/arch/sparc64/kernel/rtrap.S @@ -153,11 +153,14 @@ __handle_signal: rtrap_irq: rtrap_clr_l6: clr %l6 rtrap: - ldub [%g6 + TI_CPU], %l0 - sethi %hi(irq_stat), %l2 ! &softirq_active - or %l2, %lo(irq_stat), %l2 ! &softirq_active -irqsz_patchme: sllx %l0, 0, %l0 - lduw [%l2 + %l0], %l1 ! softirq_pending +#ifndef CONFIG_SMP + sethi %hi(per_cpu____cpu_data), %l0 + lduw [%l0 + %lo(per_cpu____cpu_data)], %l1 +#else + sethi %hi(per_cpu____cpu_data), %l0 + or %l0, %lo(per_cpu____cpu_data), %l0 + lduw [%l0 + %g5], %l1 +#endif cmp %l1, 0 /* mm/ultra.S:xcall_report_regs KNOWS about this load. */ @@ -253,9 +256,8 @@ rt_continue: ldx [%sp + PTREGS_OFF + PT_V9_G1], %g1 brnz,pn %l3, kern_rtt mov PRIMARY_CONTEXT, %l7 ldxa [%l7 + %l7] ASI_DMMU, %l0 -cplus_rtrap_insn_1: - sethi %hi(0), %l1 - sllx %l1, 32, %l1 + sethi %hi(sparc64_kern_pri_nuc_bits), %l1 + ldx [%l1 + %lo(sparc64_kern_pri_nuc_bits)], %l1 or %l0, %l1, %l0 stxa %l0, [%l7] ASI_DMMU flush %g6 @@ -310,53 +312,36 @@ kern_fpucheck: ldub [%g6 + TI_FPDEPTH], %l5 wr %g1, FPRS_FEF, %fprs ldx [%o1 + %o5], %g1 add %g6, TI_XFSR, %o1 - membar #StoreLoad | #LoadLoad sll %o0, 8, %o2 add %g6, TI_FPREGS, %o3 brz,pn %l6, 1f add %g6, TI_FPREGS+0x40, %o4 + membar #Sync ldda [%o3 + %o2] ASI_BLK_P, %f0 ldda [%o4 + %o2] ASI_BLK_P, %f16 + membar #Sync 1: andcc %l2, FPRS_DU, %g0 be,pn %icc, 1f wr %g1, 0, %gsr add %o2, 0x80, %o2 + membar #Sync ldda [%o3 + %o2] ASI_BLK_P, %f32 ldda [%o4 + %o2] ASI_BLK_P, %f48 - 1: membar #Sync ldx [%o1 + %o5], %fsr 2: stb %l5, [%g6 + TI_FPDEPTH] ba,pt %xcc, rt_continue nop 5: wr %g0, FPRS_FEF, %fprs - membar #StoreLoad | #LoadLoad sll %o0, 8, %o2 add %g6, TI_FPREGS+0x80, %o3 add %g6, TI_FPREGS+0xc0, %o4 + membar #Sync ldda [%o3 + %o2] ASI_BLK_P, %f32 ldda [%o4 + %o2] ASI_BLK_P, %f48 membar #Sync wr %g0, FPRS_DU, %fprs ba,pt %xcc, rt_continue stb %l5, [%g6 + TI_FPDEPTH] - -cplus_rinsn_1: - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %l1 - - .globl cheetah_plus_patch_rtrap -cheetah_plus_patch_rtrap: - /* We configure the dTLB512_0 for 4MB pages and the - * dTLB512_1 for 8K pages when in context zero. - */ - sethi %hi(cplus_rinsn_1), %o0 - sethi %hi(cplus_rtrap_insn_1), %o2 - lduw [%o0 + %lo(cplus_rinsn_1)], %o1 - or %o2, %lo(cplus_rtrap_insn_1), %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop diff --git a/arch/sparc64/kernel/sbus.c b/arch/sparc64/kernel/sbus.c index 89f5e019f24c..e09ddf927655 100644 --- a/arch/sparc64/kernel/sbus.c +++ b/arch/sparc64/kernel/sbus.c @@ -147,7 +147,7 @@ static void sbus_strbuf_flush(struct sbus_iommu *iommu, u32 base, unsigned long if (!limit) break; udelay(1); - membar("#LoadLoad"); + rmb(); } if (!limit) printk(KERN_WARNING "sbus_strbuf_flush: flushflag timeout " diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c index b7e6a91952b2..c1f34237cdf2 100644 --- a/arch/sparc64/kernel/setup.c +++ b/arch/sparc64/kernel/setup.c @@ -33,7 +33,6 @@ #include <linux/cpu.h> #include <linux/initrd.h> -#include <asm/segment.h> #include <asm/system.h> #include <asm/io.h> #include <asm/processor.h> @@ -188,17 +187,13 @@ int prom_callback(long *args) } if ((va >= KERNBASE) && (va < (KERNBASE + (4 * 1024 * 1024)))) { - unsigned long kernel_pctx = 0; - - if (tlb_type == cheetah_plus) - kernel_pctx |= (CTX_CHEETAH_PLUS_NUC | - CTX_CHEETAH_PLUS_CTX0); + extern unsigned long sparc64_kern_pri_context; /* Spitfire Errata #32 workaround */ __asm__ __volatile__("stxa %0, [%1] %2\n\t" "flush %%g6" : /* No outputs */ - : "r" (kernel_pctx), + : "r" (sparc64_kern_pri_context), "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU)); @@ -465,8 +460,6 @@ static void __init boot_flags_init(char *commands) } } -extern int prom_probe_memory(void); -extern unsigned long start, end; extern void panic_setup(char *, int *); extern unsigned short root_flags; @@ -493,13 +486,8 @@ void register_prom_callbacks(void) "' linux-.soft2 to .soft2"); } -extern void paging_init(void); - void __init setup_arch(char **cmdline_p) { - unsigned long highest_paddr; - int i; - /* Initialize PROM console and command line. */ *cmdline_p = prom_getbootargs(); strcpy(saved_command_line, *cmdline_p); @@ -512,58 +500,12 @@ void __init setup_arch(char **cmdline_p) conswitchp = &prom_con; #endif -#ifdef CONFIG_SMP - i = (unsigned long)&irq_stat[1] - (unsigned long)&irq_stat[0]; - if ((i == SMP_CACHE_BYTES) || (i == (2 * SMP_CACHE_BYTES))) { - extern unsigned int irqsz_patchme[1]; - irqsz_patchme[0] |= ((i == SMP_CACHE_BYTES) ? SMP_CACHE_BYTES_SHIFT : \ - SMP_CACHE_BYTES_SHIFT + 1); - flushi((long)&irqsz_patchme[0]); - } else { - prom_printf("Unexpected size of irq_stat[] elements\n"); - prom_halt(); - } -#endif /* Work out if we are starfire early on */ check_if_starfire(); boot_flags_init(*cmdline_p); idprom_init(); - (void) prom_probe_memory(); - - /* In paging_init() we tip off this value to see if we need - * to change init_mm.pgd to point to the real alias mapping. - */ - phys_base = 0xffffffffffffffffUL; - highest_paddr = 0UL; - for (i = 0; sp_banks[i].num_bytes != 0; i++) { - unsigned long top; - - if (sp_banks[i].base_addr < phys_base) - phys_base = sp_banks[i].base_addr; - top = sp_banks[i].base_addr + - sp_banks[i].num_bytes; - if (highest_paddr < top) - highest_paddr = top; - } - pfn_base = phys_base >> PAGE_SHIFT; - - switch (tlb_type) { - default: - case spitfire: - kern_base = spitfire_get_itlb_data(sparc64_highest_locked_tlbent()); - kern_base &= _PAGE_PADDR_SF; - break; - - case cheetah: - case cheetah_plus: - kern_base = cheetah_get_litlb_data(sparc64_highest_locked_tlbent()); - kern_base &= _PAGE_PADDR; - break; - }; - - kern_size = (unsigned long)&_end - (unsigned long)KERNBASE; if (!root_flags) root_mountflags &= ~MS_RDONLY; @@ -638,6 +580,9 @@ extern void smp_info(struct seq_file *); extern void smp_bogo(struct seq_file *); extern void mmu_info(struct seq_file *); +unsigned int dcache_parity_tl1_occurred; +unsigned int icache_parity_tl1_occurred; + static int show_cpuinfo(struct seq_file *m, void *__unused) { seq_printf(m, @@ -648,6 +593,8 @@ static int show_cpuinfo(struct seq_file *m, void *__unused) "type\t\t: sun4u\n" "ncpus probed\t: %ld\n" "ncpus active\t: %ld\n" + "D$ parity tl1\t: %u\n" + "I$ parity tl1\t: %u\n" #ifndef CONFIG_SMP "Cpu0Bogo\t: %lu.%02lu\n" "Cpu0ClkTck\t: %016lx\n" @@ -660,7 +607,9 @@ static int show_cpuinfo(struct seq_file *m, void *__unused) (prom_prev >> 8) & 0xff, prom_prev & 0xff, (long)num_possible_cpus(), - (long)num_online_cpus() + (long)num_online_cpus(), + dcache_parity_tl1_occurred, + icache_parity_tl1_occurred #ifndef CONFIG_SMP , cpu_data(0).udelay_val/(500000/HZ), (cpu_data(0).udelay_val/(5000/HZ)) % 100, diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c index b27934671c35..60f5dfabb1e1 100644 --- a/arch/sparc64/kernel/signal.c +++ b/arch/sparc64/kernel/signal.c @@ -574,13 +574,12 @@ static inline void handle_signal(unsigned long signr, struct k_sigaction *ka, { setup_rt_frame(ka, regs, signr, oldset, (ka->sa.sa_flags & SA_SIGINFO) ? info : NULL); - if (!(ka->sa.sa_flags & SA_NOMASK)) { - spin_lock_irq(¤t->sighand->siglock); - sigorsets(¤t->blocked,¤t->blocked,&ka->sa.sa_mask); + spin_lock_irq(¤t->sighand->siglock); + sigorsets(¤t->blocked,¤t->blocked,&ka->sa.sa_mask); + if (!(ka->sa.sa_flags & SA_NOMASK)) sigaddset(¤t->blocked,signr); - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - } + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); } static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs, diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c index f28428f4170e..aecccd0df1d1 100644 --- a/arch/sparc64/kernel/signal32.c +++ b/arch/sparc64/kernel/signal32.c @@ -877,11 +877,12 @@ static void new_setup_frame32(struct k_sigaction *ka, struct pt_regs *regs, unsigned long page = (unsigned long) page_address(pte_page(*ptep)); - __asm__ __volatile__( - " membar #StoreStore\n" - " flush %0 + %1" - : : "r" (page), "r" (address & (PAGE_SIZE - 1)) - : "memory"); + wmb(); + __asm__ __volatile__("flush %0 + %1" + : /* no outputs */ + : "r" (page), + "r" (address & (PAGE_SIZE - 1)) + : "memory"); } pte_unmap(ptep); preempt_enable(); @@ -1292,11 +1293,12 @@ static void setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs, unsigned long page = (unsigned long) page_address(pte_page(*ptep)); - __asm__ __volatile__( - " membar #StoreStore\n" - " flush %0 + %1" - : : "r" (page), "r" (address & (PAGE_SIZE - 1)) - : "memory"); + wmb(); + __asm__ __volatile__("flush %0 + %1" + : /* no outputs */ + : "r" (page), + "r" (address & (PAGE_SIZE - 1)) + : "memory"); } pte_unmap(ptep); preempt_enable(); @@ -1325,13 +1327,12 @@ static inline void handle_signal32(unsigned long signr, struct k_sigaction *ka, else setup_frame32(&ka->sa, regs, signr, oldset, info); } - if (!(ka->sa.sa_flags & SA_NOMASK)) { - spin_lock_irq(¤t->sighand->siglock); - sigorsets(¤t->blocked,¤t->blocked,&ka->sa.sa_mask); + spin_lock_irq(¤t->sighand->siglock); + sigorsets(¤t->blocked,¤t->blocked,&ka->sa.sa_mask); + if (!(ka->sa.sa_flags & SA_NOMASK)) sigaddset(¤t->blocked,signr); - recalc_sigpending(); - spin_unlock_irq(¤t->sighand->siglock); - } + recalc_sigpending(); + spin_unlock_irq(¤t->sighand->siglock); } static inline void syscall_restart32(unsigned long orig_i0, struct pt_regs *regs, diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index b9b42491e118..b137fd63f5e1 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -93,6 +93,27 @@ void __init smp_store_cpu_info(int id) cpu_data(id).pte_cache[1] = NULL; cpu_data(id).pgd_cache = NULL; cpu_data(id).idle_volume = 1; + + cpu_data(id).dcache_size = prom_getintdefault(cpu_node, "dcache-size", + 16 * 1024); + cpu_data(id).dcache_line_size = + prom_getintdefault(cpu_node, "dcache-line-size", 32); + cpu_data(id).icache_size = prom_getintdefault(cpu_node, "icache-size", + 16 * 1024); + cpu_data(id).icache_line_size = + prom_getintdefault(cpu_node, "icache-line-size", 32); + cpu_data(id).ecache_size = prom_getintdefault(cpu_node, "ecache-size", + 4 * 1024 * 1024); + cpu_data(id).ecache_line_size = + prom_getintdefault(cpu_node, "ecache-line-size", 64); + printk("CPU[%d]: Caches " + "D[sz(%d):line_sz(%d)] " + "I[sz(%d):line_sz(%d)] " + "E[sz(%d):line_sz(%d)]\n", + id, + cpu_data(id).dcache_size, cpu_data(id).dcache_line_size, + cpu_data(id).icache_size, cpu_data(id).icache_line_size, + cpu_data(id).ecache_size, cpu_data(id).ecache_line_size); } static void smp_setup_percpu_timer(void); @@ -144,7 +165,7 @@ void __init smp_callin(void) current->active_mm = &init_mm; while (!cpu_isset(cpuid, smp_commenced_mask)) - membar("#LoadLoad"); + rmb(); cpu_set(cpuid, cpu_online_map); } @@ -184,11 +205,11 @@ static inline long get_delta (long *rt, long *master) for (i = 0; i < NUM_ITERS; i++) { t0 = tick_ops->get_tick(); go[MASTER] = 1; - membar("#StoreLoad"); + membar_storeload(); while (!(tm = go[SLAVE])) - membar("#LoadLoad"); + rmb(); go[SLAVE] = 0; - membar("#StoreStore"); + wmb(); t1 = tick_ops->get_tick(); if (t1 - t0 < best_t1 - best_t0) @@ -221,7 +242,7 @@ void smp_synchronize_tick_client(void) go[MASTER] = 1; while (go[MASTER]) - membar("#LoadLoad"); + rmb(); local_irq_save(flags); { @@ -273,21 +294,21 @@ static void smp_synchronize_one_tick(int cpu) /* wait for client to be ready */ while (!go[MASTER]) - membar("#LoadLoad"); + rmb(); /* now let the client proceed into his loop */ go[MASTER] = 0; - membar("#StoreLoad"); + membar_storeload(); spin_lock_irqsave(&itc_sync_lock, flags); { for (i = 0; i < NUM_ROUNDS*NUM_ITERS; i++) { while (!go[MASTER]) - membar("#LoadLoad"); + rmb(); go[MASTER] = 0; - membar("#StoreStore"); + wmb(); go[SLAVE] = tick_ops->get_tick(); - membar("#StoreLoad"); + membar_storeload(); } } spin_unlock_irqrestore(&itc_sync_lock, flags); @@ -927,11 +948,11 @@ void smp_capture(void) smp_processor_id()); #endif penguins_are_doing_time = 1; - membar("#StoreStore | #LoadStore"); + membar_storestore_loadstore(); atomic_inc(&smp_capture_registry); smp_cross_call(&xcall_capture, 0, 0, 0); while (atomic_read(&smp_capture_registry) != ncpus) - membar("#LoadLoad"); + rmb(); #ifdef CAPTURE_DEBUG printk("done\n"); #endif @@ -947,7 +968,7 @@ void smp_release(void) smp_processor_id()); #endif penguins_are_doing_time = 0; - membar("#StoreStore | #StoreLoad"); + membar_storeload_storestore(); atomic_dec(&smp_capture_registry); } } @@ -970,9 +991,9 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs) save_alternate_globals(global_save); prom_world(1); atomic_inc(&smp_capture_registry); - membar("#StoreLoad | #StoreStore"); + membar_storeload_storestore(); while (penguins_are_doing_time) - membar("#LoadLoad"); + rmb(); restore_alternate_globals(global_save); atomic_dec(&smp_capture_registry); prom_world(0); @@ -980,13 +1001,6 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs) preempt_enable(); } -extern unsigned long xcall_promstop; - -void smp_promstop_others(void) -{ - smp_cross_call(&xcall_promstop, 0, 0, 0); -} - #define prof_multiplier(__cpu) cpu_data(__cpu).multiplier #define prof_counter(__cpu) cpu_data(__cpu).counter diff --git a/arch/sparc64/kernel/sparc64_ksyms.c b/arch/sparc64/kernel/sparc64_ksyms.c index 9202d925a9ce..fb7a5370dbfc 100644 --- a/arch/sparc64/kernel/sparc64_ksyms.c +++ b/arch/sparc64/kernel/sparc64_ksyms.c @@ -88,8 +88,6 @@ extern int svr4_setcontext(svr4_ucontext_t *uc, struct pt_regs *regs); extern int compat_sys_ioctl(unsigned int fd, unsigned int cmd, u32 arg); extern int (*handle_mathemu)(struct pt_regs *, struct fpustate *); extern long sparc32_open(const char __user * filename, int flags, int mode); -extern int io_remap_page_range(struct vm_area_struct *vma, unsigned long from, - unsigned long offset, unsigned long size, pgprot_t prot, int space); extern int io_remap_pfn_range(struct vm_area_struct *vma, unsigned long from, unsigned long pfn, unsigned long size, pgprot_t prot); extern void (*prom_palette)(int); @@ -99,17 +97,6 @@ extern int __ashrdi3(int, int); extern void dump_thread(struct pt_regs *, struct user *); extern int dump_fpu (struct pt_regs * regs, elf_fpregset_t * fpregs); -#if defined(CONFIG_SMP) && defined(CONFIG_DEBUG_SPINLOCK) -extern void _do_spin_lock (spinlock_t *lock, char *str); -extern void _do_spin_unlock (spinlock_t *lock); -extern int _spin_trylock (spinlock_t *lock); -extern void _do_read_lock(rwlock_t *rw, char *str); -extern void _do_read_unlock(rwlock_t *rw, char *str); -extern void _do_write_lock(rwlock_t *rw, char *str); -extern void _do_write_unlock(rwlock_t *rw); -extern int _do_write_trylock(rwlock_t *rw, char *str); -#endif - extern unsigned long phys_base; extern unsigned long pfn_base; @@ -128,17 +115,12 @@ EXPORT_PER_CPU_SYMBOL(__cpu_data); /* used by various drivers */ #ifdef CONFIG_SMP -#ifndef CONFIG_DEBUG_SPINLOCK /* Out of line rw-locking implementation. */ EXPORT_SYMBOL(__read_lock); EXPORT_SYMBOL(__read_unlock); EXPORT_SYMBOL(__write_lock); EXPORT_SYMBOL(__write_unlock); EXPORT_SYMBOL(__write_trylock); -/* Out of line spin-locking implementation. */ -EXPORT_SYMBOL(_raw_spin_lock); -EXPORT_SYMBOL(_raw_spin_lock_flags); -#endif /* Hard IRQ locking */ EXPORT_SYMBOL(synchronize_irq); @@ -152,18 +134,6 @@ EXPORT_SYMBOL(_mcount); EXPORT_SYMBOL(cpu_online_map); EXPORT_SYMBOL(phys_cpu_present_map); -/* Spinlock debugging library, optional. */ -#ifdef CONFIG_DEBUG_SPINLOCK -EXPORT_SYMBOL(_do_spin_lock); -EXPORT_SYMBOL(_do_spin_unlock); -EXPORT_SYMBOL(_spin_trylock); -EXPORT_SYMBOL(_do_read_lock); -EXPORT_SYMBOL(_do_read_unlock); -EXPORT_SYMBOL(_do_write_lock); -EXPORT_SYMBOL(_do_write_unlock); -EXPORT_SYMBOL(_do_write_trylock); -#endif - EXPORT_SYMBOL(smp_call_function); #endif /* CONFIG_SMP */ @@ -193,9 +163,6 @@ EXPORT_SYMBOL(atomic64_add); EXPORT_SYMBOL(atomic64_add_ret); EXPORT_SYMBOL(atomic64_sub); EXPORT_SYMBOL(atomic64_sub_ret); -#ifdef CONFIG_SMP -EXPORT_SYMBOL(_atomic_dec_and_lock); -#endif /* Atomic bit operations. */ EXPORT_SYMBOL(test_and_set_bit); @@ -268,7 +235,6 @@ EXPORT_SYMBOL(pci_dma_supported); #endif /* I/O device mmaping on Sparc64. */ -EXPORT_SYMBOL(io_remap_page_range); EXPORT_SYMBOL(io_remap_pfn_range); /* Solaris/SunOS binary compatibility */ diff --git a/arch/sparc64/kernel/sunos_ioctl32.c b/arch/sparc64/kernel/sunos_ioctl32.c index 87c1aeb02220..7654b8a7f03a 100644 --- a/arch/sparc64/kernel/sunos_ioctl32.c +++ b/arch/sparc64/kernel/sunos_ioctl32.c @@ -152,11 +152,12 @@ asmlinkage int sunos_ioctl (int fd, u32 cmd, u32 arg) ret = compat_sys_ioctl(fd, SIOCGIFCONF, arg); goto out; - case _IOW('i', 21, struct ifreq): /* SIOCSIFMTU */ - ret = sys_ioctl(fd, SIOCSIFMTU, arg); + case _IOW('i', 21, struct ifreq32): + ret = compat_sys_ioctl(fd, SIOCSIFMTU, arg); goto out; - case _IOWR('i', 22, struct ifreq): /* SIOCGIFMTU */ - ret = sys_ioctl(fd, SIOCGIFMTU, arg); + + case _IOWR('i', 22, struct ifreq32): + ret = compat_sys_ioctl(fd, SIOCGIFMTU, arg); goto out; case _IOWR('i', 23, struct ifreq32): diff --git a/arch/sparc64/kernel/sys32.S b/arch/sparc64/kernel/sys32.S index 5f9e4fae612e..9cd272ac3ac1 100644 --- a/arch/sparc64/kernel/sys32.S +++ b/arch/sparc64/kernel/sys32.S @@ -157,173 +157,199 @@ sys32_socketcall: /* %o0=call, %o1=args */ or %g2, %lo(__socketcall_table_begin), %g2 jmpl %g2 + %o0, %g0 nop +do_einval: + retl + mov -EINVAL, %o0 - /* Each entry is exactly 32 bytes. */ .align 32 __socketcall_table_begin: + + /* Each entry is exactly 32 bytes. */ do_sys_socket: /* sys_socket(int, int, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +1: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_socket), %g1 - ldswa [%o1 + 0x8] %asi, %o2 +2: ldswa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_socket), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +3: ldswa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_bind: /* sys_bind(int fd, struct sockaddr *, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +4: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_bind), %g1 - ldswa [%o1 + 0x8] %asi, %o2 +5: ldswa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_bind), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +6: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_connect: /* sys_connect(int, struct sockaddr *, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +7: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_connect), %g1 - ldswa [%o1 + 0x8] %asi, %o2 +8: ldswa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_connect), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +9: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_listen: /* sys_listen(int, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +10: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_listen), %g1 jmpl %g1 + %lo(sys_listen), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +11: ldswa [%o1 + 0x4] %asi, %o1 nop nop nop nop do_sys_accept: /* sys_accept(int, struct sockaddr *, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +12: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_accept), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +13: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_accept), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +14: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_getsockname: /* sys_getsockname(int, struct sockaddr *, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +15: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_getsockname), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +16: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_getsockname), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +17: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_getpeername: /* sys_getpeername(int, struct sockaddr *, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +18: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_getpeername), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +19: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(sys_getpeername), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +20: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_socketpair: /* sys_socketpair(int, int, int, int *) */ - ldswa [%o1 + 0x0] %asi, %o0 +21: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_socketpair), %g1 - ldswa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 +22: ldswa [%o1 + 0x8] %asi, %o2 +23: lduwa [%o1 + 0xc] %asi, %o3 jmpl %g1 + %lo(sys_socketpair), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +24: ldswa [%o1 + 0x4] %asi, %o1 nop nop do_sys_send: /* sys_send(int, void *, size_t, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +25: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_send), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 +26: lduwa [%o1 + 0x8] %asi, %o2 +27: lduwa [%o1 + 0xc] %asi, %o3 jmpl %g1 + %lo(sys_send), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +28: lduwa [%o1 + 0x4] %asi, %o1 nop nop do_sys_recv: /* sys_recv(int, void *, size_t, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +29: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_recv), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 +30: lduwa [%o1 + 0x8] %asi, %o2 +31: lduwa [%o1 + 0xc] %asi, %o3 jmpl %g1 + %lo(sys_recv), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +32: lduwa [%o1 + 0x4] %asi, %o1 nop nop do_sys_sendto: /* sys_sendto(int, u32, compat_size_t, unsigned int, u32, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +33: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_sendto), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - lduwa [%o1 + 0x10] %asi, %o4 - ldswa [%o1 + 0x14] %asi, %o5 +34: lduwa [%o1 + 0x8] %asi, %o2 +35: lduwa [%o1 + 0xc] %asi, %o3 +36: lduwa [%o1 + 0x10] %asi, %o4 +37: ldswa [%o1 + 0x14] %asi, %o5 jmpl %g1 + %lo(sys_sendto), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +38: lduwa [%o1 + 0x4] %asi, %o1 do_sys_recvfrom: /* sys_recvfrom(int, u32, compat_size_t, unsigned int, u32, u32) */ - ldswa [%o1 + 0x0] %asi, %o0 +39: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_recvfrom), %g1 - lduwa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - lduwa [%o1 + 0x10] %asi, %o4 - lduwa [%o1 + 0x14] %asi, %o5 +40: lduwa [%o1 + 0x8] %asi, %o2 +41: lduwa [%o1 + 0xc] %asi, %o3 +42: lduwa [%o1 + 0x10] %asi, %o4 +43: lduwa [%o1 + 0x14] %asi, %o5 jmpl %g1 + %lo(sys_recvfrom), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +44: lduwa [%o1 + 0x4] %asi, %o1 do_sys_shutdown: /* sys_shutdown(int, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +45: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(sys_shutdown), %g1 jmpl %g1 + %lo(sys_shutdown), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +46: ldswa [%o1 + 0x4] %asi, %o1 nop nop nop nop do_sys_setsockopt: /* compat_sys_setsockopt(int, int, int, char *, int) */ - ldswa [%o1 + 0x0] %asi, %o0 +47: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_setsockopt), %g1 - ldswa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - ldswa [%o1 + 0x10] %asi, %o4 +48: ldswa [%o1 + 0x8] %asi, %o2 +49: lduwa [%o1 + 0xc] %asi, %o3 +50: ldswa [%o1 + 0x10] %asi, %o4 jmpl %g1 + %lo(compat_sys_setsockopt), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +51: ldswa [%o1 + 0x4] %asi, %o1 nop do_sys_getsockopt: /* compat_sys_getsockopt(int, int, int, u32, u32) */ - ldswa [%o1 + 0x0] %asi, %o0 +52: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_getsockopt), %g1 - ldswa [%o1 + 0x8] %asi, %o2 - lduwa [%o1 + 0xc] %asi, %o3 - lduwa [%o1 + 0x10] %asi, %o4 +53: ldswa [%o1 + 0x8] %asi, %o2 +54: lduwa [%o1 + 0xc] %asi, %o3 +55: lduwa [%o1 + 0x10] %asi, %o4 jmpl %g1 + %lo(compat_sys_getsockopt), %g0 - ldswa [%o1 + 0x4] %asi, %o1 +56: ldswa [%o1 + 0x4] %asi, %o1 nop do_sys_sendmsg: /* compat_sys_sendmsg(int, struct compat_msghdr *, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +57: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_sendmsg), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +58: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(compat_sys_sendmsg), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +59: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop do_sys_recvmsg: /* compat_sys_recvmsg(int, struct compat_msghdr *, unsigned int) */ - ldswa [%o1 + 0x0] %asi, %o0 +60: ldswa [%o1 + 0x0] %asi, %o0 sethi %hi(compat_sys_recvmsg), %g1 - lduwa [%o1 + 0x8] %asi, %o2 +61: lduwa [%o1 + 0x8] %asi, %o2 jmpl %g1 + %lo(compat_sys_recvmsg), %g0 - lduwa [%o1 + 0x4] %asi, %o1 +62: lduwa [%o1 + 0x4] %asi, %o1 nop nop nop -__socketcall_table_end: - -do_einval: - retl - mov -EINVAL, %o0 -do_efault: - retl - mov -EFAULT, %o0 .section __ex_table .align 4 - .word __socketcall_table_begin, 0, __socketcall_table_end, do_efault + .word 1b, __retl_efault, 2b, __retl_efault + .word 3b, __retl_efault, 4b, __retl_efault + .word 5b, __retl_efault, 6b, __retl_efault + .word 7b, __retl_efault, 8b, __retl_efault + .word 9b, __retl_efault, 10b, __retl_efault + .word 11b, __retl_efault, 12b, __retl_efault + .word 13b, __retl_efault, 14b, __retl_efault + .word 15b, __retl_efault, 16b, __retl_efault + .word 17b, __retl_efault, 18b, __retl_efault + .word 19b, __retl_efault, 20b, __retl_efault + .word 21b, __retl_efault, 22b, __retl_efault + .word 23b, __retl_efault, 24b, __retl_efault + .word 25b, __retl_efault, 26b, __retl_efault + .word 27b, __retl_efault, 28b, __retl_efault + .word 29b, __retl_efault, 30b, __retl_efault + .word 31b, __retl_efault, 32b, __retl_efault + .word 33b, __retl_efault, 34b, __retl_efault + .word 35b, __retl_efault, 36b, __retl_efault + .word 37b, __retl_efault, 38b, __retl_efault + .word 39b, __retl_efault, 40b, __retl_efault + .word 41b, __retl_efault, 42b, __retl_efault + .word 43b, __retl_efault, 44b, __retl_efault + .word 45b, __retl_efault, 46b, __retl_efault + .word 47b, __retl_efault, 48b, __retl_efault + .word 49b, __retl_efault, 50b, __retl_efault + .word 51b, __retl_efault, 52b, __retl_efault + .word 53b, __retl_efault, 54b, __retl_efault + .word 55b, __retl_efault, 56b, __retl_efault + .word 57b, __retl_efault, 58b, __retl_efault + .word 59b, __retl_efault, 60b, __retl_efault + .word 61b, __retl_efault, 62b, __retl_efault .previous diff --git a/arch/sparc64/kernel/sys_sparc32.c b/arch/sparc64/kernel/sys_sparc32.c index 1d3aa588df8a..7f6239ed2521 100644 --- a/arch/sparc64/kernel/sys_sparc32.c +++ b/arch/sparc64/kernel/sys_sparc32.c @@ -1002,29 +1002,7 @@ asmlinkage long sys32_adjtimex(struct timex32 __user *utp) asmlinkage long sparc32_open(const char __user *filename, int flags, int mode) { - char * tmp; - int fd, error; - - tmp = getname(filename); - fd = PTR_ERR(tmp); - if (!IS_ERR(tmp)) { - fd = get_unused_fd(); - if (fd >= 0) { - struct file * f = filp_open(tmp, flags, mode); - error = PTR_ERR(f); - if (IS_ERR(f)) - goto out_error; - fd_install(fd, f); - } -out: - putname(tmp); - } - return fd; - -out_error: - put_unused_fd(fd); - fd = error; - goto out; + return do_sys_open(filename, flags, mode); } extern unsigned long do_mremap(unsigned long addr, diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c index 362b9c26871b..38c5525087a2 100644 --- a/arch/sparc64/kernel/time.c +++ b/arch/sparc64/kernel/time.c @@ -55,10 +55,6 @@ unsigned long ds1287_regs = 0UL; extern unsigned long wall_jiffies; -u64 jiffies_64 = INITIAL_JIFFIES; - -EXPORT_SYMBOL(jiffies_64); - static void __iomem *mstk48t08_regs; static void __iomem *mstk48t59_regs; @@ -449,7 +445,7 @@ static inline void timer_check_rtc(void) static long last_rtc_update; /* Determine when to update the Mostek clock. */ - if ((time_status & STA_UNSYNC) == 0 && + if (ntp_synced() && xtime.tv_sec > last_rtc_update + 660 && (xtime.tv_nsec / 1000) >= 500000 - ((unsigned) TICK_SIZE) / 2 && (xtime.tv_nsec / 1000) <= 500000 + ((unsigned) TICK_SIZE) / 2) { diff --git a/arch/sparc64/kernel/trampoline.S b/arch/sparc64/kernel/trampoline.S index 3a145fc39cf2..9478551cb020 100644 --- a/arch/sparc64/kernel/trampoline.S +++ b/arch/sparc64/kernel/trampoline.S @@ -119,8 +119,8 @@ startup_continue: sethi %hi(itlb_load), %g2 or %g2, %lo(itlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] @@ -156,8 +156,8 @@ startup_continue: sethi %hi(itlb_load), %g2 or %g2, %lo(itlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE + 0x400000), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] @@ -190,8 +190,8 @@ do_dtlb: sethi %hi(dtlb_load), %g2 or %g2, %lo(dtlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] @@ -228,8 +228,8 @@ do_dtlb: sethi %hi(dtlb_load), %g2 or %g2, %lo(dtlb_load), %g2 stx %g2, [%sp + 2047 + 128 + 0x18] - sethi %hi(mmu_ihandle_cache), %g2 - lduw [%g2 + %lo(mmu_ihandle_cache)], %g2 + sethi %hi(prom_mmu_ihandle_cache), %g2 + lduw [%g2 + %lo(prom_mmu_ihandle_cache)], %g2 stx %g2, [%sp + 2047 + 128 + 0x20] sethi %hi(KERNBASE + 0x400000), %g2 stx %g2, [%sp + 2047 + 128 + 0x28] @@ -336,20 +336,13 @@ do_unlock: call init_irqwork_curcpu nop - BRANCH_IF_CHEETAH_PLUS_OR_FOLLOWON(g2,g3,1f) - ba,pt %xcc, 2f - nop - -1: /* Start using proper page size encodings in ctx register. */ - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %g3 + /* Start using proper page size encodings in ctx register. */ + sethi %hi(sparc64_kern_pri_context), %g3 + ldx [%g3 + %lo(sparc64_kern_pri_context)], %g2 mov PRIMARY_CONTEXT, %g1 - sllx %g3, 32, %g3 - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - or %g3, %g2, %g3 - stxa %g3, [%g1] ASI_DMMU + stxa %g2, [%g1] ASI_DMMU membar #Sync -2: rdpr %pstate, %o1 or %o1, PSTATE_IE, %o1 wrpr %o1, 0, %pstate diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c index 100b0107c4be..5570e7bb22bb 100644 --- a/arch/sparc64/kernel/traps.c +++ b/arch/sparc64/kernel/traps.c @@ -33,6 +33,7 @@ #include <asm/dcu.h> #include <asm/estate.h> #include <asm/chafsr.h> +#include <asm/sfafsr.h> #include <asm/psrcompat.h> #include <asm/processor.h> #include <asm/timer.h> @@ -143,8 +144,7 @@ void do_BUG(const char *file, int line) } #endif -void instruction_access_exception(struct pt_regs *regs, - unsigned long sfsr, unsigned long sfar) +void spitfire_insn_access_exception(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar) { siginfo_t info; @@ -153,8 +153,8 @@ void instruction_access_exception(struct pt_regs *regs, return; if (regs->tstate & TSTATE_PRIV) { - printk("instruction_access_exception: SFSR[%016lx] SFAR[%016lx], going.\n", - sfsr, sfar); + printk("spitfire_insn_access_exception: SFSR[%016lx] " + "SFAR[%016lx], going.\n", sfsr, sfar); die_if_kernel("Iax", regs); } if (test_thread_flag(TIF_32BIT)) { @@ -169,19 +169,17 @@ void instruction_access_exception(struct pt_regs *regs, force_sig_info(SIGSEGV, &info, current); } -void instruction_access_exception_tl1(struct pt_regs *regs, - unsigned long sfsr, unsigned long sfar) +void spitfire_insn_access_exception_tl1(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar) { if (notify_die(DIE_TRAP_TL1, "instruction access exception tl1", regs, 0, 0x8, SIGTRAP) == NOTIFY_STOP) return; dump_tl1_traplog((struct tl1_traplog *)(regs + 1)); - instruction_access_exception(regs, sfsr, sfar); + spitfire_insn_access_exception(regs, sfsr, sfar); } -void data_access_exception(struct pt_regs *regs, - unsigned long sfsr, unsigned long sfar) +void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar) { siginfo_t info; @@ -191,24 +189,23 @@ void data_access_exception(struct pt_regs *regs, if (regs->tstate & TSTATE_PRIV) { /* Test if this comes from uaccess places. */ - unsigned long fixup; - unsigned long g2 = regs->u_regs[UREG_G2]; + const struct exception_table_entry *entry; - if ((fixup = search_extables_range(regs->tpc, &g2))) { - /* Ouch, somebody is trying ugly VM hole tricks on us... */ + entry = search_exception_tables(regs->tpc); + if (entry) { + /* Ouch, somebody is trying VM hole tricks on us... */ #ifdef DEBUG_EXCEPTIONS printk("Exception: PC<%016lx> faddr<UNKNOWN>\n", regs->tpc); - printk("EX_TABLE: insn<%016lx> fixup<%016lx> " - "g2<%016lx>\n", regs->tpc, fixup, g2); + printk("EX_TABLE: insn<%016lx> fixup<%016lx>\n", + regs->tpc, entry->fixup); #endif - regs->tpc = fixup; + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs[UREG_G2] = g2; return; } /* Shit... */ - printk("data_access_exception: SFSR[%016lx] SFAR[%016lx], going.\n", - sfsr, sfar); + printk("spitfire_data_access_exception: SFSR[%016lx] " + "SFAR[%016lx], going.\n", sfsr, sfar); die_if_kernel("Dax", regs); } @@ -220,6 +217,16 @@ void data_access_exception(struct pt_regs *regs, force_sig_info(SIGSEGV, &info, current); } +void spitfire_data_access_exception_tl1(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar) +{ + if (notify_die(DIE_TRAP_TL1, "data access exception tl1", regs, + 0, 0x30, SIGTRAP) == NOTIFY_STOP) + return; + + dump_tl1_traplog((struct tl1_traplog *)(regs + 1)); + spitfire_data_access_exception(regs, sfsr, sfar); +} + #ifdef CONFIG_PCI /* This is really pathetic... */ extern volatile int pci_poke_in_progress; @@ -253,54 +260,13 @@ static void spitfire_clean_and_reenable_l1_caches(void) : "memory"); } -void do_iae(struct pt_regs *regs) +static void spitfire_enable_estate_errors(void) { - siginfo_t info; - - spitfire_clean_and_reenable_l1_caches(); - - if (notify_die(DIE_TRAP, "instruction access exception", regs, - 0, 0x8, SIGTRAP) == NOTIFY_STOP) - return; - - info.si_signo = SIGBUS; - info.si_errno = 0; - info.si_code = BUS_OBJERR; - info.si_addr = (void *)0; - info.si_trapno = 0; - force_sig_info(SIGBUS, &info, current); -} - -void do_dae(struct pt_regs *regs) -{ - siginfo_t info; - -#ifdef CONFIG_PCI - if (pci_poke_in_progress && pci_poke_cpu == smp_processor_id()) { - spitfire_clean_and_reenable_l1_caches(); - - pci_poke_faulted = 1; - - /* Why the fuck did they have to change this? */ - if (tlb_type == cheetah || tlb_type == cheetah_plus) - regs->tpc += 4; - - regs->tnpc = regs->tpc + 4; - return; - } -#endif - spitfire_clean_and_reenable_l1_caches(); - - if (notify_die(DIE_TRAP, "data access exception", regs, - 0, 0x30, SIGTRAP) == NOTIFY_STOP) - return; - - info.si_signo = SIGBUS; - info.si_errno = 0; - info.si_code = BUS_OBJERR; - info.si_addr = (void *)0; - info.si_trapno = 0; - force_sig_info(SIGBUS, &info, current); + __asm__ __volatile__("stxa %0, [%%g0] %1\n\t" + "membar #Sync" + : /* no outputs */ + : "r" (ESTATE_ERR_ALL), + "i" (ASI_ESTATE_ERROR_EN)); } static char ecc_syndrome_table[] = { @@ -338,65 +304,15 @@ static char ecc_syndrome_table[] = { 0x0b, 0x48, 0x48, 0x4b, 0x48, 0x4b, 0x4b, 0x4a }; -/* cee_trap in entry.S encodes AFSR/UDBH/UDBL error status - * in the following format. The AFAR is left as is, with - * reserved bits cleared, and is a raw 40-bit physical - * address. - */ -#define CE_STATUS_UDBH_UE (1UL << (43 + 9)) -#define CE_STATUS_UDBH_CE (1UL << (43 + 8)) -#define CE_STATUS_UDBH_ESYNDR (0xffUL << 43) -#define CE_STATUS_UDBH_SHIFT 43 -#define CE_STATUS_UDBL_UE (1UL << (33 + 9)) -#define CE_STATUS_UDBL_CE (1UL << (33 + 8)) -#define CE_STATUS_UDBL_ESYNDR (0xffUL << 33) -#define CE_STATUS_UDBL_SHIFT 33 -#define CE_STATUS_AFSR_MASK (0x1ffffffffUL) -#define CE_STATUS_AFSR_ME (1UL << 32) -#define CE_STATUS_AFSR_PRIV (1UL << 31) -#define CE_STATUS_AFSR_ISAP (1UL << 30) -#define CE_STATUS_AFSR_ETP (1UL << 29) -#define CE_STATUS_AFSR_IVUE (1UL << 28) -#define CE_STATUS_AFSR_TO (1UL << 27) -#define CE_STATUS_AFSR_BERR (1UL << 26) -#define CE_STATUS_AFSR_LDP (1UL << 25) -#define CE_STATUS_AFSR_CP (1UL << 24) -#define CE_STATUS_AFSR_WP (1UL << 23) -#define CE_STATUS_AFSR_EDP (1UL << 22) -#define CE_STATUS_AFSR_UE (1UL << 21) -#define CE_STATUS_AFSR_CE (1UL << 20) -#define CE_STATUS_AFSR_ETS (0xfUL << 16) -#define CE_STATUS_AFSR_ETS_SHIFT 16 -#define CE_STATUS_AFSR_PSYND (0xffffUL << 0) -#define CE_STATUS_AFSR_PSYND_SHIFT 0 - -/* Layout of Ecache TAG Parity Syndrome of AFSR */ -#define AFSR_ETSYNDROME_7_0 0x1UL /* E$-tag bus bits <7:0> */ -#define AFSR_ETSYNDROME_15_8 0x2UL /* E$-tag bus bits <15:8> */ -#define AFSR_ETSYNDROME_21_16 0x4UL /* E$-tag bus bits <21:16> */ -#define AFSR_ETSYNDROME_24_22 0x8UL /* E$-tag bus bits <24:22> */ - static char *syndrome_unknown = "<Unknown>"; -asmlinkage void cee_log(unsigned long ce_status, - unsigned long afar, - struct pt_regs *regs) +static void spitfire_log_udb_syndrome(unsigned long afar, unsigned long udbh, unsigned long udbl, unsigned long bit) { - char memmod_str[64]; - char *p; - unsigned short scode, udb_reg; + unsigned short scode; + char memmod_str[64], *p; - printk(KERN_WARNING "CPU[%d]: Correctable ECC Error " - "AFSR[%lx] AFAR[%016lx] UDBL[%lx] UDBH[%lx]\n", - smp_processor_id(), - (ce_status & CE_STATUS_AFSR_MASK), - afar, - ((ce_status >> CE_STATUS_UDBL_SHIFT) & 0x3ffUL), - ((ce_status >> CE_STATUS_UDBH_SHIFT) & 0x3ffUL)); - - udb_reg = ((ce_status >> CE_STATUS_UDBL_SHIFT) & 0x3ffUL); - if (udb_reg & (1 << 8)) { - scode = ecc_syndrome_table[udb_reg & 0xff]; + if (udbl & bit) { + scode = ecc_syndrome_table[udbl & 0xff]; if (prom_getunumber(scode, afar, memmod_str, sizeof(memmod_str)) == -1) p = syndrome_unknown; @@ -407,9 +323,8 @@ asmlinkage void cee_log(unsigned long ce_status, smp_processor_id(), scode, p); } - udb_reg = ((ce_status >> CE_STATUS_UDBH_SHIFT) & 0x3ffUL); - if (udb_reg & (1 << 8)) { - scode = ecc_syndrome_table[udb_reg & 0xff]; + if (udbh & bit) { + scode = ecc_syndrome_table[udbh & 0xff]; if (prom_getunumber(scode, afar, memmod_str, sizeof(memmod_str)) == -1) p = syndrome_unknown; @@ -419,6 +334,127 @@ asmlinkage void cee_log(unsigned long ce_status, "Memory Module \"%s\"\n", smp_processor_id(), scode, p); } + +} + +static void spitfire_cee_log(unsigned long afsr, unsigned long afar, unsigned long udbh, unsigned long udbl, int tl1, struct pt_regs *regs) +{ + + printk(KERN_WARNING "CPU[%d]: Correctable ECC Error " + "AFSR[%lx] AFAR[%016lx] UDBL[%lx] UDBH[%lx] TL>1[%d]\n", + smp_processor_id(), afsr, afar, udbl, udbh, tl1); + + spitfire_log_udb_syndrome(afar, udbh, udbl, UDBE_CE); + + /* We always log it, even if someone is listening for this + * trap. + */ + notify_die(DIE_TRAP, "Correctable ECC Error", regs, + 0, TRAP_TYPE_CEE, SIGTRAP); + + /* The Correctable ECC Error trap does not disable I/D caches. So + * we only have to restore the ESTATE Error Enable register. + */ + spitfire_enable_estate_errors(); +} + +static void spitfire_ue_log(unsigned long afsr, unsigned long afar, unsigned long udbh, unsigned long udbl, unsigned long tt, int tl1, struct pt_regs *regs) +{ + siginfo_t info; + + printk(KERN_WARNING "CPU[%d]: Uncorrectable Error AFSR[%lx] " + "AFAR[%lx] UDBL[%lx] UDBH[%ld] TT[%lx] TL>1[%d]\n", + smp_processor_id(), afsr, afar, udbl, udbh, tt, tl1); + + /* XXX add more human friendly logging of the error status + * XXX as is implemented for cheetah + */ + + spitfire_log_udb_syndrome(afar, udbh, udbl, UDBE_UE); + + /* We always log it, even if someone is listening for this + * trap. + */ + notify_die(DIE_TRAP, "Uncorrectable Error", regs, + 0, tt, SIGTRAP); + + if (regs->tstate & TSTATE_PRIV) { + if (tl1) + dump_tl1_traplog((struct tl1_traplog *)(regs + 1)); + die_if_kernel("UE", regs); + } + + /* XXX need more intelligent processing here, such as is implemented + * XXX for cheetah errors, in fact if the E-cache still holds the + * XXX line with bad parity this will loop + */ + + spitfire_clean_and_reenable_l1_caches(); + spitfire_enable_estate_errors(); + + if (test_thread_flag(TIF_32BIT)) { + regs->tpc &= 0xffffffff; + regs->tnpc &= 0xffffffff; + } + info.si_signo = SIGBUS; + info.si_errno = 0; + info.si_code = BUS_OBJERR; + info.si_addr = (void *)0; + info.si_trapno = 0; + force_sig_info(SIGBUS, &info, current); +} + +void spitfire_access_error(struct pt_regs *regs, unsigned long status_encoded, unsigned long afar) +{ + unsigned long afsr, tt, udbh, udbl; + int tl1; + + afsr = (status_encoded & SFSTAT_AFSR_MASK) >> SFSTAT_AFSR_SHIFT; + tt = (status_encoded & SFSTAT_TRAP_TYPE) >> SFSTAT_TRAP_TYPE_SHIFT; + tl1 = (status_encoded & SFSTAT_TL_GT_ONE) ? 1 : 0; + udbl = (status_encoded & SFSTAT_UDBL_MASK) >> SFSTAT_UDBL_SHIFT; + udbh = (status_encoded & SFSTAT_UDBH_MASK) >> SFSTAT_UDBH_SHIFT; + +#ifdef CONFIG_PCI + if (tt == TRAP_TYPE_DAE && + pci_poke_in_progress && pci_poke_cpu == smp_processor_id()) { + spitfire_clean_and_reenable_l1_caches(); + spitfire_enable_estate_errors(); + + pci_poke_faulted = 1; + regs->tnpc = regs->tpc + 4; + return; + } +#endif + + if (afsr & SFAFSR_UE) + spitfire_ue_log(afsr, afar, udbh, udbl, tt, tl1, regs); + + if (tt == TRAP_TYPE_CEE) { + /* Handle the case where we took a CEE trap, but ACK'd + * only the UE state in the UDB error registers. + */ + if (afsr & SFAFSR_UE) { + if (udbh & UDBE_CE) { + __asm__ __volatile__( + "stxa %0, [%1] %2\n\t" + "membar #Sync" + : /* no outputs */ + : "r" (udbh & UDBE_CE), + "r" (0x0), "i" (ASI_UDB_ERROR_W)); + } + if (udbl & UDBE_CE) { + __asm__ __volatile__( + "stxa %0, [%1] %2\n\t" + "membar #Sync" + : /* no outputs */ + : "r" (udbl & UDBE_CE), + "r" (0x18), "i" (ASI_UDB_ERROR_W)); + } + } + + spitfire_cee_log(afsr, afar, udbh, udbl, tl1, regs); + } } int cheetah_pcache_forced_on; @@ -721,26 +757,12 @@ void __init cheetah_ecache_flush_init(void) ecache_flush_size = (2 * largest_size); ecache_flush_linesize = smallest_linesize; - /* Discover a physically contiguous chunk of physical - * memory in 'sp_banks' of size ecache_flush_size calculated - * above. Store the physical base of this area at - * ecache_flush_physbase. - */ - for (node = 0; ; node++) { - if (sp_banks[node].num_bytes == 0) - break; - if (sp_banks[node].num_bytes >= ecache_flush_size) { - ecache_flush_physbase = sp_banks[node].base_addr; - break; - } - } + ecache_flush_physbase = find_ecache_flush_span(ecache_flush_size); - /* Note: Zero would be a valid value of ecache_flush_physbase so - * don't use that as the success test. :-) - */ - if (sp_banks[node].num_bytes == 0) { + if (ecache_flush_physbase == ~0UL) { prom_printf("cheetah_ecache_flush_init: Cannot find %d byte " - "contiguous physical memory.\n", ecache_flush_size); + "contiguous physical memory.\n", + ecache_flush_size); prom_halt(); } @@ -832,14 +854,19 @@ static void cheetah_flush_ecache_line(unsigned long physaddr) */ static void __cheetah_flush_icache(void) { - unsigned long i; + unsigned int icache_size, icache_line_size; + unsigned long addr; + + icache_size = local_cpu_data().icache_size; + icache_line_size = local_cpu_data().icache_line_size; /* Clear the valid bits in all the tags. */ - for (i = 0; i < (1 << 15); i += (1 << 5)) { + for (addr = 0; addr < icache_size; addr += icache_line_size) { __asm__ __volatile__("stxa %%g0, [%0] %1\n\t" "membar #Sync" : /* no outputs */ - : "r" (i | (2 << 3)), "i" (ASI_IC_TAG)); + : "r" (addr | (2 << 3)), + "i" (ASI_IC_TAG)); } } @@ -867,13 +894,17 @@ static void cheetah_flush_icache(void) static void cheetah_flush_dcache(void) { - unsigned long i; + unsigned int dcache_size, dcache_line_size; + unsigned long addr; + + dcache_size = local_cpu_data().dcache_size; + dcache_line_size = local_cpu_data().dcache_line_size; - for (i = 0; i < (1 << 16); i += (1 << 5)) { + for (addr = 0; addr < dcache_size; addr += dcache_line_size) { __asm__ __volatile__("stxa %%g0, [%0] %1\n\t" "membar #Sync" : /* no outputs */ - : "r" (i), "i" (ASI_DCACHE_TAG)); + : "r" (addr), "i" (ASI_DCACHE_TAG)); } } @@ -884,24 +915,29 @@ static void cheetah_flush_dcache(void) */ static void cheetah_plus_zap_dcache_parity(void) { - unsigned long i; + unsigned int dcache_size, dcache_line_size; + unsigned long addr; - for (i = 0; i < (1 << 16); i += (1 << 5)) { - unsigned long tag = (i >> 14); - unsigned long j; + dcache_size = local_cpu_data().dcache_size; + dcache_line_size = local_cpu_data().dcache_line_size; + + for (addr = 0; addr < dcache_size; addr += dcache_line_size) { + unsigned long tag = (addr >> 14); + unsigned long line; __asm__ __volatile__("membar #Sync\n\t" "stxa %0, [%1] %2\n\t" "membar #Sync" : /* no outputs */ - : "r" (tag), "r" (i), + : "r" (tag), "r" (addr), "i" (ASI_DCACHE_UTAG)); - for (j = i; j < i + (1 << 5); j += (1 << 3)) + for (line = addr; line < addr + dcache_line_size; line += 8) __asm__ __volatile__("membar #Sync\n\t" "stxa %%g0, [%0] %1\n\t" "membar #Sync" : /* no outputs */ - : "r" (j), "i" (ASI_DCACHE_DATA)); + : "r" (line), + "i" (ASI_DCACHE_DATA)); } } @@ -1295,16 +1331,12 @@ static int cheetah_fix_ce(unsigned long physaddr) /* Return non-zero if PADDR is a valid physical memory address. */ static int cheetah_check_main_memory(unsigned long paddr) { - int i; + unsigned long vaddr = PAGE_OFFSET + paddr; - for (i = 0; ; i++) { - if (sp_banks[i].num_bytes == 0) - break; - if (paddr >= sp_banks[i].base_addr && - paddr < (sp_banks[i].base_addr + sp_banks[i].num_bytes)) - return 1; - } - return 0; + if (vaddr > (unsigned long) high_memory) + return 0; + + return kern_addr_valid(vaddr); } void cheetah_cee_handler(struct pt_regs *regs, unsigned long afsr, unsigned long afar) @@ -1559,10 +1591,10 @@ void cheetah_deferred_handler(struct pt_regs *regs, unsigned long afsr, unsigned /* OK, usermode access. */ recoverable = 1; } else { - unsigned long g2 = regs->u_regs[UREG_G2]; - unsigned long fixup = search_extables_range(regs->tpc, &g2); + const struct exception_table_entry *entry; - if (fixup != 0UL) { + entry = search_exception_tables(regs->tpc); + if (entry) { /* OK, kernel access to userspace. */ recoverable = 1; @@ -1581,9 +1613,8 @@ void cheetah_deferred_handler(struct pt_regs *regs, unsigned long afsr, unsigned * recoverable condition. */ if (recoverable) { - regs->tpc = fixup; + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs[UREG_G2] = g2; } } } @@ -2127,6 +2158,9 @@ void __init trap_init(void) TI_PRE_COUNT != offsetof(struct thread_info, preempt_count) || TI_NEW_CHILD != offsetof(struct thread_info, new_child) || TI_SYS_NOERROR != offsetof(struct thread_info, syscall_noerror) || + TI_RESTART_BLOCK != offsetof(struct thread_info, restart_block) || + TI_KUNA_REGS != offsetof(struct thread_info, kern_una_regs) || + TI_KUNA_INSN != offsetof(struct thread_info, kern_una_insn) || TI_FPREGS != offsetof(struct thread_info, fpregs) || (TI_FPREGS & (64 - 1))) thread_info_offsets_are_bolixed_dave(); diff --git a/arch/sparc64/kernel/ttable.S b/arch/sparc64/kernel/ttable.S index 491bb3681f9d..8365bc1f81f3 100644 --- a/arch/sparc64/kernel/ttable.S +++ b/arch/sparc64/kernel/ttable.S @@ -18,9 +18,10 @@ sparc64_ttable_tl0: tl0_resv000: BOOT_KERNEL BTRAP(0x1) BTRAP(0x2) BTRAP(0x3) tl0_resv004: BTRAP(0x4) BTRAP(0x5) BTRAP(0x6) BTRAP(0x7) tl0_iax: membar #Sync - TRAP_NOSAVE_7INSNS(__do_instruction_access_exception) + TRAP_NOSAVE_7INSNS(__spitfire_insn_access_exception) tl0_resv009: BTRAP(0x9) -tl0_iae: TRAP(do_iae) +tl0_iae: membar #Sync + TRAP_NOSAVE_7INSNS(__spitfire_access_error) tl0_resv00b: BTRAP(0xb) BTRAP(0xc) BTRAP(0xd) BTRAP(0xe) BTRAP(0xf) tl0_ill: membar #Sync TRAP_7INSNS(do_illegal_instruction) @@ -36,9 +37,10 @@ tl0_cwin: CLEAN_WINDOW tl0_div0: TRAP(do_div0) tl0_resv029: BTRAP(0x29) BTRAP(0x2a) BTRAP(0x2b) BTRAP(0x2c) BTRAP(0x2d) BTRAP(0x2e) tl0_resv02f: BTRAP(0x2f) -tl0_dax: TRAP_NOSAVE(__do_data_access_exception) +tl0_dax: TRAP_NOSAVE(__spitfire_data_access_exception) tl0_resv031: BTRAP(0x31) -tl0_dae: TRAP(do_dae) +tl0_dae: membar #Sync + TRAP_NOSAVE_7INSNS(__spitfire_access_error) tl0_resv033: BTRAP(0x33) tl0_mna: TRAP_NOSAVE(do_mna) tl0_lddfmna: TRAP_NOSAVE(do_lddfmna) @@ -73,7 +75,8 @@ tl0_resv05c: BTRAP(0x5c) BTRAP(0x5d) BTRAP(0x5e) BTRAP(0x5f) tl0_ivec: TRAP_IVEC tl0_paw: TRAP(do_paw) tl0_vaw: TRAP(do_vaw) -tl0_cee: TRAP_NOSAVE(cee_trap) +tl0_cee: membar #Sync + TRAP_NOSAVE_7INSNS(__spitfire_cee_trap) tl0_iamiss: #include "itlb_base.S" tl0_damiss: @@ -175,9 +178,10 @@ tl0_resv1f0: BTRAPS(0x1f0) BTRAPS(0x1f8) sparc64_ttable_tl1: tl1_resv000: BOOT_KERNEL BTRAPTL1(0x1) BTRAPTL1(0x2) BTRAPTL1(0x3) tl1_resv004: BTRAPTL1(0x4) BTRAPTL1(0x5) BTRAPTL1(0x6) BTRAPTL1(0x7) -tl1_iax: TRAP_NOSAVE(__do_instruction_access_exception_tl1) +tl1_iax: TRAP_NOSAVE(__spitfire_insn_access_exception_tl1) tl1_resv009: BTRAPTL1(0x9) -tl1_iae: TRAPTL1(do_iae_tl1) +tl1_iae: membar #Sync + TRAP_NOSAVE_7INSNS(__spitfire_access_error) tl1_resv00b: BTRAPTL1(0xb) BTRAPTL1(0xc) BTRAPTL1(0xd) BTRAPTL1(0xe) BTRAPTL1(0xf) tl1_ill: TRAPTL1(do_ill_tl1) tl1_privop: BTRAPTL1(0x11) @@ -193,9 +197,10 @@ tl1_cwin: CLEAN_WINDOW tl1_div0: TRAPTL1(do_div0_tl1) tl1_resv029: BTRAPTL1(0x29) BTRAPTL1(0x2a) BTRAPTL1(0x2b) BTRAPTL1(0x2c) tl1_resv02d: BTRAPTL1(0x2d) BTRAPTL1(0x2e) BTRAPTL1(0x2f) -tl1_dax: TRAP_NOSAVE(__do_data_access_exception_tl1) +tl1_dax: TRAP_NOSAVE(__spitfire_data_access_exception_tl1) tl1_resv031: BTRAPTL1(0x31) -tl1_dae: TRAPTL1(do_dae_tl1) +tl1_dae: membar #Sync + TRAP_NOSAVE_7INSNS(__spitfire_access_error) tl1_resv033: BTRAPTL1(0x33) tl1_mna: TRAP_NOSAVE(do_mna) tl1_lddfmna: TRAPTL1(do_lddfmna_tl1) @@ -219,8 +224,8 @@ tl1_paw: TRAPTL1(do_paw_tl1) tl1_vaw: TRAPTL1(do_vaw_tl1) /* The grotty trick to save %g1 into current->thread.cee_stuff - * is because when we take this trap we could be interrupting trap - * code already using the trap alternate global registers. + * is because when we take this trap we could be interrupting + * trap code already using the trap alternate global registers. * * We cross our fingers and pray that this store/load does * not cause yet another CEE trap. diff --git a/arch/sparc64/kernel/una_asm.S b/arch/sparc64/kernel/una_asm.S new file mode 100644 index 000000000000..1f5b5b708ce7 --- /dev/null +++ b/arch/sparc64/kernel/una_asm.S @@ -0,0 +1,146 @@ +/* una_asm.S: Kernel unaligned trap assembler helpers. + * + * Copyright (C) 1996,2005 David S. Miller (davem@davemloft.net) + * Copyright (C) 1996,1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz) + */ + + .text + + .globl __do_int_store +__do_int_store: + rd %asi, %o4 + wr %o3, 0, %asi + mov %o2, %g3 + cmp %o1, 2 + be,pn %icc, 2f + cmp %o1, 4 + be,pt %icc, 1f + srlx %g3, 24, %g2 + srlx %g3, 56, %g1 + srlx %g3, 48, %g7 +4: stba %g1, [%o0] %asi + srlx %g3, 40, %g1 +5: stba %g7, [%o0 + 1] %asi + srlx %g3, 32, %g7 +6: stba %g1, [%o0 + 2] %asi +7: stba %g7, [%o0 + 3] %asi + srlx %g3, 16, %g1 +8: stba %g2, [%o0 + 4] %asi + srlx %g3, 8, %g7 +9: stba %g1, [%o0 + 5] %asi +10: stba %g7, [%o0 + 6] %asi + ba,pt %xcc, 0f +11: stba %g3, [%o0 + 7] %asi +1: srl %g3, 16, %g7 +12: stba %g2, [%o0] %asi + srl %g3, 8, %g2 +13: stba %g7, [%o0 + 1] %asi +14: stba %g2, [%o0 + 2] %asi + ba,pt %xcc, 0f +15: stba %g3, [%o0 + 3] %asi +2: srl %g3, 8, %g2 +16: stba %g2, [%o0] %asi +17: stba %g3, [%o0 + 1] %asi +0: + wr %o4, 0x0, %asi + retl + mov 0, %o0 + .size __do_int_store, .-__do_int_store + + .section __ex_table + .word 4b, __retl_efault + .word 5b, __retl_efault + .word 6b, __retl_efault + .word 7b, __retl_efault + .word 8b, __retl_efault + .word 9b, __retl_efault + .word 10b, __retl_efault + .word 11b, __retl_efault + .word 12b, __retl_efault + .word 13b, __retl_efault + .word 14b, __retl_efault + .word 15b, __retl_efault + .word 16b, __retl_efault + .word 17b, __retl_efault + .previous + + .globl do_int_load +do_int_load: + rd %asi, %o5 + wr %o4, 0, %asi + cmp %o1, 8 + bge,pn %icc, 9f + cmp %o1, 4 + be,pt %icc, 6f +4: lduba [%o2] %asi, %g2 +5: lduba [%o2 + 1] %asi, %g3 + sll %g2, 8, %g2 + brz,pt %o3, 3f + add %g2, %g3, %g2 + sllx %g2, 48, %g2 + srax %g2, 48, %g2 +3: ba,pt %xcc, 0f + stx %g2, [%o0] +6: lduba [%o2 + 1] %asi, %g3 + sll %g2, 24, %g2 +7: lduba [%o2 + 2] %asi, %g7 + sll %g3, 16, %g3 +8: lduba [%o2 + 3] %asi, %g1 + sll %g7, 8, %g7 + or %g2, %g3, %g2 + or %g7, %g1, %g7 + or %g2, %g7, %g2 + brnz,a,pt %o3, 3f + sra %g2, 0, %g2 +3: ba,pt %xcc, 0f + stx %g2, [%o0] +9: lduba [%o2] %asi, %g2 +10: lduba [%o2 + 1] %asi, %g3 + sllx %g2, 56, %g2 +11: lduba [%o2 + 2] %asi, %g7 + sllx %g3, 48, %g3 +12: lduba [%o2 + 3] %asi, %g1 + sllx %g7, 40, %g7 + sllx %g1, 32, %g1 + or %g2, %g3, %g2 + or %g7, %g1, %g7 +13: lduba [%o2 + 4] %asi, %g3 + or %g2, %g7, %g7 +14: lduba [%o2 + 5] %asi, %g1 + sllx %g3, 24, %g3 +15: lduba [%o2 + 6] %asi, %g2 + sllx %g1, 16, %g1 + or %g7, %g3, %g7 +16: lduba [%o2 + 7] %asi, %g3 + sllx %g2, 8, %g2 + or %g7, %g1, %g7 + or %g2, %g3, %g2 + or %g7, %g2, %g7 + cmp %o1, 8 + be,a,pt %icc, 0f + stx %g7, [%o0] + srlx %g7, 32, %g2 + sra %g7, 0, %g7 + stx %g2, [%o0] + stx %g7, [%o0 + 8] +0: + wr %o5, 0x0, %asi + retl + mov 0, %o0 + .size __do_int_load, .-__do_int_load + + .section __ex_table + .word 4b, __retl_efault + .word 5b, __retl_efault + .word 6b, __retl_efault + .word 7b, __retl_efault + .word 8b, __retl_efault + .word 9b, __retl_efault + .word 10b, __retl_efault + .word 11b, __retl_efault + .word 12b, __retl_efault + .word 13b, __retl_efault + .word 14b, __retl_efault + .word 15b, __retl_efault + .word 16b, __retl_efault + .previous diff --git a/arch/sparc64/kernel/unaligned.c b/arch/sparc64/kernel/unaligned.c index 4372bf32ecf6..70faf630603b 100644 --- a/arch/sparc64/kernel/unaligned.c +++ b/arch/sparc64/kernel/unaligned.c @@ -180,169 +180,47 @@ static void __attribute_used__ unaligned_panic(char *str, struct pt_regs *regs) die_if_kernel(str, regs); } -#define do_integer_load(dest_reg, size, saddr, is_signed, asi, errh) ({ \ -__asm__ __volatile__ ( \ - "wr %4, 0, %%asi\n\t" \ - "cmp %1, 8\n\t" \ - "bge,pn %%icc, 9f\n\t" \ - " cmp %1, 4\n\t" \ - "be,pt %%icc, 6f\n" \ -"4:\t" " lduba [%2] %%asi, %%l1\n" \ -"5:\t" "lduba [%2 + 1] %%asi, %%l2\n\t" \ - "sll %%l1, 8, %%l1\n\t" \ - "brz,pt %3, 3f\n\t" \ - " add %%l1, %%l2, %%l1\n\t" \ - "sllx %%l1, 48, %%l1\n\t" \ - "srax %%l1, 48, %%l1\n" \ -"3:\t" "ba,pt %%xcc, 0f\n\t" \ - " stx %%l1, [%0]\n" \ -"6:\t" "lduba [%2 + 1] %%asi, %%l2\n\t" \ - "sll %%l1, 24, %%l1\n" \ -"7:\t" "lduba [%2 + 2] %%asi, %%g7\n\t" \ - "sll %%l2, 16, %%l2\n" \ -"8:\t" "lduba [%2 + 3] %%asi, %%g1\n\t" \ - "sll %%g7, 8, %%g7\n\t" \ - "or %%l1, %%l2, %%l1\n\t" \ - "or %%g7, %%g1, %%g7\n\t" \ - "or %%l1, %%g7, %%l1\n\t" \ - "brnz,a,pt %3, 3f\n\t" \ - " sra %%l1, 0, %%l1\n" \ -"3:\t" "ba,pt %%xcc, 0f\n\t" \ - " stx %%l1, [%0]\n" \ -"9:\t" "lduba [%2] %%asi, %%l1\n" \ -"10:\t" "lduba [%2 + 1] %%asi, %%l2\n\t" \ - "sllx %%l1, 56, %%l1\n" \ -"11:\t" "lduba [%2 + 2] %%asi, %%g7\n\t" \ - "sllx %%l2, 48, %%l2\n" \ -"12:\t" "lduba [%2 + 3] %%asi, %%g1\n\t" \ - "sllx %%g7, 40, %%g7\n\t" \ - "sllx %%g1, 32, %%g1\n\t" \ - "or %%l1, %%l2, %%l1\n\t" \ - "or %%g7, %%g1, %%g7\n" \ -"13:\t" "lduba [%2 + 4] %%asi, %%l2\n\t" \ - "or %%l1, %%g7, %%g7\n" \ -"14:\t" "lduba [%2 + 5] %%asi, %%g1\n\t" \ - "sllx %%l2, 24, %%l2\n" \ -"15:\t" "lduba [%2 + 6] %%asi, %%l1\n\t" \ - "sllx %%g1, 16, %%g1\n\t" \ - "or %%g7, %%l2, %%g7\n" \ -"16:\t" "lduba [%2 + 7] %%asi, %%l2\n\t" \ - "sllx %%l1, 8, %%l1\n\t" \ - "or %%g7, %%g1, %%g7\n\t" \ - "or %%l1, %%l2, %%l1\n\t" \ - "or %%g7, %%l1, %%g7\n\t" \ - "cmp %1, 8\n\t" \ - "be,a,pt %%icc, 0f\n\t" \ - " stx %%g7, [%0]\n\t" \ - "srlx %%g7, 32, %%l1\n\t" \ - "sra %%g7, 0, %%g7\n\t" \ - "stx %%l1, [%0]\n\t" \ - "stx %%g7, [%0 + 8]\n" \ -"0:\n\t" \ - "wr %%g0, %5, %%asi\n\n\t" \ - ".section __ex_table\n\t" \ - ".word 4b, " #errh "\n\t" \ - ".word 5b, " #errh "\n\t" \ - ".word 6b, " #errh "\n\t" \ - ".word 7b, " #errh "\n\t" \ - ".word 8b, " #errh "\n\t" \ - ".word 9b, " #errh "\n\t" \ - ".word 10b, " #errh "\n\t" \ - ".word 11b, " #errh "\n\t" \ - ".word 12b, " #errh "\n\t" \ - ".word 13b, " #errh "\n\t" \ - ".word 14b, " #errh "\n\t" \ - ".word 15b, " #errh "\n\t" \ - ".word 16b, " #errh "\n\n\t" \ - ".previous\n\t" \ - : : "r" (dest_reg), "r" (size), "r" (saddr), "r" (is_signed), \ - "r" (asi), "i" (ASI_AIUS) \ - : "l1", "l2", "g7", "g1", "cc"); \ -}) +extern int do_int_load(unsigned long *dest_reg, int size, + unsigned long *saddr, int is_signed, int asi); -#define store_common(dst_addr, size, src_val, asi, errh) ({ \ -__asm__ __volatile__ ( \ - "wr %3, 0, %%asi\n\t" \ - "ldx [%2], %%l1\n" \ - "cmp %1, 2\n\t" \ - "be,pn %%icc, 2f\n\t" \ - " cmp %1, 4\n\t" \ - "be,pt %%icc, 1f\n\t" \ - " srlx %%l1, 24, %%l2\n\t" \ - "srlx %%l1, 56, %%g1\n\t" \ - "srlx %%l1, 48, %%g7\n" \ -"4:\t" "stba %%g1, [%0] %%asi\n\t" \ - "srlx %%l1, 40, %%g1\n" \ -"5:\t" "stba %%g7, [%0 + 1] %%asi\n\t" \ - "srlx %%l1, 32, %%g7\n" \ -"6:\t" "stba %%g1, [%0 + 2] %%asi\n" \ -"7:\t" "stba %%g7, [%0 + 3] %%asi\n\t" \ - "srlx %%l1, 16, %%g1\n" \ -"8:\t" "stba %%l2, [%0 + 4] %%asi\n\t" \ - "srlx %%l1, 8, %%g7\n" \ -"9:\t" "stba %%g1, [%0 + 5] %%asi\n" \ -"10:\t" "stba %%g7, [%0 + 6] %%asi\n\t" \ - "ba,pt %%xcc, 0f\n" \ -"11:\t" " stba %%l1, [%0 + 7] %%asi\n" \ -"1:\t" "srl %%l1, 16, %%g7\n" \ -"12:\t" "stba %%l2, [%0] %%asi\n\t" \ - "srl %%l1, 8, %%l2\n" \ -"13:\t" "stba %%g7, [%0 + 1] %%asi\n" \ -"14:\t" "stba %%l2, [%0 + 2] %%asi\n\t" \ - "ba,pt %%xcc, 0f\n" \ -"15:\t" " stba %%l1, [%0 + 3] %%asi\n" \ -"2:\t" "srl %%l1, 8, %%l2\n" \ -"16:\t" "stba %%l2, [%0] %%asi\n" \ -"17:\t" "stba %%l1, [%0 + 1] %%asi\n" \ -"0:\n\t" \ - "wr %%g0, %4, %%asi\n\n\t" \ - ".section __ex_table\n\t" \ - ".word 4b, " #errh "\n\t" \ - ".word 5b, " #errh "\n\t" \ - ".word 6b, " #errh "\n\t" \ - ".word 7b, " #errh "\n\t" \ - ".word 8b, " #errh "\n\t" \ - ".word 9b, " #errh "\n\t" \ - ".word 10b, " #errh "\n\t" \ - ".word 11b, " #errh "\n\t" \ - ".word 12b, " #errh "\n\t" \ - ".word 13b, " #errh "\n\t" \ - ".word 14b, " #errh "\n\t" \ - ".word 15b, " #errh "\n\t" \ - ".word 16b, " #errh "\n\t" \ - ".word 17b, " #errh "\n\n\t" \ - ".previous\n\t" \ - : : "r" (dst_addr), "r" (size), "r" (src_val), "r" (asi), "i" (ASI_AIUS)\ - : "l1", "l2", "g7", "g1", "cc"); \ -}) - -#define do_integer_store(reg_num, size, dst_addr, regs, asi, errh) ({ \ - unsigned long zero = 0; \ - unsigned long *src_val = &zero; \ - \ - if (size == 16) { \ - size = 8; \ - zero = (((long)(reg_num ? \ - (unsigned)fetch_reg(reg_num, regs) : 0)) << 32) | \ - (unsigned)fetch_reg(reg_num + 1, regs); \ - } else if (reg_num) src_val = fetch_reg_addr(reg_num, regs); \ - store_common(dst_addr, size, src_val, asi, errh); \ -}) - -extern void smp_capture(void); -extern void smp_release(void); - -#define do_atomic(srcdest_reg, mem, errh) ({ \ - unsigned long flags, tmp; \ - \ - smp_capture(); \ - local_irq_save(flags); \ - tmp = *srcdest_reg; \ - do_integer_load(srcdest_reg, 4, mem, 0, errh); \ - store_common(mem, 4, &tmp, errh); \ - local_irq_restore(flags); \ - smp_release(); \ -}) +extern int __do_int_store(unsigned long *dst_addr, int size, + unsigned long src_val, int asi); + +static inline int do_int_store(int reg_num, int size, unsigned long *dst_addr, + struct pt_regs *regs, int asi, int orig_asi) +{ + unsigned long zero = 0; + unsigned long *src_val_p = &zero; + unsigned long src_val; + + if (size == 16) { + size = 8; + zero = (((long)(reg_num ? + (unsigned)fetch_reg(reg_num, regs) : 0)) << 32) | + (unsigned)fetch_reg(reg_num + 1, regs); + } else if (reg_num) { + src_val_p = fetch_reg_addr(reg_num, regs); + } + src_val = *src_val_p; + if (unlikely(asi != orig_asi)) { + switch (size) { + case 2: + src_val = swab16(src_val); + break; + case 4: + src_val = swab32(src_val); + break; + case 8: + src_val = swab64(src_val); + break; + case 16: + default: + BUG(); + break; + }; + } + return __do_int_store(dst_addr, size, src_val, asi); +} static inline void advance(struct pt_regs *regs) { @@ -364,32 +242,36 @@ static inline int ok_for_kernel(unsigned int insn) return !floating_point_load_or_store_p(insn); } -void kernel_mna_trap_fault(struct pt_regs *regs, unsigned int insn) __asm__ ("kernel_mna_trap_fault"); - -void kernel_mna_trap_fault(struct pt_regs *regs, unsigned int insn) +static void kernel_mna_trap_fault(void) { - unsigned long g2 = regs->u_regs [UREG_G2]; - unsigned long fixup = search_extables_range(regs->tpc, &g2); + struct pt_regs *regs = current_thread_info()->kern_una_regs; + unsigned int insn = current_thread_info()->kern_una_insn; + const struct exception_table_entry *entry; - if (!fixup) { - unsigned long address = compute_effective_address(regs, insn, ((insn >> 25) & 0x1f)); + entry = search_exception_tables(regs->tpc); + if (!entry) { + unsigned long address; + + address = compute_effective_address(regs, insn, + ((insn >> 25) & 0x1f)); if (address < PAGE_SIZE) { - printk(KERN_ALERT "Unable to handle kernel NULL pointer dereference in mna handler"); + printk(KERN_ALERT "Unable to handle kernel NULL " + "pointer dereference in mna handler"); } else - printk(KERN_ALERT "Unable to handle kernel paging request in mna handler"); + printk(KERN_ALERT "Unable to handle kernel paging " + "request in mna handler"); printk(KERN_ALERT " at virtual address %016lx\n",address); - printk(KERN_ALERT "current->{mm,active_mm}->context = %016lx\n", + printk(KERN_ALERT "current->{active_,}mm->context = %016lx\n", (current->mm ? CTX_HWBITS(current->mm->context) : CTX_HWBITS(current->active_mm->context))); - printk(KERN_ALERT "current->{mm,active_mm}->pgd = %016lx\n", + printk(KERN_ALERT "current->{active_,}mm->pgd = %016lx\n", (current->mm ? (unsigned long) current->mm->pgd : (unsigned long) current->active_mm->pgd)); die_if_kernel("Oops", regs); /* Not reached */ } - regs->tpc = fixup; + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs [UREG_G2] = g2; regs->tstate &= ~TSTATE_ASI; regs->tstate |= (ASI_AIUS << 24UL); @@ -400,53 +282,81 @@ asmlinkage void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn, u enum direction dir = decode_direction(insn); int size = decode_access_size(insn); + current_thread_info()->kern_una_regs = regs; + current_thread_info()->kern_una_insn = insn; + if (!ok_for_kernel(insn) || dir == both) { - printk("Unsupported unaligned load/store trap for kernel at <%016lx>.\n", - regs->tpc); - unaligned_panic("Kernel does fpu/atomic unaligned load/store.", regs); - - __asm__ __volatile__ ("\n" -"kernel_unaligned_trap_fault:\n\t" - "mov %0, %%o0\n\t" - "call kernel_mna_trap_fault\n\t" - " mov %1, %%o1\n\t" - : - : "r" (regs), "r" (insn) - : "o0", "o1", "o2", "o3", "o4", "o5", "o7", - "g1", "g2", "g3", "g4", "g7", "cc"); + printk("Unsupported unaligned load/store trap for kernel " + "at <%016lx>.\n", regs->tpc); + unaligned_panic("Kernel does fpu/atomic " + "unaligned load/store.", regs); + + kernel_mna_trap_fault(); } else { - unsigned long addr = compute_effective_address(regs, insn, ((insn >> 25) & 0x1f)); + unsigned long addr, *reg_addr; + int orig_asi, asi, err; + addr = compute_effective_address(regs, insn, + ((insn >> 25) & 0x1f)); #ifdef DEBUG_MNA - printk("KMNA: pc=%016lx [dir=%s addr=%016lx size=%d] retpc[%016lx]\n", - regs->tpc, dirstrings[dir], addr, size, regs->u_regs[UREG_RETPC]); + printk("KMNA: pc=%016lx [dir=%s addr=%016lx size=%d] " + "retpc[%016lx]\n", + regs->tpc, dirstrings[dir], addr, size, + regs->u_regs[UREG_RETPC]); #endif + orig_asi = asi = decode_asi(insn, regs); + switch (asi) { + case ASI_NL: + case ASI_AIUPL: + case ASI_AIUSL: + case ASI_PL: + case ASI_SL: + case ASI_PNFL: + case ASI_SNFL: + asi &= ~0x08; + break; + }; switch (dir) { case load: - do_integer_load(fetch_reg_addr(((insn>>25)&0x1f), regs), - size, (unsigned long *) addr, - decode_signedness(insn), decode_asi(insn, regs), - kernel_unaligned_trap_fault); + reg_addr = fetch_reg_addr(((insn>>25)&0x1f), regs); + err = do_int_load(reg_addr, size, + (unsigned long *) addr, + decode_signedness(insn), asi); + if (likely(!err) && unlikely(asi != orig_asi)) { + unsigned long val_in = *reg_addr; + switch (size) { + case 2: + val_in = swab16(val_in); + break; + case 4: + val_in = swab32(val_in); + break; + case 8: + val_in = swab64(val_in); + break; + case 16: + default: + BUG(); + break; + }; + *reg_addr = val_in; + } break; case store: - do_integer_store(((insn>>25)&0x1f), size, - (unsigned long *) addr, regs, - decode_asi(insn, regs), - kernel_unaligned_trap_fault); - break; -#if 0 /* unsupported */ - case both: - do_atomic(fetch_reg_addr(((insn>>25)&0x1f), regs), - (unsigned long *) addr, - kernel_unaligned_trap_fault); + err = do_int_store(((insn>>25)&0x1f), size, + (unsigned long *) addr, regs, + asi, orig_asi); break; -#endif + default: panic("Impossible kernel unaligned trap."); /* Not reached... */ } - advance(regs); + if (unlikely(err)) + kernel_mna_trap_fault(); + else + advance(regs); } } @@ -492,9 +402,9 @@ int handle_popc(u32 insn, struct pt_regs *regs) extern void do_fpother(struct pt_regs *regs); extern void do_privact(struct pt_regs *regs); -extern void data_access_exception(struct pt_regs *regs, - unsigned long sfsr, - unsigned long sfar); +extern void spitfire_data_access_exception(struct pt_regs *regs, + unsigned long sfsr, + unsigned long sfar); int handle_ldf_stq(u32 insn, struct pt_regs *regs) { @@ -537,14 +447,14 @@ int handle_ldf_stq(u32 insn, struct pt_regs *regs) break; } default: - data_access_exception(regs, 0, addr); + spitfire_data_access_exception(regs, 0, addr); return 1; } if (put_user (first >> 32, (u32 __user *)addr) || __put_user ((u32)first, (u32 __user *)(addr + 4)) || __put_user (second >> 32, (u32 __user *)(addr + 8)) || __put_user ((u32)second, (u32 __user *)(addr + 12))) { - data_access_exception(regs, 0, addr); + spitfire_data_access_exception(regs, 0, addr); return 1; } } else { @@ -557,7 +467,7 @@ int handle_ldf_stq(u32 insn, struct pt_regs *regs) do_privact(regs); return 1; } else if (asi > ASI_SNFL) { - data_access_exception(regs, 0, addr); + spitfire_data_access_exception(regs, 0, addr); return 1; } switch (insn & 0x180000) { @@ -574,7 +484,7 @@ int handle_ldf_stq(u32 insn, struct pt_regs *regs) err |= __get_user (data[i], (u32 __user *)(addr + 4*i)); } if (err && !(asi & 0x2 /* NF */)) { - data_access_exception(regs, 0, addr); + spitfire_data_access_exception(regs, 0, addr); return 1; } if (asi & 0x8) /* Little */ { @@ -677,7 +587,7 @@ void handle_lddfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr *(u64 *)(f->regs + freg) = value; current_thread_info()->fpsaved[0] |= flag; } else { -daex: data_access_exception(regs, sfsr, sfar); +daex: spitfire_data_access_exception(regs, sfsr, sfar); return; } advance(regs); @@ -721,7 +631,7 @@ void handle_stdfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr __put_user ((u32)value, (u32 __user *)(sfar + 4))) goto daex; } else { -daex: data_access_exception(regs, sfsr, sfar); +daex: spitfire_data_access_exception(regs, sfsr, sfar); return; } advance(regs); diff --git a/arch/sparc64/kernel/us2e_cpufreq.c b/arch/sparc64/kernel/us2e_cpufreq.c index 7aae0a18aabe..686e526bec04 100644 --- a/arch/sparc64/kernel/us2e_cpufreq.c +++ b/arch/sparc64/kernel/us2e_cpufreq.c @@ -88,7 +88,6 @@ static void frob_mem_refresh(int cpu_slowing_down, { unsigned long old_refr_count, refr_count, mctrl; - refr_count = (clock_tick * MCTRL0_REFR_INTERVAL); refr_count /= (MCTRL0_REFR_CLKS_P_CNT * divisor * 1000000000UL); @@ -230,6 +229,25 @@ static unsigned long estar_to_divisor(unsigned long estar) return ret; } +static unsigned int us2e_freq_get(unsigned int cpu) +{ + cpumask_t cpus_allowed; + unsigned long clock_tick, estar; + + if (!cpu_online(cpu)) + return 0; + + cpus_allowed = current->cpus_allowed; + set_cpus_allowed(current, cpumask_of_cpu(cpu)); + + clock_tick = sparc64_get_clock_tick(cpu) / 1000; + estar = read_hbreg(HBIRD_ESTAR_MODE_ADDR); + + set_cpus_allowed(current, cpus_allowed); + + return clock_tick / estar_to_divisor(estar); +} + static void us2e_set_cpu_divider_index(unsigned int cpu, unsigned int index) { unsigned long new_bits, new_freq; @@ -243,7 +261,7 @@ static void us2e_set_cpu_divider_index(unsigned int cpu, unsigned int index) cpus_allowed = current->cpus_allowed; set_cpus_allowed(current, cpumask_of_cpu(cpu)); - new_freq = clock_tick = sparc64_get_clock_tick(cpu); + new_freq = clock_tick = sparc64_get_clock_tick(cpu) / 1000; new_bits = index_to_estar_mode(index); divisor = index_to_divisor(index); new_freq /= divisor; @@ -258,7 +276,8 @@ static void us2e_set_cpu_divider_index(unsigned int cpu, unsigned int index) cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE); if (old_divisor != divisor) - us2e_transition(estar, new_bits, clock_tick, old_divisor, divisor); + us2e_transition(estar, new_bits, clock_tick * 1000, + old_divisor, divisor); cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE); @@ -272,10 +291,8 @@ static int us2e_freq_target(struct cpufreq_policy *policy, unsigned int new_index = 0; if (cpufreq_frequency_table_target(policy, - &us2e_freq_table[policy->cpu].table[0], - target_freq, - relation, - &new_index)) + &us2e_freq_table[policy->cpu].table[0], + target_freq, relation, &new_index)) return -EINVAL; us2e_set_cpu_divider_index(policy->cpu, new_index); @@ -292,7 +309,7 @@ static int us2e_freq_verify(struct cpufreq_policy *policy) static int __init us2e_freq_cpu_init(struct cpufreq_policy *policy) { unsigned int cpu = policy->cpu; - unsigned long clock_tick = sparc64_get_clock_tick(cpu); + unsigned long clock_tick = sparc64_get_clock_tick(cpu) / 1000; struct cpufreq_frequency_table *table = &us2e_freq_table[cpu].table[0]; @@ -351,9 +368,10 @@ static int __init us2e_freq_init(void) memset(us2e_freq_table, 0, (NR_CPUS * sizeof(struct us2e_freq_percpu_info))); + driver->init = us2e_freq_cpu_init; driver->verify = us2e_freq_verify; driver->target = us2e_freq_target; - driver->init = us2e_freq_cpu_init; + driver->get = us2e_freq_get; driver->exit = us2e_freq_cpu_exit; driver->owner = THIS_MODULE, strcpy(driver->name, "UltraSPARC-IIe"); diff --git a/arch/sparc64/kernel/us3_cpufreq.c b/arch/sparc64/kernel/us3_cpufreq.c index 18fe54b8aa55..0340041f6143 100644 --- a/arch/sparc64/kernel/us3_cpufreq.c +++ b/arch/sparc64/kernel/us3_cpufreq.c @@ -56,7 +56,7 @@ static void write_safari_cfg(unsigned long val) static unsigned long get_current_freq(unsigned int cpu, unsigned long safari_cfg) { - unsigned long clock_tick = sparc64_get_clock_tick(cpu); + unsigned long clock_tick = sparc64_get_clock_tick(cpu) / 1000; unsigned long ret; switch (safari_cfg & SAFARI_CFG_DIV_MASK) { @@ -76,6 +76,26 @@ static unsigned long get_current_freq(unsigned int cpu, unsigned long safari_cfg return ret; } +static unsigned int us3_freq_get(unsigned int cpu) +{ + cpumask_t cpus_allowed; + unsigned long reg; + unsigned int ret; + + if (!cpu_online(cpu)) + return 0; + + cpus_allowed = current->cpus_allowed; + set_cpus_allowed(current, cpumask_of_cpu(cpu)); + + reg = read_safari_cfg(); + ret = get_current_freq(cpu, reg); + + set_cpus_allowed(current, cpus_allowed); + + return ret; +} + static void us3_set_cpu_divider_index(unsigned int cpu, unsigned int index) { unsigned long new_bits, new_freq, reg; @@ -88,7 +108,7 @@ static void us3_set_cpu_divider_index(unsigned int cpu, unsigned int index) cpus_allowed = current->cpus_allowed; set_cpus_allowed(current, cpumask_of_cpu(cpu)); - new_freq = sparc64_get_clock_tick(cpu); + new_freq = sparc64_get_clock_tick(cpu) / 1000; switch (index) { case 0: new_bits = SAFARI_CFG_DIV_1; @@ -150,7 +170,7 @@ static int us3_freq_verify(struct cpufreq_policy *policy) static int __init us3_freq_cpu_init(struct cpufreq_policy *policy) { unsigned int cpu = policy->cpu; - unsigned long clock_tick = sparc64_get_clock_tick(cpu); + unsigned long clock_tick = sparc64_get_clock_tick(cpu) / 1000; struct cpufreq_frequency_table *table = &us3_freq_table[cpu].table[0]; @@ -188,7 +208,10 @@ static int __init us3_freq_init(void) impl = ((ver >> 32) & 0xffff); if (manuf == CHEETAH_MANUF && - (impl == CHEETAH_IMPL || impl == CHEETAH_PLUS_IMPL)) { + (impl == CHEETAH_IMPL || + impl == CHEETAH_PLUS_IMPL || + impl == JAGUAR_IMPL || + impl == PANTHER_IMPL)) { struct cpufreq_driver *driver; ret = -ENOMEM; @@ -206,9 +229,10 @@ static int __init us3_freq_init(void) memset(us3_freq_table, 0, (NR_CPUS * sizeof(struct us3_freq_percpu_info))); + driver->init = us3_freq_cpu_init; driver->verify = us3_freq_verify; driver->target = us3_freq_target; - driver->init = us3_freq_cpu_init; + driver->get = us3_freq_get; driver->exit = us3_freq_cpu_exit; driver->owner = THIS_MODULE, strcpy(driver->name, "UltraSPARC-III"); diff --git a/arch/sparc64/kernel/vmlinux.lds.S b/arch/sparc64/kernel/vmlinux.lds.S index 950423da8a6a..2af0cf0a8640 100644 --- a/arch/sparc64/kernel/vmlinux.lds.S +++ b/arch/sparc64/kernel/vmlinux.lds.S @@ -9,14 +9,14 @@ ENTRY(_start) jiffies = jiffies_64; SECTIONS { - swapper_pmd_dir = 0x0000000000402000; - empty_pg_dir = 0x0000000000403000; + swapper_low_pmd_dir = 0x0000000000402000; . = 0x4000; .text 0x0000000000404000 : { *(.text) SCHED_TEXT LOCK_TEXT + KPROBES_TEXT *(.gnu.warning) } =0 _etext = .; diff --git a/arch/sparc64/kernel/winfixup.S b/arch/sparc64/kernel/winfixup.S index dfbc7e0dcf70..39160926267b 100644 --- a/arch/sparc64/kernel/winfixup.S +++ b/arch/sparc64/kernel/winfixup.S @@ -16,23 +16,14 @@ .text set_pcontext: -cplus_winfixup_insn_1: - sethi %hi(0), %l1 + sethi %hi(sparc64_kern_pri_context), %l1 + ldx [%l1 + %lo(sparc64_kern_pri_context)], %l1 mov PRIMARY_CONTEXT, %g1 - sllx %l1, 32, %l1 -cplus_winfixup_insn_2: - sethi %hi(0), %g2 - or %l1, %g2, %l1 stxa %l1, [%g1] ASI_DMMU flush %g6 retl nop -cplus_wfinsn_1: - sethi %uhi(CTX_CHEETAH_PLUS_NUC), %l1 -cplus_wfinsn_2: - sethi %hi(CTX_CHEETAH_PLUS_CTX0), %g2 - .align 32 /* Here are the rules, pay attention. @@ -318,7 +309,7 @@ fill_fixup_dax: nop rdpr %pstate, %l1 ! Prepare to change globals. mov %g4, %o1 ! Setup args for - mov %g5, %o2 ! final call to data_access_exception. + mov %g5, %o2 ! final call to spitfire_data_access_exception. andn %l1, PSTATE_MM, %l1 ! We want to be in RMO mov %g6, %o7 ! Stash away current. @@ -330,7 +321,7 @@ fill_fixup_dax: mov TSB_REG, %g1 ldxa [%g1] ASI_IMMU, %g5 #endif - call data_access_exception + call spitfire_data_access_exception add %sp, PTREGS_OFF, %o0 b,pt %xcc, rtrap @@ -391,27 +382,7 @@ window_dax_from_user_common: 109: or %g7, %lo(109b), %g7 mov %l4, %o1 mov %l5, %o2 - call data_access_exception + call spitfire_data_access_exception add %sp, PTREGS_OFF, %o0 ba,pt %xcc, rtrap clr %l6 - - - .globl cheetah_plus_patch_winfixup -cheetah_plus_patch_winfixup: - sethi %hi(cplus_wfinsn_1), %o0 - sethi %hi(cplus_winfixup_insn_1), %o2 - lduw [%o0 + %lo(cplus_wfinsn_1)], %o1 - or %o2, %lo(cplus_winfixup_insn_1), %o2 - stw %o1, [%o2] - flush %o2 - - sethi %hi(cplus_wfinsn_2), %o0 - sethi %hi(cplus_winfixup_insn_2), %o2 - lduw [%o0 + %lo(cplus_wfinsn_2)], %o1 - or %o2, %lo(cplus_winfixup_insn_2), %o2 - stw %o1, [%o2] - flush %o2 - - retl - nop diff --git a/arch/sparc64/lib/Makefile b/arch/sparc64/lib/Makefile index 40dbeec7e5d6..c295806500f7 100644 --- a/arch/sparc64/lib/Makefile +++ b/arch/sparc64/lib/Makefile @@ -14,7 +14,4 @@ lib-y := PeeCeeI.o copy_page.o clear_page.o strlen.o strncmp.o \ copy_in_user.o user_fixup.o memmove.o \ mcount.o ipcsum.o rwsem.o xor.o find_bit.o delay.o -lib-$(CONFIG_DEBUG_SPINLOCK) += debuglocks.o -lib-$(CONFIG_HAVE_DEC_LOCK) += dec_and_lock.o - obj-y += iomap.o diff --git a/arch/sparc64/lib/PeeCeeI.c b/arch/sparc64/lib/PeeCeeI.c index 3008d536e8c2..3c6cfbb20360 100644 --- a/arch/sparc64/lib/PeeCeeI.c +++ b/arch/sparc64/lib/PeeCeeI.c @@ -7,28 +7,31 @@ #include <asm/io.h> #include <asm/byteorder.h> -void outsb(void __iomem *addr, const void *src, unsigned long count) +void outsb(unsigned long __addr, const void *src, unsigned long count) { + void __iomem *addr = (void __iomem *) __addr; const u8 *p = src; - while(count--) + while (count--) outb(*p++, addr); } -void outsw(void __iomem *addr, const void *src, unsigned long count) +void outsw(unsigned long __addr, const void *src, unsigned long count) { - if(count) { + void __iomem *addr = (void __iomem *) __addr; + + if (count) { u16 *ps = (u16 *)src; u32 *pi; - if(((u64)src) & 0x2) { + if (((u64)src) & 0x2) { u16 val = le16_to_cpup(ps); outw(val, addr); ps++; count--; } pi = (u32 *)ps; - while(count >= 2) { + while (count >= 2) { u32 w = le32_to_cpup(pi); pi++; @@ -37,19 +40,21 @@ void outsw(void __iomem *addr, const void *src, unsigned long count) count -= 2; } ps = (u16 *)pi; - if(count) { + if (count) { u16 val = le16_to_cpup(ps); outw(val, addr); } } } -void outsl(void __iomem *addr, const void *src, unsigned long count) +void outsl(unsigned long __addr, const void *src, unsigned long count) { - if(count) { - if((((u64)src) & 0x3) == 0) { + void __iomem *addr = (void __iomem *) __addr; + + if (count) { + if ((((u64)src) & 0x3) == 0) { u32 *p = (u32 *)src; - while(count--) { + while (count--) { u32 val = cpu_to_le32p(p); outl(val, addr); p++; @@ -60,13 +65,13 @@ void outsl(void __iomem *addr, const void *src, unsigned long count) u32 l = 0, l2; u32 *pi; - switch(((u64)src) & 0x3) { + switch (((u64)src) & 0x3) { case 0x2: count -= 1; l = cpu_to_le16p(ps) << 16; ps++; pi = (u32 *)ps; - while(count--) { + while (count--) { l2 = cpu_to_le32p(pi); pi++; outl(((l >> 16) | (l2 << 16)), addr); @@ -86,7 +91,7 @@ void outsl(void __iomem *addr, const void *src, unsigned long count) ps++; l |= (l2 << 16); pi = (u32 *)ps; - while(count--) { + while (count--) { l2 = cpu_to_le32p(pi); pi++; outl(((l >> 8) | (l2 << 24)), addr); @@ -101,7 +106,7 @@ void outsl(void __iomem *addr, const void *src, unsigned long count) pb = (u8 *)src; l = (*pb++ << 24); pi = (u32 *)pb; - while(count--) { + while (count--) { l2 = cpu_to_le32p(pi); pi++; outl(((l >> 24) | (l2 << 8)), addr); @@ -119,16 +124,18 @@ void outsl(void __iomem *addr, const void *src, unsigned long count) } } -void insb(void __iomem *addr, void *dst, unsigned long count) +void insb(unsigned long __addr, void *dst, unsigned long count) { - if(count) { + void __iomem *addr = (void __iomem *) __addr; + + if (count) { u32 *pi; u8 *pb = dst; - while((((unsigned long)pb) & 0x3) && count--) + while ((((unsigned long)pb) & 0x3) && count--) *pb++ = inb(addr); pi = (u32 *)pb; - while(count >= 4) { + while (count >= 4) { u32 w; w = (inb(addr) << 24); @@ -139,23 +146,25 @@ void insb(void __iomem *addr, void *dst, unsigned long count) count -= 4; } pb = (u8 *)pi; - while(count--) + while (count--) *pb++ = inb(addr); } } -void insw(void __iomem *addr, void *dst, unsigned long count) +void insw(unsigned long __addr, void *dst, unsigned long count) { - if(count) { + void __iomem *addr = (void __iomem *) __addr; + + if (count) { u16 *ps = dst; u32 *pi; - if(((unsigned long)ps) & 0x2) { + if (((unsigned long)ps) & 0x2) { *ps++ = le16_to_cpu(inw(addr)); count--; } pi = (u32 *)ps; - while(count >= 2) { + while (count >= 2) { u32 w; w = (le16_to_cpu(inw(addr)) << 16); @@ -164,31 +173,33 @@ void insw(void __iomem *addr, void *dst, unsigned long count) count -= 2; } ps = (u16 *)pi; - if(count) + if (count) *ps = le16_to_cpu(inw(addr)); } } -void insl(void __iomem *addr, void *dst, unsigned long count) +void insl(unsigned long __addr, void *dst, unsigned long count) { - if(count) { - if((((unsigned long)dst) & 0x3) == 0) { + void __iomem *addr = (void __iomem *) __addr; + + if (count) { + if ((((unsigned long)dst) & 0x3) == 0) { u32 *pi = dst; - while(count--) + while (count--) *pi++ = le32_to_cpu(inl(addr)); } else { u32 l = 0, l2, *pi; u16 *ps; u8 *pb; - switch(((unsigned long)dst) & 3) { + switch (((unsigned long)dst) & 3) { case 0x2: ps = dst; count -= 1; l = le32_to_cpu(inl(addr)); *ps++ = l; pi = (u32 *)ps; - while(count--) { + while (count--) { l2 = le32_to_cpu(inl(addr)); *pi++ = (l << 16) | (l2 >> 16); l = l2; @@ -205,7 +216,7 @@ void insl(void __iomem *addr, void *dst, unsigned long count) ps = (u16 *)pb; *ps++ = ((l >> 8) & 0xffff); pi = (u32 *)ps; - while(count--) { + while (count--) { l2 = le32_to_cpu(inl(addr)); *pi++ = (l << 24) | (l2 >> 8); l = l2; @@ -220,7 +231,7 @@ void insl(void __iomem *addr, void *dst, unsigned long count) l = le32_to_cpu(inl(addr)); *pb++ = l >> 24; pi = (u32 *)pb; - while(count--) { + while (count--) { l2 = le32_to_cpu(inl(addr)); *pi++ = (l << 8) | (l2 >> 24); l = l2; diff --git a/arch/sparc64/lib/VISsave.S b/arch/sparc64/lib/VISsave.S index 4e18989bd602..a0ded5c5aa5c 100644 --- a/arch/sparc64/lib/VISsave.S +++ b/arch/sparc64/lib/VISsave.S @@ -59,15 +59,17 @@ vis1: ldub [%g6 + TI_FPSAVED], %g3 be,pn %icc, 9b add %g6, TI_FPREGS, %g2 andcc %o5, FPRS_DL, %g0 - membar #StoreStore | #LoadStore be,pn %icc, 4f add %g6, TI_FPREGS+0x40, %g3 + membar #Sync stda %f0, [%g2 + %g1] ASI_BLK_P stda %f16, [%g3 + %g1] ASI_BLK_P + membar #Sync andcc %o5, FPRS_DU, %g0 be,pn %icc, 5f 4: add %g1, 128, %g1 + membar #Sync stda %f32, [%g2 + %g1] ASI_BLK_P stda %f48, [%g3 + %g1] ASI_BLK_P @@ -87,7 +89,7 @@ vis1: ldub [%g6 + TI_FPSAVED], %g3 sll %g1, 5, %g1 add %g6, TI_FPREGS+0xc0, %g3 wr %g0, FPRS_FEF, %fprs - membar #StoreStore | #LoadStore + membar #Sync stda %f32, [%g2 + %g1] ASI_BLK_P stda %f48, [%g3 + %g1] ASI_BLK_P membar #Sync @@ -128,8 +130,8 @@ VISenterhalf: be,pn %icc, 4f add %g6, TI_FPREGS, %g2 - membar #StoreStore | #LoadStore add %g6, TI_FPREGS+0x40, %g3 + membar #Sync stda %f0, [%g2 + %g1] ASI_BLK_P stda %f16, [%g3 + %g1] ASI_BLK_P membar #Sync diff --git a/arch/sparc64/lib/copy_page.S b/arch/sparc64/lib/copy_page.S index 23ebf2c970b7..feebb14fd27a 100644 --- a/arch/sparc64/lib/copy_page.S +++ b/arch/sparc64/lib/copy_page.S @@ -87,7 +87,7 @@ copy_user_page: /* %o0=dest, %o1=src, %o2=vaddr */ membar #Sync wrpr %o2, 0x0, %pstate - BRANCH_IF_ANY_CHEETAH(g3,o2,1f) +cheetah_copy_page_insn: ba,pt %xcc, 9f nop @@ -240,3 +240,14 @@ copy_user_page: /* %o0=dest, %o1=src, %o2=vaddr */ stw %o4, [%g6 + TI_PRE_COUNT] .size copy_user_page, .-copy_user_page + + .globl cheetah_patch_copy_page +cheetah_patch_copy_page: + sethi %hi(0x01000000), %o1 ! NOP + sethi %hi(cheetah_copy_page_insn), %o0 + or %o0, %lo(cheetah_copy_page_insn), %o0 + stw %o1, [%o0] + membar #StoreStore + flush %o0 + retl + nop diff --git a/arch/sparc64/lib/debuglocks.c b/arch/sparc64/lib/debuglocks.c deleted file mode 100644 index f03344cf784e..000000000000 --- a/arch/sparc64/lib/debuglocks.c +++ /dev/null @@ -1,378 +0,0 @@ -/* $Id: debuglocks.c,v 1.9 2001/11/17 00:10:48 davem Exp $ - * debuglocks.c: Debugging versions of SMP locking primitives. - * - * Copyright (C) 1998 David S. Miller (davem@redhat.com) - */ - -#include <linux/config.h> -#include <linux/kernel.h> -#include <linux/sched.h> -#include <linux/spinlock.h> -#include <asm/system.h> - -#ifdef CONFIG_SMP - -#define GET_CALLER(PC) __asm__ __volatile__("mov %%i7, %0" : "=r" (PC)) - -static inline void show (char *str, spinlock_t *lock, unsigned long caller) -{ - int cpu = smp_processor_id(); - - printk("%s(%p) CPU#%d stuck at %08x, owner PC(%08x):CPU(%x)\n", - str, lock, cpu, (unsigned int) caller, - lock->owner_pc, lock->owner_cpu); -} - -static inline void show_read (char *str, rwlock_t *lock, unsigned long caller) -{ - int cpu = smp_processor_id(); - - printk("%s(%p) CPU#%d stuck at %08x, writer PC(%08x):CPU(%x)\n", - str, lock, cpu, (unsigned int) caller, - lock->writer_pc, lock->writer_cpu); -} - -static inline void show_write (char *str, rwlock_t *lock, unsigned long caller) -{ - int cpu = smp_processor_id(); - int i; - - printk("%s(%p) CPU#%d stuck at %08x\n", - str, lock, cpu, (unsigned int) caller); - printk("Writer: PC(%08x):CPU(%x)\n", - lock->writer_pc, lock->writer_cpu); - printk("Readers:"); - for (i = 0; i < NR_CPUS; i++) - if (lock->reader_pc[i]) - printk(" %d[%08x]", i, lock->reader_pc[i]); - printk("\n"); -} - -#undef INIT_STUCK -#define INIT_STUCK 100000000 - -void _do_spin_lock(spinlock_t *lock, char *str) -{ - unsigned long caller, val; - int stuck = INIT_STUCK; - int cpu = get_cpu(); - int shown = 0; - - GET_CALLER(caller); -again: - __asm__ __volatile__("ldstub [%1], %0" - : "=r" (val) - : "r" (&(lock->lock)) - : "memory"); - membar("#StoreLoad | #StoreStore"); - if (val) { - while (lock->lock) { - if (!--stuck) { - if (shown++ <= 2) - show(str, lock, caller); - stuck = INIT_STUCK; - } - membar("#LoadLoad"); - } - goto again; - } - lock->owner_pc = ((unsigned int)caller); - lock->owner_cpu = cpu; - current->thread.smp_lock_count++; - current->thread.smp_lock_pc = ((unsigned int)caller); - - put_cpu(); -} - -int _do_spin_trylock(spinlock_t *lock) -{ - unsigned long val, caller; - int cpu = get_cpu(); - - GET_CALLER(caller); - __asm__ __volatile__("ldstub [%1], %0" - : "=r" (val) - : "r" (&(lock->lock)) - : "memory"); - membar("#StoreLoad | #StoreStore"); - if (!val) { - lock->owner_pc = ((unsigned int)caller); - lock->owner_cpu = cpu; - current->thread.smp_lock_count++; - current->thread.smp_lock_pc = ((unsigned int)caller); - } - - put_cpu(); - - return val == 0; -} - -void _do_spin_unlock(spinlock_t *lock) -{ - lock->owner_pc = 0; - lock->owner_cpu = NO_PROC_ID; - membar("#StoreStore | #LoadStore"); - lock->lock = 0; - current->thread.smp_lock_count--; -} - -/* Keep INIT_STUCK the same... */ - -void _do_read_lock (rwlock_t *rw, char *str) -{ - unsigned long caller, val; - int stuck = INIT_STUCK; - int cpu = get_cpu(); - int shown = 0; - - GET_CALLER(caller); -wlock_again: - /* Wait for any writer to go away. */ - while (((long)(rw->lock)) < 0) { - if (!--stuck) { - if (shown++ <= 2) - show_read(str, rw, caller); - stuck = INIT_STUCK; - } - membar("#LoadLoad"); - } - /* Try once to increment the counter. */ - __asm__ __volatile__( -" ldx [%0], %%g1\n" -" brlz,a,pn %%g1, 2f\n" -" mov 1, %0\n" -" add %%g1, 1, %%g7\n" -" casx [%0], %%g1, %%g7\n" -" sub %%g1, %%g7, %0\n" -"2:" : "=r" (val) - : "0" (&(rw->lock)) - : "g1", "g7", "memory"); - membar("#StoreLoad | #StoreStore"); - if (val) - goto wlock_again; - rw->reader_pc[cpu] = ((unsigned int)caller); - current->thread.smp_lock_count++; - current->thread.smp_lock_pc = ((unsigned int)caller); - - put_cpu(); -} - -void _do_read_unlock (rwlock_t *rw, char *str) -{ - unsigned long caller, val; - int stuck = INIT_STUCK; - int cpu = get_cpu(); - int shown = 0; - - GET_CALLER(caller); - - /* Drop our identity _first_. */ - rw->reader_pc[cpu] = 0; - current->thread.smp_lock_count--; -runlock_again: - /* Spin trying to decrement the counter using casx. */ - __asm__ __volatile__( -" membar #StoreLoad | #LoadLoad\n" -" ldx [%0], %%g1\n" -" sub %%g1, 1, %%g7\n" -" casx [%0], %%g1, %%g7\n" -" membar #StoreLoad | #StoreStore\n" -" sub %%g1, %%g7, %0\n" - : "=r" (val) - : "0" (&(rw->lock)) - : "g1", "g7", "memory"); - if (val) { - if (!--stuck) { - if (shown++ <= 2) - show_read(str, rw, caller); - stuck = INIT_STUCK; - } - goto runlock_again; - } - - put_cpu(); -} - -void _do_write_lock (rwlock_t *rw, char *str) -{ - unsigned long caller, val; - int stuck = INIT_STUCK; - int cpu = get_cpu(); - int shown = 0; - - GET_CALLER(caller); -wlock_again: - /* Spin while there is another writer. */ - while (((long)rw->lock) < 0) { - if (!--stuck) { - if (shown++ <= 2) - show_write(str, rw, caller); - stuck = INIT_STUCK; - } - membar("#LoadLoad"); - } - - /* Try to acuire the write bit. */ - __asm__ __volatile__( -" mov 1, %%g3\n" -" sllx %%g3, 63, %%g3\n" -" ldx [%0], %%g1\n" -" brlz,pn %%g1, 1f\n" -" or %%g1, %%g3, %%g7\n" -" casx [%0], %%g1, %%g7\n" -" membar #StoreLoad | #StoreStore\n" -" ba,pt %%xcc, 2f\n" -" sub %%g1, %%g7, %0\n" -"1: mov 1, %0\n" -"2:" : "=r" (val) - : "0" (&(rw->lock)) - : "g3", "g1", "g7", "memory"); - if (val) { - /* We couldn't get the write bit. */ - if (!--stuck) { - if (shown++ <= 2) - show_write(str, rw, caller); - stuck = INIT_STUCK; - } - goto wlock_again; - } - if ((rw->lock & ((1UL<<63)-1UL)) != 0UL) { - /* Readers still around, drop the write - * lock, spin, and try again. - */ - if (!--stuck) { - if (shown++ <= 2) - show_write(str, rw, caller); - stuck = INIT_STUCK; - } - __asm__ __volatile__( -" mov 1, %%g3\n" -" sllx %%g3, 63, %%g3\n" -"1: ldx [%0], %%g1\n" -" andn %%g1, %%g3, %%g7\n" -" casx [%0], %%g1, %%g7\n" -" cmp %%g1, %%g7\n" -" membar #StoreLoad | #StoreStore\n" -" bne,pn %%xcc, 1b\n" -" nop" - : /* no outputs */ - : "r" (&(rw->lock)) - : "g3", "g1", "g7", "cc", "memory"); - while(rw->lock != 0) { - if (!--stuck) { - if (shown++ <= 2) - show_write(str, rw, caller); - stuck = INIT_STUCK; - } - membar("#LoadLoad"); - } - goto wlock_again; - } - - /* We have it, say who we are. */ - rw->writer_pc = ((unsigned int)caller); - rw->writer_cpu = cpu; - current->thread.smp_lock_count++; - current->thread.smp_lock_pc = ((unsigned int)caller); - - put_cpu(); -} - -void _do_write_unlock(rwlock_t *rw) -{ - unsigned long caller, val; - int stuck = INIT_STUCK; - int shown = 0; - - GET_CALLER(caller); - - /* Drop our identity _first_ */ - rw->writer_pc = 0; - rw->writer_cpu = NO_PROC_ID; - current->thread.smp_lock_count--; -wlock_again: - __asm__ __volatile__( -" membar #StoreLoad | #LoadLoad\n" -" mov 1, %%g3\n" -" sllx %%g3, 63, %%g3\n" -" ldx [%0], %%g1\n" -" andn %%g1, %%g3, %%g7\n" -" casx [%0], %%g1, %%g7\n" -" membar #StoreLoad | #StoreStore\n" -" sub %%g1, %%g7, %0\n" - : "=r" (val) - : "0" (&(rw->lock)) - : "g3", "g1", "g7", "memory"); - if (val) { - if (!--stuck) { - if (shown++ <= 2) - show_write("write_unlock", rw, caller); - stuck = INIT_STUCK; - } - goto wlock_again; - } -} - -int _do_write_trylock (rwlock_t *rw, char *str) -{ - unsigned long caller, val; - int cpu = get_cpu(); - - GET_CALLER(caller); - - /* Try to acuire the write bit. */ - __asm__ __volatile__( -" mov 1, %%g3\n" -" sllx %%g3, 63, %%g3\n" -" ldx [%0], %%g1\n" -" brlz,pn %%g1, 1f\n" -" or %%g1, %%g3, %%g7\n" -" casx [%0], %%g1, %%g7\n" -" membar #StoreLoad | #StoreStore\n" -" ba,pt %%xcc, 2f\n" -" sub %%g1, %%g7, %0\n" -"1: mov 1, %0\n" -"2:" : "=r" (val) - : "0" (&(rw->lock)) - : "g3", "g1", "g7", "memory"); - - if (val) { - put_cpu(); - return 0; - } - - if ((rw->lock & ((1UL<<63)-1UL)) != 0UL) { - /* Readers still around, drop the write - * lock, return failure. - */ - __asm__ __volatile__( -" mov 1, %%g3\n" -" sllx %%g3, 63, %%g3\n" -"1: ldx [%0], %%g1\n" -" andn %%g1, %%g3, %%g7\n" -" casx [%0], %%g1, %%g7\n" -" cmp %%g1, %%g7\n" -" membar #StoreLoad | #StoreStore\n" -" bne,pn %%xcc, 1b\n" -" nop" - : /* no outputs */ - : "r" (&(rw->lock)) - : "g3", "g1", "g7", "cc", "memory"); - - put_cpu(); - - return 0; - } - - /* We have it, say who we are. */ - rw->writer_pc = ((unsigned int)caller); - rw->writer_cpu = cpu; - current->thread.smp_lock_count++; - current->thread.smp_lock_pc = ((unsigned int)caller); - - put_cpu(); - - return 1; -} - -#endif /* CONFIG_SMP */ diff --git a/arch/sparc64/lib/dec_and_lock.S b/arch/sparc64/lib/dec_and_lock.S deleted file mode 100644 index 8ee288dd0afc..000000000000 --- a/arch/sparc64/lib/dec_and_lock.S +++ /dev/null @@ -1,80 +0,0 @@ -/* $Id: dec_and_lock.S,v 1.5 2001/11/18 00:12:56 davem Exp $ - * dec_and_lock.S: Sparc64 version of "atomic_dec_and_lock()" - * using cas and ldstub instructions. - * - * Copyright (C) 2000 David S. Miller (davem@redhat.com) - */ -#include <linux/config.h> -#include <asm/thread_info.h> - - .text - .align 64 - - /* CAS basically works like this: - * - * void CAS(MEM, REG1, REG2) - * { - * START_ATOMIC(); - * if (*(MEM) == REG1) { - * TMP = *(MEM); - * *(MEM) = REG2; - * REG2 = TMP; - * } else - * REG2 = *(MEM); - * END_ATOMIC(); - * } - */ - - .globl _atomic_dec_and_lock -_atomic_dec_and_lock: /* %o0 = counter, %o1 = lock */ -loop1: lduw [%o0], %g2 - subcc %g2, 1, %g7 - be,pn %icc, start_to_zero - nop -nzero: cas [%o0], %g2, %g7 - cmp %g2, %g7 - bne,pn %icc, loop1 - mov 0, %g1 - -out: - membar #StoreLoad | #StoreStore - retl - mov %g1, %o0 -start_to_zero: -#ifdef CONFIG_PREEMPT - ldsw [%g6 + TI_PRE_COUNT], %g3 - add %g3, 1, %g3 - stw %g3, [%g6 + TI_PRE_COUNT] -#endif -to_zero: - ldstub [%o1], %g3 - membar #StoreLoad | #StoreStore - brnz,pn %g3, spin_on_lock - nop -loop2: cas [%o0], %g2, %g7 /* ASSERT(g7 == 0) */ - cmp %g2, %g7 - - be,pt %icc, out - mov 1, %g1 - lduw [%o0], %g2 - subcc %g2, 1, %g7 - be,pn %icc, loop2 - nop - membar #StoreStore | #LoadStore - stb %g0, [%o1] -#ifdef CONFIG_PREEMPT - ldsw [%g6 + TI_PRE_COUNT], %g3 - sub %g3, 1, %g3 - stw %g3, [%g6 + TI_PRE_COUNT] -#endif - - b,pt %xcc, nzero - nop -spin_on_lock: - ldub [%o1], %g3 - membar #LoadLoad - brnz,pt %g3, spin_on_lock - nop - ba,pt %xcc, to_zero - nop - nop diff --git a/arch/sparc64/lib/strncpy_from_user.S b/arch/sparc64/lib/strncpy_from_user.S index 09cbbaa0ebf4..e1264650ca7a 100644 --- a/arch/sparc64/lib/strncpy_from_user.S +++ b/arch/sparc64/lib/strncpy_from_user.S @@ -125,15 +125,11 @@ __strncpy_from_user: add %o2, %o3, %o0 .size __strncpy_from_user, .-__strncpy_from_user - .section .fixup,#alloc,#execinstr - .align 4 -4: retl - mov -EFAULT, %o0 - .section __ex_table,#alloc .align 4 - .word 60b, 4b - .word 61b, 4b - .word 62b, 4b - .word 63b, 4b - .word 64b, 4b + .word 60b, __retl_efault + .word 61b, __retl_efault + .word 62b, __retl_efault + .word 63b, __retl_efault + .word 64b, __retl_efault + .previous diff --git a/arch/sparc64/lib/user_fixup.c b/arch/sparc64/lib/user_fixup.c index 0278e34125db..19d1fdb17d0e 100644 --- a/arch/sparc64/lib/user_fixup.c +++ b/arch/sparc64/lib/user_fixup.c @@ -11,61 +11,56 @@ /* Calculating the exact fault address when using * block loads and stores can be very complicated. + * * Instead of trying to be clever and handling all * of the cases, just fix things up simply here. */ -unsigned long copy_from_user_fixup(void *to, const void __user *from, unsigned long size) +static unsigned long compute_size(unsigned long start, unsigned long size, unsigned long *offset) { - char *dst = to; - const char __user *src = from; + unsigned long fault_addr = current_thread_info()->fault_address; + unsigned long end = start + size; - while (size) { - if (__get_user(*dst, src)) - break; - dst++; - src++; - size--; + if (fault_addr < start || fault_addr >= end) { + *offset = 0; + } else { + *offset = start - fault_addr; + size = end - fault_addr; } + return size; +} - if (size) - memset(dst, 0, size); +unsigned long copy_from_user_fixup(void *to, const void __user *from, unsigned long size) +{ + unsigned long offset; + + size = compute_size((unsigned long) from, size, &offset); + if (likely(size)) + memset(to + offset, 0, size); return size; } unsigned long copy_to_user_fixup(void __user *to, const void *from, unsigned long size) { - char __user *dst = to; - const char *src = from; - - while (size) { - if (__put_user(*src, dst)) - break; - dst++; - src++; - size--; - } + unsigned long offset; - return size; + return compute_size((unsigned long) to, size, &offset); } unsigned long copy_in_user_fixup(void __user *to, void __user *from, unsigned long size) { - char __user *dst = to; - char __user *src = from; + unsigned long fault_addr = current_thread_info()->fault_address; + unsigned long start = (unsigned long) to; + unsigned long end = start + size; - while (size) { - char tmp; + if (fault_addr >= start && fault_addr < end) + return end - fault_addr; - if (__get_user(tmp, src)) - break; - if (__put_user(tmp, dst)) - break; - dst++; - src++; - size--; - } + start = (unsigned long) from; + end = start + size; + if (fault_addr >= start && fault_addr < end) + return end - fault_addr; return size; } diff --git a/arch/sparc64/mm/Makefile b/arch/sparc64/mm/Makefile index cda87333a77b..9d0960e69f48 100644 --- a/arch/sparc64/mm/Makefile +++ b/arch/sparc64/mm/Makefile @@ -5,6 +5,6 @@ EXTRA_AFLAGS := -ansi EXTRA_CFLAGS := -Werror -obj-y := ultra.o tlb.o fault.o init.o generic.o extable.o +obj-y := ultra.o tlb.o fault.o init.o generic.o obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o diff --git a/arch/sparc64/mm/extable.c b/arch/sparc64/mm/extable.c deleted file mode 100644 index ec334297ff4f..000000000000 --- a/arch/sparc64/mm/extable.c +++ /dev/null @@ -1,80 +0,0 @@ -/* - * linux/arch/sparc64/mm/extable.c - */ - -#include <linux/config.h> -#include <linux/module.h> -#include <asm/uaccess.h> - -extern const struct exception_table_entry __start___ex_table[]; -extern const struct exception_table_entry __stop___ex_table[]; - -void sort_extable(struct exception_table_entry *start, - struct exception_table_entry *finish) -{ -} - -/* Caller knows they are in a range if ret->fixup == 0 */ -const struct exception_table_entry * -search_extable(const struct exception_table_entry *start, - const struct exception_table_entry *last, - unsigned long value) -{ - const struct exception_table_entry *walk; - - /* Single insn entries are encoded as: - * word 1: insn address - * word 2: fixup code address - * - * Range entries are encoded as: - * word 1: first insn address - * word 2: 0 - * word 3: last insn address + 4 bytes - * word 4: fixup code address - * - * See asm/uaccess.h for more details. - */ - - /* 1. Try to find an exact match. */ - for (walk = start; walk <= last; walk++) { - if (walk->fixup == 0) { - /* A range entry, skip both parts. */ - walk++; - continue; - } - - if (walk->insn == value) - return walk; - } - - /* 2. Try to find a range match. */ - for (walk = start; walk <= (last - 1); walk++) { - if (walk->fixup) - continue; - - if (walk[0].insn <= value && walk[1].insn > value) - return walk; - - walk++; - } - - return NULL; -} - -/* Special extable search, which handles ranges. Returns fixup */ -unsigned long search_extables_range(unsigned long addr, unsigned long *g2) -{ - const struct exception_table_entry *entry; - - entry = search_exception_tables(addr); - if (!entry) - return 0; - - /* Inside range? Fix g2 and return correct fixup */ - if (!entry->fixup) { - *g2 = (addr - entry->insn) / 4; - return (entry + 1)->fixup; - } - - return entry->fixup; -} diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c index 52e9375288a9..31fbc67719a1 100644 --- a/arch/sparc64/mm/fault.c +++ b/arch/sparc64/mm/fault.c @@ -18,6 +18,7 @@ #include <linux/smp_lock.h> #include <linux/init.h> #include <linux/interrupt.h> +#include <linux/kprobes.h> #include <asm/page.h> #include <asm/pgtable.h> @@ -31,8 +32,6 @@ #define ELEMENTS(arr) (sizeof (arr)/sizeof (arr[0])) -extern struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; - /* * To debug kernel to catch accesses to certain virtual/physical addresses. * Mode = 0 selects physical watchpoints, mode = 1 selects virtual watchpoints. @@ -70,55 +69,9 @@ void set_brkpt(unsigned long addr, unsigned char mask, int flags, int mode) : "memory"); } -/* Nice, simple, prom library does all the sweating for us. ;) */ -unsigned long __init prom_probe_memory (void) -{ - register struct linux_mlist_p1275 *mlist; - register unsigned long bytes, base_paddr, tally; - register int i; - - i = 0; - mlist = *prom_meminfo()->p1275_available; - bytes = tally = mlist->num_bytes; - base_paddr = mlist->start_adr; - - sp_banks[0].base_addr = base_paddr; - sp_banks[0].num_bytes = bytes; - - while (mlist->theres_more != (void *) 0) { - i++; - mlist = mlist->theres_more; - bytes = mlist->num_bytes; - tally += bytes; - if (i >= SPARC_PHYS_BANKS-1) { - printk ("The machine has more banks than " - "this kernel can support\n" - "Increase the SPARC_PHYS_BANKS " - "setting (currently %d)\n", - SPARC_PHYS_BANKS); - i = SPARC_PHYS_BANKS-1; - break; - } - - sp_banks[i].base_addr = mlist->start_adr; - sp_banks[i].num_bytes = mlist->num_bytes; - } - - i++; - sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL; - sp_banks[i].num_bytes = 0; - - /* Now mask all bank sizes on a page boundary, it is all we can - * use anyways. - */ - for (i = 0; sp_banks[i].num_bytes != 0; i++) - sp_banks[i].num_bytes &= PAGE_MASK; - - return tally; -} - -static void unhandled_fault(unsigned long address, struct task_struct *tsk, - struct pt_regs *regs) +static void __kprobes unhandled_fault(unsigned long address, + struct task_struct *tsk, + struct pt_regs *regs) { if ((unsigned long) address < PAGE_SIZE) { printk(KERN_ALERT "Unable to handle kernel NULL " @@ -240,7 +193,6 @@ static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn) static void do_kernel_fault(struct pt_regs *regs, int si_code, int fault_code, unsigned int insn, unsigned long address) { - unsigned long g2; unsigned char asi = ASI_P; if ((!insn) && (regs->tstate & TSTATE_PRIV)) @@ -271,11 +223,9 @@ static void do_kernel_fault(struct pt_regs *regs, int si_code, int fault_code, } } - g2 = regs->u_regs[UREG_G2]; - /* Is this in ex_table? */ if (regs->tstate & TSTATE_PRIV) { - unsigned long fixup; + const struct exception_table_entry *entry; if (asi == ASI_P && (insn & 0xc0800000) == 0xc0800000) { if (insn & 0x2000) @@ -286,10 +236,9 @@ static void do_kernel_fault(struct pt_regs *regs, int si_code, int fault_code, /* Look in asi.h: All _S asis have LS bit set */ if ((asi & 0x1) && - (fixup = search_extables_range(regs->tpc, &g2))) { - regs->tpc = fixup; + (entry = search_exception_tables(regs->tpc))) { + regs->tpc = entry->fixup; regs->tnpc = regs->tpc + 4; - regs->u_regs[UREG_G2] = g2; return; } } else { @@ -304,7 +253,7 @@ cannot_handle: unhandled_fault (address, current, regs); } -asmlinkage void do_sparc64_fault(struct pt_regs *regs) +asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs) { struct mm_struct *mm = current->mm; struct vm_area_struct *vma; @@ -459,7 +408,7 @@ good_area: } up_read(&mm->mmap_sem); - goto fault_done; + return; /* * Something tried to access memory that isn't in our memory map.. @@ -471,8 +420,7 @@ bad_area: handle_kernel_fault: do_kernel_fault(regs, si_code, fault_code, insn, address); - - goto fault_done; + return; /* * We ran out of memory, or some other thing happened to us that made @@ -503,9 +451,4 @@ do_sigbus: /* Kernel mode? Handle exceptions or die */ if (regs->tstate & TSTATE_PRIV) goto handle_kernel_fault; - -fault_done: - /* These values are no longer needed, clear them. */ - set_thread_fault_code(0); - current_thread_info()->fault_address = 0; } diff --git a/arch/sparc64/mm/generic.c b/arch/sparc64/mm/generic.c index 6b31f6117a95..112c316e7cd2 100644 --- a/arch/sparc64/mm/generic.c +++ b/arch/sparc64/mm/generic.c @@ -116,37 +116,6 @@ static inline int io_remap_pud_range(struct mm_struct *mm, pud_t * pud, unsigned return 0; } -int io_remap_page_range(struct vm_area_struct *vma, unsigned long from, unsigned long offset, unsigned long size, pgprot_t prot, int space) -{ - int error = 0; - pgd_t * dir; - unsigned long beg = from; - unsigned long end = from + size; - struct mm_struct *mm = vma->vm_mm; - - prot = __pgprot(pg_iobits); - offset -= from; - dir = pgd_offset(mm, from); - flush_cache_range(vma, beg, end); - - spin_lock(&mm->page_table_lock); - while (from < end) { - pud_t *pud = pud_alloc(mm, dir, from); - error = -ENOMEM; - if (!pud) - break; - error = io_remap_pud_range(mm, pud, from, end - from, offset + from, prot, space); - if (error) - break; - from = (from + PGDIR_SIZE) & PGDIR_MASK; - dir++; - } - flush_tlb_range(vma, beg, end); - spin_unlock(&mm->page_table_lock); - - return error; -} - int io_remap_pfn_range(struct vm_area_struct *vma, unsigned long from, unsigned long pfn, unsigned long size, pgprot_t prot) { @@ -158,14 +127,16 @@ int io_remap_pfn_range(struct vm_area_struct *vma, unsigned long from, int space = GET_IOSPACE(pfn); unsigned long offset = GET_PFN(pfn) << PAGE_SHIFT; + /* See comment in mm/memory.c remap_pfn_range */ + vma->vm_flags |= VM_IO | VM_RESERVED; + prot = __pgprot(pg_iobits); offset -= from; dir = pgd_offset(mm, from); flush_cache_range(vma, beg, end); - spin_lock(&mm->page_table_lock); while (from < end) { - pud_t *pud = pud_alloc(current->mm, dir, from); + pud_t *pud = pud_alloc(mm, dir, from); error = -ENOMEM; if (!pud) break; @@ -175,8 +146,7 @@ int io_remap_pfn_range(struct vm_area_struct *vma, unsigned long from, from = (from + PGDIR_SIZE) & PGDIR_MASK; dir++; } - flush_tlb_range(vma, beg, end); - spin_unlock(&mm->page_table_lock); + flush_tlb_range(vma, beg, end); return error; } diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index 3fbaf342a452..1e44ee26cee8 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -19,6 +19,9 @@ #include <linux/pagemap.h> #include <linux/fs.h> #include <linux/seq_file.h> +#include <linux/kprobes.h> +#include <linux/cache.h> +#include <linux/sort.h> #include <asm/head.h> #include <asm/system.h> @@ -39,24 +42,80 @@ extern void device_scan(void); -struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS]; +#define MAX_BANKS 32 -unsigned long *sparc64_valid_addr_bitmap; +static struct linux_prom64_registers pavail[MAX_BANKS] __initdata; +static struct linux_prom64_registers pavail_rescan[MAX_BANKS] __initdata; +static int pavail_ents __initdata; +static int pavail_rescan_ents __initdata; + +static int cmp_p64(const void *a, const void *b) +{ + const struct linux_prom64_registers *x = a, *y = b; + + if (x->phys_addr > y->phys_addr) + return 1; + if (x->phys_addr < y->phys_addr) + return -1; + return 0; +} + +static void __init read_obp_memory(const char *property, + struct linux_prom64_registers *regs, + int *num_ents) +{ + int node = prom_finddevice("/memory"); + int prop_size = prom_getproplen(node, property); + int ents, ret, i; + + ents = prop_size / sizeof(struct linux_prom64_registers); + if (ents > MAX_BANKS) { + prom_printf("The machine has more %s property entries than " + "this kernel can support (%d).\n", + property, MAX_BANKS); + prom_halt(); + } + + ret = prom_getproperty(node, property, (char *) regs, prop_size); + if (ret == -1) { + prom_printf("Couldn't get %s property from /memory.\n"); + prom_halt(); + } + + *num_ents = ents; + + /* Sanitize what we got from the firmware, by page aligning + * everything. + */ + for (i = 0; i < ents; i++) { + unsigned long base, size; + + base = regs[i].phys_addr; + size = regs[i].reg_size; + + size &= PAGE_MASK; + if (base & ~PAGE_MASK) { + unsigned long new_base = PAGE_ALIGN(base); + + size -= new_base - base; + if ((long) size < 0L) + size = 0UL; + base = new_base; + } + regs[i].phys_addr = base; + regs[i].reg_size = size; + } + sort(regs, ents, sizeof(struct linux_prom64_registers), + cmp_p64, NULL); +} + +unsigned long *sparc64_valid_addr_bitmap __read_mostly; /* Ugly, but necessary... -DaveM */ -unsigned long phys_base; -unsigned long kern_base; -unsigned long kern_size; -unsigned long pfn_base; - -/* This is even uglier. We have a problem where the kernel may not be - * located at phys_base. However, initial __alloc_bootmem() calls need to - * be adjusted to be within the 4-8Megs that the kernel is mapped to, else - * those page mappings wont work. Things are ok after inherit_prom_mappings - * is called though. Dave says he'll clean this up some other time. - * -- BenC - */ -static unsigned long bootmap_base; +unsigned long phys_base __read_mostly; +unsigned long kern_base __read_mostly; +unsigned long kern_size __read_mostly; +unsigned long pfn_base __read_mostly; /* get_new_mmu_context() uses "cache + 1". */ DEFINE_SPINLOCK(ctx_alloc_lock); @@ -72,7 +131,13 @@ extern unsigned long sparc_ramdisk_image64; extern unsigned int sparc_ramdisk_image; extern unsigned int sparc_ramdisk_size; -struct page *mem_map_zero; +struct page *mem_map_zero __read_mostly; + +unsigned int sparc64_highest_unlocked_tlb_ent __read_mostly; + +unsigned long sparc64_kern_pri_context __read_mostly; +unsigned long sparc64_kern_pri_nuc_bits __read_mostly; +unsigned long sparc64_kern_sec_context __read_mostly; int bigkernel = 0; @@ -178,8 +243,6 @@ static __inline__ void clear_dcache_dirty_cpu(struct page *page, unsigned long c : "g1", "g7"); } -extern void __update_mmu_cache(unsigned long mmu_context_hw, unsigned long address, pte_t pte, int code); - void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t pte) { struct page *page; @@ -206,10 +269,6 @@ void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t p put_cpu(); } - - if (get_thread_fault_code()) - __update_mmu_cache(CTX_NRBITS(vma->vm_mm->context), - address, pte, get_thread_fault_code()); } void flush_dcache_page(struct page *page) @@ -250,7 +309,7 @@ out: put_cpu(); } -void flush_icache_range(unsigned long start, unsigned long end) +void __kprobes flush_icache_range(unsigned long start, unsigned long end) { /* Cheetah has coherent I-cache. */ if (tlb_type == spitfire) { @@ -309,6 +368,11 @@ struct linux_prom_translation { unsigned long data; }; +/* Exported for kernel TLB miss handling in ktlb.S */ +struct linux_prom_translation prom_trans[512] __read_mostly; +unsigned int prom_trans_ents __read_mostly; +unsigned int swapper_pgd_zero __read_mostly; + extern unsigned long prom_boot_page; extern void prom_remap(unsigned long physpage, unsigned long virtpage, int mmu_ihandle); extern int prom_get_mmu_ihandle(void); @@ -317,297 +381,162 @@ extern void register_prom_callbacks(void); /* Exported for SMP bootup purposes. */ unsigned long kern_locked_tte_data; -void __init early_pgtable_allocfail(char *type) -{ - prom_printf("inherit_prom_mappings: Cannot alloc kernel %s.\n", type); - prom_halt(); -} - -#define BASE_PAGE_SIZE 8192 -static pmd_t *prompmd; - /* * Translate PROM's mapping we capture at boot time into physical address. * The second parameter is only set from prom_callback() invocations. */ unsigned long prom_virt_to_phys(unsigned long promva, int *error) { - pmd_t *pmdp = prompmd + ((promva >> 23) & 0x7ff); - pte_t *ptep; - unsigned long base; - - if (pmd_none(*pmdp)) { - if (error) - *error = 1; - return(0); - } - ptep = (pte_t *)__pmd_page(*pmdp) + ((promva >> 13) & 0x3ff); - if (!pte_present(*ptep)) { - if (error) - *error = 1; - return(0); - } - if (error) { - *error = 0; - return(pte_val(*ptep)); + int i; + + for (i = 0; i < prom_trans_ents; i++) { + struct linux_prom_translation *p = &prom_trans[i]; + + if (promva >= p->virt && + promva < (p->virt + p->size)) { + unsigned long base = p->data & _PAGE_PADDR; + + if (error) + *error = 0; + return base + (promva & (8192 - 1)); + } } - base = pte_val(*ptep) & _PAGE_PADDR; - return(base + (promva & (BASE_PAGE_SIZE - 1))); + if (error) + *error = 1; + return 0UL; } -static void inherit_prom_mappings(void) +/* The obp translations are saved based on 8k pagesize, since obp can + * use a mixture of pagesizes. Misses to the LOW_OBP_ADDRESS -> + * HI_OBP_ADDRESS range are handled in ktlb.S and do not use the vpte + * scheme (also, see rant in inherit_locked_prom_mappings()). + */ +static inline int in_obp_range(unsigned long vaddr) { - struct linux_prom_translation *trans; - unsigned long phys_page, tte_vaddr, tte_data; - void (*remap_func)(unsigned long, unsigned long, int); - pmd_t *pmdp; - pte_t *ptep; - int node, n, i, tsz; - extern unsigned int obp_iaddr_patch[2], obp_daddr_patch[2]; + return (vaddr >= LOW_OBP_ADDRESS && + vaddr < HI_OBP_ADDRESS); +} + +static int cmp_ptrans(const void *a, const void *b) +{ + const struct linux_prom_translation *x = a, *y = b; + + if (x->virt > y->virt) + return 1; + if (x->virt < y->virt) + return -1; + return 0; +} + +/* Read OBP translations property into 'prom_trans[]'. */ +static void __init read_obp_translations(void) +{ + int n, node, ents, first, last, i; node = prom_finddevice("/virtual-memory"); n = prom_getproplen(node, "translations"); - if (n == 0 || n == -1) { - prom_printf("Couldn't get translation property\n"); + if (unlikely(n == 0 || n == -1)) { + prom_printf("prom_mappings: Couldn't get size.\n"); prom_halt(); } - n += 5 * sizeof(struct linux_prom_translation); - for (tsz = 1; tsz < n; tsz <<= 1) - /* empty */; - trans = __alloc_bootmem(tsz, SMP_CACHE_BYTES, bootmap_base); - if (trans == NULL) { - prom_printf("inherit_prom_mappings: Cannot alloc translations.\n"); + if (unlikely(n > sizeof(prom_trans))) { + prom_printf("prom_mappings: Size %Zd is too big.\n", n); prom_halt(); } - memset(trans, 0, tsz); - if ((n = prom_getproperty(node, "translations", (char *)trans, tsz)) == -1) { - prom_printf("Couldn't get translation property\n"); + if ((n = prom_getproperty(node, "translations", + (char *)&prom_trans[0], + sizeof(prom_trans))) == -1) { + prom_printf("prom_mappings: Couldn't get property.\n"); prom_halt(); } - n = n / sizeof(*trans); - /* - * The obp translations are saved based on 8k pagesize, since obp can - * use a mixture of pagesizes. Misses to the 0xf0000000 - 0x100000000, - * ie obp range, are handled in entry.S and do not use the vpte scheme - * (see rant in inherit_locked_prom_mappings()). - */ -#define OBP_PMD_SIZE 2048 - prompmd = __alloc_bootmem(OBP_PMD_SIZE, OBP_PMD_SIZE, bootmap_base); - if (prompmd == NULL) - early_pgtable_allocfail("pmd"); - memset(prompmd, 0, OBP_PMD_SIZE); - for (i = 0; i < n; i++) { - unsigned long vaddr; - - if (trans[i].virt >= LOW_OBP_ADDRESS && trans[i].virt < HI_OBP_ADDRESS) { - for (vaddr = trans[i].virt; - ((vaddr < trans[i].virt + trans[i].size) && - (vaddr < HI_OBP_ADDRESS)); - vaddr += BASE_PAGE_SIZE) { - unsigned long val; - - pmdp = prompmd + ((vaddr >> 23) & 0x7ff); - if (pmd_none(*pmdp)) { - ptep = __alloc_bootmem(BASE_PAGE_SIZE, - BASE_PAGE_SIZE, - bootmap_base); - if (ptep == NULL) - early_pgtable_allocfail("pte"); - memset(ptep, 0, BASE_PAGE_SIZE); - pmd_set(pmdp, ptep); - } - ptep = (pte_t *)__pmd_page(*pmdp) + - ((vaddr >> 13) & 0x3ff); + n = n / sizeof(struct linux_prom_translation); - val = trans[i].data; + ents = n; - /* Clear diag TTE bits. */ - if (tlb_type == spitfire) - val &= ~0x0003fe0000000000UL; + sort(prom_trans, ents, sizeof(struct linux_prom_translation), + cmp_ptrans, NULL); - set_pte_at(&init_mm, vaddr, - ptep, __pte(val | _PAGE_MODIFIED)); - trans[i].data += BASE_PAGE_SIZE; - } - } + /* Now kick out all the non-OBP entries. */ + for (i = 0; i < ents; i++) { + if (in_obp_range(prom_trans[i].virt)) + break; + } + first = i; + for (; i < ents; i++) { + if (!in_obp_range(prom_trans[i].virt)) + break; } - phys_page = __pa(prompmd); - obp_iaddr_patch[0] |= (phys_page >> 10); - obp_iaddr_patch[1] |= (phys_page & 0x3ff); - flushi((long)&obp_iaddr_patch[0]); - obp_daddr_patch[0] |= (phys_page >> 10); - obp_daddr_patch[1] |= (phys_page & 0x3ff); - flushi((long)&obp_daddr_patch[0]); + last = i; - /* Now fixup OBP's idea about where we really are mapped. */ - prom_printf("Remapping the kernel... "); + for (i = 0; i < (last - first); i++) { + struct linux_prom_translation *src = &prom_trans[i + first]; + struct linux_prom_translation *dest = &prom_trans[i]; - /* Spitfire Errata #32 workaround */ - /* NOTE: Using plain zero for the context value is - * correct here, we are not using the Linux trap - * tables yet so we should not use the special - * UltraSPARC-III+ page size encodings yet. - */ - __asm__ __volatile__("stxa %0, [%1] %2\n\t" - "flush %%g6" - : /* No outputs */ - : "r" (0), "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU)); - - switch (tlb_type) { - default: - case spitfire: - phys_page = spitfire_get_dtlb_data(sparc64_highest_locked_tlbent()); - break; - - case cheetah: - case cheetah_plus: - phys_page = cheetah_get_litlb_data(sparc64_highest_locked_tlbent()); - break; - }; - - phys_page &= _PAGE_PADDR; - phys_page += ((unsigned long)&prom_boot_page - - (unsigned long)KERNBASE); + *dest = *src; + } + for (; i < ents; i++) { + struct linux_prom_translation *dest = &prom_trans[i]; + dest->virt = dest->size = dest->data = 0x0UL; + } + + prom_trans_ents = last - first; if (tlb_type == spitfire) { - /* Lock this into i/d tlb entry 59 */ - __asm__ __volatile__( - "stxa %%g0, [%2] %3\n\t" - "stxa %0, [%1] %4\n\t" - "membar #Sync\n\t" - "flush %%g6\n\t" - "stxa %%g0, [%2] %5\n\t" - "stxa %0, [%1] %6\n\t" - "membar #Sync\n\t" - "flush %%g6" - : : "r" (phys_page | _PAGE_VALID | _PAGE_SZ8K | _PAGE_CP | - _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W), - "r" (59 << 3), "r" (TLB_TAG_ACCESS), - "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), - "i" (ASI_IMMU), "i" (ASI_ITLB_DATA_ACCESS) - : "memory"); - } else if (tlb_type == cheetah || tlb_type == cheetah_plus) { - /* Lock this into i/d tlb-0 entry 11 */ - __asm__ __volatile__( - "stxa %%g0, [%2] %3\n\t" - "stxa %0, [%1] %4\n\t" - "membar #Sync\n\t" - "flush %%g6\n\t" - "stxa %%g0, [%2] %5\n\t" - "stxa %0, [%1] %6\n\t" - "membar #Sync\n\t" - "flush %%g6" - : : "r" (phys_page | _PAGE_VALID | _PAGE_SZ8K | _PAGE_CP | - _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W), - "r" ((0 << 16) | (11 << 3)), "r" (TLB_TAG_ACCESS), - "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), - "i" (ASI_IMMU), "i" (ASI_ITLB_DATA_ACCESS) - : "memory"); - } else { - /* Implement me :-) */ - BUG(); + /* Clear diag TTE bits. */ + for (i = 0; i < prom_trans_ents; i++) + prom_trans[i].data &= ~0x0003fe0000000000UL; } +} - tte_vaddr = (unsigned long) KERNBASE; +static void __init remap_kernel(void) +{ + unsigned long phys_page, tte_vaddr, tte_data; + int tlb_ent = sparc64_highest_locked_tlbent(); - /* Spitfire Errata #32 workaround */ - /* NOTE: Using plain zero for the context value is - * correct here, we are not using the Linux trap - * tables yet so we should not use the special - * UltraSPARC-III+ page size encodings yet. - */ - __asm__ __volatile__("stxa %0, [%1] %2\n\t" - "flush %%g6" - : /* No outputs */ - : "r" (0), - "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU)); - - if (tlb_type == spitfire) - tte_data = spitfire_get_dtlb_data(sparc64_highest_locked_tlbent()); - else - tte_data = cheetah_get_ldtlb_data(sparc64_highest_locked_tlbent()); + tte_vaddr = (unsigned long) KERNBASE; + phys_page = (prom_boot_mapping_phys_low >> 22UL) << 22UL; + tte_data = (phys_page | (_PAGE_VALID | _PAGE_SZ4MB | + _PAGE_CP | _PAGE_CV | _PAGE_P | + _PAGE_L | _PAGE_W)); kern_locked_tte_data = tte_data; - remap_func = (void *) ((unsigned long) &prom_remap - - (unsigned long) &prom_boot_page); - - - /* Spitfire Errata #32 workaround */ - /* NOTE: Using plain zero for the context value is - * correct here, we are not using the Linux trap - * tables yet so we should not use the special - * UltraSPARC-III+ page size encodings yet. - */ - __asm__ __volatile__("stxa %0, [%1] %2\n\t" - "flush %%g6" - : /* No outputs */ - : "r" (0), - "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU)); - - remap_func((tlb_type == spitfire ? - (spitfire_get_dtlb_data(sparc64_highest_locked_tlbent()) & _PAGE_PADDR) : - (cheetah_get_litlb_data(sparc64_highest_locked_tlbent()) & _PAGE_PADDR)), - (unsigned long) KERNBASE, - prom_get_mmu_ihandle()); - - if (bigkernel) - remap_func(((tte_data + 0x400000) & _PAGE_PADDR), - (unsigned long) KERNBASE + 0x400000, prom_get_mmu_ihandle()); - - /* Flush out that temporary mapping. */ - spitfire_flush_dtlb_nucleus_page(0x0); - spitfire_flush_itlb_nucleus_page(0x0); - - /* Now lock us back into the TLBs via OBP. */ - prom_dtlb_load(sparc64_highest_locked_tlbent(), tte_data, tte_vaddr); - prom_itlb_load(sparc64_highest_locked_tlbent(), tte_data, tte_vaddr); + /* Now lock us into the TLBs via OBP. */ + prom_dtlb_load(tlb_ent, tte_data, tte_vaddr); + prom_itlb_load(tlb_ent, tte_data, tte_vaddr); if (bigkernel) { - prom_dtlb_load(sparc64_highest_locked_tlbent()-1, tte_data + 0x400000, - tte_vaddr + 0x400000); - prom_itlb_load(sparc64_highest_locked_tlbent()-1, tte_data + 0x400000, - tte_vaddr + 0x400000); + tlb_ent -= 1; + prom_dtlb_load(tlb_ent, + tte_data + 0x400000, + tte_vaddr + 0x400000); + prom_itlb_load(tlb_ent, + tte_data + 0x400000, + tte_vaddr + 0x400000); } - - /* Re-read translations property. */ - if ((n = prom_getproperty(node, "translations", (char *)trans, tsz)) == -1) { - prom_printf("Couldn't get translation property\n"); - prom_halt(); + sparc64_highest_unlocked_tlb_ent = tlb_ent - 1; + if (tlb_type == cheetah_plus) { + sparc64_kern_pri_context = (CTX_CHEETAH_PLUS_CTX0 | + CTX_CHEETAH_PLUS_NUC); + sparc64_kern_pri_nuc_bits = CTX_CHEETAH_PLUS_NUC; + sparc64_kern_sec_context = CTX_CHEETAH_PLUS_CTX0; } - n = n / sizeof(*trans); - - for (i = 0; i < n; i++) { - unsigned long vaddr = trans[i].virt; - unsigned long size = trans[i].size; - - if (vaddr < 0xf0000000UL) { - unsigned long avoid_start = (unsigned long) KERNBASE; - unsigned long avoid_end = avoid_start + (4 * 1024 * 1024); - - if (bigkernel) - avoid_end += (4 * 1024 * 1024); - if (vaddr < avoid_start) { - unsigned long top = vaddr + size; +} - if (top > avoid_start) - top = avoid_start; - prom_unmap(top - vaddr, vaddr); - } - if ((vaddr + size) > avoid_end) { - unsigned long bottom = vaddr; - if (bottom < avoid_end) - bottom = avoid_end; - prom_unmap((vaddr + size) - bottom, bottom); - } - } - } +static void __init inherit_prom_mappings(void) +{ + read_obp_translations(); + /* Now fixup OBP's idea about where we really are mapped. */ + prom_printf("Remapping the kernel... "); + remap_kernel(); prom_printf("done.\n"); + prom_printf("Registering callbacks... "); register_prom_callbacks(); + prom_printf("done.\n"); } /* The OBP specifications for sun4u mark 0xfffffffc00000000 and @@ -791,8 +720,8 @@ void inherit_locked_prom_mappings(int save_p) } } if (tlb_type == spitfire) { - int high = SPITFIRE_HIGHEST_LOCKED_TLBENT - bigkernel; - for (i = 0; i < high; i++) { + int high = sparc64_highest_unlocked_tlb_ent; + for (i = 0; i <= high; i++) { unsigned long data; /* Spitfire Errata #32 workaround */ @@ -880,9 +809,9 @@ void inherit_locked_prom_mappings(int save_p) } } } else if (tlb_type == cheetah || tlb_type == cheetah_plus) { - int high = CHEETAH_HIGHEST_LOCKED_TLBENT - bigkernel; + int high = sparc64_highest_unlocked_tlb_ent; - for (i = 0; i < high; i++) { + for (i = 0; i <= high; i++) { unsigned long data; data = cheetah_get_ldtlb_data(i); @@ -1275,14 +1204,14 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) int i; #ifdef CONFIG_DEBUG_BOOTMEM - prom_printf("bootmem_init: Scan sp_banks, "); + prom_printf("bootmem_init: Scan pavail, "); #endif bytes_avail = 0UL; - for (i = 0; sp_banks[i].num_bytes != 0; i++) { - end_of_phys_memory = sp_banks[i].base_addr + - sp_banks[i].num_bytes; - bytes_avail += sp_banks[i].num_bytes; + for (i = 0; i < pavail_ents; i++) { + end_of_phys_memory = pavail[i].phys_addr + + pavail[i].reg_size; + bytes_avail += pavail[i].reg_size; if (cmdline_memory_size) { if (bytes_avail > cmdline_memory_size) { unsigned long slack = bytes_avail - cmdline_memory_size; @@ -1290,12 +1219,15 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) bytes_avail -= slack; end_of_phys_memory -= slack; - sp_banks[i].num_bytes -= slack; - if (sp_banks[i].num_bytes == 0) { - sp_banks[i].base_addr = 0xdeadbeef; + pavail[i].reg_size -= slack; + if ((long)pavail[i].reg_size <= 0L) { + pavail[i].phys_addr = 0xdeadbeefUL; + pavail[i].reg_size = 0UL; + pavail_ents = i; } else { - sp_banks[i+1].num_bytes = 0; - sp_banks[i+1].base_addr = 0xdeadbeef; + pavail[i+1].reg_size = 0Ul; + pavail[i+1].phys_addr = 0xdeadbeefUL; + pavail_ents = i + 1; } break; } @@ -1346,17 +1278,15 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) #endif bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap_pfn, pfn_base, end_pfn); - bootmap_base = bootmap_pfn << PAGE_SHIFT; - /* Now register the available physical memory with the * allocator. */ - for (i = 0; sp_banks[i].num_bytes != 0; i++) { + for (i = 0; i < pavail_ents; i++) { #ifdef CONFIG_DEBUG_BOOTMEM - prom_printf("free_bootmem(sp_banks:%d): base[%lx] size[%lx]\n", - i, sp_banks[i].base_addr, sp_banks[i].num_bytes); + prom_printf("free_bootmem(pavail:%d): base[%lx] size[%lx]\n", + i, pavail[i].phys_addr, pavail[i].reg_size); #endif - free_bootmem(sp_banks[i].base_addr, sp_banks[i].num_bytes); + free_bootmem(pavail[i].phys_addr, pavail[i].reg_size); } #ifdef CONFIG_BLK_DEV_INITRD @@ -1397,121 +1327,167 @@ unsigned long __init bootmem_init(unsigned long *pages_avail) return end_pfn; } +#ifdef CONFIG_DEBUG_PAGEALLOC +static unsigned long kernel_map_range(unsigned long pstart, unsigned long pend, pgprot_t prot) +{ + unsigned long vstart = PAGE_OFFSET + pstart; + unsigned long vend = PAGE_OFFSET + pend; + unsigned long alloc_bytes = 0UL; + + if ((vstart & ~PAGE_MASK) || (vend & ~PAGE_MASK)) { + prom_printf("kernel_map: Unaligned physmem[%lx:%lx]\n", + vstart, vend); + prom_halt(); + } + + while (vstart < vend) { + unsigned long this_end, paddr = __pa(vstart); + pgd_t *pgd = pgd_offset_k(vstart); + pud_t *pud; + pmd_t *pmd; + pte_t *pte; + + pud = pud_offset(pgd, vstart); + if (pud_none(*pud)) { + pmd_t *new; + + new = __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, PAGE_SIZE); + alloc_bytes += PAGE_SIZE; + pud_populate(&init_mm, pud, new); + } + + pmd = pmd_offset(pud, vstart); + if (!pmd_present(*pmd)) { + pte_t *new; + + new = __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, PAGE_SIZE); + alloc_bytes += PAGE_SIZE; + pmd_populate_kernel(&init_mm, pmd, new); + } + + pte = pte_offset_kernel(pmd, vstart); + this_end = (vstart + PMD_SIZE) & PMD_MASK; + if (this_end > vend) + this_end = vend; + + while (vstart < this_end) { + pte_val(*pte) = (paddr | pgprot_val(prot)); + + vstart += PAGE_SIZE; + paddr += PAGE_SIZE; + pte++; + } + } + + return alloc_bytes; +} + +static struct linux_prom64_registers pall[MAX_BANKS] __initdata; +static int pall_ents __initdata; + +extern unsigned int kvmap_linear_patch[1]; + +static void __init kernel_physical_mapping_init(void) +{ + unsigned long i, mem_alloced = 0UL; + + read_obp_memory("reg", &pall[0], &pall_ents); + + for (i = 0; i < pall_ents; i++) { + unsigned long phys_start, phys_end; + + phys_start = pall[i].phys_addr; + phys_end = phys_start + pall[i].reg_size; + mem_alloced += kernel_map_range(phys_start, phys_end, + PAGE_KERNEL); + } + + printk("Allocated %ld bytes for kernel page tables.\n", + mem_alloced); + + kvmap_linear_patch[0] = 0x01000000; /* nop */ + flushi(&kvmap_linear_patch[0]); + + __flush_tlb_all(); +} + +void kernel_map_pages(struct page *page, int numpages, int enable) +{ + unsigned long phys_start = page_to_pfn(page) << PAGE_SHIFT; + unsigned long phys_end = phys_start + (numpages * PAGE_SIZE); + + kernel_map_range(phys_start, phys_end, + (enable ? PAGE_KERNEL : __pgprot(0))); + + /* we should perform an IPI and flush all tlbs, + * but that can deadlock->flush only current cpu. + */ + __flush_tlb_kernel_range(PAGE_OFFSET + phys_start, + PAGE_OFFSET + phys_end); +} +#endif + +unsigned long __init find_ecache_flush_span(unsigned long size) +{ + int i; + + for (i = 0; i < pavail_ents; i++) { + if (pavail[i].reg_size >= size) + return pavail[i].phys_addr; + } + + return ~0UL; +} + /* paging_init() sets up the page tables */ extern void cheetah_ecache_flush_init(void); static unsigned long last_valid_pfn; +pgd_t swapper_pg_dir[2048]; void __init paging_init(void) { - extern pmd_t swapper_pmd_dir[1024]; - extern unsigned int sparc64_vpte_patchme1[1]; - extern unsigned int sparc64_vpte_patchme2[1]; - unsigned long alias_base = kern_base + PAGE_OFFSET; - unsigned long second_alias_page = 0; - unsigned long pt, flags, end_pfn, pages_avail; - unsigned long shift = alias_base - ((unsigned long)KERNBASE); - unsigned long real_end; + unsigned long end_pfn, pages_avail, shift; + unsigned long real_end, i; + + /* Find available physical memory... */ + read_obp_memory("available", &pavail[0], &pavail_ents); + + phys_base = 0xffffffffffffffffUL; + for (i = 0; i < pavail_ents; i++) + phys_base = min(phys_base, pavail[i].phys_addr); + + pfn_base = phys_base >> PAGE_SHIFT; + + kern_base = (prom_boot_mapping_phys_low >> 22UL) << 22UL; + kern_size = (unsigned long)&_end - (unsigned long)KERNBASE; set_bit(0, mmu_context_bmap); + shift = kern_base + PAGE_OFFSET - ((unsigned long)KERNBASE); + real_end = (unsigned long)_end; if ((real_end > ((unsigned long)KERNBASE + 0x400000))) bigkernel = 1; -#ifdef CONFIG_BLK_DEV_INITRD - if (sparc_ramdisk_image || sparc_ramdisk_image64) - real_end = (PAGE_ALIGN(real_end) + PAGE_ALIGN(sparc_ramdisk_size)); -#endif - - /* We assume physical memory starts at some 4mb multiple, - * if this were not true we wouldn't boot up to this point - * anyways. - */ - pt = kern_base | _PAGE_VALID | _PAGE_SZ4MB; - pt |= _PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W; - local_irq_save(flags); - if (tlb_type == spitfire) { - __asm__ __volatile__( - " stxa %1, [%0] %3\n" - " stxa %2, [%5] %4\n" - " membar #Sync\n" - " flush %%g6\n" - " nop\n" - " nop\n" - " nop\n" - : /* No outputs */ - : "r" (TLB_TAG_ACCESS), "r" (alias_base), "r" (pt), - "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" (61 << 3) - : "memory"); - if (real_end >= KERNBASE + 0x340000) { - second_alias_page = alias_base + 0x400000; - __asm__ __volatile__( - " stxa %1, [%0] %3\n" - " stxa %2, [%5] %4\n" - " membar #Sync\n" - " flush %%g6\n" - " nop\n" - " nop\n" - " nop\n" - : /* No outputs */ - : "r" (TLB_TAG_ACCESS), "r" (second_alias_page), "r" (pt + 0x400000), - "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" (60 << 3) - : "memory"); - } - } else if (tlb_type == cheetah || tlb_type == cheetah_plus) { - __asm__ __volatile__( - " stxa %1, [%0] %3\n" - " stxa %2, [%5] %4\n" - " membar #Sync\n" - " flush %%g6\n" - " nop\n" - " nop\n" - " nop\n" - : /* No outputs */ - : "r" (TLB_TAG_ACCESS), "r" (alias_base), "r" (pt), - "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" ((0<<16) | (13<<3)) - : "memory"); - if (real_end >= KERNBASE + 0x340000) { - second_alias_page = alias_base + 0x400000; - __asm__ __volatile__( - " stxa %1, [%0] %3\n" - " stxa %2, [%5] %4\n" - " membar #Sync\n" - " flush %%g6\n" - " nop\n" - " nop\n" - " nop\n" - : /* No outputs */ - : "r" (TLB_TAG_ACCESS), "r" (second_alias_page), "r" (pt + 0x400000), - "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" ((0<<16) | (12<<3)) - : "memory"); - } + if ((real_end > ((unsigned long)KERNBASE + 0x800000))) { + prom_printf("paging_init: Kernel > 8MB, too large.\n"); + prom_halt(); } - local_irq_restore(flags); - - /* Now set kernel pgd to upper alias so physical page computations + + /* Set kernel pgd to upper alias so physical page computations * work. */ init_mm.pgd += ((shift) / (sizeof(pgd_t))); - memset(swapper_pmd_dir, 0, sizeof(swapper_pmd_dir)); + memset(swapper_low_pmd_dir, 0, sizeof(swapper_low_pmd_dir)); /* Now can init the kernel/bad page tables. */ pud_set(pud_offset(&swapper_pg_dir[0], 0), - swapper_pmd_dir + (shift / sizeof(pgd_t))); + swapper_low_pmd_dir + (shift / sizeof(pgd_t))); - sparc64_vpte_patchme1[0] |= - (((unsigned long)pgd_val(init_mm.pgd[0])) >> 10); - sparc64_vpte_patchme2[0] |= - (((unsigned long)pgd_val(init_mm.pgd[0])) & 0x3ff); - flushi((long)&sparc64_vpte_patchme1[0]); + swapper_pgd_zero = pgd_val(swapper_pg_dir[0]); - /* Setup bootmem... */ - pages_avail = 0; - last_valid_pfn = end_pfn = bootmem_init(&pages_avail); - - /* Inherit non-locked OBP mappings. */ inherit_prom_mappings(); /* Ok, we can use our TLB miss and window trap handlers safely. @@ -1526,13 +1502,16 @@ void __init paging_init(void) inherit_locked_prom_mappings(1); - /* We only created DTLB mapping of this stuff. */ - spitfire_flush_dtlb_nucleus_page(alias_base); - if (second_alias_page) - spitfire_flush_dtlb_nucleus_page(second_alias_page); - __flush_tlb_all(); + /* Setup bootmem... */ + pages_avail = 0; + last_valid_pfn = end_pfn = bootmem_init(&pages_avail); + +#ifdef CONFIG_DEBUG_PAGEALLOC + kernel_physical_mapping_init(); +#endif + { unsigned long zones_size[MAX_NR_ZONES]; unsigned long zholes_size[MAX_NR_ZONES]; @@ -1553,128 +1532,35 @@ void __init paging_init(void) device_scan(); } -/* Ok, it seems that the prom can allocate some more memory chunks - * as a side effect of some prom calls we perform during the - * boot sequence. My most likely theory is that it is from the - * prom_set_traptable() call, and OBP is allocating a scratchpad - * for saving client program register state etc. - */ -static void __init sort_memlist(struct linux_mlist_p1275 *thislist) -{ - int swapi = 0; - int i, mitr; - unsigned long tmpaddr, tmpsize; - unsigned long lowest; - - for (i = 0; thislist[i].theres_more != 0; i++) { - lowest = thislist[i].start_adr; - for (mitr = i+1; thislist[mitr-1].theres_more != 0; mitr++) - if (thislist[mitr].start_adr < lowest) { - lowest = thislist[mitr].start_adr; - swapi = mitr; - } - if (lowest == thislist[i].start_adr) - continue; - tmpaddr = thislist[swapi].start_adr; - tmpsize = thislist[swapi].num_bytes; - for (mitr = swapi; mitr > i; mitr--) { - thislist[mitr].start_adr = thislist[mitr-1].start_adr; - thislist[mitr].num_bytes = thislist[mitr-1].num_bytes; - } - thislist[i].start_adr = tmpaddr; - thislist[i].num_bytes = tmpsize; - } -} - -void __init rescan_sp_banks(void) -{ - struct linux_prom64_registers memlist[64]; - struct linux_mlist_p1275 avail[64], *mlist; - unsigned long bytes, base_paddr; - int num_regs, node = prom_finddevice("/memory"); - int i; - - num_regs = prom_getproperty(node, "available", - (char *) memlist, sizeof(memlist)); - num_regs = (num_regs / sizeof(struct linux_prom64_registers)); - for (i = 0; i < num_regs; i++) { - avail[i].start_adr = memlist[i].phys_addr; - avail[i].num_bytes = memlist[i].reg_size; - avail[i].theres_more = &avail[i + 1]; - } - avail[i - 1].theres_more = NULL; - sort_memlist(avail); - - mlist = &avail[0]; - i = 0; - bytes = mlist->num_bytes; - base_paddr = mlist->start_adr; - - sp_banks[0].base_addr = base_paddr; - sp_banks[0].num_bytes = bytes; - - while (mlist->theres_more != NULL){ - i++; - mlist = mlist->theres_more; - bytes = mlist->num_bytes; - if (i >= SPARC_PHYS_BANKS-1) { - printk ("The machine has more banks than " - "this kernel can support\n" - "Increase the SPARC_PHYS_BANKS " - "setting (currently %d)\n", - SPARC_PHYS_BANKS); - i = SPARC_PHYS_BANKS-1; - break; - } - - sp_banks[i].base_addr = mlist->start_adr; - sp_banks[i].num_bytes = mlist->num_bytes; - } - - i++; - sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL; - sp_banks[i].num_bytes = 0; - - for (i = 0; sp_banks[i].num_bytes != 0; i++) - sp_banks[i].num_bytes &= PAGE_MASK; -} - static void __init taint_real_pages(void) { - struct sparc_phys_banks saved_sp_banks[SPARC_PHYS_BANKS]; int i; - for (i = 0; i < SPARC_PHYS_BANKS; i++) { - saved_sp_banks[i].base_addr = - sp_banks[i].base_addr; - saved_sp_banks[i].num_bytes = - sp_banks[i].num_bytes; - } - - rescan_sp_banks(); + read_obp_memory("available", &pavail_rescan[0], &pavail_rescan_ents); - /* Find changes discovered in the sp_bank rescan and + /* Find changes discovered in the physmem available rescan and * reserve the lost portions in the bootmem maps. */ - for (i = 0; saved_sp_banks[i].num_bytes; i++) { + for (i = 0; i < pavail_ents; i++) { unsigned long old_start, old_end; - old_start = saved_sp_banks[i].base_addr; + old_start = pavail[i].phys_addr; old_end = old_start + - saved_sp_banks[i].num_bytes; + pavail[i].reg_size; while (old_start < old_end) { int n; - for (n = 0; sp_banks[n].num_bytes; n++) { + for (n = 0; pavail_rescan_ents; n++) { unsigned long new_start, new_end; - new_start = sp_banks[n].base_addr; - new_end = new_start + sp_banks[n].num_bytes; + new_start = pavail_rescan[n].phys_addr; + new_end = new_start + + pavail_rescan[n].reg_size; if (new_start <= old_start && new_end >= (old_start + PAGE_SIZE)) { - set_bit (old_start >> 22, - sparc64_valid_addr_bitmap); + set_bit(old_start >> 22, + sparc64_valid_addr_bitmap); goto do_next_page; } } @@ -1694,8 +1580,7 @@ void __init mem_init(void) i = last_valid_pfn >> ((22 - PAGE_SHIFT) + 6); i += 1; - sparc64_valid_addr_bitmap = (unsigned long *) - __alloc_bootmem(i << 3, SMP_CACHE_BYTES, bootmap_base); + sparc64_valid_addr_bitmap = (unsigned long *) alloc_bootmem(i << 3); if (sparc64_valid_addr_bitmap == NULL) { prom_printf("mem_init: Cannot alloc valid_addr_bitmap.\n"); prom_halt(); @@ -1748,7 +1633,7 @@ void __init mem_init(void) cheetah_ecache_flush_init(); } -void free_initmem (void) +void free_initmem(void) { unsigned long addr, initend; diff --git a/arch/sparc64/mm/tlb.c b/arch/sparc64/mm/tlb.c index 90ca99d0b89c..8b104be4662b 100644 --- a/arch/sparc64/mm/tlb.c +++ b/arch/sparc64/mm/tlb.c @@ -18,8 +18,7 @@ /* Heavily inspired by the ppc64 code. */ -DEFINE_PER_CPU(struct mmu_gather, mmu_gathers) = - { NULL, 0, 0, 0, 0, 0, { 0 }, { NULL }, }; +DEFINE_PER_CPU(struct mmu_gather, mmu_gathers) = { 0, }; void flush_tlb_pending(void) { @@ -72,7 +71,7 @@ void tlb_batch_add(struct mm_struct *mm, unsigned long vaddr, pte_t *ptep, pte_t no_cache_flush: - if (mp->tlb_frozen) + if (mp->fullmm) return; nr = mp->tlb_nr; @@ -97,7 +96,7 @@ void flush_tlb_pgtables(struct mm_struct *mm, unsigned long start, unsigned long unsigned long nr = mp->tlb_nr; long s = start, e = end, vpte_base; - if (mp->tlb_frozen) + if (mp->fullmm) return; /* If start is greater than end, that is a real problem. */ diff --git a/arch/sparc64/mm/ultra.S b/arch/sparc64/mm/ultra.S index 363770893797..e4c9151fa116 100644 --- a/arch/sparc64/mm/ultra.S +++ b/arch/sparc64/mm/ultra.S @@ -10,6 +10,7 @@ #include <asm/page.h> #include <asm/spitfire.h> #include <asm/mmu_context.h> +#include <asm/mmu.h> #include <asm/pil.h> #include <asm/head.h> #include <asm/thread_info.h> @@ -45,6 +46,8 @@ __flush_tlb_mm: /* %o0=(ctx & TAG_CONTEXT_BITS), %o1=SECONDARY_CONTEXT */ nop nop nop + nop + nop .align 32 .globl __flush_tlb_pending @@ -73,6 +76,9 @@ __flush_tlb_pending: retl wrpr %g7, 0x0, %pstate nop + nop + nop + nop .align 32 .globl __flush_tlb_kernel_range @@ -113,6 +119,7 @@ __spitfire_flush_tlb_mm_slow: #else #error unsupported PAGE_SIZE #endif + .section .kprobes.text, "ax" .align 32 .globl __flush_icache_page __flush_icache_page: /* %o0 = phys_page */ @@ -137,42 +144,29 @@ __flush_icache_page: /* %o0 = phys_page */ #define DTAG_MASK 0x3 + /* This routine is Spitfire specific so the hardcoded + * D-cache size and line-size are OK. + */ .align 64 .globl __flush_dcache_page __flush_dcache_page: /* %o0=kaddr, %o1=flush_icache */ sethi %uhi(PAGE_OFFSET), %g1 sllx %g1, 32, %g1 - sub %o0, %g1, %o0 - clr %o4 - srlx %o0, 11, %o0 - sethi %hi(1 << 14), %o2 -1: ldxa [%o4] ASI_DCACHE_TAG, %o3 ! LSU Group - add %o4, (1 << 5), %o4 ! IEU0 - ldxa [%o4] ASI_DCACHE_TAG, %g1 ! LSU Group - add %o4, (1 << 5), %o4 ! IEU0 - ldxa [%o4] ASI_DCACHE_TAG, %g2 ! LSU Group o3 available - add %o4, (1 << 5), %o4 ! IEU0 - andn %o3, DTAG_MASK, %o3 ! IEU1 - ldxa [%o4] ASI_DCACHE_TAG, %g3 ! LSU Group - add %o4, (1 << 5), %o4 ! IEU0 - andn %g1, DTAG_MASK, %g1 ! IEU1 - cmp %o0, %o3 ! IEU1 Group - be,a,pn %xcc, dflush1 ! CTI - sub %o4, (4 << 5), %o4 ! IEU0 (Group) - cmp %o0, %g1 ! IEU1 Group - andn %g2, DTAG_MASK, %g2 ! IEU0 - be,a,pn %xcc, dflush2 ! CTI - sub %o4, (3 << 5), %o4 ! IEU0 (Group) - cmp %o0, %g2 ! IEU1 Group - andn %g3, DTAG_MASK, %g3 ! IEU0 - be,a,pn %xcc, dflush3 ! CTI - sub %o4, (2 << 5), %o4 ! IEU0 (Group) - cmp %o0, %g3 ! IEU1 Group - be,a,pn %xcc, dflush4 ! CTI - sub %o4, (1 << 5), %o4 ! IEU0 -2: cmp %o4, %o2 ! IEU1 Group - bne,pt %xcc, 1b ! CTI - nop ! IEU0 + sub %o0, %g1, %o0 ! physical address + srlx %o0, 11, %o0 ! make D-cache TAG + sethi %hi(1 << 14), %o2 ! D-cache size + sub %o2, (1 << 5), %o2 ! D-cache line size +1: ldxa [%o2] ASI_DCACHE_TAG, %o3 ! load D-cache TAG + andcc %o3, DTAG_MASK, %g0 ! Valid? + be,pn %xcc, 2f ! Nope, branch + andn %o3, DTAG_MASK, %o3 ! Clear valid bits + cmp %o3, %o0 ! TAG match? + bne,pt %xcc, 2f ! Nope, branch + nop + stxa %g0, [%o2] ASI_DCACHE_TAG ! Invalidate TAG + membar #Sync +2: brnz,pt %o2, 1b + sub %o2, (1 << 5), %o2 ! D-cache line size /* The I-cache does not snoop local stores so we * better flush that too when necessary. @@ -182,58 +176,12 @@ __flush_dcache_page: /* %o0=kaddr, %o1=flush_icache */ retl nop -dflush1:stxa %g0, [%o4] ASI_DCACHE_TAG - add %o4, (1 << 5), %o4 -dflush2:stxa %g0, [%o4] ASI_DCACHE_TAG - add %o4, (1 << 5), %o4 -dflush3:stxa %g0, [%o4] ASI_DCACHE_TAG - add %o4, (1 << 5), %o4 -dflush4:stxa %g0, [%o4] ASI_DCACHE_TAG - add %o4, (1 << 5), %o4 - membar #Sync - ba,pt %xcc, 2b - nop #endif /* DCACHE_ALIASING_POSSIBLE */ - .align 32 -__prefill_dtlb: - rdpr %pstate, %g7 - wrpr %g7, PSTATE_IE, %pstate - mov TLB_TAG_ACCESS, %g1 - stxa %o5, [%g1] ASI_DMMU - stxa %o2, [%g0] ASI_DTLB_DATA_IN - flush %g6 - retl - wrpr %g7, %pstate -__prefill_itlb: - rdpr %pstate, %g7 - wrpr %g7, PSTATE_IE, %pstate - mov TLB_TAG_ACCESS, %g1 - stxa %o5, [%g1] ASI_IMMU - stxa %o2, [%g0] ASI_ITLB_DATA_IN - flush %g6 - retl - wrpr %g7, %pstate - - .globl __update_mmu_cache -__update_mmu_cache: /* %o0=hw_context, %o1=address, %o2=pte, %o3=fault_code */ - srlx %o1, PAGE_SHIFT, %o1 - andcc %o3, FAULT_CODE_DTLB, %g0 - sllx %o1, PAGE_SHIFT, %o5 - bne,pt %xcc, __prefill_dtlb - or %o5, %o0, %o5 - ba,a,pt %xcc, __prefill_itlb - - /* Cheetah specific versions, patched at boot time. - * - * This writes of the PRIMARY_CONTEXT register in this file are - * safe even on Cheetah+ and later wrt. the page size fields. - * The nucleus page size fields do not matter because we make - * no data references, and these instructions execute out of a - * locked I-TLB entry sitting in the fully assosciative I-TLB. - * This sequence should also never trap. - */ -__cheetah_flush_tlb_mm: /* 15 insns */ + .previous + + /* Cheetah specific versions, patched at boot time. */ +__cheetah_flush_tlb_mm: /* 18 insns */ rdpr %pstate, %g7 andn %g7, PSTATE_IE, %g2 wrpr %g2, 0x0, %pstate @@ -241,6 +189,9 @@ __cheetah_flush_tlb_mm: /* 15 insns */ mov PRIMARY_CONTEXT, %o2 mov 0x40, %g3 ldxa [%o2] ASI_DMMU, %g2 + srlx %g2, CTX_PGSZ1_NUC_SHIFT, %o1 + sllx %o1, CTX_PGSZ1_NUC_SHIFT, %o1 + or %o0, %o1, %o0 /* Preserve nucleus page size fields */ stxa %o0, [%o2] ASI_DMMU stxa %g0, [%g3] ASI_DMMU_DEMAP stxa %g0, [%g3] ASI_IMMU_DEMAP @@ -250,7 +201,7 @@ __cheetah_flush_tlb_mm: /* 15 insns */ retl wrpr %g7, 0x0, %pstate -__cheetah_flush_tlb_pending: /* 23 insns */ +__cheetah_flush_tlb_pending: /* 26 insns */ /* %o0 = context, %o1 = nr, %o2 = vaddrs[] */ rdpr %pstate, %g7 sllx %o1, 3, %o1 @@ -259,6 +210,9 @@ __cheetah_flush_tlb_pending: /* 23 insns */ wrpr %g0, 1, %tl mov PRIMARY_CONTEXT, %o4 ldxa [%o4] ASI_DMMU, %g2 + srlx %g2, CTX_PGSZ1_NUC_SHIFT, %o3 + sllx %o3, CTX_PGSZ1_NUC_SHIFT, %o3 + or %o0, %o3, %o0 /* Preserve nucleus page size fields */ stxa %o0, [%o4] ASI_DMMU 1: sub %o1, (1 << 3), %o1 ldx [%o2 + %o1], %o3 @@ -277,7 +231,7 @@ __cheetah_flush_tlb_pending: /* 23 insns */ wrpr %g7, 0x0, %pstate #ifdef DCACHE_ALIASING_POSSIBLE -flush_dcpage_cheetah: /* 11 insns */ +__cheetah_flush_dcache_page: /* 11 insns */ sethi %uhi(PAGE_OFFSET), %g1 sllx %g1, 32, %g1 sub %o0, %g1, %o0 @@ -311,20 +265,20 @@ cheetah_patch_cachetlbops: sethi %hi(__cheetah_flush_tlb_mm), %o1 or %o1, %lo(__cheetah_flush_tlb_mm), %o1 call cheetah_patch_one - mov 15, %o2 + mov 18, %o2 sethi %hi(__flush_tlb_pending), %o0 or %o0, %lo(__flush_tlb_pending), %o0 sethi %hi(__cheetah_flush_tlb_pending), %o1 or %o1, %lo(__cheetah_flush_tlb_pending), %o1 call cheetah_patch_one - mov 23, %o2 + mov 26, %o2 #ifdef DCACHE_ALIASING_POSSIBLE sethi %hi(__flush_dcache_page), %o0 or %o0, %lo(__flush_dcache_page), %o0 - sethi %hi(flush_dcpage_cheetah), %o1 - or %o1, %lo(flush_dcpage_cheetah), %o1 + sethi %hi(__cheetah_flush_dcache_page), %o1 + or %o1, %lo(__cheetah_flush_dcache_page), %o1 call cheetah_patch_one mov 11, %o2 #endif /* DCACHE_ALIASING_POSSIBLE */ @@ -352,9 +306,12 @@ cheetah_patch_cachetlbops: .globl xcall_flush_tlb_mm xcall_flush_tlb_mm: mov PRIMARY_CONTEXT, %g2 - mov 0x40, %g4 ldxa [%g2] ASI_DMMU, %g3 + srlx %g3, CTX_PGSZ1_NUC_SHIFT, %g4 + sllx %g4, CTX_PGSZ1_NUC_SHIFT, %g4 + or %g5, %g4, %g5 /* Preserve nucleus page size fields */ stxa %g5, [%g2] ASI_DMMU + mov 0x40, %g4 stxa %g0, [%g4] ASI_DMMU_DEMAP stxa %g0, [%g4] ASI_IMMU_DEMAP stxa %g3, [%g2] ASI_DMMU @@ -366,6 +323,10 @@ xcall_flush_tlb_pending: sllx %g1, 3, %g1 mov PRIMARY_CONTEXT, %g4 ldxa [%g4] ASI_DMMU, %g2 + srlx %g2, CTX_PGSZ1_NUC_SHIFT, %g4 + sllx %g4, CTX_PGSZ1_NUC_SHIFT, %g4 + or %g5, %g4, %g5 + mov PRIMARY_CONTEXT, %g4 stxa %g5, [%g4] ASI_DMMU 1: sub %g1, (1 << 3), %g1 ldx [%g7 + %g1], %g5 @@ -492,22 +453,6 @@ xcall_flush_dcache_page_spitfire: /* %g1 == physical page address nop nop - .globl xcall_promstop -xcall_promstop: - rdpr %pstate, %g2 - wrpr %g2, PSTATE_IG | PSTATE_AG, %pstate - rdpr %pil, %g2 - wrpr %g0, 15, %pil - sethi %hi(109f), %g7 - b,pt %xcc, etrap_irq -109: or %g7, %lo(109b), %g7 - flushw - call prom_stopself - nop - /* We should not return, just spin if we do... */ -1: b,a,pt %xcc, 1b - nop - .data errata32_hwbug: diff --git a/arch/sparc64/prom/Makefile b/arch/sparc64/prom/Makefile index 8f2420d9e9e6..3d33ed27bc27 100644 --- a/arch/sparc64/prom/Makefile +++ b/arch/sparc64/prom/Makefile @@ -6,5 +6,5 @@ EXTRA_AFLAGS := -ansi EXTRA_CFLAGS := -Werror -lib-y := bootstr.o devops.o init.o memory.o misc.o \ - tree.o console.o printf.o p1275.o map.o cif.o +lib-y := bootstr.o devops.o init.o misc.o \ + tree.o console.o printf.o p1275.o cif.o diff --git a/arch/sparc64/prom/console.c b/arch/sparc64/prom/console.c index 028a53fcb1ec..eae5db8dda56 100644 --- a/arch/sparc64/prom/console.c +++ b/arch/sparc64/prom/console.c @@ -67,7 +67,7 @@ prom_putchar(char c) } void -prom_puts(char *s, int len) +prom_puts(const char *s, int len) { p1275_cmd("write", P1275_ARG(1,P1275_ARG_IN_BUF)| P1275_INOUT(3,1), diff --git a/arch/sparc64/prom/devops.c b/arch/sparc64/prom/devops.c index 2c99b21b6981..4641839eb39a 100644 --- a/arch/sparc64/prom/devops.c +++ b/arch/sparc64/prom/devops.c @@ -16,7 +16,7 @@ * Returns 0 on failure. */ int -prom_devopen(char *dstr) +prom_devopen(const char *dstr) { return p1275_cmd ("open", P1275_ARG(0,P1275_ARG_IN_STRING)| P1275_INOUT(1,1), diff --git a/arch/sparc64/prom/init.c b/arch/sparc64/prom/init.c index 817faae058cd..f3cc2d8578b2 100644 --- a/arch/sparc64/prom/init.c +++ b/arch/sparc64/prom/init.c @@ -27,7 +27,6 @@ int prom_chosen_node; * failure. It gets passed the pointer to the PROM vector. */ -extern void prom_meminit(void); extern void prom_cif_init(void *, void *); void __init prom_init(void *cif_handler, void *cif_stack) @@ -46,7 +45,7 @@ void __init prom_init(void *cif_handler, void *cif_stack) if((prom_root_node == 0) || (prom_root_node == -1)) prom_halt(); - prom_chosen_node = prom_finddevice("/chosen"); + prom_chosen_node = prom_finddevice(prom_chosen_path); if (!prom_chosen_node || prom_chosen_node == -1) prom_halt(); @@ -90,8 +89,6 @@ void __init prom_init(void *cif_handler, void *cif_stack) printk ("PROMLIB: Sun IEEE Boot Prom %s\n", buffer + bufadjust); - prom_meminit(); - /* Initialization successful. */ return; diff --git a/arch/sparc64/prom/map.S b/arch/sparc64/prom/map.S deleted file mode 100644 index 21b3f9c99ea7..000000000000 --- a/arch/sparc64/prom/map.S +++ /dev/null @@ -1,72 +0,0 @@ -/* $Id: map.S,v 1.2 1999/11/19 05:53:02 davem Exp $ - * map.S: Tricky coding required to fixup the kernel OBP maps - * properly. - * - * Copyright (C) 1999 David S. Miller (davem@redhat.com) - */ - - .text - .align 8192 - .globl prom_boot_page -prom_boot_page: -call_method: - .asciz "call-method" - .align 8 -map: - .asciz "map" - .align 8 - - /* When we are invoked, our caller has remapped us to - * page zero, therefore we must use PC relative addressing - * for everything after we begin performing the unmap/map - * calls. - */ - .globl prom_remap -prom_remap: /* %o0 = physpage, %o1 = virtpage, %o2 = mmu_ihandle */ - rd %pc, %g1 - srl %o2, 0, %o2 ! kill sign extension - sethi %hi(p1275buf), %g2 - or %g2, %lo(p1275buf), %g2 - ldx [%g2 + 0x10], %g3 ! prom_cif_stack - save %g3, -(192 + 128), %sp - ldx [%g2 + 0x08], %l0 ! prom_cif_handler - mov %g6, %i3 - mov %g4, %i4 - mov %g5, %i5 - flushw - - sethi %hi(prom_remap - call_method), %g7 - or %g7, %lo(prom_remap - call_method), %g7 - sub %g1, %g7, %l2 ! call-method string - sethi %hi(prom_remap - map), %g7 - or %g7, %lo(prom_remap - map), %g7 - sub %g1, %g7, %l4 ! map string - - /* OK, map the 4MB region we really live at. */ - stx %l2, [%sp + 2047 + 128 + 0x00] ! call-method - mov 7, %l5 - stx %l5, [%sp + 2047 + 128 + 0x08] ! num_args - mov 1, %l5 - stx %l5, [%sp + 2047 + 128 + 0x10] ! num_rets - stx %l4, [%sp + 2047 + 128 + 0x18] ! map - stx %i2, [%sp + 2047 + 128 + 0x20] ! mmu_ihandle - mov -1, %l5 - stx %l5, [%sp + 2047 + 128 + 0x28] ! mode == default - sethi %hi(4 * 1024 * 1024), %l5 - stx %l5, [%sp + 2047 + 128 + 0x30] ! size - stx %i1, [%sp + 2047 + 128 + 0x38] ! vaddr - stx %g0, [%sp + 2047 + 128 + 0x40] ! filler - stx %i0, [%sp + 2047 + 128 + 0x48] ! paddr - call %l0 - add %sp, (2047 + 128), %o0 ! argument array - - /* Restore hard-coded globals. */ - mov %i3, %g6 - mov %i4, %g4 - mov %i5, %g5 - - /* Wheee.... we are done. */ - ret - restore - - .align 8192 diff --git a/arch/sparc64/prom/memory.c b/arch/sparc64/prom/memory.c deleted file mode 100644 index f4a8143e052c..000000000000 --- a/arch/sparc64/prom/memory.c +++ /dev/null @@ -1,152 +0,0 @@ -/* $Id: memory.c,v 1.5 1999/08/31 06:55:04 davem Exp $ - * memory.c: Prom routine for acquiring various bits of information - * about RAM on the machine, both virtual and physical. - * - * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu) - * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz) - */ - -#include <linux/kernel.h> -#include <linux/init.h> - -#include <asm/openprom.h> -#include <asm/oplib.h> - -/* This routine, for consistency, returns the ram parameters in the - * V0 prom memory descriptor format. I choose this format because I - * think it was the easiest to work with. I feel the religious - * arguments now... ;) Also, I return the linked lists sorted to - * prevent paging_init() upset stomach as I have not yet written - * the pepto-bismol kernel module yet. - */ - -struct linux_prom64_registers prom_reg_memlist[64]; -struct linux_prom64_registers prom_reg_tmp[64]; - -struct linux_mlist_p1275 prom_phys_total[64]; -struct linux_mlist_p1275 prom_prom_taken[64]; -struct linux_mlist_p1275 prom_phys_avail[64]; - -struct linux_mlist_p1275 *prom_ptot_ptr = prom_phys_total; -struct linux_mlist_p1275 *prom_ptak_ptr = prom_prom_taken; -struct linux_mlist_p1275 *prom_pavl_ptr = prom_phys_avail; - -struct linux_mem_p1275 prom_memlist; - - -/* Internal Prom library routine to sort a linux_mlist_p1275 memory - * list. Used below in initialization. - */ -static void __init -prom_sortmemlist(struct linux_mlist_p1275 *thislist) -{ - int swapi = 0; - int i, mitr; - unsigned long tmpaddr, tmpsize; - unsigned long lowest; - - for(i=0; thislist[i].theres_more; i++) { - lowest = thislist[i].start_adr; - for(mitr = i+1; thislist[mitr-1].theres_more; mitr++) - if(thislist[mitr].start_adr < lowest) { - lowest = thislist[mitr].start_adr; - swapi = mitr; - } - if(lowest == thislist[i].start_adr) continue; - tmpaddr = thislist[swapi].start_adr; - tmpsize = thislist[swapi].num_bytes; - for(mitr = swapi; mitr > i; mitr--) { - thislist[mitr].start_adr = thislist[mitr-1].start_adr; - thislist[mitr].num_bytes = thislist[mitr-1].num_bytes; - } - thislist[i].start_adr = tmpaddr; - thislist[i].num_bytes = tmpsize; - } -} - -/* Initialize the memory lists based upon the prom version. */ -void __init prom_meminit(void) -{ - int node = 0; - unsigned int iter, num_regs; - - node = prom_finddevice("/memory"); - num_regs = prom_getproperty(node, "available", - (char *) prom_reg_memlist, - sizeof(prom_reg_memlist)); - num_regs = (num_regs/sizeof(struct linux_prom64_registers)); - for(iter=0; iter<num_regs; iter++) { - prom_phys_avail[iter].start_adr = - prom_reg_memlist[iter].phys_addr; - prom_phys_avail[iter].num_bytes = - prom_reg_memlist[iter].reg_size; - prom_phys_avail[iter].theres_more = - &prom_phys_avail[iter+1]; - } - prom_phys_avail[iter-1].theres_more = NULL; - - num_regs = prom_getproperty(node, "reg", - (char *) prom_reg_memlist, - sizeof(prom_reg_memlist)); - num_regs = (num_regs/sizeof(struct linux_prom64_registers)); - for(iter=0; iter<num_regs; iter++) { - prom_phys_total[iter].start_adr = - prom_reg_memlist[iter].phys_addr; - prom_phys_total[iter].num_bytes = - prom_reg_memlist[iter].reg_size; - prom_phys_total[iter].theres_more = - &prom_phys_total[iter+1]; - } - prom_phys_total[iter-1].theres_more = NULL; - - node = prom_finddevice("/virtual-memory"); - num_regs = prom_getproperty(node, "available", - (char *) prom_reg_memlist, - sizeof(prom_reg_memlist)); - num_regs = (num_regs/sizeof(struct linux_prom64_registers)); - - /* Convert available virtual areas to taken virtual - * areas. First sort, then convert. - */ - for(iter=0; iter<num_regs; iter++) { - prom_prom_taken[iter].start_adr = - prom_reg_memlist[iter].phys_addr; - prom_prom_taken[iter].num_bytes = - prom_reg_memlist[iter].reg_size; - prom_prom_taken[iter].theres_more = - &prom_prom_taken[iter+1]; - } - prom_prom_taken[iter-1].theres_more = NULL; - - prom_sortmemlist(prom_prom_taken); - - /* Finally, convert. */ - for(iter=0; iter<num_regs; iter++) { - prom_prom_taken[iter].start_adr = - prom_prom_taken[iter].start_adr + - prom_prom_taken[iter].num_bytes; - prom_prom_taken[iter].num_bytes = - prom_prom_taken[iter+1].start_adr - - prom_prom_taken[iter].start_adr; - } - prom_prom_taken[iter-1].num_bytes = - -1UL - prom_prom_taken[iter-1].start_adr; - - /* Sort the other two lists. */ - prom_sortmemlist(prom_phys_total); - prom_sortmemlist(prom_phys_avail); - - /* Link all the lists into the top-level descriptor. */ - prom_memlist.p1275_totphys=&prom_ptot_ptr; - prom_memlist.p1275_prommap=&prom_ptak_ptr; - prom_memlist.p1275_available=&prom_pavl_ptr; -} - -/* This returns a pointer to our libraries internal p1275 format - * memory descriptor. - */ -struct linux_mem_p1275 * -prom_meminfo(void) -{ - return &prom_memlist; -} diff --git a/arch/sparc64/prom/misc.c b/arch/sparc64/prom/misc.c index 19c44e97e9ee..87f5cfce23bb 100644 --- a/arch/sparc64/prom/misc.c +++ b/arch/sparc64/prom/misc.c @@ -17,14 +17,14 @@ #include <asm/system.h> /* Reset and reboot the machine with the command 'bcommand'. */ -void prom_reboot(char *bcommand) +void prom_reboot(const char *bcommand) { p1275_cmd("boot", P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_INOUT(1, 0), bcommand); } /* Forth evaluate the expression contained in 'fstring'. */ -void prom_feval(char *fstring) +void prom_feval(const char *fstring) { if (!fstring || fstring[0] == 0) return; @@ -68,19 +68,11 @@ void prom_cmdline(void) local_irq_restore(flags); } -#ifdef CONFIG_SMP -extern void smp_promstop_others(void); -#endif - /* Drop into the prom, but completely terminate the program. * No chance of continuing. */ void prom_halt(void) { -#ifdef CONFIG_SMP - smp_promstop_others(); - udelay(8000); -#endif again: p1275_cmd("exit", P1275_INOUT(0, 0)); goto again; /* PROM is out to get me -DaveM */ @@ -88,10 +80,6 @@ again: void prom_halt_power_off(void) { -#ifdef CONFIG_SMP - smp_promstop_others(); - udelay(8000); -#endif p1275_cmd("SUNW,power-off", P1275_INOUT(0, 0)); /* if nothing else helps, we just halt */ @@ -148,21 +136,19 @@ void prom_set_trap_table(unsigned long tba) p1275_cmd("SUNW,set-trap-table", P1275_INOUT(1, 0), tba); } -int mmu_ihandle_cache = 0; - int prom_get_mmu_ihandle(void) { int node, ret; - if (mmu_ihandle_cache != 0) - return mmu_ihandle_cache; + if (prom_mmu_ihandle_cache != 0) + return prom_mmu_ihandle_cache; - node = prom_finddevice("/chosen"); - ret = prom_getint(node, "mmu"); + node = prom_finddevice(prom_chosen_path); + ret = prom_getint(node, prom_mmu_name); if (ret == -1 || ret == 0) - mmu_ihandle_cache = -1; + prom_mmu_ihandle_cache = -1; else - mmu_ihandle_cache = ret; + prom_mmu_ihandle_cache = ret; return ret; } @@ -190,7 +176,7 @@ long prom_itlb_load(unsigned long index, unsigned long tte_data, unsigned long vaddr) { - return p1275_cmd("call-method", + return p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(2, P1275_ARG_IN_64B) | P1275_ARG(3, P1275_ARG_IN_64B) | @@ -207,7 +193,7 @@ long prom_dtlb_load(unsigned long index, unsigned long tte_data, unsigned long vaddr) { - return p1275_cmd("call-method", + return p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(2, P1275_ARG_IN_64B) | P1275_ARG(3, P1275_ARG_IN_64B) | @@ -223,13 +209,13 @@ long prom_dtlb_load(unsigned long index, int prom_map(int mode, unsigned long size, unsigned long vaddr, unsigned long paddr) { - int ret = p1275_cmd("call-method", + int ret = p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(3, P1275_ARG_IN_64B) | P1275_ARG(4, P1275_ARG_IN_64B) | P1275_ARG(6, P1275_ARG_IN_64B) | P1275_INOUT(7, 1)), - "map", + prom_map_name, prom_get_mmu_ihandle(), mode, size, @@ -244,12 +230,12 @@ int prom_map(int mode, unsigned long size, void prom_unmap(unsigned long size, unsigned long vaddr) { - p1275_cmd("call-method", + p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(2, P1275_ARG_IN_64B) | P1275_ARG(3, P1275_ARG_IN_64B) | P1275_INOUT(4, 0)), - "unmap", + prom_unmap_name, prom_get_mmu_ihandle(), size, vaddr); @@ -258,7 +244,7 @@ void prom_unmap(unsigned long size, unsigned long vaddr) /* Set aside physical memory which is not touched or modified * across soft resets. */ -unsigned long prom_retain(char *name, +unsigned long prom_retain(const char *name, unsigned long pa_low, unsigned long pa_high, long size, long align) { @@ -290,7 +276,7 @@ int prom_getunumber(int syndrome_code, unsigned long phys_addr, char *buf, int buflen) { - return p1275_cmd("call-method", + return p1275_cmd(prom_callmethod_name, (P1275_ARG(0, P1275_ARG_IN_STRING) | P1275_ARG(3, P1275_ARG_OUT_BUF) | P1275_ARG(6, P1275_ARG_IN_64B) | diff --git a/arch/sparc64/prom/p1275.c b/arch/sparc64/prom/p1275.c index 59fe38bba39e..a5a7c5712028 100644 --- a/arch/sparc64/prom/p1275.c +++ b/arch/sparc64/prom/p1275.c @@ -46,7 +46,7 @@ static inline unsigned long spitfire_get_primary_context(void) */ DEFINE_SPINLOCK(prom_entry_lock); -long p1275_cmd (char *service, long fmt, ...) +long p1275_cmd(const char *service, long fmt, ...) { char *p, *q; unsigned long flags; diff --git a/arch/sparc64/prom/printf.c b/arch/sparc64/prom/printf.c index a6df82cafa0d..660943ee4c2a 100644 --- a/arch/sparc64/prom/printf.c +++ b/arch/sparc64/prom/printf.c @@ -34,7 +34,7 @@ prom_write(const char *buf, unsigned int n) } void -prom_printf(char *fmt, ...) +prom_printf(const char *fmt, ...) { va_list args; int i; diff --git a/arch/sparc64/prom/tree.c b/arch/sparc64/prom/tree.c index ccf73258ebf7..b1ff9e87dcc6 100644 --- a/arch/sparc64/prom/tree.c +++ b/arch/sparc64/prom/tree.c @@ -69,7 +69,7 @@ prom_getsibling(int node) * Return -1 on error. */ __inline__ int -prom_getproplen(int node, char *prop) +prom_getproplen(int node, const char *prop) { if((!node) || (!prop)) return -1; return p1275_cmd ("getproplen", @@ -83,20 +83,20 @@ prom_getproplen(int node, char *prop) * was successful the length will be returned, else -1 is returned. */ __inline__ int -prom_getproperty(int node, char *prop, char *buffer, int bufsize) +prom_getproperty(int node, const char *prop, char *buffer, int bufsize) { int plen; plen = prom_getproplen(node, prop); - if((plen > bufsize) || (plen == 0) || (plen == -1)) + if ((plen > bufsize) || (plen == 0) || (plen == -1)) { return -1; - else { + } else { /* Ok, things seem all right. */ - return p1275_cmd ("getprop", - P1275_ARG(1,P1275_ARG_IN_STRING)| - P1275_ARG(2,P1275_ARG_OUT_BUF)| - P1275_INOUT(4, 1), - node, prop, buffer, P1275_SIZE(plen)); + return p1275_cmd(prom_getprop_name, + P1275_ARG(1,P1275_ARG_IN_STRING)| + P1275_ARG(2,P1275_ARG_OUT_BUF)| + P1275_INOUT(4, 1), + node, prop, buffer, P1275_SIZE(plen)); } } @@ -104,7 +104,7 @@ prom_getproperty(int node, char *prop, char *buffer, int bufsize) * on failure. */ __inline__ int -prom_getint(int node, char *prop) +prom_getint(int node, const char *prop) { int intprop; @@ -119,7 +119,7 @@ prom_getint(int node, char *prop) */ int -prom_getintdefault(int node, char *property, int deflt) +prom_getintdefault(int node, const char *property, int deflt) { int retval; @@ -131,7 +131,7 @@ prom_getintdefault(int node, char *property, int deflt) /* Acquire a boolean property, 1=TRUE 0=FALSE. */ int -prom_getbool(int node, char *prop) +prom_getbool(int node, const char *prop) { int retval; @@ -145,7 +145,7 @@ prom_getbool(int node, char *prop) * buffer. */ void -prom_getstring(int node, char *prop, char *user_buf, int ubuf_size) +prom_getstring(int node, const char *prop, char *user_buf, int ubuf_size) { int len; @@ -160,7 +160,7 @@ prom_getstring(int node, char *prop, char *user_buf, int ubuf_size) * YES = 1 NO = 0 */ int -prom_nodematch(int node, char *name) +prom_nodematch(int node, const char *name) { char namebuf[128]; prom_getproperty(node, "name", namebuf, sizeof(namebuf)); @@ -172,7 +172,7 @@ prom_nodematch(int node, char *name) * 'nodename'. Return node if successful, zero if not. */ int -prom_searchsiblings(int node_start, char *nodename) +prom_searchsiblings(int node_start, const char *nodename) { int thisnode, error; @@ -294,7 +294,7 @@ prom_firstprop(int node, char *buffer) * property types for this node. */ __inline__ char * -prom_nextprop(int node, char *oprop, char *buffer) +prom_nextprop(int node, const char *oprop, char *buffer) { char buf[32]; @@ -314,15 +314,17 @@ prom_nextprop(int node, char *oprop, char *buffer) } int -prom_finddevice(char *name) +prom_finddevice(const char *name) { - if(!name) return 0; - return p1275_cmd ("finddevice", P1275_ARG(0,P1275_ARG_IN_STRING)| - P1275_INOUT(1, 1), - name); + if (!name) + return 0; + return p1275_cmd(prom_finddev_name, + P1275_ARG(0,P1275_ARG_IN_STRING)| + P1275_INOUT(1, 1), + name); } -int prom_node_has_property(int node, char *prop) +int prom_node_has_property(int node, const char *prop) { char buf [32]; @@ -339,7 +341,7 @@ int prom_node_has_property(int node, char *prop) * of 'size' bytes. Return the number of bytes the prom accepted. */ int -prom_setprop(int node, char *pname, char *value, int size) +prom_setprop(int node, const char *pname, char *value, int size) { if(size == 0) return 0; if((pname == 0) || (value == 0)) return 0; @@ -364,7 +366,7 @@ prom_inst2pkg(int inst) * FIXME: Should work for v0 as well */ int -prom_pathtoinode(char *path) +prom_pathtoinode(const char *path) { int node, inst; diff --git a/arch/sparc64/solaris/ioctl.c b/arch/sparc64/solaris/ioctl.c index cac0a1cf0050..be0a054e3ed6 100644 --- a/arch/sparc64/solaris/ioctl.c +++ b/arch/sparc64/solaris/ioctl.c @@ -24,6 +24,7 @@ #include <linux/netdevice.h> #include <linux/mtio.h> #include <linux/time.h> +#include <linux/rcupdate.h> #include <linux/compat.h> #include <net/sock.h> @@ -293,16 +294,18 @@ static struct module_info { static inline int solaris_sockmod(unsigned int fd, unsigned int cmd, u32 arg) { struct inode *ino; + struct fdtable *fdt; /* I wonder which of these tests are superfluous... --patrik */ - spin_lock(¤t->files->file_lock); - if (! current->files->fd[fd] || - ! current->files->fd[fd]->f_dentry || - ! (ino = current->files->fd[fd]->f_dentry->d_inode) || + rcu_read_lock(); + fdt = files_fdtable(current->files); + if (! fdt->fd[fd] || + ! fdt->fd[fd]->f_dentry || + ! (ino = fdt->fd[fd]->f_dentry->d_inode) || ! S_ISSOCK(ino->i_mode)) { - spin_unlock(¤t->files->file_lock); + rcu_read_unlock(); return TBADF; } - spin_unlock(¤t->files->file_lock); + rcu_read_unlock(); switch (cmd & 0xff) { case 109: /* SI_SOCKPARAMS */ diff --git a/arch/sparc64/solaris/misc.c b/arch/sparc64/solaris/misc.c index 15b4cfe07557..302efbcba70e 100644 --- a/arch/sparc64/solaris/misc.c +++ b/arch/sparc64/solaris/misc.c @@ -737,7 +737,8 @@ MODULE_LICENSE("GPL"); extern u32 tl0_solaris[8]; #define update_ttable(x) \ tl0_solaris[3] = (((long)(x) - (long)tl0_solaris - 3) >> 2) | 0x40000000; \ - __asm__ __volatile__ ("membar #StoreStore; flush %0" : : "r" (&tl0_solaris[3])) + wmb(); \ + __asm__ __volatile__ ("flush %0" : : "r" (&tl0_solaris[3])) #else #endif @@ -761,7 +762,8 @@ int init_module(void) entry64_personality_patch |= (offsetof(struct task_struct, personality) + (sizeof(unsigned long) - 1)); - __asm__ __volatile__("membar #StoreStore; flush %0" + wmb(); + __asm__ __volatile__("flush %0" : : "r" (&entry64_personality_patch)); return 0; } diff --git a/arch/sparc64/solaris/socket.c b/arch/sparc64/solaris/socket.c index 06740582717e..d3a66ea74a7f 100644 --- a/arch/sparc64/solaris/socket.c +++ b/arch/sparc64/solaris/socket.c @@ -16,6 +16,7 @@ #include <linux/net.h> #include <linux/compat.h> #include <net/compat.h> +#include <net/sock.h> #include <asm/uaccess.h> #include <asm/string.h> @@ -297,121 +298,165 @@ asmlinkage int solaris_sendmsg(int fd, struct sol_nmsghdr __user *user_msg, unsi { struct socket *sock; char address[MAX_SOCK_ADDR]; - struct iovec iov[UIO_FASTIOV]; + struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; unsigned char ctl[sizeof(struct cmsghdr) + 20]; unsigned char *ctl_buf = ctl; - struct msghdr kern_msg; - int err, total_len; + struct msghdr msg_sys; + int err, ctl_len, iov_size, total_len; - if(msghdr_from_user32_to_kern(&kern_msg, user_msg)) - return -EFAULT; - if(kern_msg.msg_iovlen > UIO_MAXIOV) - return -EINVAL; - err = verify_compat_iovec(&kern_msg, iov, address, VERIFY_READ); - if (err < 0) + err = -EFAULT; + if (msghdr_from_user32_to_kern(&msg_sys, user_msg)) + goto out; + + sock = sockfd_lookup(fd, &err); + if (!sock) goto out; + + /* do not move before msg_sys is valid */ + err = -EMSGSIZE; + if (msg_sys.msg_iovlen > UIO_MAXIOV) + goto out_put; + + /* Check whether to allocate the iovec area*/ + err = -ENOMEM; + iov_size = msg_sys.msg_iovlen * sizeof(struct iovec); + if (msg_sys.msg_iovlen > UIO_FASTIOV) { + iov = sock_kmalloc(sock->sk, iov_size, GFP_KERNEL); + if (!iov) + goto out_put; + } + + err = verify_compat_iovec(&msg_sys, iov, address, VERIFY_READ); + if (err < 0) + goto out_freeiov; total_len = err; - if(kern_msg.msg_controllen) { - struct sol_cmsghdr __user *ucmsg = kern_msg.msg_control; + err = -ENOBUFS; + if (msg_sys.msg_controllen > INT_MAX) + goto out_freeiov; + + ctl_len = msg_sys.msg_controllen; + if (ctl_len) { + struct sol_cmsghdr __user *ucmsg = msg_sys.msg_control; unsigned long *kcmsg; compat_size_t cmlen; - if (kern_msg.msg_controllen <= sizeof(compat_size_t)) - return -EINVAL; + err = -EINVAL; + if (ctl_len <= sizeof(compat_size_t)) + goto out_freeiov; - if(kern_msg.msg_controllen > sizeof(ctl)) { + if (ctl_len > sizeof(ctl)) { err = -ENOBUFS; - ctl_buf = kmalloc(kern_msg.msg_controllen, GFP_KERNEL); - if(!ctl_buf) + ctl_buf = kmalloc(ctl_len, GFP_KERNEL); + if (!ctl_buf) goto out_freeiov; } __get_user(cmlen, &ucmsg->cmsg_len); kcmsg = (unsigned long *) ctl_buf; *kcmsg++ = (unsigned long)cmlen; err = -EFAULT; - if(copy_from_user(kcmsg, &ucmsg->cmsg_level, - kern_msg.msg_controllen - sizeof(compat_size_t))) + if (copy_from_user(kcmsg, &ucmsg->cmsg_level, + ctl_len - sizeof(compat_size_t))) goto out_freectl; - kern_msg.msg_control = ctl_buf; + msg_sys.msg_control = ctl_buf; } - kern_msg.msg_flags = solaris_to_linux_msgflags(user_flags); + msg_sys.msg_flags = solaris_to_linux_msgflags(user_flags); - lock_kernel(); - sock = sockfd_lookup(fd, &err); - if (sock != NULL) { - if (sock->file->f_flags & O_NONBLOCK) - kern_msg.msg_flags |= MSG_DONTWAIT; - err = sock_sendmsg(sock, &kern_msg, total_len); - sockfd_put(sock); - } - unlock_kernel(); + if (sock->file->f_flags & O_NONBLOCK) + msg_sys.msg_flags |= MSG_DONTWAIT; + err = sock_sendmsg(sock, &msg_sys, total_len); out_freectl: - /* N.B. Use kfree here, as kern_msg.msg_controllen might change? */ - if(ctl_buf != ctl) - kfree(ctl_buf); + if (ctl_buf != ctl) + sock_kfree_s(sock->sk, ctl_buf, ctl_len); out_freeiov: - if(kern_msg.msg_iov != iov) - kfree(kern_msg.msg_iov); -out: + if (iov != iovstack) + sock_kfree_s(sock->sk, iov, iov_size); +out_put: + sockfd_put(sock); +out: return err; } asmlinkage int solaris_recvmsg(int fd, struct sol_nmsghdr __user *user_msg, unsigned int user_flags) { - struct iovec iovstack[UIO_FASTIOV]; - struct msghdr kern_msg; - char addr[MAX_SOCK_ADDR]; struct socket *sock; + struct iovec iovstack[UIO_FASTIOV]; struct iovec *iov = iovstack; + struct msghdr msg_sys; + unsigned long cmsg_ptr; + int err, iov_size, total_len, len; + + /* kernel mode address */ + char addr[MAX_SOCK_ADDR]; + + /* user mode address pointers */ struct sockaddr __user *uaddr; int __user *uaddr_len; - unsigned long cmsg_ptr; - int err, total_len, len = 0; - if(msghdr_from_user32_to_kern(&kern_msg, user_msg)) + if (msghdr_from_user32_to_kern(&msg_sys, user_msg)) return -EFAULT; - if(kern_msg.msg_iovlen > UIO_MAXIOV) - return -EINVAL; - uaddr = kern_msg.msg_name; + sock = sockfd_lookup(fd, &err); + if (!sock) + goto out; + + err = -EMSGSIZE; + if (msg_sys.msg_iovlen > UIO_MAXIOV) + goto out_put; + + /* Check whether to allocate the iovec area*/ + err = -ENOMEM; + iov_size = msg_sys.msg_iovlen * sizeof(struct iovec); + if (msg_sys.msg_iovlen > UIO_FASTIOV) { + iov = sock_kmalloc(sock->sk, iov_size, GFP_KERNEL); + if (!iov) + goto out_put; + } + + /* + * Save the user-mode address (verify_iovec will change the + * kernel msghdr to use the kernel address space) + */ + + uaddr = (void __user *) msg_sys.msg_name; uaddr_len = &user_msg->msg_namelen; - err = verify_compat_iovec(&kern_msg, iov, addr, VERIFY_WRITE); + err = verify_compat_iovec(&msg_sys, iov, addr, VERIFY_WRITE); if (err < 0) - goto out; + goto out_freeiov; total_len = err; - cmsg_ptr = (unsigned long) kern_msg.msg_control; - kern_msg.msg_flags = 0; + cmsg_ptr = (unsigned long) msg_sys.msg_control; + msg_sys.msg_flags = MSG_CMSG_COMPAT; - lock_kernel(); - sock = sockfd_lookup(fd, &err); - if (sock != NULL) { - if (sock->file->f_flags & O_NONBLOCK) - user_flags |= MSG_DONTWAIT; - err = sock_recvmsg(sock, &kern_msg, total_len, user_flags); - if(err >= 0) - len = err; - sockfd_put(sock); - } - unlock_kernel(); - - if(uaddr != NULL && err >= 0) - err = move_addr_to_user(addr, kern_msg.msg_namelen, uaddr, uaddr_len); - if(err >= 0) { - err = __put_user(linux_to_solaris_msgflags(kern_msg.msg_flags), &user_msg->msg_flags); - if(!err) { - /* XXX Convert cmsg back into userspace 32-bit format... */ - err = __put_user((unsigned long)kern_msg.msg_control - cmsg_ptr, - &user_msg->msg_controllen); - } + if (sock->file->f_flags & O_NONBLOCK) + user_flags |= MSG_DONTWAIT; + + err = sock_recvmsg(sock, &msg_sys, total_len, user_flags); + if(err < 0) + goto out_freeiov; + + len = err; + + if (uaddr != NULL) { + err = move_addr_to_user(addr, msg_sys.msg_namelen, uaddr, uaddr_len); + if (err < 0) + goto out_freeiov; } + err = __put_user(linux_to_solaris_msgflags(msg_sys.msg_flags), &user_msg->msg_flags); + if (err) + goto out_freeiov; + err = __put_user((unsigned long)msg_sys.msg_control - cmsg_ptr, + &user_msg->msg_controllen); + if (err) + goto out_freeiov; + err = len; - if(kern_msg.msg_iov != iov) - kfree(kern_msg.msg_iov); +out_freeiov: + if (iov != iovstack) + sock_kfree_s(sock->sk, iov, iov_size); +out_put: + sockfd_put(sock); out: - if(err < 0) - return err; - return len; + return err; } diff --git a/arch/sparc64/solaris/socksys.c b/arch/sparc64/solaris/socksys.c index d7c1c76582cc..fc6669e8dde1 100644 --- a/arch/sparc64/solaris/socksys.c +++ b/arch/sparc64/solaris/socksys.c @@ -49,7 +49,7 @@ IPPROTO_EGP, IPPROTO_PUP, IPPROTO_UDP, IPPROTO_IDP, IPPROTO_RAW, #else -extern void * mykmalloc(size_t s, int gfp); +extern void * mykmalloc(size_t s, gfp_t gfp); extern void mykfree(void *); #endif diff --git a/arch/sparc64/solaris/timod.c b/arch/sparc64/solaris/timod.c index 022c80f43392..b84e5456b025 100644 --- a/arch/sparc64/solaris/timod.c +++ b/arch/sparc64/solaris/timod.c @@ -39,7 +39,7 @@ static char * page = NULL ; #else -void * mykmalloc(size_t s, int gfp) +void * mykmalloc(size_t s, gfp_t gfp) { static char * page; static size_t free; @@ -143,9 +143,11 @@ static struct T_primsg *timod_mkctl(int size) static void timod_wake_socket(unsigned int fd) { struct socket *sock; + struct fdtable *fdt; SOLD("wakeing socket"); - sock = SOCKET_I(current->files->fd[fd]->f_dentry->d_inode); + fdt = files_fdtable(current->files); + sock = SOCKET_I(fdt->fd[fd]->f_dentry->d_inode); wake_up_interruptible(&sock->wait); read_lock(&sock->sk->sk_callback_lock); if (sock->fasync_list && !test_bit(SOCK_ASYNC_WAITDATA, &sock->flags)) @@ -157,9 +159,11 @@ static void timod_wake_socket(unsigned int fd) static void timod_queue(unsigned int fd, struct T_primsg *it) { struct sol_socket_struct *sock; + struct fdtable *fdt; SOLD("queuing primsg"); - sock = (struct sol_socket_struct *)current->files->fd[fd]->private_data; + fdt = files_fdtable(current->files); + sock = (struct sol_socket_struct *)fdt->fd[fd]->private_data; it->next = sock->pfirst; sock->pfirst = it; if (!sock->plast) @@ -171,9 +175,11 @@ static void timod_queue(unsigned int fd, struct T_primsg *it) static void timod_queue_end(unsigned int fd, struct T_primsg *it) { struct sol_socket_struct *sock; + struct fdtable *fdt; SOLD("queuing primsg at end"); - sock = (struct sol_socket_struct *)current->files->fd[fd]->private_data; + fdt = files_fdtable(current->files); + sock = (struct sol_socket_struct *)fdt->fd[fd]->private_data; it->next = NULL; if (sock->plast) sock->plast->next = it; @@ -344,6 +350,7 @@ int timod_putmsg(unsigned int fd, char __user *ctl_buf, int ctl_len, char *buf; struct file *filp; struct inode *ino; + struct fdtable *fdt; struct sol_socket_struct *sock; mm_segment_t old_fs = get_fs(); long args[6]; @@ -351,7 +358,9 @@ int timod_putmsg(unsigned int fd, char __user *ctl_buf, int ctl_len, (int (*)(int, unsigned long __user *))SYS(socketcall); int (*sys_sendto)(int, void __user *, size_t, unsigned, struct sockaddr __user *, int) = (int (*)(int, void __user *, size_t, unsigned, struct sockaddr __user *, int))SYS(sendto); - filp = current->files->fd[fd]; + + fdt = files_fdtable(current->files); + filp = fdt->fd[fd]; ino = filp->f_dentry->d_inode; sock = (struct sol_socket_struct *)filp->private_data; SOLD("entry"); @@ -620,6 +629,7 @@ int timod_getmsg(unsigned int fd, char __user *ctl_buf, int ctl_maxlen, s32 __us int oldflags; struct file *filp; struct inode *ino; + struct fdtable *fdt; struct sol_socket_struct *sock; struct T_unitdata_ind udi; mm_segment_t old_fs = get_fs(); @@ -632,7 +642,8 @@ int timod_getmsg(unsigned int fd, char __user *ctl_buf, int ctl_maxlen, s32 __us SOLD("entry"); SOLDD(("%u %p %d %p %p %d %p %d\n", fd, ctl_buf, ctl_maxlen, ctl_len, data_buf, data_maxlen, data_len, *flags_p)); - filp = current->files->fd[fd]; + fdt = files_fdtable(current->files); + filp = fdt->fd[fd]; ino = filp->f_dentry->d_inode; sock = (struct sol_socket_struct *)filp->private_data; SOLDD(("%p %p\n", sock->pfirst, sock->pfirst ? sock->pfirst->next : NULL)); @@ -844,12 +855,14 @@ asmlinkage int solaris_getmsg(unsigned int fd, u32 arg1, u32 arg2, u32 arg3) int __user *flgptr; int flags; int error = -EBADF; + struct fdtable *fdt; SOLD("entry"); lock_kernel(); if(fd >= NR_OPEN) goto out; - filp = current->files->fd[fd]; + fdt = files_fdtable(current->files); + filp = fdt->fd[fd]; if(!filp) goto out; ino = filp->f_dentry->d_inode; @@ -910,12 +923,14 @@ asmlinkage int solaris_putmsg(unsigned int fd, u32 arg1, u32 arg2, u32 arg3) struct strbuf ctl, dat; int flags = (int) arg3; int error = -EBADF; + struct fdtable *fdt; SOLD("entry"); lock_kernel(); if(fd >= NR_OPEN) goto out; - filp = current->files->fd[fd]; + fdt = files_fdtable(current->files); + filp = fdt->fd[fd]; if(!filp) goto out; ino = filp->f_dentry->d_inode; |