From 25a3cd8189c8832c04225e6f1d41228fd6cc64cc Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 3 Oct 2019 11:18:54 -0700 Subject: net/tls: move TOE-related structures to a separate header Move tls_device structure and register/unregister functions to a new header to avoid confusion with normal, non-TOE offload. Signed-off-by: Jakub Kicinski Reviewed-by: John Hurley Reviewed-by: Simon Horman Signed-off-by: David S. Miller --- drivers/crypto/chelsio/chtls/chtls.h | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/crypto') diff --git a/drivers/crypto/chelsio/chtls/chtls.h b/drivers/crypto/chelsio/chtls/chtls.h index 025c831d0899..e353c42fea91 100644 --- a/drivers/crypto/chelsio/chtls/chtls.h +++ b/drivers/crypto/chelsio/chtls/chtls.h @@ -21,6 +21,7 @@ #include #include #include +#include #include "t4fw_api.h" #include "t4_msg.h" -- cgit v1.2.3-59-g8ed1b From f21912edd1570818cbcb16bd1da7d7a2b122d66b Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 3 Oct 2019 11:18:55 -0700 Subject: net/tls: rename tls_device to tls_toe_device Rename struct tls_device to struct tls_toe_device to avoid confusion with normal, non-TOE offload. No functional changes. Signed-off-by: Jakub Kicinski Reviewed-by: John Hurley Reviewed-by: Simon Horman Signed-off-by: David S. Miller --- drivers/crypto/chelsio/chtls/chtls.h | 4 ++-- drivers/crypto/chelsio/chtls/chtls_main.c | 20 ++++++++++---------- include/net/tls_toe.h | 24 ++++++++++++------------ net/tls/tls_main.c | 14 +++++++------- 4 files changed, 31 insertions(+), 31 deletions(-) (limited to 'drivers/crypto') diff --git a/drivers/crypto/chelsio/chtls/chtls.h b/drivers/crypto/chelsio/chtls/chtls.h index e353c42fea91..d2bc655ab931 100644 --- a/drivers/crypto/chelsio/chtls/chtls.h +++ b/drivers/crypto/chelsio/chtls/chtls.h @@ -119,7 +119,7 @@ struct tls_scmd { }; struct chtls_dev { - struct tls_device tlsdev; + struct tls_toe_device tlsdev; struct list_head list; struct cxgb4_lld_info *lldi; struct pci_dev *pdev; @@ -363,7 +363,7 @@ enum { #define TCP_PAGE(sk) (sk->sk_frag.page) #define TCP_OFF(sk) (sk->sk_frag.offset) -static inline struct chtls_dev *to_chtls_dev(struct tls_device *tlsdev) +static inline struct chtls_dev *to_chtls_dev(struct tls_toe_device *tlsdev) { return container_of(tlsdev, struct chtls_dev, tlsdev); } diff --git a/drivers/crypto/chelsio/chtls/chtls_main.c b/drivers/crypto/chelsio/chtls/chtls_main.c index e6df5b95ed47..18996935d8ba 100644 --- a/drivers/crypto/chelsio/chtls/chtls_main.c +++ b/drivers/crypto/chelsio/chtls/chtls_main.c @@ -124,7 +124,7 @@ static void chtls_stop_listen(struct chtls_dev *cdev, struct sock *sk) mutex_unlock(¬ify_mutex); } -static int chtls_inline_feature(struct tls_device *dev) +static int chtls_inline_feature(struct tls_toe_device *dev) { struct net_device *netdev; struct chtls_dev *cdev; @@ -140,7 +140,7 @@ static int chtls_inline_feature(struct tls_device *dev) return 0; } -static int chtls_create_hash(struct tls_device *dev, struct sock *sk) +static int chtls_create_hash(struct tls_toe_device *dev, struct sock *sk) { struct chtls_dev *cdev = to_chtls_dev(dev); @@ -149,7 +149,7 @@ static int chtls_create_hash(struct tls_device *dev, struct sock *sk) return 0; } -static void chtls_destroy_hash(struct tls_device *dev, struct sock *sk) +static void chtls_destroy_hash(struct tls_toe_device *dev, struct sock *sk) { struct chtls_dev *cdev = to_chtls_dev(dev); @@ -161,7 +161,7 @@ static void chtls_free_uld(struct chtls_dev *cdev) { int i; - tls_unregister_device(&cdev->tlsdev); + tls_toe_unregister_device(&cdev->tlsdev); kvfree(cdev->kmap.addr); idr_destroy(&cdev->hwtid_idr); for (i = 0; i < (1 << RSPQ_HASH_BITS); i++) @@ -173,27 +173,27 @@ static void chtls_free_uld(struct chtls_dev *cdev) static inline void chtls_dev_release(struct kref *kref) { + struct tls_toe_device *dev; struct chtls_dev *cdev; - struct tls_device *dev; - dev = container_of(kref, struct tls_device, kref); + dev = container_of(kref, struct tls_toe_device, kref); cdev = to_chtls_dev(dev); chtls_free_uld(cdev); } static void chtls_register_dev(struct chtls_dev *cdev) { - struct tls_device *tlsdev = &cdev->tlsdev; + struct tls_toe_device *tlsdev = &cdev->tlsdev; - strlcpy(tlsdev->name, "chtls", TLS_DEVICE_NAME_MAX); + strlcpy(tlsdev->name, "chtls", TLS_TOE_DEVICE_NAME_MAX); strlcat(tlsdev->name, cdev->lldi->ports[0]->name, - TLS_DEVICE_NAME_MAX); + TLS_TOE_DEVICE_NAME_MAX); tlsdev->feature = chtls_inline_feature; tlsdev->hash = chtls_create_hash; tlsdev->unhash = chtls_destroy_hash; tlsdev->release = chtls_dev_release; kref_init(&tlsdev->kref); - tls_register_device(tlsdev); + tls_toe_register_device(tlsdev); cdev->cdev_state = CHTLS_CDEV_STATE_UP; } diff --git a/include/net/tls_toe.h b/include/net/tls_toe.h index 81b66c76b31f..b56d30a5bd6d 100644 --- a/include/net/tls_toe.h +++ b/include/net/tls_toe.h @@ -36,7 +36,7 @@ struct sock; -#define TLS_DEVICE_NAME_MAX 32 +#define TLS_TOE_DEVICE_NAME_MAX 32 /* * This structure defines the routines for Inline TLS driver. @@ -45,29 +45,29 @@ struct sock; * * @name: Its the name of registered Inline tls device * @dev_list: Inline tls device list - * int (*feature)(struct tls_device *device); + * int (*feature)(struct tls_toe_device *device); * Called to return Inline TLS driver capability * - * int (*hash)(struct tls_device *device, struct sock *sk); + * int (*hash)(struct tls_toe_device *device, struct sock *sk); * This function sets Inline driver for listen and program * device specific functioanlity as required * - * void (*unhash)(struct tls_device *device, struct sock *sk); + * void (*unhash)(struct tls_toe_device *device, struct sock *sk); * This function cleans listen state set by Inline TLS driver * * void (*release)(struct kref *kref); * Release the registered device and allocated resources - * @kref: Number of reference to tls_device + * @kref: Number of reference to tls_toe_device */ -struct tls_device { - char name[TLS_DEVICE_NAME_MAX]; +struct tls_toe_device { + char name[TLS_TOE_DEVICE_NAME_MAX]; struct list_head dev_list; - int (*feature)(struct tls_device *device); - int (*hash)(struct tls_device *device, struct sock *sk); - void (*unhash)(struct tls_device *device, struct sock *sk); + int (*feature)(struct tls_toe_device *device); + int (*hash)(struct tls_toe_device *device, struct sock *sk); + void (*unhash)(struct tls_toe_device *device, struct sock *sk); void (*release)(struct kref *kref); struct kref kref; }; -void tls_register_device(struct tls_device *device); -void tls_unregister_device(struct tls_device *device); +void tls_toe_register_device(struct tls_toe_device *device); +void tls_toe_unregister_device(struct tls_toe_device *device); diff --git a/net/tls/tls_main.c b/net/tls/tls_main.c index a19c6a1e034a..a1203807a3ef 100644 --- a/net/tls/tls_main.c +++ b/net/tls/tls_main.c @@ -657,8 +657,8 @@ static void tls_hw_sk_destruct(struct sock *sk) static int tls_hw_prot(struct sock *sk) { + struct tls_toe_device *dev; struct tls_context *ctx; - struct tls_device *dev; int rc = 0; spin_lock_bh(&device_spinlock); @@ -688,7 +688,7 @@ out: static void tls_hw_unhash(struct sock *sk) { struct tls_context *ctx = tls_get_ctx(sk); - struct tls_device *dev; + struct tls_toe_device *dev; spin_lock_bh(&device_spinlock); list_for_each_entry(dev, &device_list, dev_list) { @@ -707,7 +707,7 @@ static void tls_hw_unhash(struct sock *sk) static int tls_hw_hash(struct sock *sk) { struct tls_context *ctx = tls_get_ctx(sk); - struct tls_device *dev; + struct tls_toe_device *dev; int err; err = ctx->sk_proto->hash(sk); @@ -878,21 +878,21 @@ static size_t tls_get_info_size(const struct sock *sk) return size; } -void tls_register_device(struct tls_device *device) +void tls_toe_register_device(struct tls_toe_device *device) { spin_lock_bh(&device_spinlock); list_add_tail(&device->dev_list, &device_list); spin_unlock_bh(&device_spinlock); } -EXPORT_SYMBOL(tls_register_device); +EXPORT_SYMBOL(tls_toe_register_device); -void tls_unregister_device(struct tls_device *device) +void tls_toe_unregister_device(struct tls_toe_device *device) { spin_lock_bh(&device_spinlock); list_del(&device->dev_list); spin_unlock_bh(&device_spinlock); } -EXPORT_SYMBOL(tls_unregister_device); +EXPORT_SYMBOL(tls_toe_unregister_device); static struct tcp_ulp_ops tcp_tls_ulp_ops __read_mostly = { .name = "tls", -- cgit v1.2.3-59-g8ed1b From 53b4414a7003099f41ab61ef9a452804c025e2c1 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 3 Oct 2019 11:18:59 -0700 Subject: net/tls: allow compiling TLS TOE out TLS "record layer offload" requires TOE, and bypasses most of the normal networking stack. It is also significantly less maintained. Allow users to compile it out to avoid issues. Signed-off-by: Jakub Kicinski Reviewed-by: John Hurley Reviewed-by: Simon Horman Signed-off-by: David S. Miller --- drivers/crypto/chelsio/Kconfig | 2 +- net/tls/Kconfig | 10 ++++++++++ net/tls/Makefile | 3 ++- net/tls/tls_main.c | 5 ++++- 4 files changed, 17 insertions(+), 3 deletions(-) (limited to 'drivers/crypto') diff --git a/drivers/crypto/chelsio/Kconfig b/drivers/crypto/chelsio/Kconfig index 250150560e68..91e424378217 100644 --- a/drivers/crypto/chelsio/Kconfig +++ b/drivers/crypto/chelsio/Kconfig @@ -35,7 +35,7 @@ config CHELSIO_IPSEC_INLINE config CRYPTO_DEV_CHELSIO_TLS tristate "Chelsio Crypto Inline TLS Driver" depends on CHELSIO_T4 - depends on TLS + depends on TLS_TOE select CRYPTO_DEV_CHELSIO ---help--- Support Chelsio Inline TLS with Chelsio crypto accelerator. diff --git a/net/tls/Kconfig b/net/tls/Kconfig index e4328b3b72eb..61ec78521a60 100644 --- a/net/tls/Kconfig +++ b/net/tls/Kconfig @@ -26,3 +26,13 @@ config TLS_DEVICE Enable kernel support for HW offload of the TLS protocol. If unsure, say N. + +config TLS_TOE + bool "Transport Layer Security TCP stack bypass" + depends on TLS + default n + help + Enable kernel support for legacy HW offload of the TLS protocol, + which is incompatible with the Linux networking stack semantics. + + If unsure, say N. diff --git a/net/tls/Makefile b/net/tls/Makefile index 322250e912db..95d8c06a14b9 100644 --- a/net/tls/Makefile +++ b/net/tls/Makefile @@ -5,6 +5,7 @@ obj-$(CONFIG_TLS) += tls.o -tls-y := tls_main.o tls_sw.o tls_toe.o +tls-y := tls_main.o tls_sw.o +tls-$(CONFIG_TLS_TOE) += tls_toe.o tls-$(CONFIG_TLS_DEVICE) += tls_device.o tls_device_fallback.o diff --git a/net/tls/tls_main.c b/net/tls/tls_main.c index 483dda6c3155..237e58e4928a 100644 --- a/net/tls/tls_main.c +++ b/net/tls/tls_main.c @@ -679,10 +679,11 @@ static void build_protos(struct proto prot[TLS_NUM_CONFIG][TLS_NUM_CONFIG], prot[TLS_HW][TLS_HW] = prot[TLS_HW][TLS_SW]; #endif - +#ifdef CONFIG_TLS_TOE prot[TLS_HW_RECORD][TLS_HW_RECORD] = *base; prot[TLS_HW_RECORD][TLS_HW_RECORD].hash = tls_toe_hash; prot[TLS_HW_RECORD][TLS_HW_RECORD].unhash = tls_toe_unhash; +#endif } static int tls_init(struct sock *sk) @@ -692,8 +693,10 @@ static int tls_init(struct sock *sk) tls_build_proto(sk); +#ifdef CONFIG_TLS_TOE if (tls_toe_bypass(sk)) return 0; +#endif /* The TLS ulp is currently supported only for TCP sockets * in ESTABLISHED state. -- cgit v1.2.3-59-g8ed1b From 9ed498c6280a2f2b51d02df96df53037272ede49 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 6 Nov 2019 10:04:11 -0800 Subject: net: silence data-races on sk_backlog.tail sk->sk_backlog.tail might be read without holding the socket spinlock, we need to add proper READ_ONCE()/WRITE_ONCE() to silence the warnings. KCSAN reported : BUG: KCSAN: data-race in tcp_add_backlog / tcp_recvmsg write to 0xffff8881265109f8 of 8 bytes by interrupt on cpu 1: __sk_add_backlog include/net/sock.h:907 [inline] sk_add_backlog include/net/sock.h:938 [inline] tcp_add_backlog+0x476/0xce0 net/ipv4/tcp_ipv4.c:1759 tcp_v4_rcv+0x1a70/0x1bd0 net/ipv4/tcp_ipv4.c:1947 ip_protocol_deliver_rcu+0x4d/0x420 net/ipv4/ip_input.c:204 ip_local_deliver_finish+0x110/0x140 net/ipv4/ip_input.c:231 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip_local_deliver+0x133/0x210 net/ipv4/ip_input.c:252 dst_input include/net/dst.h:442 [inline] ip_rcv_finish+0x121/0x160 net/ipv4/ip_input.c:413 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip_rcv+0x18f/0x1a0 net/ipv4/ip_input.c:523 __netif_receive_skb_one_core+0xa7/0xe0 net/core/dev.c:4929 __netif_receive_skb+0x37/0xf0 net/core/dev.c:5043 netif_receive_skb_internal+0x59/0x190 net/core/dev.c:5133 napi_skb_finish net/core/dev.c:5596 [inline] napi_gro_receive+0x28f/0x330 net/core/dev.c:5629 receive_buf+0x284/0x30b0 drivers/net/virtio_net.c:1061 virtnet_receive drivers/net/virtio_net.c:1323 [inline] virtnet_poll+0x436/0x7d0 drivers/net/virtio_net.c:1428 napi_poll net/core/dev.c:6311 [inline] net_rx_action+0x3ae/0xa90 net/core/dev.c:6379 __do_softirq+0x115/0x33f kernel/softirq.c:292 invoke_softirq kernel/softirq.c:373 [inline] irq_exit+0xbb/0xe0 kernel/softirq.c:413 exiting_irq arch/x86/include/asm/apic.h:536 [inline] do_IRQ+0xa6/0x180 arch/x86/kernel/irq.c:263 ret_from_intr+0x0/0x19 native_safe_halt+0xe/0x10 arch/x86/kernel/paravirt.c:71 arch_cpu_idle+0x1f/0x30 arch/x86/kernel/process.c:571 default_idle_call+0x1e/0x40 kernel/sched/idle.c:94 cpuidle_idle_call kernel/sched/idle.c:154 [inline] do_idle+0x1af/0x280 kernel/sched/idle.c:263 cpu_startup_entry+0x1b/0x20 kernel/sched/idle.c:355 start_secondary+0x208/0x260 arch/x86/kernel/smpboot.c:264 secondary_startup_64+0xa4/0xb0 arch/x86/kernel/head_64.S:241 read to 0xffff8881265109f8 of 8 bytes by task 8057 on cpu 0: tcp_recvmsg+0x46e/0x1b40 net/ipv4/tcp.c:2050 inet_recvmsg+0xbb/0x250 net/ipv4/af_inet.c:838 sock_recvmsg_nosec net/socket.c:871 [inline] sock_recvmsg net/socket.c:889 [inline] sock_recvmsg+0x92/0xb0 net/socket.c:885 sock_read_iter+0x15f/0x1e0 net/socket.c:967 call_read_iter include/linux/fs.h:1889 [inline] new_sync_read+0x389/0x4f0 fs/read_write.c:414 __vfs_read+0xb1/0xc0 fs/read_write.c:427 vfs_read fs/read_write.c:461 [inline] vfs_read+0x143/0x2c0 fs/read_write.c:446 ksys_read+0xd5/0x1b0 fs/read_write.c:587 __do_sys_read fs/read_write.c:597 [inline] __se_sys_read fs/read_write.c:595 [inline] __x64_sys_read+0x4c/0x60 fs/read_write.c:595 do_syscall_64+0xcc/0x370 arch/x86/entry/common.c:290 entry_SYSCALL_64_after_hwframe+0x44/0xa9 Reported by Kernel Concurrency Sanitizer on: CPU: 0 PID: 8057 Comm: syz-fuzzer Not tainted 5.4.0-rc6+ #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- drivers/crypto/chelsio/chtls/chtls_io.c | 10 +++++----- include/net/sock.h | 4 ++-- net/ipv4/tcp.c | 2 +- net/llc/af_llc.c | 2 +- 4 files changed, 9 insertions(+), 9 deletions(-) (limited to 'drivers/crypto') diff --git a/drivers/crypto/chelsio/chtls/chtls_io.c b/drivers/crypto/chelsio/chtls/chtls_io.c index 98bc5a4cd5e7..599dec59c6cc 100644 --- a/drivers/crypto/chelsio/chtls/chtls_io.c +++ b/drivers/crypto/chelsio/chtls/chtls_io.c @@ -1437,7 +1437,7 @@ static int chtls_pt_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, csk->wr_max_credits)) sk->sk_write_space(sk); - if (copied >= target && !sk->sk_backlog.tail) + if (copied >= target && !READ_ONCE(sk->sk_backlog.tail)) break; if (copied) { @@ -1470,7 +1470,7 @@ static int chtls_pt_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, break; } } - if (sk->sk_backlog.tail) { + if (READ_ONCE(sk->sk_backlog.tail)) { release_sock(sk); lock_sock(sk); chtls_cleanup_rbuf(sk, copied); @@ -1615,7 +1615,7 @@ static int peekmsg(struct sock *sk, struct msghdr *msg, break; } - if (sk->sk_backlog.tail) { + if (READ_ONCE(sk->sk_backlog.tail)) { /* Do not sleep, just process backlog. */ release_sock(sk); lock_sock(sk); @@ -1743,7 +1743,7 @@ int chtls_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, csk->wr_max_credits)) sk->sk_write_space(sk); - if (copied >= target && !sk->sk_backlog.tail) + if (copied >= target && !READ_ONCE(sk->sk_backlog.tail)) break; if (copied) { @@ -1774,7 +1774,7 @@ int chtls_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, } } - if (sk->sk_backlog.tail) { + if (READ_ONCE(sk->sk_backlog.tail)) { release_sock(sk); lock_sock(sk); chtls_cleanup_rbuf(sk, copied); diff --git a/include/net/sock.h b/include/net/sock.h index d4d3ef5ba049..bd210c78dc9d 100644 --- a/include/net/sock.h +++ b/include/net/sock.h @@ -899,11 +899,11 @@ static inline void __sk_add_backlog(struct sock *sk, struct sk_buff *skb) skb_dst_force(skb); if (!sk->sk_backlog.tail) - sk->sk_backlog.head = skb; + WRITE_ONCE(sk->sk_backlog.head, skb); else sk->sk_backlog.tail->next = skb; - sk->sk_backlog.tail = skb; + WRITE_ONCE(sk->sk_backlog.tail, skb); skb->next = NULL; } diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index fb1666440e10..8fb4fefcfd54 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -2047,7 +2047,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int nonblock, /* Well, if we have backlog, try to process it now yet. */ - if (copied >= target && !sk->sk_backlog.tail) + if (copied >= target && !READ_ONCE(sk->sk_backlog.tail)) break; if (copied) { diff --git a/net/llc/af_llc.c b/net/llc/af_llc.c index 50d2c9749db3..2922d4150d88 100644 --- a/net/llc/af_llc.c +++ b/net/llc/af_llc.c @@ -780,7 +780,7 @@ static int llc_ui_recvmsg(struct socket *sock, struct msghdr *msg, size_t len, } /* Well, if we have backlog, try to process it now yet. */ - if (copied >= target && !sk->sk_backlog.tail) + if (copied >= target && !READ_ONCE(sk->sk_backlog.tail)) break; if (copied) { -- cgit v1.2.3-59-g8ed1b From 0ed96b46c0ac26ebcdd2ee95c2479ef8cf94bbd6 Mon Sep 17 00:00:00 2001 From: Rahul Lakkireddy Date: Fri, 22 Nov 2019 06:30:01 +0530 Subject: cxgb4/chcr: update SGL DMA unmap for USO The FW_ETH_TX_EO_WR used for sending UDP Segmentation Offload (USO) requests expects the headers to be part of the descriptor and the payload to be part of the SGL containing the DMA mapped addresses. Hence, the DMA address in the first entry of the SGL can start after the packet headers. Currently, unmap_sgl() tries to unmap from this wrong offset, instead of the originally mapped DMA address. So, use existing unmap_skb() instead, which takes originally saved DMA addresses as input. Update all necessary Tx paths to save the original DMA addresses, so that unmap_skb() can unmap them properly. v2: - No change. Signed-off-by: Rahul Lakkireddy Signed-off-by: Jakub Kicinski --- drivers/crypto/chelsio/chcr_ipsec.c | 27 +++-- drivers/net/ethernet/chelsio/cxgb4/cxgb4.h | 19 ++-- .../net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c | 2 +- drivers/net/ethernet/chelsio/cxgb4/sge.c | 119 ++++++--------------- 4 files changed, 52 insertions(+), 115 deletions(-) (limited to 'drivers/crypto') diff --git a/drivers/crypto/chelsio/chcr_ipsec.c b/drivers/crypto/chelsio/chcr_ipsec.c index 24355680f30a..9da0f93a330b 100644 --- a/drivers/crypto/chelsio/chcr_ipsec.c +++ b/drivers/crypto/chelsio/chcr_ipsec.c @@ -673,16 +673,16 @@ static inline void txq_advance(struct sge_txq *q, unsigned int n) int chcr_ipsec_xmit(struct sk_buff *skb, struct net_device *dev) { struct xfrm_state *x = xfrm_input_state(skb); + unsigned int last_desc, ndesc, flits = 0; struct ipsec_sa_entry *sa_entry; u64 *pos, *end, *before, *sgl; + struct tx_sw_desc *sgl_sdesc; int qidx, left, credits; - unsigned int flits = 0, ndesc; - struct adapter *adap; + bool immediate = false; struct sge_eth_txq *q; + struct adapter *adap; struct port_info *pi; - dma_addr_t addr[MAX_SKB_FRAGS + 1]; struct sec_path *sp; - bool immediate = false; if (!x->xso.offload_handle) return NETDEV_TX_BUSY; @@ -715,8 +715,14 @@ out_free: dev_kfree_skb_any(skb); return NETDEV_TX_BUSY; } + last_desc = q->q.pidx + ndesc - 1; + if (last_desc >= q->q.size) + last_desc -= q->q.size; + sgl_sdesc = &q->q.sdesc[last_desc]; + if (!immediate && - unlikely(cxgb4_map_skb(adap->pdev_dev, skb, addr) < 0)) { + unlikely(cxgb4_map_skb(adap->pdev_dev, skb, sgl_sdesc->addr) < 0)) { + memset(sgl_sdesc->addr, 0, sizeof(sgl_sdesc->addr)); q->mapping_err++; goto out_free; } @@ -742,17 +748,10 @@ out_free: dev_kfree_skb_any(skb); cxgb4_inline_tx_skb(skb, &q->q, sgl); dev_consume_skb_any(skb); } else { - int last_desc; - cxgb4_write_sgl(skb, &q->q, (void *)sgl, end, - 0, addr); + 0, sgl_sdesc->addr); skb_orphan(skb); - - last_desc = q->q.pidx + ndesc - 1; - if (last_desc >= q->q.size) - last_desc -= q->q.size; - q->q.sdesc[last_desc].skb = skb; - q->q.sdesc[last_desc].sgl = (struct ulptx_sgl *)sgl; + sgl_sdesc->skb = skb; } txq_advance(&q->q, ndesc); diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h index 3121ed83d8e2..61a2cf62f694 100644 --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h @@ -735,7 +735,12 @@ struct tx_desc { __be64 flit[8]; }; -struct tx_sw_desc; +struct ulptx_sgl; + +struct tx_sw_desc { + struct sk_buff *skb; /* SKB to free after getting completion */ + dma_addr_t addr[MAX_SKB_FRAGS + 1]; /* DMA mapped addresses */ +}; struct sge_txq { unsigned int in_use; /* # of in-use Tx descriptors */ @@ -814,15 +819,10 @@ enum sge_eosw_state { CXGB4_EO_STATE_FLOWC_CLOSE_REPLY, /* Waiting for FLOWC close reply */ }; -struct sge_eosw_desc { - struct sk_buff *skb; /* SKB to free after getting completion */ - dma_addr_t addr[MAX_SKB_FRAGS + 1]; /* DMA mapped addresses */ -}; - struct sge_eosw_txq { spinlock_t lock; /* Per queue lock to synchronize completions */ enum sge_eosw_state state; /* Current ETHOFLD State */ - struct sge_eosw_desc *desc; /* Descriptor ring to hold packets */ + struct tx_sw_desc *desc; /* Descriptor ring to hold packets */ u32 ndesc; /* Number of descriptors */ u32 pidx; /* Current Producer Index */ u32 last_pidx; /* Last successfully transmitted Producer Index */ @@ -1151,11 +1151,6 @@ enum { SCHED_CLASS_RATEMODE_ABS = 1, /* Kb/s */ }; -struct tx_sw_desc { /* SW state per Tx descriptor */ - struct sk_buff *skb; - struct ulptx_sgl *sgl; -}; - /* Support for "sched_queue" command to allow one or more NIC TX Queues * to be bound to a TX Scheduling Class. */ diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c index db55673b77bd..477973d2e341 100644 --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c @@ -70,7 +70,7 @@ static int cxgb4_init_eosw_txq(struct net_device *dev, u32 eotid, u32 hwqid) { struct adapter *adap = netdev2adap(dev); - struct sge_eosw_desc *ring; + struct tx_sw_desc *ring; memset(eosw_txq, 0, sizeof(*eosw_txq)); diff --git a/drivers/net/ethernet/chelsio/cxgb4/sge.c b/drivers/net/ethernet/chelsio/cxgb4/sge.c index a0400b9a11e9..5ad54493f825 100644 --- a/drivers/net/ethernet/chelsio/cxgb4/sge.c +++ b/drivers/net/ethernet/chelsio/cxgb4/sge.c @@ -300,65 +300,6 @@ static void deferred_unmap_destructor(struct sk_buff *skb) } #endif -static void unmap_sgl(struct device *dev, const struct sk_buff *skb, - const struct ulptx_sgl *sgl, const struct sge_txq *q) -{ - const struct ulptx_sge_pair *p; - unsigned int nfrags = skb_shinfo(skb)->nr_frags; - - if (likely(skb_headlen(skb))) - dma_unmap_single(dev, be64_to_cpu(sgl->addr0), ntohl(sgl->len0), - DMA_TO_DEVICE); - else { - dma_unmap_page(dev, be64_to_cpu(sgl->addr0), ntohl(sgl->len0), - DMA_TO_DEVICE); - nfrags--; - } - - /* - * the complexity below is because of the possibility of a wrap-around - * in the middle of an SGL - */ - for (p = sgl->sge; nfrags >= 2; nfrags -= 2) { - if (likely((u8 *)(p + 1) <= (u8 *)q->stat)) { -unmap: dma_unmap_page(dev, be64_to_cpu(p->addr[0]), - ntohl(p->len[0]), DMA_TO_DEVICE); - dma_unmap_page(dev, be64_to_cpu(p->addr[1]), - ntohl(p->len[1]), DMA_TO_DEVICE); - p++; - } else if ((u8 *)p == (u8 *)q->stat) { - p = (const struct ulptx_sge_pair *)q->desc; - goto unmap; - } else if ((u8 *)p + 8 == (u8 *)q->stat) { - const __be64 *addr = (const __be64 *)q->desc; - - dma_unmap_page(dev, be64_to_cpu(addr[0]), - ntohl(p->len[0]), DMA_TO_DEVICE); - dma_unmap_page(dev, be64_to_cpu(addr[1]), - ntohl(p->len[1]), DMA_TO_DEVICE); - p = (const struct ulptx_sge_pair *)&addr[2]; - } else { - const __be64 *addr = (const __be64 *)q->desc; - - dma_unmap_page(dev, be64_to_cpu(p->addr[0]), - ntohl(p->len[0]), DMA_TO_DEVICE); - dma_unmap_page(dev, be64_to_cpu(addr[0]), - ntohl(p->len[1]), DMA_TO_DEVICE); - p = (const struct ulptx_sge_pair *)&addr[1]; - } - } - if (nfrags) { - __be64 addr; - - if ((u8 *)p == (u8 *)q->stat) - p = (const struct ulptx_sge_pair *)q->desc; - addr = (u8 *)p + 16 <= (u8 *)q->stat ? p->addr[0] : - *(const __be64 *)q->desc; - dma_unmap_page(dev, be64_to_cpu(addr), ntohl(p->len[0]), - DMA_TO_DEVICE); - } -} - /** * free_tx_desc - reclaims Tx descriptors and their buffers * @adapter: the adapter @@ -372,15 +313,16 @@ unmap: dma_unmap_page(dev, be64_to_cpu(p->addr[0]), void free_tx_desc(struct adapter *adap, struct sge_txq *q, unsigned int n, bool unmap) { - struct tx_sw_desc *d; unsigned int cidx = q->cidx; - struct device *dev = adap->pdev_dev; + struct tx_sw_desc *d; d = &q->sdesc[cidx]; while (n--) { if (d->skb) { /* an SGL is present */ - if (unmap) - unmap_sgl(dev, d->skb, d->sgl, q); + if (unmap && d->addr[0]) { + unmap_skb(adap->pdev_dev, d->skb, d->addr); + memset(d->addr, 0, sizeof(d->addr)); + } dev_consume_skb_any(d->skb); d->skb = NULL; } @@ -1414,13 +1356,13 @@ static netdev_tx_t cxgb4_eth_xmit(struct sk_buff *skb, struct net_device *dev) { enum cpl_tx_tnl_lso_type tnl_type = TX_TNL_TYPE_OPAQUE; bool ptp_enabled = is_ptp_enabled(skb, dev); - dma_addr_t addr[MAX_SKB_FRAGS + 1]; + unsigned int last_desc, flits, ndesc; const struct skb_shared_info *ssi; + struct tx_sw_desc *sgl_sdesc; struct fw_eth_tx_pkt_wr *wr; struct cpl_tx_pkt_core *cpl; int len, qidx, credits, ret; const struct port_info *pi; - unsigned int flits, ndesc; bool immediate = false; u32 wr_mid, ctrl0, op; u64 cntrl, *end, *sgl; @@ -1489,8 +1431,14 @@ static netdev_tx_t cxgb4_eth_xmit(struct sk_buff *skb, struct net_device *dev) if (skb->encapsulation && chip_ver > CHELSIO_T5) tnl_type = cxgb_encap_offload_supported(skb); + last_desc = q->q.pidx + ndesc - 1; + if (last_desc >= q->q.size) + last_desc -= q->q.size; + sgl_sdesc = &q->q.sdesc[last_desc]; + if (!immediate && - unlikely(cxgb4_map_skb(adap->pdev_dev, skb, addr) < 0)) { + unlikely(cxgb4_map_skb(adap->pdev_dev, skb, sgl_sdesc->addr) < 0)) { + memset(sgl_sdesc->addr, 0, sizeof(sgl_sdesc->addr)); q->mapping_err++; if (ptp_enabled) spin_unlock(&adap->ptp_lock); @@ -1618,16 +1566,10 @@ static netdev_tx_t cxgb4_eth_xmit(struct sk_buff *skb, struct net_device *dev) cxgb4_inline_tx_skb(skb, &q->q, sgl); dev_consume_skb_any(skb); } else { - int last_desc; - - cxgb4_write_sgl(skb, &q->q, (void *)sgl, end, 0, addr); + cxgb4_write_sgl(skb, &q->q, (void *)sgl, end, 0, + sgl_sdesc->addr); skb_orphan(skb); - - last_desc = q->q.pidx + ndesc - 1; - if (last_desc >= q->q.size) - last_desc -= q->q.size; - q->q.sdesc[last_desc].skb = skb; - q->q.sdesc[last_desc].sgl = (struct ulptx_sgl *)sgl; + sgl_sdesc->skb = skb; } txq_advance(&q->q, ndesc); @@ -1725,12 +1667,12 @@ static inline unsigned int t4vf_calc_tx_flits(const struct sk_buff *skb) static netdev_tx_t cxgb4_vf_eth_xmit(struct sk_buff *skb, struct net_device *dev) { - dma_addr_t addr[MAX_SKB_FRAGS + 1]; + unsigned int last_desc, flits, ndesc; const struct skb_shared_info *ssi; struct fw_eth_tx_pkt_vm_wr *wr; + struct tx_sw_desc *sgl_sdesc; struct cpl_tx_pkt_core *cpl; const struct port_info *pi; - unsigned int flits, ndesc; struct sge_eth_txq *txq; struct adapter *adapter; int qidx, credits, ret; @@ -1782,12 +1724,19 @@ static netdev_tx_t cxgb4_vf_eth_xmit(struct sk_buff *skb, return NETDEV_TX_BUSY; } + last_desc = txq->q.pidx + ndesc - 1; + if (last_desc >= txq->q.size) + last_desc -= txq->q.size; + sgl_sdesc = &txq->q.sdesc[last_desc]; + if (!t4vf_is_eth_imm(skb) && - unlikely(cxgb4_map_skb(adapter->pdev_dev, skb, addr) < 0)) { + unlikely(cxgb4_map_skb(adapter->pdev_dev, skb, + sgl_sdesc->addr) < 0)) { /* We need to map the skb into PCI DMA space (because it can't * be in-lined directly into the Work Request) and the mapping * operation failed. Record the error and drop the packet. */ + memset(sgl_sdesc->addr, 0, sizeof(sgl_sdesc->addr)); txq->mapping_err++; goto out_free; } @@ -1962,7 +1911,6 @@ static netdev_tx_t cxgb4_vf_eth_xmit(struct sk_buff *skb, */ struct ulptx_sgl *sgl = (struct ulptx_sgl *)(cpl + 1); struct sge_txq *tq = &txq->q; - int last_desc; /* If the Work Request header was an exact multiple of our TX * Descriptor length, then it's possible that the starting SGL @@ -1976,14 +1924,9 @@ static netdev_tx_t cxgb4_vf_eth_xmit(struct sk_buff *skb, ((void *)end - (void *)tq->stat)); } - cxgb4_write_sgl(skb, tq, sgl, end, 0, addr); + cxgb4_write_sgl(skb, tq, sgl, end, 0, sgl_sdesc->addr); skb_orphan(skb); - - last_desc = tq->pidx + ndesc - 1; - if (last_desc >= tq->size) - last_desc -= tq->size; - tq->sdesc[last_desc].skb = skb; - tq->sdesc[last_desc].sgl = sgl; + sgl_sdesc->skb = skb; } /* Advance our internal TX Queue state, tell the hardware about @@ -2035,7 +1978,7 @@ static inline void eosw_txq_advance_index(u32 *idx, u32 n, u32 max) void cxgb4_eosw_txq_free_desc(struct adapter *adap, struct sge_eosw_txq *eosw_txq, u32 ndesc) { - struct sge_eosw_desc *d; + struct tx_sw_desc *d; d = &eosw_txq->desc[eosw_txq->last_cidx]; while (ndesc--) { @@ -2167,7 +2110,7 @@ static void ethofld_hard_xmit(struct net_device *dev, struct cpl_tx_pkt_core *cpl; struct fw_eth_tx_eo_wr *wr; bool skip_eotx_wr = false; - struct sge_eosw_desc *d; + struct tx_sw_desc *d; struct sk_buff *skb; u8 flits, ndesc; int left; -- cgit v1.2.3-59-g8ed1b