From 3f1baa91a1fdf3de9dbad4bd615b35fab347874b Mon Sep 17 00:00:00 2001 From: Sankararaman Jayaraman Date: Fri, 31 Jan 2025 09:53:41 +0530 Subject: [PATCH 01/16] vmxnet3: Fix tx queue race condition with XDP MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit If XDP traffic runs on a CPU which is greater than or equal to the number of the Tx queues of the NIC, then vmxnet3_xdp_get_tq() always picks up queue 0 for transmission as it uses reciprocal scale instead of simple modulo operation. vmxnet3_xdp_xmit() and vmxnet3_xdp_xmit_frame() use the above returned queue without any locking which can lead to race conditions when multiple XDP xmits run in parallel on different CPU's. This patch uses a simple module scheme when the current CPU equals or exceeds the number of Tx queues on the NIC. It also adds locking in vmxnet3_xdp_xmit() and vmxnet3_xdp_xmit_frame() functions. Fixes: 54f00cce1178 ("vmxnet3: Add XDP support.") Signed-off-by: Sankararaman Jayaraman Signed-off-by: Ronak Doshi Reviewed-by: Simon Horman Link: https://patch.msgid.link/20250131042340.156547-1-sankararaman.jayaraman@broadcom.com Signed-off-by: Jakub Kicinski --- drivers/net/vmxnet3/vmxnet3_xdp.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) diff --git a/drivers/net/vmxnet3/vmxnet3_xdp.c b/drivers/net/vmxnet3/vmxnet3_xdp.c index 1341374a4588..616ecc38d172 100644 --- a/drivers/net/vmxnet3/vmxnet3_xdp.c +++ b/drivers/net/vmxnet3/vmxnet3_xdp.c @@ -28,7 +28,7 @@ vmxnet3_xdp_get_tq(struct vmxnet3_adapter *adapter) if (likely(cpu < tq_number)) tq = &adapter->tx_queue[cpu]; else - tq = &adapter->tx_queue[reciprocal_scale(cpu, tq_number)]; + tq = &adapter->tx_queue[cpu % tq_number]; return tq; } @@ -124,6 +124,7 @@ vmxnet3_xdp_xmit_frame(struct vmxnet3_adapter *adapter, u32 buf_size; u32 dw2; + spin_lock_irq(&tq->tx_lock); dw2 = (tq->tx_ring.gen ^ 0x1) << VMXNET3_TXD_GEN_SHIFT; dw2 |= xdpf->len; ctx.sop_txd = tq->tx_ring.base + tq->tx_ring.next2fill; @@ -134,6 +135,7 @@ vmxnet3_xdp_xmit_frame(struct vmxnet3_adapter *adapter, if (vmxnet3_cmd_ring_desc_avail(&tq->tx_ring) == 0) { tq->stats.tx_ring_full++; + spin_unlock_irq(&tq->tx_lock); return -ENOSPC; } @@ -142,8 +144,10 @@ vmxnet3_xdp_xmit_frame(struct vmxnet3_adapter *adapter, tbi->dma_addr = dma_map_single(&adapter->pdev->dev, xdpf->data, buf_size, DMA_TO_DEVICE); - if (dma_mapping_error(&adapter->pdev->dev, tbi->dma_addr)) + if (dma_mapping_error(&adapter->pdev->dev, tbi->dma_addr)) { + spin_unlock_irq(&tq->tx_lock); return -EFAULT; + } tbi->map_type |= VMXNET3_MAP_SINGLE; } else { /* XDP buffer from page pool */ page = virt_to_page(xdpf->data); @@ -182,6 +186,7 @@ vmxnet3_xdp_xmit_frame(struct vmxnet3_adapter *adapter, dma_wmb(); gdesc->dword[2] = cpu_to_le32(le32_to_cpu(gdesc->dword[2]) ^ VMXNET3_TXD_GEN); + spin_unlock_irq(&tq->tx_lock); /* No need to handle the case when tx_num_deferred doesn't reach * threshold. Backend driver at hypervisor side will poll and reset @@ -225,6 +230,7 @@ vmxnet3_xdp_xmit(struct net_device *dev, { struct vmxnet3_adapter *adapter = netdev_priv(dev); struct vmxnet3_tx_queue *tq; + struct netdev_queue *nq; int i; if (unlikely(test_bit(VMXNET3_STATE_BIT_QUIESCED, &adapter->state))) @@ -236,6 +242,9 @@ vmxnet3_xdp_xmit(struct net_device *dev, if (tq->stopped) return -ENETDOWN; + nq = netdev_get_tx_queue(adapter->netdev, tq->qid); + + __netif_tx_lock(nq, smp_processor_id()); for (i = 0; i < n; i++) { if (vmxnet3_xdp_xmit_frame(adapter, frames[i], tq, true)) { tq->stats.xdp_xmit_err++; @@ -243,6 +252,7 @@ vmxnet3_xdp_xmit(struct net_device *dev, } } tq->stats.xdp_xmit += i; + __netif_tx_unlock(nq); return i; } -- 2.51.0 From e0efe83ed325277bb70f9435d4d9fc70bebdcca8 Mon Sep 17 00:00:00 2001 From: Lenny Szubowicz Date: Thu, 30 Jan 2025 16:57:54 -0500 Subject: [PATCH 02/16] tg3: Disable tg3 PCIe AER on system reboot Disable PCIe AER on the tg3 device on system reboot on a limited list of Dell PowerEdge systems. This prevents a fatal PCIe AER event on the tg3 device during the ACPI _PTS (prepare to sleep) method for S5 on those systems. The _PTS is invoked by acpi_enter_sleep_state_prep() as part of the kernel's reboot sequence as a result of commit 38f34dba806a ("PM: ACPI: reboot: Reinstate S5 for reboot"). There was an earlier fix for this problem by commit 2ca1c94ce0b6 ("tg3: Disable tg3 device on system reboot to avoid triggering AER"). But it was discovered that this earlier fix caused a reboot hang when some Dell PowerEdge servers were booted via ipxe. To address this reboot hang, the earlier fix was essentially reverted by commit 9fc3bc764334 ("tg3: power down device only on SYSTEM_POWER_OFF"). This re-exposed the tg3 PCIe AER on reboot problem. This fix is not an ideal solution because the root cause of the AER is in system firmware. Instead, it's a targeted work-around in the tg3 driver. Note also that the PCIe AER must be disabled on the tg3 device even if the system is configured to use "firmware first" error handling. V3: - Fix sparse warning on improper comparison of pdev->current_state - Adhere to netdev comment style Fixes: 9fc3bc764334 ("tg3: power down device only on SYSTEM_POWER_OFF") Signed-off-by: Lenny Szubowicz Reviewed-by: Pavan Chebbi Reviewed-by: Simon Horman Signed-off-by: David S. Miller --- drivers/net/ethernet/broadcom/tg3.c | 58 +++++++++++++++++++++++++++++ 1 file changed, 58 insertions(+) diff --git a/drivers/net/ethernet/broadcom/tg3.c b/drivers/net/ethernet/broadcom/tg3.c index 1c94bf1db718..d9d675f1ebfe 100644 --- a/drivers/net/ethernet/broadcom/tg3.c +++ b/drivers/net/ethernet/broadcom/tg3.c @@ -55,6 +55,7 @@ #include #include #include +#include #include #include @@ -18212,6 +18213,50 @@ unlock: static SIMPLE_DEV_PM_OPS(tg3_pm_ops, tg3_suspend, tg3_resume); +/* Systems where ACPI _PTS (Prepare To Sleep) S5 will result in a fatal + * PCIe AER event on the tg3 device if the tg3 device is not, or cannot + * be, powered down. + */ +static const struct dmi_system_id tg3_restart_aer_quirk_table[] = { + { + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R440"), + }, + }, + { + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R540"), + }, + }, + { + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R640"), + }, + }, + { + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R650"), + }, + }, + { + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R740"), + }, + }, + { + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R750"), + }, + }, + {} +}; + static void tg3_shutdown(struct pci_dev *pdev) { struct net_device *dev = pci_get_drvdata(pdev); @@ -18228,6 +18273,19 @@ static void tg3_shutdown(struct pci_dev *pdev) if (system_state == SYSTEM_POWER_OFF) tg3_power_down(tp); + else if (system_state == SYSTEM_RESTART && + dmi_first_match(tg3_restart_aer_quirk_table) && + pdev->current_state != PCI_D3cold && + pdev->current_state != PCI_UNKNOWN) { + /* Disable PCIe AER on the tg3 to avoid a fatal + * error during this system restart. + */ + pcie_capability_clear_word(pdev, PCI_EXP_DEVCTL, + PCI_EXP_DEVCTL_CERE | + PCI_EXP_DEVCTL_NFERE | + PCI_EXP_DEVCTL_FERE | + PCI_EXP_DEVCTL_URRE); + } rtnl_unlock(); -- 2.51.0 From 235174b2bed88501fda689c113c55737f99332d8 Mon Sep 17 00:00:00 2001 From: Yan Zhai Date: Fri, 31 Jan 2025 00:31:39 -0800 Subject: [PATCH 03/16] udp: gso: do not drop small packets when PMTU reduces Commit 4094871db1d6 ("udp: only do GSO if # of segs > 1") avoided GSO for small packets. But the kernel currently dismisses GSO requests only after checking MTU/PMTU on gso_size. This means any packets, regardless of their payload sizes, could be dropped when PMTU becomes smaller than requested gso_size. We encountered this issue in production and it caused a reliability problem that new QUIC connection cannot be established before PMTU cache expired, while non GSO sockets still worked fine at the same time. Ideally, do not check any GSO related constraints when payload size is smaller than requested gso_size, and return EMSGSIZE instead of EINVAL on MTU/PMTU check failure to be more specific on the error cause. Fixes: 4094871db1d6 ("udp: only do GSO if # of segs > 1") Signed-off-by: Yan Zhai Suggested-by: Willem de Bruijn Reviewed-by: Willem de Bruijn Signed-off-by: David S. Miller --- net/ipv4/udp.c | 4 ++-- net/ipv6/udp.c | 4 ++-- tools/testing/selftests/net/udpgso.c | 26 ++++++++++++++++++++++++++ 3 files changed, 30 insertions(+), 4 deletions(-) diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index c472c9a57cf6..a9bb9ce5438e 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -1141,9 +1141,9 @@ static int udp_send_skb(struct sk_buff *skb, struct flowi4 *fl4, const int hlen = skb_network_header_len(skb) + sizeof(struct udphdr); - if (hlen + cork->gso_size > cork->fragsize) { + if (hlen + min(datalen, cork->gso_size) > cork->fragsize) { kfree_skb(skb); - return -EINVAL; + return -EMSGSIZE; } if (datalen > cork->gso_size * UDP_MAX_SEGMENTS) { kfree_skb(skb); diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index 6671daa67f4f..c6ea438b5c75 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -1389,9 +1389,9 @@ static int udp_v6_send_skb(struct sk_buff *skb, struct flowi6 *fl6, const int hlen = skb_network_header_len(skb) + sizeof(struct udphdr); - if (hlen + cork->gso_size > cork->fragsize) { + if (hlen + min(datalen, cork->gso_size) > cork->fragsize) { kfree_skb(skb); - return -EINVAL; + return -EMSGSIZE; } if (datalen > cork->gso_size * UDP_MAX_SEGMENTS) { kfree_skb(skb); diff --git a/tools/testing/selftests/net/udpgso.c b/tools/testing/selftests/net/udpgso.c index 3f2fca02fec5..36ff28af4b19 100644 --- a/tools/testing/selftests/net/udpgso.c +++ b/tools/testing/selftests/net/udpgso.c @@ -102,6 +102,19 @@ struct testcase testcases_v4[] = { .gso_len = CONST_MSS_V4, .r_num_mss = 1, }, + { + /* datalen <= MSS < gso_len: will fall back to no GSO */ + .tlen = CONST_MSS_V4, + .gso_len = CONST_MSS_V4 + 1, + .r_num_mss = 0, + .r_len_last = CONST_MSS_V4, + }, + { + /* MSS < datalen < gso_len: fail */ + .tlen = CONST_MSS_V4 + 1, + .gso_len = CONST_MSS_V4 + 2, + .tfail = true, + }, { /* send a single MSS + 1B */ .tlen = CONST_MSS_V4 + 1, @@ -205,6 +218,19 @@ struct testcase testcases_v6[] = { .gso_len = CONST_MSS_V6, .r_num_mss = 1, }, + { + /* datalen <= MSS < gso_len: will fall back to no GSO */ + .tlen = CONST_MSS_V6, + .gso_len = CONST_MSS_V6 + 1, + .r_num_mss = 0, + .r_len_last = CONST_MSS_V6, + }, + { + /* MSS < datalen < gso_len: fail */ + .tlen = CONST_MSS_V6 + 1, + .gso_len = CONST_MSS_V6 + 2, + .tfail = true + }, { /* send a single MSS + 1B */ .tlen = CONST_MSS_V6 + 1, -- 2.51.0 From 3a4e7193ec37ee2476ce726589de4495a066b565 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Sat, 1 Feb 2025 16:50:24 -0800 Subject: [PATCH 04/16] MAINTAINERS: list openvswitch docs under its entry Submissions to the docs seem to not get properly CCed. Acked-by: Ilya Maximets Link: https://patch.msgid.link/20250202005024.964262-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index d1086e53a317..c7b8c6535a1e 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -17706,6 +17706,7 @@ L: netdev@vger.kernel.org L: dev@openvswitch.org S: Maintained W: http://openvswitch.org +F: Documentation/networking/openvswitch.rst F: include/uapi/linux/openvswitch.h F: net/openvswitch/ F: tools/testing/selftests/net/openvswitch/ -- 2.51.0 From 4d896b35394144c246daaeb5280a015a630958e7 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Sat, 1 Feb 2025 17:47:26 -0800 Subject: [PATCH 05/16] MAINTAINERS: add Kuniyuki Iwashima to TCP reviewers List Kuniyuki as an official TCP reviewer. Reviewed-by: Eric Dumazet Reviewed-by: Kuniyuki Iwashima Link: https://patch.msgid.link/20250202014728.1005003-2-kuba@kernel.org Signed-off-by: Jakub Kicinski --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index c7b8c6535a1e..48677d61c97b 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -16614,6 +16614,7 @@ F: tools/testing/selftests/net/mptcp/ NETWORKING [TCP] M: Eric Dumazet M: Neal Cardwell +R: Kuniyuki Iwashima L: netdev@vger.kernel.org S: Maintained F: Documentation/networking/net_cachelines/tcp_sock.rst -- 2.51.0 From ae0585b04ab741b536b0db20c12baf24bf7118d2 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Sat, 1 Feb 2025 17:47:27 -0800 Subject: [PATCH 06/16] MAINTAINERS: add a general entry for BSD sockets Create a MAINTAINERS entry for BSD sockets. List the top 3 reviewers as maintainers. The entry is meant to cover core socket code (of which there isn't much) but also reviews of any new socket families. Reviewed-by: Simon Horman Acked-by: Willem de Bruijn Reviewed-by: Kuniyuki Iwashima Link: https://patch.msgid.link/20250202014728.1005003-3-kuba@kernel.org Signed-off-by: Jakub Kicinski --- MAINTAINERS | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/MAINTAINERS b/MAINTAINERS index 48677d61c97b..438d85bb97a2 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -16642,6 +16642,22 @@ F: include/net/tls.h F: include/uapi/linux/tls.h F: net/tls/* +NETWORKING [SOCKETS] +M: Eric Dumazet +M: Kuniyuki Iwashima +M: Paolo Abeni +M: Willem de Bruijn +S: Maintained +F: include/linux/sock_diag.h +F: include/linux/socket.h +F: include/linux/sockptr.h +F: include/net/sock.h +F: include/net/sock_reuseport.h +F: include/uapi/linux/socket.h +F: net/core/*sock* +F: net/core/scm.c +F: net/socket.c + NETXEN (1/10) GbE SUPPORT M: Manish Chopra M: Rahul Verma -- 2.51.0 From 8a2e22f665a0b5c212057031e94b75cfdc11a4a6 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Sat, 1 Feb 2025 17:47:28 -0800 Subject: [PATCH 07/16] MAINTAINERS: add entry for UNIX sockets Add a MAINTAINERS entry for UNIX socket, Kuniyuki has been the de-facto maintainer of this code for a while. Reviewed-by: Simon Horman Reviewed-by: Kuniyuki Iwashima Link: https://patch.msgid.link/20250202014728.1005003-4-kuba@kernel.org Signed-off-by: Jakub Kicinski --- MAINTAINERS | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/MAINTAINERS b/MAINTAINERS index 438d85bb97a2..74b09dad4662 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -16658,6 +16658,15 @@ F: net/core/*sock* F: net/core/scm.c F: net/socket.c +NETWORKING [UNIX SOCKETS] +M: Kuniyuki Iwashima +S: Maintained +F: include/net/af_unix.h +F: include/net/netns/unix.h +F: include/uapi/linux/unix_diag.h +F: net/unix/ +F: tools/testing/selftests/net/af_unix/ + NETXEN (1/10) GbE SUPPORT M: Manish Chopra M: Rahul Verma -- 2.51.0 From 244f8aa46fa9e2f4ea5fe0e04988b395d5e30fc7 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Fri, 31 Jan 2025 17:30:37 -0800 Subject: [PATCH 08/16] ethtool: rss: fix hiding unsupported fields in dumps Commit ec6e57beaf8b ("ethtool: rss: don't report key if device doesn't support it") intended to stop reporting key fields for additional rss contexts if device has a global hashing key. Later we added dump support and the filtering wasn't properly added there. So we end up reporting the key fields in dumps but not in dos: # ./pyynl/cli.py --spec netlink/specs/ethtool.yaml --do rss-get \ --json '{"header": {"dev-index":2}, "context": 1 }' { "header": { ... }, "context": 1, "indir": [0, 1, 2, 3, ...]] } # ./pyynl/cli.py --spec netlink/specs/ethtool.yaml --dump rss-get [ ... snip context 0 ... { "header": { ... }, "context": 1, "indir": [0, 1, 2, 3, ...], -> "input_xfrm": 255, -> "hfunc": 1, -> "hkey": "000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000" } ] Hide these fields correctly. The drivers/net/hw/rss_ctx.py selftest catches this when run on a device with single key, already: # Check| At /root/./ksft-net-drv/drivers/net/hw/rss_ctx.py, line 381, in test_rss_context_dump: # Check| ksft_ne(set(data.get('hkey', [1])), {0}, "key is all zero") # Check failed {0} == {0} key is all zero not ok 8 rss_ctx.test_rss_context_dump Fixes: f6122900f4e2 ("ethtool: rss: support dumping RSS contexts") Reviewed-by: Gal Pressman Reviewed-by: Joe Damato Link: https://patch.msgid.link/20250201013040.725123-2-kuba@kernel.org Signed-off-by: Jakub Kicinski --- net/ethtool/rss.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/net/ethtool/rss.c b/net/ethtool/rss.c index 7cb106b590ab..58df9ad02ce8 100644 --- a/net/ethtool/rss.c +++ b/net/ethtool/rss.c @@ -107,6 +107,8 @@ rss_prepare_ctx(const struct rss_req_info *request, struct net_device *dev, u32 total_size, indir_bytes; u8 *rss_config; + data->no_key_fields = !dev->ethtool_ops->rxfh_per_ctx_key; + ctx = xa_load(&dev->ethtool->rss_ctx, request->rss_context); if (!ctx) return -ENOENT; @@ -153,7 +155,6 @@ rss_prepare_data(const struct ethnl_req_info *req_base, if (!ops->cap_rss_ctx_supported && !ops->create_rxfh_context) return -EOPNOTSUPP; - data->no_key_fields = !ops->rxfh_per_ctx_key; return rss_prepare_ctx(request, dev, data, info); } -- 2.51.0 From 2b91cc1214b165c25ac9b0885db89a0d3224028a Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Fri, 31 Jan 2025 17:30:38 -0800 Subject: [PATCH 09/16] ethtool: ntuple: fix rss + ring_cookie check The info.flow_type is for RXFH commands, ntuple flow_type is inside the flow spec. The check currently does nothing, as info.flow_type is 0 (or even uninitialized by user space) for ETHTOOL_SRXCLSRLINS. Fixes: 9e43ad7a1ede ("net: ethtool: only allow set_rxnfc with rss + ring_cookie if driver opts in") Reviewed-by: Gal Pressman Reviewed-by: Joe Damato Link: https://patch.msgid.link/20250201013040.725123-3-kuba@kernel.org Signed-off-by: Jakub Kicinski --- net/ethtool/ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c index 34bee42e1247..7609ce2b2c5e 100644 --- a/net/ethtool/ioctl.c +++ b/net/ethtool/ioctl.c @@ -993,7 +993,7 @@ static noinline_for_stack int ethtool_set_rxnfc(struct net_device *dev, return rc; /* Nonzero ring with RSS only makes sense if NIC adds them together */ - if (cmd == ETHTOOL_SRXCLSRLINS && info.flow_type & FLOW_RSS && + if (cmd == ETHTOOL_SRXCLSRLINS && info.fs.flow_type & FLOW_RSS && !ops->cap_rss_rxnfc_adds && ethtool_get_flow_spec_ring(info.fs.ring_cookie)) return -EINVAL; -- 2.51.0 From de379dfd9ada2995699052f4a1ecebe5d8f8d70f Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Fri, 31 Jan 2025 17:30:39 -0800 Subject: [PATCH 10/16] selftests: drv-net: rss_ctx: add missing cleanup in queue reconfigure Commit under Fixes adds ntuple rules but never deletes them. Fixes: 29a4bc1fe961 ("selftest: extend test_rss_context_queue_reconfigure for action addition") Reviewed-by: Joe Damato Link: https://patch.msgid.link/20250201013040.725123-4-kuba@kernel.org Signed-off-by: Jakub Kicinski --- tools/testing/selftests/drivers/net/hw/rss_ctx.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tools/testing/selftests/drivers/net/hw/rss_ctx.py b/tools/testing/selftests/drivers/net/hw/rss_ctx.py index ca8a7edff3dd..27e24e20749f 100755 --- a/tools/testing/selftests/drivers/net/hw/rss_ctx.py +++ b/tools/testing/selftests/drivers/net/hw/rss_ctx.py @@ -252,6 +252,7 @@ def test_rss_queue_reconfigure(cfg, main_ctx=True): try: # this targets queue 4, which doesn't exist ntuple2 = ethtool_create(cfg, "-N", flow) + defer(ethtool, f"-N {cfg.ifname} delete {ntuple2}") except CmdExitFailure: pass else: @@ -260,6 +261,7 @@ def test_rss_queue_reconfigure(cfg, main_ctx=True): ethtool(f"-X {cfg.ifname} {ctx_ref} weight 1 0 1 0") # ntuple rule therefore targets queues 1 and 3 ntuple2 = ethtool_create(cfg, "-N", flow) + defer(ethtool, f"-N {cfg.ifname} delete {ntuple2}") # should replace existing filter ksft_eq(ntuple, ntuple2) _send_traffic_check(cfg, port, ctx_ref, { 'target': (1, 3), -- 2.51.0 From c3da585509aeb8476886adf75a266c81a9b0df6c Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Fri, 31 Jan 2025 17:30:40 -0800 Subject: [PATCH 11/16] selftests: drv-net: rss_ctx: don't fail reconfigure test if queue offset not supported Vast majority of drivers does not support queue offset. Simply return if the rss context + queue ntuple fails. Reviewed-by: Joe Damato Link: https://patch.msgid.link/20250201013040.725123-5-kuba@kernel.org Signed-off-by: Jakub Kicinski --- tools/testing/selftests/drivers/net/hw/rss_ctx.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/tools/testing/selftests/drivers/net/hw/rss_ctx.py b/tools/testing/selftests/drivers/net/hw/rss_ctx.py index 27e24e20749f..319aaa004c40 100755 --- a/tools/testing/selftests/drivers/net/hw/rss_ctx.py +++ b/tools/testing/selftests/drivers/net/hw/rss_ctx.py @@ -260,7 +260,12 @@ def test_rss_queue_reconfigure(cfg, main_ctx=True): # change the table to target queues 0 and 2 ethtool(f"-X {cfg.ifname} {ctx_ref} weight 1 0 1 0") # ntuple rule therefore targets queues 1 and 3 - ntuple2 = ethtool_create(cfg, "-N", flow) + try: + ntuple2 = ethtool_create(cfg, "-N", flow) + except CmdExitFailure: + ksft_pr("Driver does not support rss + queue offset") + return + defer(ethtool, f"-N {cfg.ifname} delete {ntuple2}") # should replace existing filter ksft_eq(ntuple, ntuple2) -- 2.51.0 From d3ed6dee73c560fad0a8e152c8e233b3fb3a2e44 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Sat, 1 Feb 2025 19:02:51 +0100 Subject: [PATCH 12/16] net: harmonize tstats and dstats After the blamed commits below, some UDP tunnel use dstats for accounting. On the xmit path, all the UDP-base tunnels ends up using iptunnel_xmit_stats() for stats accounting, and the latter assumes the relevant (tunnel) network device uses tstats. The end result is some 'funny' stat report for the mentioned UDP tunnel, e.g. when no packet is actually dropped and a bunch of packets are transmitted: gnv2: mtu 1450 qdisc noqueue \ state UNKNOWN mode DEFAULT group default qlen 1000 link/ether ee:7d:09:87:90:ea brd ff:ff:ff:ff:ff:ff RX: bytes packets errors dropped missed mcast 14916 23 0 15 0 0 TX: bytes packets errors dropped carrier collsns 0 1566 0 0 0 0 Address the issue ensuring the same binary layout for the overlapping fields of dstats and tstats. While this solution is a bit hackish, is smaller and with no performance pitfall compared to other alternatives i.e. supporting both dstat and tstat in iptunnel_xmit_stats() or reverting the blamed commit. With time we should possibly move all the IP-based tunnel (and virtual devices) to dstats. Fixes: c77200c07491 ("bareudp: Handle stats using NETDEV_PCPU_STAT_DSTATS.") Fixes: 6fa6de302246 ("geneve: Handle stats using NETDEV_PCPU_STAT_DSTATS.") Fixes: be226352e8dc ("vxlan: Handle stats using NETDEV_PCPU_STAT_DSTATS.") Signed-off-by: Paolo Abeni Reviewed-by: Guillaume Nault Link: https://patch.msgid.link/2e1c444cf0f63ae472baff29862c4c869be17031.1738432804.git.pabeni@redhat.com Signed-off-by: Jakub Kicinski --- include/linux/netdevice.h | 2 +- net/core/dev.c | 14 ++++++++++++++ 2 files changed, 15 insertions(+), 1 deletion(-) diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 2a59034a5fa2..03bb584c62cf 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -2904,9 +2904,9 @@ struct pcpu_sw_netstats { struct pcpu_dstats { u64_stats_t rx_packets; u64_stats_t rx_bytes; - u64_stats_t rx_drops; u64_stats_t tx_packets; u64_stats_t tx_bytes; + u64_stats_t rx_drops; u64_stats_t tx_drops; struct u64_stats_sync syncp; } __aligned(8 * sizeof(u64)); diff --git a/net/core/dev.c b/net/core/dev.c index c0021cbd28fc..b91658e8aedb 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -11286,6 +11286,20 @@ struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev, const struct net_device_ops *ops = dev->netdev_ops; const struct net_device_core_stats __percpu *p; + /* + * IPv{4,6} and udp tunnels share common stat helpers and use + * different stat type (NETDEV_PCPU_STAT_TSTATS vs + * NETDEV_PCPU_STAT_DSTATS). Ensure the accounting is consistent. + */ + BUILD_BUG_ON(offsetof(struct pcpu_sw_netstats, rx_bytes) != + offsetof(struct pcpu_dstats, rx_bytes)); + BUILD_BUG_ON(offsetof(struct pcpu_sw_netstats, rx_packets) != + offsetof(struct pcpu_dstats, rx_packets)); + BUILD_BUG_ON(offsetof(struct pcpu_sw_netstats, tx_bytes) != + offsetof(struct pcpu_dstats, tx_bytes)); + BUILD_BUG_ON(offsetof(struct pcpu_sw_netstats, tx_packets) != + offsetof(struct pcpu_dstats, tx_packets)); + if (ops->ndo_get_stats64) { memset(storage, 0, sizeof(*storage)); ops->ndo_get_stats64(dev, storage); -- 2.51.0 From 4241a702e0d0c2ca9364cfac08dbf134264962de Mon Sep 17 00:00:00 2001 From: David Howells Date: Mon, 3 Feb 2025 11:03:04 +0000 Subject: [PATCH 13/16] rxrpc: Fix the rxrpc_connection attend queue handling The rxrpc_connection attend queue is never used because conn::attend_link is never initialised and so is always NULL'd out and thus always appears to be busy. This requires the following fix: (1) Fix this the attend queue problem by initialising conn::attend_link. And, consequently, two further fixes for things masked by the above bug: (2) Fix rxrpc_input_conn_event() to handle being invoked with a NULL sk_buff pointer - something that can now happen with the above change. (3) Fix the RXRPC_SKB_MARK_SERVICE_CONN_SECURED message to carry a pointer to the connection and a ref on it. Signed-off-by: David Howells cc: Marc Dionne cc: Jakub Kicinski cc: "David S. Miller" cc: Eric Dumazet cc: Paolo Abeni cc: Simon Horman cc: linux-afs@lists.infradead.org cc: netdev@vger.kernel.org Fixes: f2cce89a074e ("rxrpc: Implement a mechanism to send an event notification to a connection") Link: https://patch.msgid.link/20250203110307.7265-3-dhowells@redhat.com Signed-off-by: Paolo Abeni --- include/trace/events/rxrpc.h | 1 + net/rxrpc/conn_event.c | 17 ++++++++++------- net/rxrpc/conn_object.c | 1 + 3 files changed, 12 insertions(+), 7 deletions(-) diff --git a/include/trace/events/rxrpc.h b/include/trace/events/rxrpc.h index 2f119d18a061..cad50d91077e 100644 --- a/include/trace/events/rxrpc.h +++ b/include/trace/events/rxrpc.h @@ -219,6 +219,7 @@ EM(rxrpc_conn_get_conn_input, "GET inp-conn") \ EM(rxrpc_conn_get_idle, "GET idle ") \ EM(rxrpc_conn_get_poke_abort, "GET pk-abort") \ + EM(rxrpc_conn_get_poke_secured, "GET secured ") \ EM(rxrpc_conn_get_poke_timer, "GET poke ") \ EM(rxrpc_conn_get_service_conn, "GET svc-conn") \ EM(rxrpc_conn_new_client, "NEW client ") \ diff --git a/net/rxrpc/conn_event.c b/net/rxrpc/conn_event.c index 713e04394ceb..74bb49b936cd 100644 --- a/net/rxrpc/conn_event.c +++ b/net/rxrpc/conn_event.c @@ -272,6 +272,7 @@ static int rxrpc_process_event(struct rxrpc_connection *conn, * we've already received the packet, put it on the * front of the queue. */ + sp->conn = rxrpc_get_connection(conn, rxrpc_conn_get_poke_secured); skb->mark = RXRPC_SKB_MARK_SERVICE_CONN_SECURED; rxrpc_get_skb(skb, rxrpc_skb_get_conn_secured); skb_queue_head(&conn->local->rx_queue, skb); @@ -437,14 +438,16 @@ void rxrpc_input_conn_event(struct rxrpc_connection *conn, struct sk_buff *skb) if (test_and_clear_bit(RXRPC_CONN_EV_ABORT_CALLS, &conn->events)) rxrpc_abort_calls(conn); - switch (skb->mark) { - case RXRPC_SKB_MARK_SERVICE_CONN_SECURED: - if (conn->state != RXRPC_CONN_SERVICE) - break; + if (skb) { + switch (skb->mark) { + case RXRPC_SKB_MARK_SERVICE_CONN_SECURED: + if (conn->state != RXRPC_CONN_SERVICE) + break; - for (loop = 0; loop < RXRPC_MAXCALLS; loop++) - rxrpc_call_is_secure(conn->channels[loop].call); - break; + for (loop = 0; loop < RXRPC_MAXCALLS; loop++) + rxrpc_call_is_secure(conn->channels[loop].call); + break; + } } /* Process delayed ACKs whose time has come. */ diff --git a/net/rxrpc/conn_object.c b/net/rxrpc/conn_object.c index 7eba4d7d9a38..2f1fd1e2e7e4 100644 --- a/net/rxrpc/conn_object.c +++ b/net/rxrpc/conn_object.c @@ -67,6 +67,7 @@ struct rxrpc_connection *rxrpc_alloc_connection(struct rxrpc_net *rxnet, INIT_WORK(&conn->destructor, rxrpc_clean_up_connection); INIT_LIST_HEAD(&conn->proc_link); INIT_LIST_HEAD(&conn->link); + INIT_LIST_HEAD(&conn->attend_link); mutex_init(&conn->security_lock); mutex_init(&conn->tx_data_alloc_lock); skb_queue_head_init(&conn->rx_queue); -- 2.51.0 From 028676bb189ed6d1b550a0fc570a9d695b6acfd3 Mon Sep 17 00:00:00 2001 From: Jacob Moroni Date: Mon, 3 Feb 2025 09:36:05 -0500 Subject: [PATCH 14/16] net: atlantic: fix warning during hot unplug Firmware deinitialization performs MMIO accesses which are not necessary if the device has already been removed. In some cases, these accesses happen via readx_poll_timeout_atomic which ends up timing out, resulting in a warning at hw_atl2_utils_fw.c:112: [ 104.595913] Call Trace: [ 104.595915] [ 104.595918] ? show_regs+0x6c/0x80 [ 104.595923] ? __warn+0x8d/0x150 [ 104.595925] ? aq_a2_fw_deinit+0xcf/0xe0 [atlantic] [ 104.595934] ? report_bug+0x182/0x1b0 [ 104.595938] ? handle_bug+0x6e/0xb0 [ 104.595940] ? exc_invalid_op+0x18/0x80 [ 104.595942] ? asm_exc_invalid_op+0x1b/0x20 [ 104.595944] ? aq_a2_fw_deinit+0xcf/0xe0 [atlantic] [ 104.595952] ? aq_a2_fw_deinit+0xcf/0xe0 [atlantic] [ 104.595959] aq_nic_deinit.part.0+0xbd/0xf0 [atlantic] [ 104.595964] aq_nic_deinit+0x17/0x30 [atlantic] [ 104.595970] aq_ndev_close+0x2b/0x40 [atlantic] [ 104.595975] __dev_close_many+0xad/0x160 [ 104.595978] dev_close_many+0x99/0x170 [ 104.595979] unregister_netdevice_many_notify+0x18b/0xb20 [ 104.595981] ? __call_rcu_common+0xcd/0x700 [ 104.595984] unregister_netdevice_queue+0xc6/0x110 [ 104.595986] unregister_netdev+0x1c/0x30 [ 104.595988] aq_pci_remove+0xb1/0xc0 [atlantic] Fix this by skipping firmware deinitialization altogether if the PCI device is no longer present. Tested with an AQC113 attached via Thunderbolt by performing repeated unplug cycles while traffic was running via iperf. Fixes: 97bde5c4f909 ("net: ethernet: aquantia: Support for NIC-specific code") Signed-off-by: Jacob Moroni Reviewed-by: Igor Russkikh Reviewed-by: Simon Horman Link: https://patch.msgid.link/20250203143604.24930-3-mail@jakemoroni.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/aquantia/atlantic/aq_nic.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/aquantia/atlantic/aq_nic.c b/drivers/net/ethernet/aquantia/atlantic/aq_nic.c index fe0e3e2a8117..71e50fc65c14 100644 --- a/drivers/net/ethernet/aquantia/atlantic/aq_nic.c +++ b/drivers/net/ethernet/aquantia/atlantic/aq_nic.c @@ -1441,7 +1441,9 @@ void aq_nic_deinit(struct aq_nic_s *self, bool link_down) aq_ptp_ring_free(self); aq_ptp_free(self); - if (likely(self->aq_fw_ops->deinit) && link_down) { + /* May be invoked during hot unplug. */ + if (pci_device_is_present(self->pdev) && + likely(self->aq_fw_ops->deinit) && link_down) { mutex_lock(&self->fwreq_mutex); self->aq_fw_ops->deinit(self->aq_hw); mutex_unlock(&self->fwreq_mutex); -- 2.51.0 From a1300691aed9ee852b0a9192e29e2bdc2411a7e6 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 3 Feb 2025 17:08:38 +0000 Subject: [PATCH 15/16] net: rose: lock the socket in rose_bind() syzbot reported a soft lockup in rose_loopback_timer(), with a repro calling bind() from multiple threads. rose_bind() must lock the socket to avoid this issue. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: syzbot+7ff41b5215f0c534534e@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/67a0f78d.050a0220.d7c5a.00a0.GAE@google.com/T/#u Signed-off-by: Eric Dumazet Acked-by: Paolo Abeni Link: https://patch.msgid.link/20250203170838.3521361-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- net/rose/af_rose.c | 24 ++++++++++++++++-------- 1 file changed, 16 insertions(+), 8 deletions(-) diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c index 72c65d938a15..a4a668b88a8f 100644 --- a/net/rose/af_rose.c +++ b/net/rose/af_rose.c @@ -701,11 +701,9 @@ static int rose_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) struct net_device *dev; ax25_address *source; ax25_uid_assoc *user; + int err = -EINVAL; int n; - if (!sock_flag(sk, SOCK_ZAPPED)) - return -EINVAL; - if (addr_len != sizeof(struct sockaddr_rose) && addr_len != sizeof(struct full_sockaddr_rose)) return -EINVAL; @@ -718,8 +716,15 @@ static int rose_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) if ((unsigned int) addr->srose_ndigis > ROSE_MAX_DIGIS) return -EINVAL; - if ((dev = rose_dev_get(&addr->srose_addr)) == NULL) - return -EADDRNOTAVAIL; + lock_sock(sk); + + if (!sock_flag(sk, SOCK_ZAPPED)) + goto out_release; + + err = -EADDRNOTAVAIL; + dev = rose_dev_get(&addr->srose_addr); + if (!dev) + goto out_release; source = &addr->srose_call; @@ -730,7 +735,8 @@ static int rose_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) } else { if (ax25_uid_policy && !capable(CAP_NET_BIND_SERVICE)) { dev_put(dev); - return -EACCES; + err = -EACCES; + goto out_release; } rose->source_call = *source; } @@ -753,8 +759,10 @@ static int rose_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) rose_insert_socket(sk); sock_reset_flag(sk, SOCK_ZAPPED); - - return 0; + err = 0; +out_release: + release_sock(sk); + return err; } static int rose_connect(struct socket *sock, struct sockaddr *uaddr, int addr_len, int flags) -- 2.51.0 From 5eddd76ec2fd1988f0a3450fde9730b10dd22992 Mon Sep 17 00:00:00 2001 From: Alexandre Cassen Date: Wed, 19 Feb 2025 12:20:37 +0200 Subject: [PATCH 16/16] xfrm: fix tunnel mode TX datapath in packet offload mode Packets that match the output xfrm policy are delivered to the netstack. In IPsec packet mode for tunnel mode, the HW is responsible for building the hard header and outer IP header. In such a situation, the inner header may refer to a network that is not directly reachable by the host, resulting in a failed neighbor resolution. The packet is then dropped. xfrm policy defines the netdevice to use for xmit so we can send packets directly to it. Makes direct xmit exclusive to tunnel mode, since some rules may apply in transport mode. Fixes: f8a70afafc17 ("xfrm: add TX datapath support for IPsec packet offload mode") Signed-off-by: Alexandre Cassen Signed-off-by: Leon Romanovsky Signed-off-by: Steffen Klassert --- net/xfrm/xfrm_output.c | 41 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 41 insertions(+) diff --git a/net/xfrm/xfrm_output.c b/net/xfrm/xfrm_output.c index f7abd42c077d..e5b3e343a5ec 100644 --- a/net/xfrm/xfrm_output.c +++ b/net/xfrm/xfrm_output.c @@ -612,6 +612,40 @@ out: } EXPORT_SYMBOL_GPL(xfrm_output_resume); +static int xfrm_dev_direct_output(struct sock *sk, struct xfrm_state *x, + struct sk_buff *skb) +{ + struct dst_entry *dst = skb_dst(skb); + struct net *net = xs_net(x); + int err; + + dst = skb_dst_pop(skb); + if (!dst) { + XFRM_INC_STATS(net, LINUX_MIB_XFRMOUTERROR); + kfree_skb(skb); + return -EHOSTUNREACH; + } + skb_dst_set(skb, dst); + nf_reset_ct(skb); + + err = skb_dst(skb)->ops->local_out(net, sk, skb); + if (unlikely(err != 1)) { + kfree_skb(skb); + return err; + } + + /* In transport mode, network destination is + * directly reachable, while in tunnel mode, + * inner packet network may not be. In packet + * offload type, HW is responsible for hard + * header packet mangling so directly xmit skb + * to netdevice. + */ + skb->dev = x->xso.dev; + __skb_push(skb, skb->dev->hard_header_len); + return dev_queue_xmit(skb); +} + static int xfrm_output2(struct net *net, struct sock *sk, struct sk_buff *skb) { return xfrm_output_resume(sk, skb, 1); @@ -735,6 +769,13 @@ int xfrm_output(struct sock *sk, struct sk_buff *skb) return -EHOSTUNREACH; } + /* Exclusive direct xmit for tunnel mode, as + * some filtering or matching rules may apply + * in transport mode. + */ + if (x->props.mode == XFRM_MODE_TUNNEL) + return xfrm_dev_direct_output(sk, x, skb); + return xfrm_output_resume(sk, skb, 0); } -- 2.51.0