From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: stable@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
patches@lists.linux.dev, Amery Hung <ameryhung@gmail.com>,
Tariq Toukan <tariqt@nvidia.com>,
Jakub Kicinski <kuba@kernel.org>, Sasha Levin <sashal@kernel.org>
Subject: [PATCH 6.6 28/84] net/mlx5e: RX, Fix generating skb from non-linear xdp_buff for striding RQ
Date: Mon, 27 Oct 2025 19:36:17 +0100 [thread overview]
Message-ID: <20251027183439.568226732@linuxfoundation.org> (raw)
In-Reply-To: <20251027183438.817309828@linuxfoundation.org>
6.6-stable review patch. If anyone has any objections, please let me know.
------------------
From: Amery Hung <ameryhung@gmail.com>
[ Upstream commit 87bcef158ac1faca1bd7e0104588e8e2956d10be ]
XDP programs can change the layout of an xdp_buff through
bpf_xdp_adjust_tail() and bpf_xdp_adjust_head(). Therefore, the driver
cannot assume the size of the linear data area nor fragments. Fix the
bug in mlx5 by generating skb according to xdp_buff after XDP programs
run.
Currently, when handling multi-buf XDP, the mlx5 driver assumes the
layout of an xdp_buff to be unchanged. That is, the linear data area
continues to be empty and fragments remain the same. This may cause
the driver to generate erroneous skb or triggering a kernel
warning. When an XDP program added linear data through
bpf_xdp_adjust_head(), the linear data will be ignored as
mlx5e_build_linear_skb() builds an skb without linear data and then
pull data from fragments to fill the linear data area. When an XDP
program has shrunk the non-linear data through bpf_xdp_adjust_tail(),
the delta passed to __pskb_pull_tail() may exceed the actual nonlinear
data size and trigger the BUG_ON in it.
To fix the issue, first record the original number of fragments. If the
number of fragments changes after the XDP program runs, rewind the end
fragment pointer by the difference and recalculate the truesize. Then,
build the skb with the linear data area matching the xdp_buff. Finally,
only pull data in if there is non-linear data and fill the linear part
up to 256 bytes.
Fixes: f52ac7028bec ("net/mlx5e: RX, Add XDP multi-buffer support in Striding RQ")
Signed-off-by: Amery Hung <ameryhung@gmail.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/1760644540-899148-3-git-send-email-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
.../net/ethernet/mellanox/mlx5/core/en_rx.c | 26 ++++++++++++++++---
1 file changed, 23 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index 54268892148d4..fcf7437174e18 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -2024,6 +2024,7 @@ mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *w
u32 byte_cnt = cqe_bcnt;
struct skb_shared_info *sinfo;
unsigned int truesize = 0;
+ u32 pg_consumed_bytes;
struct bpf_prog *prog;
struct sk_buff *skb;
u32 linear_frame_sz;
@@ -2075,7 +2076,8 @@ mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *w
while (byte_cnt) {
/* Non-linear mode, hence non-XSK, which always uses PAGE_SIZE. */
- u32 pg_consumed_bytes = min_t(u32, PAGE_SIZE - frag_offset, byte_cnt);
+ pg_consumed_bytes =
+ min_t(u32, PAGE_SIZE - frag_offset, byte_cnt);
if (test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state))
truesize += pg_consumed_bytes;
@@ -2091,10 +2093,15 @@ mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *w
}
if (prog) {
+ u8 nr_frags_free, old_nr_frags = sinfo->nr_frags;
+ u32 len;
+
if (mlx5e_xdp_handle(rq, prog, mxbuf)) {
if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
struct mlx5e_frag_page *pfp;
+ frag_page -= old_nr_frags - sinfo->nr_frags;
+
for (pfp = head_page; pfp < frag_page; pfp++)
pfp->frags++;
@@ -2104,9 +2111,19 @@ mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *w
return NULL; /* page/packet was consumed by XDP */
}
+ nr_frags_free = old_nr_frags - sinfo->nr_frags;
+ if (unlikely(nr_frags_free)) {
+ frag_page -= nr_frags_free;
+ truesize -= (nr_frags_free - 1) * PAGE_SIZE +
+ ALIGN(pg_consumed_bytes,
+ BIT(rq->mpwqe.log_stride_sz));
+ }
+
+ len = mxbuf->xdp.data_end - mxbuf->xdp.data;
+
skb = mlx5e_build_linear_skb(
rq, mxbuf->xdp.data_hard_start, linear_frame_sz,
- mxbuf->xdp.data - mxbuf->xdp.data_hard_start, 0,
+ mxbuf->xdp.data - mxbuf->xdp.data_hard_start, len,
mxbuf->xdp.data - mxbuf->xdp.data_meta);
if (unlikely(!skb)) {
mlx5e_page_release_fragmented(rq, &wi->linear_page);
@@ -2130,8 +2147,11 @@ mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *w
do
pagep->frags++;
while (++pagep < frag_page);
+
+ headlen = min_t(u16, MLX5E_RX_MAX_HEAD - len,
+ skb->data_len);
+ __pskb_pull_tail(skb, headlen);
}
- __pskb_pull_tail(skb, headlen);
} else {
dma_addr_t addr;
--
2.51.0
next prev parent reply other threads:[~2025-10-27 19:17 UTC|newest]
Thread overview: 95+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-10-27 18:35 [PATCH 6.6 00/84] 6.6.115-rc1 review Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 01/84] exec: Fix incorrect type for ret Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 02/84] nios2: ensure that memblock.current_limit is set when setting pfn limits Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 03/84] hfs: clear offset and space out of valid records in b-tree node Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 04/84] hfs: make proper initalization of struct hfs_find_data Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 05/84] hfsplus: fix KMSAN uninit-value issue in __hfsplus_ext_cache_extent() Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 06/84] hfs: validate record offset in hfsplus_bmap_alloc Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 07/84] hfsplus: fix KMSAN uninit-value issue in hfsplus_delete_cat() Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 08/84] dlm: check for defined force value in dlm_lockspace_release Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 09/84] hfs: fix KMSAN uninit-value issue in hfs_find_set_zero_bits() Greg Kroah-Hartman
2025-10-27 18:35 ` [PATCH 6.6 10/84] hfsplus: return EIO when type of hidden directory mismatch in hfsplus_fill_super() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 11/84] lkdtm: fortify: Fix potential NULL dereference on kmalloc failure Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 12/84] m68k: bitops: Fix find_*_bit() signatures Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 13/84] powerpc/32: Remove PAGE_KERNEL_TEXT to fix startup failure Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 14/84] drivers/perf: hisi: Relax the event ID check in the framework Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 15/84] smb: server: let smb_direct_flush_send_list() invalidate a remote key first Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 16/84] Unbreak make tools/* for user-space targets Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 17/84] net/mlx5e: Return 1 instead of 0 in invalid case in mlx5e_mpwrq_umr_entry_size() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 18/84] rtnetlink: Allow deleting FDB entries in user namespace Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 19/84] net: Tree wide: Replace xdp_do_flush_map() with xdp_do_flush() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 20/84] net: enetc: fix the deadlock of enetc_mdio_lock Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 21/84] net: enetc: correct the value of ENETC_RXB_TRUESIZE Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 22/84] dpaa2-eth: fix the pointer passed to PTR_ALIGN on Tx path Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 23/84] can: bxcan: bxcan_start_xmit(): use can_dev_dropped_skb() instead of can_dropped_invalid_skb() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 24/84] selftests/net: convert sctp_vrf.sh to run it in unique namespace Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 25/84] selftests: net: fix server bind failure in sctp_vrf.sh Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 26/84] net/mlx5e: Reuse per-RQ XDP buffer to avoid stack zeroing overhead Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 27/84] net/mlx5e: RX, Fix generating skb from non-linear xdp_buff for legacy RQ Greg Kroah-Hartman
2025-10-27 18:36 ` Greg Kroah-Hartman [this message]
2025-10-27 18:36 ` [PATCH 6.6 29/84] arm64, mm: avoid always making PTE dirty in pte_mkwrite() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 30/84] sctp: avoid NULL dereference when chunk data buffer is missing Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 31/84] net: bonding: fix possible peer notify event loss or dup issue Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 32/84] dma-debug: dont report false positives with DMA_BOUNCE_UNALIGNED_KMALLOC Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 33/84] arch_topology: Fix incorrect error check in topology_parse_cpu_capacity() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 34/84] gpio: pci-idio-16: Define maximum valid register address offset Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 35/84] gpio: 104-idio-16: " Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 36/84] Revert "cpuidle: menu: Avoid discarding useful information" Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 37/84] ACPICA: Work around bogus -Wstringop-overread warning since GCC 11 Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 38/84] can: netlink: can_changelink(): allow disabling of automatic restart Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 39/84] cifs: Fix TCP_Server_Info::credits to be signed Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 40/84] MIPS: Malta: Fix keyboard resource preventing i8042 driver from registering Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 41/84] ocfs2: clear extent cache after moving/defragmenting extents Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 42/84] vsock: fix lock inversion in vsock_assign_transport() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 43/84] net: stmmac: dwmac-rk: Fix disabling set_clock_selection Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 44/84] net: usb: rtl8150: Fix frame padding Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 45/84] net: ravb: Enforce descriptor type ordering Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 46/84] net: ravb: Ensure memory write completes before ringing TX doorbell Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 47/84] selftests: mptcp: join: mark flush re-add as skipped if not supported Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 48/84] selftests: mptcp: join: mark implicit tests " Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 49/84] spi: spi-nxp-fspi: add extra delay after dll locked Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 50/84] firmware: arm_scmi: Account for failed debug initialization Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 51/84] firmware: arm_scmi: Fix premature SCMI_XFER_FLAG_IS_RAW clearing in raw mode Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 52/84] RISC-V: Define pgprot_dmacoherent() for non-coherent devices Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 53/84] RISC-V: Dont print details of CPUs disabled in DT Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 54/84] hwmon: (sht3x) Fix error handling Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 55/84] gpio: update Intel LJCA USB GPIO driver Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 56/84] gpio: ljca: Fix duplicated IRQ mapping Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 57/84] io_uring: correct __must_hold annotation in io_install_fixed_file Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 58/84] sched: Remove never used code in mm_cid_get() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 59/84] USB: serial: option: add UNISOC UIS7720 Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 60/84] USB: serial: option: add Quectel RG255C Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 61/84] USB: serial: option: add Telit FN920C04 ECM compositions Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 62/84] usb/core/quirks: Add Huawei ME906S to wakeup quirk Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 63/84] usb: raw-gadget: do not limit transfer length Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 64/84] xhci: dbc: enable back DbC in resume if it was enabled before suspend Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 65/84] x86/microcode: Fix Entrysign revision check for Zen1/Naples Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 66/84] binder: remove "invalid inc weak" check Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 67/84] comedi: fix divide-by-zero in comedi_buf_munge() Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 68/84] mei: me: add wildcat lake P DID Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 69/84] misc: fastrpc: Fix dma_buf object leak in fastrpc_map_lookup Greg Kroah-Hartman
2025-10-27 18:36 ` [PATCH 6.6 70/84] most: usb: Fix use-after-free in hdm_disconnect Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 71/84] most: usb: hdm_probe: Fix calling put_device() before device initialization Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 72/84] tcpm: switch check for role_sw device with fw_node Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 73/84] dt-bindings: usb: dwc3-imx8mp: dma-range is required only for imx8mp Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 74/84] serial: 8250_dw: handle reset control deassert error Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 75/84] serial: 8250_exar: add support for Advantech 2 port card with Device ID 0x0018 Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 76/84] serial: 8250_mtk: Enable baud clock and manage in runtime PM Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 77/84] devcoredump: Fix circular locking dependency with devcd->mutex Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 78/84] xfs: always warn about deprecated mount options Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 79/84] fs/notify: call exportfs_encode_fid with s_umount Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 80/84] x86/resctrl: Fix miscount of bandwidth event when reactivating previously unavailable RMID Greg Kroah-Hartman
2025-10-28 16:00 ` Babu Moger
2025-10-27 18:37 ` [PATCH 6.6 81/84] s390/cio: Update purge function to unregister the unused subchannels Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 82/84] fuse: allocate ff->release_args only if release is needed Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 83/84] fuse: fix livelock in synchronous file put from fuseblk workers Greg Kroah-Hartman
2025-10-27 18:37 ` [PATCH 6.6 84/84] gpio: ljca: Initialize num before accessing item in ljca_gpio_config Greg Kroah-Hartman
2025-10-27 21:50 ` [PATCH 6.6 00/84] 6.6.115-rc1 review Florian Fainelli
2025-10-28 3:45 ` Peter Schneider
2025-10-28 11:28 ` Jon Hunter
2025-10-28 11:51 ` Ron Economos
2025-10-28 13:42 ` Naresh Kamboju
2025-10-28 13:53 ` Brett A C Sheffield
2025-10-28 19:25 ` Shuah Khan
2025-10-28 22:11 ` Slade Watkins
2025-10-29 11:21 ` Miguel Ojeda
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20251027183439.568226732@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--cc=ameryhung@gmail.com \
--cc=kuba@kernel.org \
--cc=patches@lists.linux.dev \
--cc=sashal@kernel.org \
--cc=stable@vger.kernel.org \
--cc=tariqt@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).