xsk: reject TX checksum requests that overflow skb->csum_start
Subsystems (4)
PATCH 1/1 xsk: reject TX checksum requests that overflow skb->csum_start
xsk_skb_metadata() sets the u16 skb->csum_start to hr + csum_start. With 64K pages, csum_start can be close to U16_MAX, so the sum can overflow. The checksum offset then becomes negative. With XDP_UMEM_TX_SW_CSUM, skb_checksum_help() triggers a warning. Without it, the device gets a CHECKSUM_PARTIAL skb with a negative offset.
Reject this with -EINVAL. Use the limit of skb_partial_csum_set(), which also rejects U16_MAX, the "not set" value of transport_header. That helper cannot be used here, as the skb built for an IFF_TX_SKB_NO_LINEAR device has no linear data.
---
Sashiko reported this in [1], and earlier in [2][3].
Returning -EINVAL leaves the descriptor in the TX ring, as the current check does [2]. That is a separate issue and out of scope here.
There is no selftest. The umem chunk size is at most PAGE_SIZE, so only a 64K-page kernel can hit this.
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
[3] https://lore.kernel.org/all/[email protected]/
net/xdp/xsk.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
net/xdp/xsk.c +2 -1
@@ -1006,7 +1006,8 @@ static int xsk_skb_metadata(struct sk_buff *skb, void *buffer, csum_offset =READ_ONCE (meta->request.csum_offset);if (unlikely (csum_start + csum_offset + -sizeof (__sum16) > desc->len)) +sizeof (__sum16) > desc->len || + hr + csum_start >=U16_MAX ))return -EINVAL ; skb->csum_start = hr + csum_start;
base-commit: aa40afa831ebfe45e40a23721711059f2ac584d4
Signature · 1 line
2.56.0
[email protected]
AI reviewThis is an automated message. This series looks like a fix, but its commit messages seem to be missing some information:
Hi!
This is an automated message. This series looks like a fix, but its commit messages seem to be missing some information:
- Whether the issue was actually triggered, or is only theoretical (e.g. found by LLM and not proven at runtime). If it was triggered please include the symptoms, like the stack trace or kernel messages.
Please do not repost the series just to address the above. Instead, reply to this email with the missing information, so that reviewers can take it into account.
*IMPORTANT*: If you do repost (for other reasons) - please include the relevant info in the cover letter or commit messages.
Signature · 2 lines
The evaluation is done by an LLM so it may be wrong, if you think that is the case please reply and explain.
Hi,
Quoted 3 lines · On Mon, Nov 16, 2026 at 04:32:14PM +0000, [email protected] wrote:
- Whether the issue was actually triggered, or is only theoretical (e.g. found by LLM and not proven at runtime). If it was triggered please include the symptoms, like the stack trace or kernel messages.
Sashiko reported this several times, and I reproduced it on a 64K-page arm64 kernel built from net at aa40afa831eb, under QEMU.
The test sends one TX descriptor from an AF_XDP socket in copy mode on a veth device, with XDP_UMEM_TX_SW_CSUM, a 64K chunk, a length of 65520, csum_start 65500 and csum_offset 6. The headroom is 64, so skb->csum_start wraps to 28:
skb len=65520 data_len=0 headroom=64 headlen=65520 tailroom=65168 csum(0x6001c start=28 offset=6 ip_summed=3 complete_sw=0 valid=0 level=0)
...
offset (-36) >= skb_headlen() (65520)
WARNING: net/core/dev.c:3696 at skb_checksum_help+0x1d0/0x238, CPU#0: xsk_csum_overfl/200 ... Call trace: skb_checksum_help+0x1d0/0x238 (P) __xsk_generic_xmit+0xb90/0xdf0 xsk_sendmsg+0xcc/0x188 __sys_sendto+0x128/0x1c0 __arm64_sys_sendto+0x2c/0x48 invoke_syscall.constprop.0+0x60/0x108 do_el0_svc+0x48/0xd8 el0_svc+0x40/0x238 el0t_64_sync_handler+0xa0/0xe8 el0t_64_sync+0x198/0x1a0
With this patch, sendmsg() returns -EINVAL and there is no warning.
Cheers,
Tianyi
Quoted 11 lines · Tianyi Gao
xsk_skb_metadata()sets the u16skb->csum_startto hr +csum_start. With 64K pages,csum_startcan be close toU16_MAX, so the sum can overflow. The checksum offset then becomes negative. WithXDP_UMEM_TX_SW_CSUM,skb_checksum_help()triggers a warning. Without it, the device gets aCHECKSUM_PARTIALskb with a negative offset.Reject this with
-EINVAL. Use the limit ofskb_partial_csum_set(), which also rejectsU16_MAX, the "not set" value oftransport_header. That helper cannot be used here, as the skb built for anIFF_TX_SKB_NO_LINEARdevice has no linear data.
Quoted 4 lines
Fixes 48eb03dd2630 ("xsk: Add TX timestamp and TX checksum offload support")Reported-by Sashiko <[email protected]>Signed-off-by Tianyi Gao <[email protected]>
This patch was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:
Hello:
This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:
Quoted 6 lines · On Mon, 16 Nov 2026 16:28:34 +0000 you wrote:
xsk_skb_metadata()sets the u16skb->csum_startto hr +csum_start. With 64K pages,csum_startcan be close toU16_MAX, so the sum can overflow. The checksum offset then becomes negative. WithXDP_UMEM_TX_SW_CSUM,skb_checksum_help()triggers a warning. Without it, the device gets aCHECKSUM_PARTIALskb with a negative offset.
Quoted 6 lines
Reject this with
-EINVAL. Use the limit ofskb_partial_csum_set(), which also rejectsU16_MAX, the "not set" value oftransport_header. That helper cannot be used here, as the skb built for anIFF_TX_SKB_NO_LINEARdevice has no linear data.[...]
Here is the summary with links:
- [net] xsk: reject TX checksum requests that overflow skb->csum_start
https://git.kernel.org/netdev/net/c/4fdfed6d0990
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html