* [PATCH net] tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append
@ 2026-11-19 9:08 Jiayuan Chen
2026-11-19 10:06 ` Eric Dumazet
2026-11-20 9:57 ` netdev-bot+sashiko
0 siblings, 2 replies; 4+ messages in thread
From: Jiayuan Chen @ 2026-11-19 9:08 UTC (permalink / raw)
To: netdev
Cc: Jiayuan Chen, Eric Dumazet, Neal Cardwell, Kuniyuki Iwashima,
David S. Miller, Jakub Kicinski, Paolo Abeni, Simon Horman,
Wei Wang, Soheil Hassas Yeganeh, Shakeel Butt, linux-kernel
Our local syzbot reported the following warning:
WARNING: ./include/linux/ucopysize.h:57 at tcp_sendmsg_locked+0x1e13/0x1f20
Modules linked in:
CPU: 3 UID: 0 PID: 377 Comm: syz.2.2248 Not tainted 7.4.0-rc2+ #398 PREEMPT
RIP: 0010:tcp_sendmsg_locked+0x1e13/0x1f20
Call Trace:
<TASK>
tcp_sendmsg+0x2c/0x50
inet_sendmsg+0x62/0xb0
__sys_sendto+0x2fe/0x350
__x64_sys_sendto+0x7f/0xb0
x64_sys_call+0x1375/0x27e0
do_syscall_64+0xe4/0x470
entry_SYSCALL_64_after_hwframe+0x76/0x7e
</TASK>
The problem is here:
if (!skb_zcopy_pure(skb)) {
copy = tcp_wmem_schedule(sk, copy);
if (!copy)
goto wait_for_space;
}
When MSG_ZEROCOPY data is appended to a skb that already holds copied
data, only the user length is scheduled, but the pinned pages are then
charged by their truesize:
skb_zerocopy_iter_stream
__zerocopy_sg_from_iter
zerocopy_fill_skb_from_iter //truesize += PAGE_ALIGN(copied + start)
sk_mem_charge(sk, truesize) //sk_forward_alloc goes negative
For example, sending 8 iovecs of 1 byte each schedules 8 bytes and
charges 32768 with 4K pages.
On a later sendmsg() without MSG_ZEROCOPY, if sk_wmem_schedule() fails
(e.g. tcp_mem hard limit is reached), tcp_wmem_schedule() returns
min(copy, sk->sk_forward_alloc), which is negative. The negative copy
is passed to skb_copy_to_page_nocache() and triggers the warning, and
sendmsg() fails with -EFAULT.
A few options were considered:
1) Schedule the truesize up front. It is not known before the pages
are pinned, it depends on the iovec layout.
2) Do not append zerocopy data to a non pure skb. This splits a small
header and its zerocopy payload into two skbs.
3) Schedule the missing part right after the pages are charged. They
are already in the skb at this point, so this can not fail.
Go with 3).
Fixes: 849b425cd091 ("tcp: fix possible freeze in tx path under memory pressure")
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
---
net/ipv4/tcp.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index d510e982dcea..8bd377542a89 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1359,6 +1359,8 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size)
if (err < 0)
goto do_error;
copy = err;
+ /* Pages were charged by truesize, not by copy. */
+ sk_forced_mem_schedule(sk, 0);
} else if (zc == MSG_SPLICE_PAGES) {
/* Splice in data if we can; copy if we can't. */
if (tcp_downgrade_zcopy_pure(sk, skb))
--
2.43.0
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH net] tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append
2026-11-19 9:08 [PATCH net] tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append Jiayuan Chen
@ 2026-11-19 10:06 ` Eric Dumazet
2026-11-19 10:23 ` Jiayuan Chen
2026-11-20 9:57 ` netdev-bot+sashiko
1 sibling, 1 reply; 4+ messages in thread
From: Eric Dumazet @ 2026-11-19 10:06 UTC (permalink / raw)
To: Jiayuan Chen
Cc: netdev, Neal Cardwell, Kuniyuki Iwashima, David S. Miller,
Jakub Kicinski, Paolo Abeni, Simon Horman, Wei Wang,
Soheil Hassas Yeganeh, Shakeel Butt, linux-kernel
Le jeu. 19 nov. 2026 à 10:08, Jiayuan Chen <jiayuan.chen@linux.dev> a écrit :
>
> Our local syzbot reported the following warning:
>
> WARNING: ./include/linux/ucopysize.h:57 at tcp_sendmsg_locked+0x1e13/0x1f20
> Modules linked in:
> CPU: 3 UID: 0 PID: 377 Comm: syz.2.2248 Not tainted 7.4.0-rc2+ #398 PREEMPT
> RIP: 0010:tcp_sendmsg_locked+0x1e13/0x1f20
> Call Trace:
> <TASK>
> tcp_sendmsg+0x2c/0x50
> inet_sendmsg+0x62/0xb0
> __sys_sendto+0x2fe/0x350
> __x64_sys_sendto+0x7f/0xb0
> x64_sys_call+0x1375/0x27e0
> do_syscall_64+0xe4/0x470
> entry_SYSCALL_64_after_hwframe+0x76/0x7e
> </TASK>
>
> The problem is here:
>
> if (!skb_zcopy_pure(skb)) {
> copy = tcp_wmem_schedule(sk, copy);
> if (!copy)
> goto wait_for_space;
> }
>
> When MSG_ZEROCOPY data is appended to a skb that already holds copied
> data, only the user length is scheduled, but the pinned pages are then
> charged by their truesize:
>
> skb_zerocopy_iter_stream
> __zerocopy_sg_from_iter
> zerocopy_fill_skb_from_iter //truesize += PAGE_ALIGN(copied + start)
> sk_mem_charge(sk, truesize) //sk_forward_alloc goes negative
>
> For example, sending 8 iovecs of 1 byte each schedules 8 bytes and
> charges 32768 with 4K pages.
This part is a separate bug: these 8 bytes sit in a single page.
zerocopy_fill_skb_from_iter() adds PAGE_ALIGN(copied + start) to
skb->truesize for each iov_iter_get_pages2() call, even when the
chunk is merged into the previous frag and the page reference is
released.
With a corked socket, one copied byte, then one MSG_ZEROCOPY sendmsg()
of 1024 bytes from a single page:
1 iovec sk_wmem_queued += 4096
1024 iovecs sk_wmem_queued += 4194304 (sk_sndbuf is 2626560)
With your patch, this over-estimation would become a forced charge,
above tcp_mem and memcg limits, with an amount controlled by the user.
I have a fix [1] for this accounting (only charge the pages a frag did not
already cover), I will send it shortly. With it, both cases above
charge 4096 bytes, and sk_forward_alloc is still negative after the
append, so your patch is still needed.
Thanks.
>
> On a later sendmsg() without MSG_ZEROCOPY, if sk_wmem_schedule() fails
> (e.g. tcp_mem hard limit is reached), tcp_wmem_schedule() returns
> min(copy, sk->sk_forward_alloc), which is negative. The negative copy
> is passed to skb_copy_to_page_nocache() and triggers the warning, and
> sendmsg() fails with -EFAULT.
>
> A few options were considered:
>
> 1) Schedule the truesize up front. It is not known before the pages
> are pinned, it depends on the iovec layout.
> 2) Do not append zerocopy data to a non pure skb. This splits a small
> header and its zerocopy payload into two skbs.
> 3) Schedule the missing part right after the pages are charged. They
> are already in the skb at this point, so this can not fail.
>
> Go with 3).
Analysis looks correct, and 3) is consistent with what we did
for tcp_collapse() recently.
>
> Fixes: 849b425cd091 ("tcp: fix possible freeze in tx path under memory pressure")
> Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
> ---
> net/ipv4/tcp.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
> index d510e982dcea..8bd377542a89 100644
> --- a/net/ipv4/tcp.c
> +++ b/net/ipv4/tcp.c
> @@ -1359,6 +1359,8 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size)
> if (err < 0)
> goto do_error;
> copy = err;
> + /* Pages were charged by truesize, not by copy. */
> + sk_forced_mem_schedule(sk, 0);
Pure zerocopy skbs are not charged, so there is nothing to fix for
them. Please avoid the extra work in this case:
if (!skb_zcopy_pure(skb))
sk_forced_mem_schedule(sk, 0);
> } else if (zc == MSG_SPLICE_PAGES) {
> /* Splice in data if we can; copy if we can't. */
> if (tcp_downgrade_zcopy_pure(sk, skb))
> --
> 2.43.0
>
[1]
diff --git a/net/core/datagram.c b/net/core/datagram.c
index 489d92fe6dae8952da08bc5cb2a58f9863938fe0..8c418c3adf1c97b41ee7999146d09aacd8cf29d4
100644
--- a/net/core/datagram.c
+++ b/net/core/datagram.c
@@ -659,7 +659,6 @@ int zerocopy_fill_skb_from_iter(struct sk_buff *skb,
skb->data_len += copied;
skb->len += copied;
- skb->truesize += PAGE_ALIGN(copied + start);
head = compound_head(pages[n]);
order = compound_order(head);
@@ -680,6 +679,11 @@ int zerocopy_fill_skb_from_iter(struct sk_buff *skb,
if (head == skb_frag_page(last) &&
start == skb_frag_off(last) +
skb_frag_size(last)) {
+ /* Only account the pages this frag
+ * did not already cover.
+ */
+ skb->truesize +=
PAGE_ALIGN(start + size) -
+ PAGE_ALIGN(start);
skb_frag_size_add(last, size);
/* We combined this page, we
need to release
* a reference. Since compound
pages refcount
@@ -695,6 +699,7 @@ int zerocopy_fill_skb_from_iter(struct sk_buff *skb,
page_ref_sub(last_head, refs);
refs = 0;
}
+ skb->truesize += PAGE_SIZE;
skb_fill_page_desc_noacc(skb, frag++, head,
start, size);
}
if (refs)
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH net] tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append
2026-11-19 10:06 ` Eric Dumazet
@ 2026-11-19 10:23 ` Jiayuan Chen
0 siblings, 0 replies; 4+ messages in thread
From: Jiayuan Chen @ 2026-11-19 10:23 UTC (permalink / raw)
To: Eric Dumazet
Cc: netdev, Neal Cardwell, Kuniyuki Iwashima, David S. Miller,
Jakub Kicinski, Paolo Abeni, Simon Horman, Wei Wang,
Soheil Hassas Yeganeh, Shakeel Butt, linux-kernel
On 11/19/26 6:06 PM, Eric Dumazet wrote:
> Le jeu. 19 nov. 2026 à 10:08, Jiayuan Chen <jiayuan.chen@linux.dev> a écrit :
>> Our local syzbot reported the following warning:
>>
>> WARNING: ./include/linux/ucopysize.h:57 at tcp_sendmsg_locked+0x1e13/0x1f20
>> Modules linked in:
>> CPU: 3 UID: 0 PID: 377 Comm: syz.2.2248 Not tainted 7.4.0-rc2+ #398 PREEMPT
>> RIP: 0010:tcp_sendmsg_locked+0x1e13/0x1f20
>> Call Trace:
>> <TASK>
>> tcp_sendmsg+0x2c/0x50
>> inet_sendmsg+0x62/0xb0
>> __sys_sendto+0x2fe/0x350
>> __x64_sys_sendto+0x7f/0xb0
>> x64_sys_call+0x1375/0x27e0
>> do_syscall_64+0xe4/0x470
>> entry_SYSCALL_64_after_hwframe+0x76/0x7e
>> </TASK>
>>
>> The problem is here:
>>
>> if (!skb_zcopy_pure(skb)) {
>> copy = tcp_wmem_schedule(sk, copy);
>> if (!copy)
>> goto wait_for_space;
>> }
>>
>> When MSG_ZEROCOPY data is appended to a skb that already holds copied
>> data, only the user length is scheduled, but the pinned pages are then
>> charged by their truesize:
>>
>> skb_zerocopy_iter_stream
>> __zerocopy_sg_from_iter
>> zerocopy_fill_skb_from_iter //truesize += PAGE_ALIGN(copied + start)
>> sk_mem_charge(sk, truesize) //sk_forward_alloc goes negative
>>
>> For example, sending 8 iovecs of 1 byte each schedules 8 bytes and
>> charges 32768 with 4K pages.
> This part is a separate bug: these 8 bytes sit in a single page.
>
> zerocopy_fill_skb_from_iter() adds PAGE_ALIGN(copied + start) to
> skb->truesize for each iov_iter_get_pages2() call, even when the
> chunk is merged into the previous frag and the page reference is
> released.
>
> With a corked socket, one copied byte, then one MSG_ZEROCOPY sendmsg()
> of 1024 bytes from a single page:
>
> 1 iovec sk_wmem_queued += 4096
> 1024 iovecs sk_wmem_queued += 4194304 (sk_sndbuf is 2626560)
>
> With your patch, this over-estimation would become a forced charge,
> above tcp_mem and memcg limits, with an amount controlled by the user.
>
> I have a fix [1] for this accounting (only charge the pages a frag did not
> already cover), I will send it shortly. With it, both cases above
> charge 4096 bytes, and sk_forward_alloc is still negative after the
> append, so your patch is still needed.
Thanks for the fix.
> Thanks.
>
>> On a later sendmsg() without MSG_ZEROCOPY, if sk_wmem_schedule() fails
>> (e.g. tcp_mem hard limit is reached), tcp_wmem_schedule() returns
>> min(copy, sk->sk_forward_alloc), which is negative. The negative copy
>> is passed to skb_copy_to_page_nocache() and triggers the warning, and
>> sendmsg() fails with -EFAULT.
>>
>> A few options were considered:
>>
>> 1) Schedule the truesize up front. It is not known before the pages
>> are pinned, it depends on the iovec layout.
>> 2) Do not append zerocopy data to a non pure skb. This splits a small
>> header and its zerocopy payload into two skbs.
>> 3) Schedule the missing part right after the pages are charged. They
>> are already in the skb at this point, so this can not fail.
>>
>> Go with 3).
> Analysis looks correct, and 3) is consistent with what we did
> for tcp_collapse() recently.
>
>> Fixes: 849b425cd091 ("tcp: fix possible freeze in tx path under memory pressure")
>> Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
>> ---
>> net/ipv4/tcp.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
>> index d510e982dcea..8bd377542a89 100644
>> --- a/net/ipv4/tcp.c
>> +++ b/net/ipv4/tcp.c
>> @@ -1359,6 +1359,8 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size)
>> if (err < 0)
>> goto do_error;
>> copy = err;
>> + /* Pages were charged by truesize, not by copy. */
>> + sk_forced_mem_schedule(sk, 0);
> Pure zerocopy skbs are not charged, so there is nothing to fix for
> them. Please avoid the extra work in this case:
>
> if (!skb_zcopy_pure(skb))
> sk_forced_mem_schedule(sk, 0);
Will do in v2.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH net] tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append
2026-11-19 9:08 [PATCH net] tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append Jiayuan Chen
2026-11-19 10:06 ` Eric Dumazet
@ 2026-11-20 9:57 ` netdev-bot+sashiko
1 sibling, 0 replies; 4+ messages in thread
From: netdev-bot+sashiko @ 2026-11-20 9:57 UTC (permalink / raw)
To: jiayuan.chen
Cc: netdev, edumazet, ncardwell, kuniyu, davem, kuba, pabeni, horms,
weiwan, soheil, shakeel.butt, linux-kernel
Thank you for your contribution! Sashiko AI review found 2 potential
issue(s) to consider.
Critical: 0 · High: 0 · Medium: 1 · Low: 1
If you believe none of the issues are legitimate, please reply to each
issue of Medium severity or higher and explain why the AI is wrong.
If any of them are legitimate and you plan to respin, please let patchwork
know by including "pw-bot: cr" as a separate line at the end of your reply
(one such reply per series is enough).
- [Medium] The new sk_forced_mem_schedule(sk, 0) only runs on the success
path of skb_zerocopy_iter_stream().
- [Low] In tcp_sendmsg_locked() (net/ipv4/tcp.c), the MSG_ZEROCOPY branch
only pre-schedules memory when the skb is not pure: `if…
--
tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append
The MSG_ZEROCOPY path in tcp_sendmsg_locked() can append to a skb that
already holds copied data. In that case only the user length is
scheduled, but the pinned pages are charged by truesize. This commit adds
sk_forced_mem_schedule(sk, 0) after a successful skb_zerocopy_iter_stream()
so that sk_forward_alloc does not go negative.
> diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
> index 832b25ddd3c2e..9f8c0954001bb 100644
> --- a/net/ipv4/tcp.c
> +++ b/net/ipv4/tcp.c
> @@ -1359,6 +1359,8 @@ int tcp_sendmsg_locked(struct sock *sk, struct msghdr *msg, size_t size)
> if (err < 0)
> goto do_error;
> copy = err;
> + /* Pages were charged by truesize, not by copy. */
> + sk_forced_mem_schedule(sk, 0);
[Severity: Medium]
Can sk_forward_alloc still go negative on the error path? The new call
only runs when skb_zerocopy_iter_stream() succeeds, because the
"if (err < 0) goto do_error" check comes first.
In skb_zerocopy_iter_stream(), the truesize charge made by
__zerocopy_sg_from_iter() is undone on -EFAULT only if ___pskb_trim()
succeeds:
net/core/skbuff.c:skb_zerocopy_iter_stream() {
...
if (err == -EFAULT || (err == -EMSGSIZE && skb->len == orig_len)) {
...
if (!___pskb_trim(skb, orig_len)) {
skb->truesize -= delta_truesize;
sk_wmem_queued_add(sk, -delta_truesize);
if (!skb_zcopy_pure(skb))
sk_mem_uncharge(sk, delta_truesize);
}
skb->sk = save_sk;
return err;
}
...
}
___pskb_trim() fails when the skb is cloned and
pskb_expand_head(skb, 0, 0, GFP_ATOMIC) fails.
Suppose the write queue tail skb is not pure, so only copy was scheduled,
and it is cloned. If the user iovec faults after some pages are pinned and
the atomic allocation fails, those pages stay charged by truesize. The
caller then jumps to do_error without running the fix-up.
A later non-zerocopy sendmsg() under tcp_mem pressure would then get a
negative copy back from tcp_wmem_schedule(). That is the same case the
commit message describes.
Would it work to call sk_forced_mem_schedule(sk, 0) right after
skb_zerocopy_iter_stream() returns, before the error checks? On the paths
that do undo the charge, the call would do nothing.
[Severity: Low]
This isn't a bug, but does the new comment hold for pure zerocopy skbs?
Earlier in this branch, the pre-schedule is guarded:
if (!skb_zcopy_pure(skb)) {
copy = tcp_wmem_schedule(sk, copy);
if (!copy)
goto wait_for_space;
}
__zerocopy_sg_from_iter() also skips the charge for pure skbs:
net/core/datagram.c:__zerocopy_sg_from_iter() {
...
sk_wmem_queued_add(sk, truesize);
if (!skb_zcopy_pure(skb))
sk_mem_charge(sk, truesize);
...
}
So when SKBFL_PURE_ZEROCOPY is set, no pages were charged to
sk_forward_alloc. sk_forced_mem_schedule(sk, 0) is still called anyway.
In that case it does nothing, unless sk_forward_alloc is already negative
for some unrelated reason. Then it would force-charge without any warning
and hide that imbalance.
Could the call use the same !skb_zcopy_pure(skb) check as the
pre-schedule? Alternatively, the comment could say it covers the non-pure
case.
> } else if (zc == MSG_SPLICE_PAGES) {
> /* Splice in data if we can; copy if we can't. */
> if (tcp_downgrade_zcopy_pure(sk, skb))
--
Sashiko AI review · https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20261119090828.343748-1-jiayuan.chen%40linux.dev
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2026-11-20 9:57 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-19 9:08 [PATCH net] tcp: fix negative sk_forward_alloc after MSG_ZEROCOPY append Jiayuan Chen
2026-11-19 10:06 ` Eric Dumazet
2026-11-19 10:23 ` Jiayuan Chen
2026-11-20 9:57 ` netdev-bot+sashiko
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.