[PATCH] RDMA/siw: drop duplicate check_app_limited in siw_tcp_sendpages

[PATCH] RDMA/siw: drop duplicate check_app_limited in siw_tcp_sendpages

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

After do_tcp_sendpages() was converted to use MSG_SPLICE_PAGES and
later inlined as direct tcp_sendmsg_locked() calls, callers that
previously needed an explicit tcp_rate_check_app_limited(sk) to cover
do_tcp_sendpages() no longer need it. tcp_sendmsg_locked() performs the
check on every path that queues data, making the outer call redundant.

The site changed here, siw_tcp_sendpages() in siw_qp_tx.c, holds the socket
lock and invokes tcp_sendmsg_locked() on every iteration. The early-return
paths in tcp_sendmsg_locked() that skip tcp_rate_check_app_limited() - the
MSG_ZEROCOPY allocation failure and MSG_FASTOPEN branches - return without
queueing any MSG_SPLICE_PAGES data, so there is no functional consequence
from omitting the outer check.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
Hi,

I'm Geliang from the MPTCP community. While developing "MPTCP sockmap" and
"MPTCP KTLS", I found this duplicate check_app_limited issue. I recently
submitted patches to remove it in BPF [1] and TLS [2], both of which have
been accepted. I noticed the same duplicate check_app_limited also exists
in siw_tcp_sendpages, so I'm sending this patch to remove it as well.

Thanks,
-Geliang

[1]
https://patchwork.kernel.org/project/netdevbpf/patch/1f7dc605b16fc0590f7cbf5a27d57271926c01ce.1790074764.git.tanggeliang@kylinos.cn/
[2]
https://patchwork.kernel.org/project/netdevbpf/patch/5bc803abb4e27054eaffc8b26db7812e6347b0dd.1789634247.git.tanggeliang@kylinos.cn/
---
 drivers/infiniband/sw/siw/siw_qp_tx.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/infiniband/sw/siw/siw_qp_tx.c b/drivers/infiniband/sw/siw/siw_qp_tx.c
index f7dd32c6e5ba..56bd009879c2 100644
--- a/drivers/infiniband/sw/siw/siw_qp_tx.c
+++ b/drivers/infiniband/sw/siw/siw_qp_tx.c
@@ -336,7 +336,6 @@ static int siw_tcp_sendpages(struct socket *s, struct page **page, int offset,
 		if (size + offset <= PAGE_SIZE)
 			msg.msg_flags &= ~MSG_MORE;
 
-		tcp_rate_check_app_limited(sk);
 		if (!sendpage_ok(page[i]))
 			msg.msg_flags &= ~MSG_SPLICE_PAGES;
 		bvec_set_page(&bvec, page[i], bytes, offset);
-- 
2.53.0

Re: [PATCH] RDMA/siw: drop duplicate check_app_limited in siw_tcp_sendpages

[sashiko-bot]

> From: Geliang Tang <tanggeliang@kylinos.cn>
> 
> After do_tcp_sendpages() was converted to use MSG_SPLICE_PAGES and
> later inlined as direct tcp_sendmsg_locked() calls, callers that
> previously needed an explicit tcp_rate_check_app_limited(sk) to cover
> do_tcp_sendpages() no longer need it. tcp_sendmsg_locked() performs the
> check on every path that queues data, making the outer call redundant.
> 
> The site changed here, siw_tcp_sendpages() in siw_qp_tx.c, holds the socket
> lock and invokes tcp_sendmsg_locked() on every iteration. The early-return
> paths in tcp_sendmsg_locked() that skip tcp_rate_check_app_limited() - the
> MSG_ZEROCOPY allocation failure and MSG_FASTOPEN branches - return without
> queueing any MSG_SPLICE_PAGES data, so there is no functional consequence
> from omitting the outer check.
> 
> Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>

Sashiko has reviewed this patch and found no issues. It looks great!

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/32ef40cc81486adb8f2d43585224ab44d7e77f73.1790234916.git.tanggeliang@kylinos.cn?part=1

Re: [PATCH] RDMA/siw: drop duplicate check_app_limited in siw_tcp_sendpages

[Bernard Metzler]

On 24.09.2026 09:39, Geliang Tang wrote:
> From: Geliang Tang <tanggeliang@kylinos.cn>
> 
> After do_tcp_sendpages() was converted to use MSG_SPLICE_PAGES and
> later inlined as direct tcp_sendmsg_locked() calls, callers that
> previously needed an explicit tcp_rate_check_app_limited(sk) to cover
> do_tcp_sendpages() no longer need it. tcp_sendmsg_locked() performs the
> check on every path that queues data, making the outer call redundant.
> 
> The site changed here, siw_tcp_sendpages() in siw_qp_tx.c, holds the socket
> lock and invokes tcp_sendmsg_locked() on every iteration. The early-return
> paths in tcp_sendmsg_locked() that skip tcp_rate_check_app_limited() - the
> MSG_ZEROCOPY allocation failure and MSG_FASTOPEN branches - return without
> queueing any MSG_SPLICE_PAGES data, so there is no functional consequence
> from omitting the outer check.
> 
> Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
> ---
> Hi,
> 
> I'm Geliang from the MPTCP community. While developing "MPTCP sockmap" and
> "MPTCP KTLS", I found this duplicate check_app_limited issue. I recently
> submitted patches to remove it in BPF [1] and TLS [2], both of which have
> been accepted. I noticed the same duplicate check_app_limited also exists
> in siw_tcp_sendpages, so I'm sending this patch to remove it as well.
> 
> Thanks,
> -Geliang
> 
> [1]
> https://patchwork.kernel.org/project/netdevbpf/patch/1f7dc605b16fc0590f7cbf5a27d57271926c01ce.1790074764.git.tanggeliang@kylinos.cn/
> [2]
> https://patchwork.kernel.org/project/netdevbpf/patch/5bc803abb4e27054eaffc8b26db7812e6347b0dd.1789634247.git.tanggeliang@kylinos.cn/
> ---
>   drivers/infiniband/sw/siw/siw_qp_tx.c | 1 -
>   1 file changed, 1 deletion(-)
> 
> diff --git a/drivers/infiniband/sw/siw/siw_qp_tx.c b/drivers/infiniband/sw/siw/siw_qp_tx.c
> index f7dd32c6e5ba..56bd009879c2 100644
> --- a/drivers/infiniband/sw/siw/siw_qp_tx.c
> +++ b/drivers/infiniband/sw/siw/siw_qp_tx.c
> @@ -336,7 +336,6 @@ static int siw_tcp_sendpages(struct socket *s, struct page **page, int offset,
>   		if (size + offset <= PAGE_SIZE)
>   			msg.msg_flags &= ~MSG_MORE;
>   
> -		tcp_rate_check_app_limited(sk);
>   		if (!sendpage_ok(page[i]))
>   			msg.msg_flags &= ~MSG_SPLICE_PAGES;
>   		bvec_set_page(&bvec, page[i], bytes, offset);

Thank you Geliang, much appreciated.


Acked-by: Bernard Metzler <bernard.metzler@linux.dev>

Re: [PATCH] RDMA/siw: drop duplicate check_app_limited in siw_tcp_sendpages

[Leon Romanovsky]

On Thu, 24 Sep 2026 15:39:19 +0800, Geliang Tang wrote:
> After do_tcp_sendpages() was converted to use MSG_SPLICE_PAGES and
> later inlined as direct tcp_sendmsg_locked() calls, callers that
> previously needed an explicit tcp_rate_check_app_limited(sk) to cover
> do_tcp_sendpages() no longer need it. tcp_sendmsg_locked() performs the
> check on every path that queues data, making the outer call redundant.
> 
> The site changed here, siw_tcp_sendpages() in siw_qp_tx.c, holds the socket
> lock and invokes tcp_sendmsg_locked() on every iteration. The early-return
> paths in tcp_sendmsg_locked() that skip tcp_rate_check_app_limited() - the
> MSG_ZEROCOPY allocation failure and MSG_FASTOPEN branches - return without
> queueing any MSG_SPLICE_PAGES data, so there is no functional consequence
> from omitting the outer check.
> 
> [...]

Applied, thanks!

[1/1] RDMA/siw: drop duplicate check_app_limited in siw_tcp_sendpages
      https://git.kernel.org/rdma/rdma/c/ad91fd9e9440ab

Best regards,
-- 
Leon Romanovsky <leon@kernel.org>