tls: prevent oversized sendfile() hangs by ignoring MSG_MORE
Subsystems (3)
Jakub Kicinski
maintainerPATCH 1/1 tls: prevent oversized sendfile() hangs by ignoring MSG_MORE
We got multiple reports that multi_chunk_sendfile test case from tls selftest fails. This was sort of expected, as the original fix was never applied (see it in the first Link:). The test in question uses sendfile() with count larger than the size of the underlying file. This will make splice set MSG_MORE on all sendpage calls, meaning TLS will never close and flush the last partial record.
Eric seem to have addressed a similar problem in
commit 35f9c09fe9c7 ("tcp: tcp_sendpages() should call tcp_push() once") by introducing MSG_SENDPAGE_NOTLAST. Unlike MSG_MORE MSG_SENDPAGE_NOTLAST is not set on the last call
of a "pipefull" of data (PIPE_DEF_BUFFERS == 16,
so every 16 pages or whenever we run out of data).
Having a break every 16 pages should be fine, TLS can pack exactly 4 pages into a record, so for aligned reads there should be no difference, unaligned may see one extra record per sendpage().
Sticking to TCP semantics seems preferable to modifying splice, but we can revisit it if real life scenarios show a regression.
--- net/tls/tls_sw.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
net/tls/tls_sw.c +1 -1
@@ -1153,7 +1153,7 @@ static int tls_sw_do_sendpage(struct sock *sk, struct page *page,int ret =0 ;bool eor; - eor = !(flags & (MSG_MORE |MSG_SENDPAGE_NOTLAST )); + eor = !(flags &MSG_SENDPAGE_NOTLAST );sk_clear_bit (SOCKWQ_ASYNC_NOSPACE , sk);/* Call the sk_stream functions to manage the sndbuf mem. */
Signature · 1 line
2.31.1
Quoted 24 lines · Jakub Kicinski
We got multiple reports that
multi_chunk_sendfiletest case from tls selftest fails. This was sort of expected, as the original fix was never applied (see it in the first Link:). The test in question usessendfile()with count larger than the size of the underlying file. This will make splice setMSG_MOREon all sendpage calls, meaning TLS will never close and flush the last partial record.Eric seem to have addressed a similar problem in
commit35f9c09fe9c7("tcp:tcp_sendpages()should calltcp_push()once") by introducingMSG_SENDPAGE_NOTLAST. UnlikeMSG_MOREMSG_SENDPAGE_NOTLASTis not set on the last call
of a "pipefull" of data (PIPE_DEF_BUFFERS== 16,
so every 16 pages or whenever we run out of data).Having a break every 16 pages should be fine, TLS can pack exactly 4 pages into a record, so for aligned reads there should be no difference, unaligned may see one extra record per
sendpage().Sticking to TCP semantics seems preferable to modifying splice, but we can revisit it if real life scenarios show a regression.
Quoted 5 lines
Reported-by Vadim Fedorenko <[email protected]>Reported-by Seth Forshee <[email protected]>Fixes 3c4d7559159b ("tls: kernel TLS support")Signed-off-by Jakub Kicinski <[email protected]>
With this patch the muilt_chunk_sendfile selftest passes. Thanks!
This patch was applied to netdev/net.git (refs/heads/master):
Hello:
This patch was applied to netdev/net.git (refs/heads/master):
Quoted 9 lines · On Fri, 18 Jun 2021 13:34:06 -0700 you wrote:
We got multiple reports that
multi_chunk_sendfiletest case from tls selftest fails. This was sort of expected, as the original fix was never applied (see it in the first Link:). The test in question usessendfile()with count larger than the size of the underlying file. This will make splice setMSG_MOREon all sendpage calls, meaning TLS will never close and flush the last partial record.[...]
Here is the summary with links:
- [net] tls: prevent oversized sendfile() hangs by ignoring MSG_MORE
https://git.kernel.org/netdev/net/c/d452d48b9f8b
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html