Re: [PATCH net v2 1/2] ipv6: fix cached route reuse for SO_MARK control messages

From
Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Date
2026-11-06 03:01:31
Also in
linux-kselftest, lkml, stable
Reply to
Xia Yan, 2026-11-06
Xia Yan wrote:
Hi,

Thanks for the review. v2 fixes SO_MARK sends, including the concurrent
sk_mark case that came up during the v1 review. Sends without an SO_MARK
cmsg keep the current behavior.

Willem, you also raised the flowlabel and dscp cases in the v1 review:
https://lore.kernel.org/netdev/willemdebruijn.kernel.6ba984cf38a@gmail.com/ (local)

I'd like to keep the SO_MARK fix targeted at net and handle the
remaining cache issues separately in net-next. Does that split make
sense to you?
Agreed that those additional concerns should not block this patch.

Are they net-next material, or similar bugs and thus also net?
 
I'll wait for your feedback on the scope before posting v3.

v3 will also add a fwmark 0/0xffffffff prohibit rule to the
connected-socket selftest, so reusing the cached route instead of doing
a fresh lookup fails the explicit-zero cases.

Thanks,
Xia Yan

pw-bot: cr
  
Thread (10 messages) 10 messages, 4 authors, last reply 7d ago
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help