Re: Weekly net upstream? - Re: [PATCH] vxcan: fix Kconfig description stating no local echo provided
Oliver Hartkopp <[email protected]>
| Newsgroups | org.kernel.vger.linux-can |
|---|---|
| Message-ID | <[email protected]> |
On 14.07.26 10:51, Marc Kleine-Budde wrote: > On 14.07.2026 08:55:32, Oliver Hartkopp wrote: >> I would be glad if you could upstream this patch together with the now >> sashiko-approved fixes for can-bcm and can-isotp: >> >> https://lore.kernel.org/linux-can/[email protected]/ > > If I understand you correctly, you say that the issue sashiko claims to > be in patch 3 (mid:[email protected]): > >>> [High] Lockless state transition in the timeout handler creates a >>> race condition that causes timeout errors to be silently dropped or >>> misattributed, leading sendmsg() to erroneously return success for >>> timed-out transfers. > > is a preexisting problem? Yes. Always broken for years but nobody ever complained as the pattern that would lead to this problem is not used. The core issues with concurrent execution UFA and probable stalls are fixed within the patch set. Fixing the in rare cases wrong (and unused) error attribution would be a huge effort. So it would be something for can-next or wont-fix. > >> https://lore.kernel.org/linux-can/[email protected]/ > > Can you clarify these claims: > > https://lore.kernel.org/all/[email protected]/ Fixed in patch 9. > https://lore.kernel.org/all/[email protected]/ Oh, this was really an open point that I have overlooked. Will send a V13 asap. Best regards, Oliver