On Sun, Apr 03, 2022 at 07:40:38PM -0400, Sasha Levin wrote:
> On Sun, Apr 03, 2022 at 08:24:04PM +0200, Oliver Hartkopp wrote:
> > Hi Sasha,
> >
> > On 03.04.22 19:28, Sasha Levin wrote:
> > > On Sun, Apr 03, 2022 at 11:26:30AM +0200, Oliver Hartkopp wrote:
> > > > Hello Sasha/Greg,
> > > >
> > > > the way how the can-isotp patches are currently applied to the
> > > > stable trees 5.17/5.16/5.15/5.10 is FINE.
> > > >
> > > > This means 5.15 now has a functional improvement (as only stable
> > > > tree). But that should be fine too.
> > >
> > > Are they all a functional improvement? or are some fixes?
> > >
> >
> > Only upstream commit 4b7fe92c06901f4563af0e36d25223a5ab343782 ("can:
> > isotp: add local echo tx processing for consecutive frames") is an
> > improvement, see:
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-5.15/can-isotp-add-local-echo-tx-processing-for-consecuti.patch?id=c0c51980438b5e5b0d758dc28eef5947e57c9c1e
> >
> > All others are fixes.
> >
> > You now added this improvement to 5.15 which then needs this original
> > fix, which you correctly applied:
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-5.15/can-isotp-sanitize-can-id-checks-in-isotp_bind.patch?id=c0c51980438b5e5b0d758dc28eef5947e57c9c1e
> >
> > The other stable kernels 5.17, 5.16, 5.10 do not contain the improvement
> > and for that reason I sent an adapted fix, which Greg applied to those
> > kernels, e.g. in 5.17:
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-5.17/can-isotp-sanitize-can-id-checks-in-isotp_bind.patch?id=c0c51980438b5e5b0d758dc28eef5947e57c9c1e
> >
> > I originally sent this adapted patch to both of you on 02.04.22 11:12
> > ("[PATCH stable v5.10-v5.17] can: isotp: sanitize CAN ID checks in
> > isotp_bind()")
> >
> > So you could either leave it as-as and only the 5.15 has the improvement
> > OR you remove the two first referenced patches from the 5.15-queue and
> > add only the *adapted* "sanitized CAN ID ..." patch which is applied in
> > 5.17 and the other series. Your choice ;-)
>
> Looks like we had a mismerge in 5.15. I've dropped the feature patch and
> picked up your backport instead of the one that was in the tree.
>
> > Sorry for this big pile of patches this time :-/
>
> Fixes are good :)
>
> > Finally another hint:
> >
> > All stable trees (5.10, 5.15, 5.16, 5.17) already contain the patch
> > "can: isotp: support MSG_TRUNC flag when reading from socket" (upstream
> > commit 42bf50a1795a185).
> >
> > E.g. see
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-5.17/can-isotp-support-msg_trunc-flag-when-reading-from-s.patch?id=c0c51980438b5e5b0d758dc28eef5947e57c9c1e
> >
> > This patch is needed but unfortunately introduced a regression which is
> > fixed with this follow up patch here:
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit?id=e382fea8ae54f5bb62869c6b69b33993d43adeca
> >
> > "can: isotp: restore accidentally removed MSG_PEEK feature"
> >
> > So this last patch needs to be added to 5.10, 5.15, 5.16, 5.17 too.
>
> I've queued it up for the above trees.
Thanks for fixing up my merge mess here, much appreciated.
greg k-h
