Re: [PATCH AUTOSEL for 4.9 36/56] drm/i915: Fix the level 0 max_wm hack on VLV/CHV

From: Ville Syrjälä
Date: Fri Nov 17 2017 - 07:54:00 EST


On Fri, Nov 17, 2017 at 01:41:23PM +0100, Greg KH wrote:
> On Fri, Nov 17, 2017 at 01:28:05PM +0200, Jani Nikula wrote:
> >
> > Cc: Greg
> >
> > On Wed, 15 Nov 2017, Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> wrote:
> > > On Wed, Nov 15, 2017 at 04:44:54PM +0000, alexander.levin@xxxxxxxxxxx wrote:
> > >> On Wed, Nov 15, 2017 at 01:08:05PM +0200, Ville Syrjälä wrote:
> > >> >On Wed, Nov 15, 2017 at 02:45:43AM +0000, alexander.levin@xxxxxxxxxxx wrote:
> > >> >> From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> > >> >>
> > >> >> [ Upstream commit 1be4d3793d5a93daddcd9be657c429b38ad750a3 ]
> > >> >>
> > >> >> The watermark should never exceed the FIFO size, so we need to
> > >> >> check against the current FIFO size instead of the theoretical
> > >> >> maximum when we clamp the level 0 watermark.
> > >> >>
> > >> >> Signed-off-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> > >> >> Link: https://urldefense.proofpoint.com/v2/url?u=http-3A__patchwork.freedesktop.org_patch_msgid_1480354637-2D14209-2D4-2Dgit-2Dsend-2Demail-2Dville.syrjala-40linux.intel.com&d=DwIDAw&c=udBTRvFvXC5Dhqg7UHpJlPps3mZ3LRxpb6__0PomBTQ&r=bUtaaC9mlBij4OjEG_D-KPul_335azYzfC4Rjgomobo&m=iuPtUar-VEGbH1jmVH_UTr4C02X8fmjHUfNYix-Yc0Y&s=ha_F0zP3A1Aztp5S5e6_bqdhiuuPXhn0dRWQ58vv3Is&e=
> > >> >> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@xxxxxxxxxxxxxxx>
> > >> >> Signed-off-by: Sasha Levin <alexander.levin@xxxxxxxxxxx>
> > >> >
> > >> >Why are these patches being proposed for stable? They're not straight up
> > >> >fixes for known issues, and there's always a chance that something will
> > >> >break. Who is doing the qa on this?
> > >>
> > >> Hi Ville,
> > >>
> > >> They were selected automatically as part of a new process we're trying
> > >> out. If you disagree with the selection I'd be happy to drop it.
> > >
> > > How does that automatic process decide that a patch should be backported?
> > >
> > > drm and i915 are very fast moving targets so unintended side effects from
> > > backported patches is a real possibility. So I would recommend against
> > > backporting anything that isn't fixing a real issue affecting users. We
> > > do try to add the cc:stable to such patches.
> >
> > Agreed.
> >
> > First, I think an automatic backport process is against the stable
> > kernel rules (e.g. "It must fix a real bug that bothers people").
>
> It's finding lots of fixes that did bother people enough to submit a fix
> for.
>
> > Second, we can't and won't take any responsibility for backports we
> > didn't indicate with Cc: stable, a Fixes: tag, or a specific backport
> > request.
>
> Ok, you all are already totally messing with my normal stable workflow,
> so might as well just trust you all completely. So let's just only take
> patches that you all do send me in the normal way. It's easy for Sasha
> to filter out the drm/i915 patches from his results.
>
> Is that ok?
>
> > If you think there's a commit that should be backported and is known to
> > fix a user visible issue (as per the stable rules!), please check with
> > us first.
>
> Um, that is what he was doing with the cc: of you all on the patch
> itself that started this whole conversation...

And what were the user visible issues fixed by those backports? We
can't judge the risk/benefit ratio of a backport without knowing what
is supposedly being fixed.

--
Ville Syrjälä
Intel OTC