LP:919350 Regression Commit Identified

[Herton Ronaldo Krzesinski]

On Thu, Jan 26, 2012 at 01:41:48PM -0700, Tim Gardner wrote:
> On 01/26/2012 01:18 PM, Joseph Salisbury wrote:
> >Hi Tim and Leann,
> >
> >The commit that caused the regression for bug 919350[1] has been
> >identified:
> >
> >e9925217e61dd3594f81b415f2b7e077426f208f

The same change on oneiric is causing an issue on Precise as well, but a
different one (reported to be on resume from suspend):
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/917330

So 3.2 also presents a regression, but in a different way, and is
already being worked upstream to be fixed (reported also here:
https://lkml.org/lkml/2012/1/21/66, initial fix here:
http://www.mail-archive.com/intel-gfx@lists.freedesktop.org/msg07955.html)

The reports on bug 919350 is a different regression (no screen on boot),
and people there says that 3.2 works fine. I didn't find any problem in
the backport, except that it still uses (doesn't change)
dev_priv->edp.bpp, while in upstream this code changed. I guess this bpp
is low to make the link value less than it should be, resulting in the
failure later (wrong link * lanes selection).

May be the backport for 3.0 requires also commits
858fa03527ded333dc5701f546bd5d1b5d7515ad and
89c6143263ef8e14e42e17324a234418d8030b10 (with this last one changed
later to not use intel_crtc->bpp as it causes the later resume from
suspend issue in 3.2).

But I tend that we should just revert the backport from oneiric, even if
bringing back bug 899598 (better a low resolution than no screen at all),
we are already bringing to many changes to a stable release and
diverging. Let me know what you think, I can send a revert for the
oneiric kernel, or ask reporters on 919350 to try the backport again
plus the additional changes.

> >
> >What do you suggest as the next step? Should I notify the patch creator,
> >privately, or maybe on LKML?
> >
> >Should I create a formatted patch and request the commit get reverted on
> >the Ubuntu Kernel Team mailing list or maybe request someone review the
> >commit? I haven't looked too closely at the changes in the commit code
> >yet, but I will.
> >
> >Also, I've been bisecting a few other bugs as well. For example, I
> >believe the commit that caused the regression in bug 917962 [2] has been
> >identified as well - just waiting for the bug commenters to test. So
> >whatever process you recommend, I'll use going forward for these types
> >of bugs.
> >
> >
> >Thanks,
> >
> >Joe
> >
> >
> >[1] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/919350
> >[2] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/917962
> 
> Maybe you should hassle the upstream dudes about this one. Also,
> since its a backport, take a really close look to make sure there
> were no errors.
> 
> Cc'ing u-k-t and the stable team since this is an Oneiric regression.
> 
> rtg
> -- 
> Tim Gardner tim.gardner at canonical.com
> 

-- 
[]'s
Herton