Bug 79461

Summary: when booting, during drm switch to native resolution, graphics garbled
Product: Drivers Reporter: tomas m (tmezzadra)
Component: Video(DRI - Intel)Assignee: Damien Lespiau (damien.lespiau)
Status: RESOLVED PATCH_ALREADY_AVAILABLE    
Severity: high CC: intel-gfx-bugs
Priority: P1    
Hardware: i386   
OS: Linux   
Kernel Version: 3.16-rc1 Subsystem:
Regression: Yes Bisected commit-id:

Description tomas m 2014-07-03 19:38:56 UTC 
Hello,

This is my first approach to report this bug. With almost no technical information since my time is limited. but im willing to bisect at the correct tree (if given one).

Since 3.16-rc1 i cannot boot with a GMA945 intel graphics. the display garbles up (similar to robocop scanlines).
computer boots up, but xorg fails and i cannot start a virtual console.

i thought this would affect more people, but i guess people have newer hardware these days :(

what tree should i bisect? usually bisecting a -rc1 bug for me has been impossible. in this case i have not yet tried.

newest version working is 3.15.2

thanks!
Comment 1 Damien Lespiau 2014-07-04 15:54:20 UTC 
I guess it's worth giving git bisect a try. Maybe you could try to restrict the bisect to drivers/gpu/drm/i915 (git bisect takes a path! I never tried it but looks handy)
Comment 2 tomas m 2014-07-06 00:52:15 UTC 
Damien,

Thanks! yes, found offending commit. reverted it on 3.16-rc1 and the screen works ok again.

98ec77397a5c68ce753dc283aaa6f4742328bcdd

but. Still computer will lockup. now building a test on top of the tree.

for sure reverting that commit fixes my graphics problems.
Comment 3 tomas m 2014-07-06 14:18:23 UTC 
yes. no top of the tree computer boots ok.

so i reconfirm 98ec77397a is the offending commit. maybe you need more info from my computer to figure what is going on?
Comment 4 tomas m 2014-07-06 16:02:27 UTC 
On the other hand. testing the tree at its last commit (110e4308f89fed7be8) has the problem fixed too.

So i guess the problem was being worked after all :)

im marking this as fixed. Sorry for your inconvenience

Tomas
Comment 5 tomas m 2014-07-12 01:49:00 UTC 
Hello.

i acted too fast. apparently i dont know how to undo changes to a revert in a local git correctly. and was always testing with the revert in place.

the bug is still there on master for my hardware and of course, the mentioned commit is the culprit.

any ideas on what to test?

Thanks!

Tomas
Comment 6 tomas m 2014-08-03 22:04:28 UTC 
Hello.

Tested RC7 and it works again (this time, certain!). I dont know where the fix came. 

closing for now.

Thanks!!