3.4.10 i915 [GM45] regression

3.4.10 i915 [GM45] regression

[Andreas Sturmlechner]

Hello,

please revert commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 - for some 
reason it breaks boot-up (screen stays blank) on my GM45 hardware despite >= 
3.5 being fine.

See bug 54575 on freedesktop.org for a full description: 
https://bugs.freedesktop.org/show_bug.cgi?id=54575

No idea how far-spread or not this might be since I couldn't find any other 
reports of that kind.

Best regards,
Andreas

Re: 3.4.10 i915 [GM45] regression

[Greg KH]

On Fri, Sep 07, 2012 at 02:27:12AM +0200, Andreas Sturmlechner wrote:
> Hello,
> 
> please revert commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 - for some 
> reason it breaks boot-up (screen stays blank) on my GM45 hardware despite >= 
> 3.5 being fine.
> 
> See bug 54575 on freedesktop.org for a full description: 
> https://bugs.freedesktop.org/show_bug.cgi?id=54575
> 
> No idea how far-spread or not this might be since I couldn't find any other 
> reports of that kind.

Can you cc: this to the people who wrote that patch?  That might help in
figuring out what is wrong here...

thanks,

greg k-h

Re: Re: 3.4.10 i915 [GM45] regression

[Andreas Sturmlechner]

> On Fri, Sep 07, 2012 at 02:27:12AM +0200, Andreas Sturmlechner wrote:
> > Hello,
> > 
> > please revert commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 - for some 
> > reason it breaks boot-up (screen stays blank) on my GM45 hardware despite 
> > >= 3.5 being fine.
> > 
> > See bug 54575 on freedesktop.org for a full description: 
> > https://bugs.freedesktop.org/show_bug.cgi?id=54575
> > 
> > No idea how far-spread or not this might be since I couldn't find any
> > other reports of that kind.
> 
> Can you cc: this to the people who wrote that patch?  That might help in
> figuring out what is wrong here...
> 
> thanks,
> 
> greg k-h

done as intended, hopefully

Re: Re: 3.4.10 i915 [GM45] regression

[Daniel Vetter]

On Fri, Sep 7, 2012 at 3:57 AM, Andreas Sturmlechner
<andreas.sturmlechner@gmail.com> wrote:
>> On Fri, Sep 07, 2012 at 02:27:12AM +0200, Andreas Sturmlechner wrote:
>> > Hello,
>> >
>> > please revert commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 - for some
>> > reason it breaks boot-up (screen stays blank) on my GM45 hardware despite
>> > >= 3.5 being fine.
>> >
>> > See bug 54575 on freedesktop.org for a full description:
>> > https://bugs.freedesktop.org/show_bug.cgi?id=54575
>> >
>> > No idea how far-spread or not this might be since I couldn't find any
>> > other reports of that kind.
>>
>> Can you cc: this to the people who wrote that patch?  That might help in
>> figuring out what is wrong here...

drm/i915 cat-herder here - I have no idea what's going on :( Things
seem to work on 3.5, I haven't gotten any regression reports for gm45
on 3.5 (and these boxes are rather popular). And all the people who
reported issues with this patch on 3.4 had no such problems on 3.5
(with the same patch applied).

And I've looked again at the patches in that area and don't see what
could be missing. Hence I guess we should just revert this patch from
3.4.

Thanks, Daniel
-- 
Daniel Vetter
daniel.vetter@ffwll.ch - +41 (0) 79 364 57 48 - http://blog.ffwll.ch

Re: Re: 3.4.10 i915 [GM45] regression

[Herton Ronaldo Krzesinski]

On Fri, Sep 07, 2012 at 11:37:27AM +0200, Daniel Vetter wrote:
> On Fri, Sep 7, 2012 at 3:57 AM, Andreas Sturmlechner
> <andreas.sturmlechner@gmail.com> wrote:
> >> On Fri, Sep 07, 2012 at 02:27:12AM +0200, Andreas Sturmlechner wrote:
> >> > Hello,
> >> >
> >> > please revert commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 - for some
> >> > reason it breaks boot-up (screen stays blank) on my GM45 hardware despite
> >> > >= 3.5 being fine.
> >> >
> >> > See bug 54575 on freedesktop.org for a full description:
> >> > https://bugs.freedesktop.org/show_bug.cgi?id=54575
> >> >
> >> > No idea how far-spread or not this might be since I couldn't find any
> >> > other reports of that kind.
> >>
> >> Can you cc: this to the people who wrote that patch?  That might help in
> >> figuring out what is wrong here...
> 
> drm/i915 cat-herder here - I have no idea what's going on :( Things
> seem to work on 3.5, I haven't gotten any regression reports for gm45
> on 3.5 (and these boxes are rather popular). And all the people who
> reported issues with this patch on 3.4 had no such problems on 3.5
> (with the same patch applied).
> 
> And I've looked again at the patches in that area and don't see what
> could be missing. Hence I guess we should just revert this patch from
> 3.4.

Couldn't it be that it happens because 3.4 doesn't have commits
b7884eb45ec98c0d34c7f49005ae9d4b4b4e38f6 and
f01db988ef6f6c70a6cc36ee71e4a98a68901229 applied? May be they are needed
after all.

Andreas, also make sure your hardware in the bios is configured within
the memory types supported by GM45 (DDR2-667/800, DDR3-800/1066, from
Intel website listed specifications). I had a similar problem on a G41
machine with same commit, with a pair of DDR3-1333 memory installed it
was configured to run at 1333, but the chipset only supports up to 1066,
when I forced it back to DDR3-1066 in the bios, the problem vanished.
"Funny" that commit 0d8957c8a9 triggered the issue reliably.

> 
> Thanks, Daniel
> -- 
> Daniel Vetter
> daniel.vetter@ffwll.ch - +41 (0) 79 364 57 48 - http://blog.ffwll.ch
> --
> To unsubscribe from this list: send the line "unsubscribe stable" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

-- 
[]'s
Herton

Re: Re: 3.4.10 i915 [GM45] regression

[Herton Ronaldo Krzesinski]

On Mon, Sep 10, 2012 at 10:58:03AM -0300, Herton Ronaldo Krzesinski wrote:
> Couldn't it be that it happens because 3.4 doesn't have commits
> b7884eb45ec98c0d34c7f49005ae9d4b4b4e38f6 and
> f01db988ef6f6c70a6cc36ee71e4a98a68901229 applied? May be they are needed
> after all.

Nevermind, the description on the freedesktop bug report states that
these commits were already tried and didn't work (if "3.4.10 + follow-up
patches in Herton's reply - failed" means it)

-- 
[]'s
Herton

Re: Re: 3.4.10 i915 [GM45] regression

[Greg KH]

On Mon, Sep 10, 2012 at 11:07:38AM -0300, Herton Ronaldo Krzesinski wrote:
> On Mon, Sep 10, 2012 at 10:58:03AM -0300, Herton Ronaldo Krzesinski wrote:
> > Couldn't it be that it happens because 3.4 doesn't have commits
> > b7884eb45ec98c0d34c7f49005ae9d4b4b4e38f6 and
> > f01db988ef6f6c70a6cc36ee71e4a98a68901229 applied? May be they are needed
> > after all.
> 
> Nevermind, the description on the freedesktop bug report states that
> these commits were already tried and didn't work (if "3.4.10 + follow-up
> patches in Herton's reply - failed" means it)

Did anything ever come from this?  I haven't reverted anything, or
applied anything different, so if something needs to be done here,
please let me know.

thanks,

greg k-h

Re: Re: 3.4.10 i915 [GM45] regression

[Herton Ronaldo Krzesinski]

On Wed, Sep 26, 2012 at 04:42:23PM -0700, Greg KH wrote:
> On Mon, Sep 10, 2012 at 11:07:38AM -0300, Herton Ronaldo Krzesinski wrote:
> > On Mon, Sep 10, 2012 at 10:58:03AM -0300, Herton Ronaldo Krzesinski wrote:
> > > Couldn't it be that it happens because 3.4 doesn't have commits
> > > b7884eb45ec98c0d34c7f49005ae9d4b4b4e38f6 and
> > > f01db988ef6f6c70a6cc36ee71e4a98a68901229 applied? May be they are needed
> > > after all.
> > 
> > Nevermind, the description on the freedesktop bug report states that
> > these commits were already tried and didn't work (if "3.4.10 + follow-up
> > patches in Herton's reply - failed" means it)
> 
> Did anything ever come from this?  I haven't reverted anything, or
> applied anything different, so if something needs to be done here,
> please let me know.

I think the conclusion looking at the bug report
(https://bugs.freedesktop.org/show_bug.cgi?id=54575) is
that the commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 should be 
reverted on 3.4, please correct me if I'm wrong.

> 
> thanks,
> 
> greg k-h
> 

-- 
[]'s
Herton

Re: Re: 3.4.10 i915 [GM45] regression

[Greg KH]

On Thu, Sep 27, 2012 at 03:59:18PM -0300, Herton Ronaldo Krzesinski wrote:
> On Wed, Sep 26, 2012 at 04:42:23PM -0700, Greg KH wrote:
> > On Mon, Sep 10, 2012 at 11:07:38AM -0300, Herton Ronaldo Krzesinski wrote:
> > > On Mon, Sep 10, 2012 at 10:58:03AM -0300, Herton Ronaldo Krzesinski wrote:
> > > > Couldn't it be that it happens because 3.4 doesn't have commits
> > > > b7884eb45ec98c0d34c7f49005ae9d4b4b4e38f6 and
> > > > f01db988ef6f6c70a6cc36ee71e4a98a68901229 applied? May be they are needed
> > > > after all.
> > > 
> > > Nevermind, the description on the freedesktop bug report states that
> > > these commits were already tried and didn't work (if "3.4.10 + follow-up
> > > patches in Herton's reply - failed" means it)
> > 
> > Did anything ever come from this?  I haven't reverted anything, or
> > applied anything different, so if something needs to be done here,
> > please let me know.
> 
> I think the conclusion looking at the bug report
> (https://bugs.freedesktop.org/show_bug.cgi?id=54575) is
> that the commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 should be 
> reverted on 3.4, please correct me if I'm wrong.

Daniel, do you agree with this?

thanks,

greg k-h

Re: Re: Re: 3.4.10 i915 [GM45] regression

[Andreas Sturmlechner]

> 
> Did anything ever come from this?  I haven't reverted anything, or
> applied anything different, so if something needs to be done here,
> please let me know.
> 

I haven't tried 3.4.11 yet, I was only coming back from 3.5 in the hunt for a different bug, so - with a lot of bad coincidence - it is 
not entirely impossible that something else is actually causing the black screen, because it doesn't happen all the time. On your request 
I could give it some more testing time.

Re: Re: Re: 3.4.10 i915 [GM45] regression

[Andreas Sturmlechner]

> 
> I haven't tried 3.4.11 yet

OK, first two tries just now, instant black screens:

- 3.4.11
- 3.0.42

Re: Re: 3.4.10 i915 [GM45] regression

[Daniel Vetter]

On Thu, Sep 27, 2012 at 10:25 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
>> I think the conclusion looking at the bug report
>> (https://bugs.freedesktop.org/show_bug.cgi?id=54575) is
>> that the commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 should be
>> reverted on 3.4, please correct me if I'm wrong.
>
> Daniel, do you agree with this?

Yeah, seems like the sanest thing to do given the situation and lack
of any understanding what's actually going on.

/me goes back to screaming silently
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

Re: Re: 3.4.10 i915 [GM45] regression

[Greg KH]

On Fri, Sep 28, 2012 at 08:16:01AM +0200, Daniel Vetter wrote:
> On Thu, Sep 27, 2012 at 10:25 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> >> I think the conclusion looking at the bug report
> >> (https://bugs.freedesktop.org/show_bug.cgi?id=54575) is
> >> that the commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 should be
> >> reverted on 3.4, please correct me if I'm wrong.
> >
> > Daniel, do you agree with this?
> 
> Yeah, seems like the sanest thing to do given the situation and lack
> of any understanding what's actually going on.

Ok, now reverted, thanks.

greg k-h

Re: Re: 3.4.10 i915 [GM45] regression

[Andreas Sturmlechner]

> Ok, now reverted, thanks.
> 
> greg k-h

The same patch also breaks 3.0 for me. A 3.0.42 image with the patch reverted was fine, I checked with several restarts.

Sort of confirms there is something in >=3.5 that makes this thing work.

Regards

Re: Re: 3.4.10 i915 [GM45] regression

[Greg KH]

On Sun, Sep 30, 2012 at 01:15:38PM +0200, Andreas Sturmlechner wrote:
> > Ok, now reverted, thanks.
> > 
> > greg k-h
> 
> The same patch also breaks 3.0 for me. A 3.0.42 image with the patch
> reverted was fine, I checked with several restarts.
> 
> Sort of confirms there is something in >=3.5 that makes this thing work.

Ok, Daniel, should I also revert this on 3.0.y?

thanks,

greg k-h

Re: Re: 3.4.10 i915 [GM45] regression

[Daniel Vetter]

On Sun, Sep 30, 2012 at 4:32 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Sun, Sep 30, 2012 at 01:15:38PM +0200, Andreas Sturmlechner wrote:
>> > Ok, now reverted, thanks.
>> >
>> > greg k-h
>>
>> The same patch also breaks 3.0 for me. A 3.0.42 image with the patch
>> reverted was fine, I checked with several restarts.
>>
>> Sort of confirms there is something in >=3.5 that makes this thing work.
>
> Ok, Daniel, should I also revert this on 3.0.y?

/me cries

Yes.

Andreas, can you please check whether 3.5.x works for you, just to
make sure that this particular elephant is only wreaking havoc in 3.4
and earlier?

Thanks, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

Re: Re: 3.4.10 i915 [GM45] regression

[Andreas Sturmlechner]

> Andreas, can you please check whether 3.5.x works for you, just to
> make sure that this particular elephant is only wreaking havoc in 3.4
> and earlier?

I've been using 3.5 all the time since .0, I never had any trouble there - a few reboots just now with a vanilla 3.5.4 image were 
successful as expected. Meanwhile I have upgraded to 3.6, zero issues as well.

Regards

Re: Re: 3.4.10 i915 [GM45] regression

[Daniel Vetter]

On Mon, Oct 1, 2012 at 11:56 PM, Andreas Sturmlechner
<andreas.sturmlechner@gmail.com> wrote:
>> Andreas, can you please check whether 3.5.x works for you, just to
>> make sure that this particular elephant is only wreaking havoc in 3.4
>> and earlier?
>
> I've been using 3.5 all the time since .0, I never had any trouble there - a few reboots just now with a vanilla 3.5.4 image were
> successful as expected. Meanwhile I have upgraded to 3.6, zero issues as well.

So 3.5 and later really don't seem to have any issue with this patch.
I'm as puzzled as ever. Thanks for testing this again.

Cheers, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

Re: Re: 3.4.10 i915 [GM45] regression

[Greg KH]

On Tue, Oct 02, 2012 at 09:17:22AM +0200, Daniel Vetter wrote:
> On Mon, Oct 1, 2012 at 11:56 PM, Andreas Sturmlechner
> <andreas.sturmlechner@gmail.com> wrote:
> >> Andreas, can you please check whether 3.5.x works for you, just to
> >> make sure that this particular elephant is only wreaking havoc in 3.4
> >> and earlier?
> >
> > I've been using 3.5 all the time since .0, I never had any trouble there - a few reboots just now with a vanilla 3.5.4 image were
> > successful as expected. Meanwhile I have upgraded to 3.6, zero issues as well.
> 
> So 3.5 and later really don't seem to have any issue with this patch.
> I'm as puzzled as ever. Thanks for testing this again.

I forgot to revert this for 3.4, I'll do so in the next 3.4-stable
release after this one, thanks.

greg k-h

Re: Re: 3.4.10 i915 [GM45] regression

[Andreas Sturmlechner]

> So 3.5 and later really don't seem to have any issue with this patch.
> I'm as puzzled as ever. Thanks for testing this again.
> 

Bad news from 3.6: I got a blank screen there too, yesterday. Maybe it had even happened in 3.5 once before I got into testing all those 3.4/3.0 images - at my weekend setup there's an external DVI display that won't be fired up by the X200s anyway if the timing isn't right, so that new bug could go by unnoticed there.

So it seems the trouble still exists in >=3.5, but it happens way, way less often (my laptop boots once or twice per day). I should get my other system back up running to be able to try and SSH into the laptop next time this happens at home - just in case I could provide any useful output?

Best regards

Re: Re: 3.4.10 i915 [GM45] regression

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 757 bytes --]

On Fri, 2012-09-28 at 08:16 +0200, Daniel Vetter wrote:
> On Thu, Sep 27, 2012 at 10:25 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> >> I think the conclusion looking at the bug report
> >> (https://bugs.freedesktop.org/show_bug.cgi?id=54575) is
> >> that the commit 0d8957c8a90bbb5d34fab9a304459448a5131e06 should be
> >> reverted on 3.4, please correct me if I'm wrong.
> >
> > Daniel, do you agree with this?
> 
> Yeah, seems like the sanest thing to do given the situation and lack
> of any understanding what's actually going on.

So presumably this will also need to be reverted for 3.0 and 3.2?  Or is
there something very weird about i915 in 3.4?

Ben.

-- 
Ben Hutchings
You can't have everything.  Where would you put it?

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]