Skip to contentSkip to conversation

PROBLEM: i915 modesetting - weird offset graphics (v2.6.37-rc1-27-gff8b16d)

DISCUSSION Jon Masters 4 participants 10 messages Nov 4, 2010 – Nov 11, 2010

lore
Share & tools
lore
#0 lore raw

Folks,

I upgraded the userspace on my EeePC 1015PEM netbook to the latest Fedora (F15) rawhide and booted a custom kernel in order to test both the RC1 and also see if the now in-staging brcm80211 behaves better (it doesn't, it still can't survive suspend at all, separate issue though).

The xrandr output is identical in both case, showing a mode of 1024x600 being the one that X is currently using. Booting with i915 modesetting disabled does avoid the weird offset but of course X doesn't start. I can bisect this if there isn't already some suggestions from the audience, in particular is to what debug information you need.

Kernel: v2.6.37-rc1-27-gff8b16d

Photos:

	http://www.flickr.com/photos/jonmasters/5145351220/
	http://www.flickr.com/photos/jonmasters/5144748543/

Said system will likely be at plumbers in the am...

Jon.

#1 lore raw
Quoted 11 lines · Jon Masters

I upgraded the userspace on my EeePC 1015PEM netbook to the latest Fedora (F15) rawhide and booted a custom kernel in order to test both the RC1 and also see if the now in-staging brcm80211 behaves better (it doesn't, it still can't survive suspend at all, separate issue though).

The xrandr output is identical in both case, showing a mode of 1024x600 being the one that X is currently using. Booting with i915 modesetting disabled does avoid the weird offset but of course X doesn't start. I can bisect this if there isn't already some suggestions from the audience, in particular is to what debug information you need.

Quoted 5 lines

Kernel: v2.6.37-rc1-27-gff8b16d

Photos:

	http://www.flickr.com/photos/jonmasters/5145351220/
	http://www.flickr.com/photos/jonmasters/5144748543/

Video:

	http://www.youtube.com/watch?v=43jc9_7n3kU

I'll bring it in to Plumbers in a few hours.

Jon.

#2 lore raw
Quoted 15 lines · Jon Masters
Quoted 13 lines · Jon Masters

I upgraded the userspace on my EeePC 1015PEM netbook to the latest Fedora (F15) rawhide and booted a custom kernel in order to test both the RC1 and also see if the now in-staging brcm80211 behaves better (it doesn't, it still can't survive suspend at all, separate issue though).

The xrandr output is identical in both case, showing a mode of 1024x600 being the one that X is currently using. Booting with i915 modesetting disabled does avoid the weird offset but of course X doesn't start. I can bisect this if there isn't already some suggestions from the audience, in particular is to what debug information you need.

Kernel: v2.6.37-rc1-27-gff8b16d

Quoted 8 lines
Quoted 3 lines

Photos:

	http://www.flickr.com/photos/jonmasters/5145351220/
	http://www.flickr.com/photos/jonmasters/5144748543/

Video:

	http://www.youtube.com/watch?v=43jc9_7n3kU

I'll bring it in to Plumbers in a few hours.

I took it to Plumbers and pinged a few people in person, but nobody was all that interested. I am bisecting this at the moment, a few builds in.

Jon.

#3 lore raw
In reply to Jon Masters
Quoted 10 lines · Jon Masters

I upgraded the userspace on my EeePC 1015PEM netbook to the latest Fedora (F15) rawhide and booted a custom kernel in order to test both the RC1 and also see if the now in-staging brcm80211 behaves better (it doesn't, it still can't survive suspend at all, separate issue though).

The xrandr output is identical in both case, showing a mode of 1024x600 being the one that X is currently using. Booting with i915 modesetting disabled does avoid the weird offset but of course X doesn't start. I can bisect this if there isn't already some suggestions from the audience, in particular is to what debug information you need.

Also, the intel-gfx list is moderated with auto-rejection. Do I really need to subscribe to the list just to get a mail through?

Jon.

#4 lore raw
Quoted 2 lines · Jon Masters

Also, the intel-gfx list is moderated with auto-rejection. Do I really need to subscribe to the list just to get a mail through?

A few of the Intel gfx engineers are at Plumbers. If you tell Jesse about it, he may be able to convince the powers-that-be to change the policy. -Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#5 lore raw
Quoted 6 lines · Chris Wilson
Quoted 2 lines · Jon Masters

Also, the intel-gfx list is moderated with auto-rejection. Do I really need to subscribe to the list just to get a mail through?

A few of the Intel gfx engineers are at Plumbers. If you tell Jesse about it, he may be able to convince the powers-that-be to change the policy.

I didn't see this mail in time. I did show a few folks the corruption, ran the kernel with drm.debug=4, etc. and checked the modelines. xrandr and the kernel seem to report the same modes each time, but obviously the display is shifted. I see various changes in git (everything from EDID reading to actual modeline changes in the i915 code - various weird margin settings seem to exist that might well cause some offset in the display) that could be responsible, so I will bisect it now for you.

My concern with the email list is that it's listed in the MAINTAINERS file, but it's essentially of no value to bug reports unless they are already subscribed to it (moderation is fine, even auto reject after 3 days - a standard mm feature, but I feel auto-reject is a little too harsh). Needless to say, I'm subscribed now :)

Thanks,

Jon.

#6 lore raw
Quoted 8 lines · Jon Masters
Quoted 6 lines · Chris Wilson
Quoted 2 lines · Jon Masters

Also, the intel-gfx list is moderated with auto-rejection. Do I really need to subscribe to the list just to get a mail through?

A few of the Intel gfx engineers are at Plumbers. If you tell Jesse about it, he may be able to convince the powers-that-be to change the policy.

Quoted 7 lines

I didn't see this mail in time. I did show a few folks the corruption, ran the kernel with drm.debug=4, etc. and checked the modelines. xrandr and the kernel seem to report the same modes each time, but obviously the display is shifted. I see various changes in git (everything from EDID reading to actual modeline changes in the i915 code - various weird margin settings seem to exist that might well cause some offset in the display) that could be responsible, so I will bisect it now for you.

Bisection is running. Should post something over the weekend.

Jon.

#7 lore raw
In reply to Chris Wilson
Quoted 7 lines · On 2010-11-04 13:41-0000 Chris Wilson wrote:
Quoted 2 lines · Jon Masters

Also, the intel-gfx list is moderated with auto-rejection. Do I really need to subscribe to the list just to get a mail through?

A few of the Intel gfx engineers are at Plumbers. If you tell Jesse about it, he may be able to convince the powers-that-be to change the policy. -Chris

I think the present list policy of posting only allowed for subscriber's only is a good one. My experience with some mailing lists for particular SourceForge projects that I admin is if you allow unsubscribed users to post, then lots of spam tends to gets posted despite every effort by SourceForge to filter out spam. Just out of curiosity I have forwarded failing spam attempts for all SourceForge mailing lists I admin to my own personal e-mail account to track what is going on. Most lists I admin have no spam attacks on them, but the most active of them (plplot-devel) has thousands of thwarted attempts to spam to it over say the last year. So it appears the spammers keep track of which mailing lists are active and don't waste their time on relatively inactive lists.

However, it should be pointed out that this list is more active than plplot-devel so I assume the spamming would start quite quickly if the present list policy were changed. Or this list may already be under lots of spam attacks which are thwarted by the present policy.

Such spam shouldn't matter that much to list subscribers if they have good spam filters (like I do with spam_bayes), but spamming does tend to fill up your mailing-list archive with a load of crap which makes it more difficult to browse or search that archive.

In sum, spammers appear to be generally lazy so that the small barrier to entry of demanding all posts be from the poster's subscription address continues to be an amazingly good tactic against spam in my experience.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); PLplot scientific plotting software package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net).
__________________________

Linux-powered Science
__________________________

#8 lore raw
In reply to Jon Masters
Quoted 13 lines · On czwartek, 4 listopada 2010 o 08:25:58 Jon Masters wrote:

Folks,

I upgraded the userspace on my EeePC 1015PEM netbook to the latest Fedora (F15) rawhide and booted a custom kernel in order to test both the RC1 and also see if the now in-staging brcm80211 behaves better (it doesn't, it still can't survive suspend at all, separate issue though).

The xrandr output is identical in both case, showing a mode of 1024x600 being the one that X is currently using. Booting with i915 modesetting disabled does avoid the weird offset but of course X doesn't start. I can bisect this if there isn't already some suggestions from the audience, in particular is to what debug information you need.

Quoted 8 lines

Kernel: v2.6.37-rc1-27-gff8b16d

Photos:

	http://www.flickr.com/photos/jonmasters/5145351220/
	http://www.flickr.com/photos/jonmasters/5144748543/

Said system will likely be at plumbers in the am...

I created a Bugzilla entry at
https://bugzilla.kernel.org/show_bug.cgi?id=22712
for your bug report, please add your address to the CC list in there, thanks!

Signature · 2 lines
Maciej Rutecki
http://www.maciek.unixy.pl
#9 lore raw
Quoted 20 lines · Maciej Rutecki
Quoted 19 lines · On czwartek, 4 listopada 2010 o 08:25:58 Jon Masters wrote:

Folks,

I upgraded the userspace on my EeePC 1015PEM netbook to the latest Fedora (F15) rawhide and booted a custom kernel in order to test both the RC1 and also see if the now in-staging brcm80211 behaves better (it doesn't, it still can't survive suspend at all, separate issue though).

The xrandr output is identical in both case, showing a mode of 1024x600 being the one that X is currently using. Booting with i915 modesetting disabled does avoid the weird offset but of course X doesn't start. I can bisect this if there isn't already some suggestions from the audience, in particular is to what debug information you need.

Kernel: v2.6.37-rc1-27-gff8b16d

Photos:

	http://www.flickr.com/photos/jonmasters/5145351220/
	http://www.flickr.com/photos/jonmasters/5144748543/
Quoted 5 lines
Quoted 2 lines

Said system will likely be at plumbers in the am...

I created a Bugzilla entry at
https://bugzilla.kernel.org/show_bug.cgi?id=22712
for your bug report, please add your address to the CC list in there, thanks!

Thanks. I requested the k.o BZ to send me a new password over the weekend, which was delayed by a few hours and then I didn't get around to this. FWIW, the bug has a fix on the intel-gfx list.

Jon.