Bug 12542

Summary: [915] bad initial mode
Product: xorg Reporter: Martin von Gagern <Martin.vGagern>
Component: Driver/intelAssignee: Martin von Gagern <Martin.vGagern>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium CC: hong.liu, michael.fu
Version: 7.3 (2007.09)Keywords: NEEDINFO
Hardware: x86 (IA32)   
OS: All   
URL: https://bugs.gentoo.org/show_bug.cgi?id=176501
Whiteboard:
i915 platform: i915 features:
Bug Depends on:    
Bug Blocks: 13027    
Attachments:
Description Flags
xorg.conf
none
Driver 1.7.4 with unmodified xorg.conf
none
Driver 2.1.1 with unmodified xorg.conf
none
Driver 2.1.1 with xorg.conf removed
none
Driver 2.1.1 with PreferredMode option
none
patch to ry
none
Output of xrandr after resolution change
none
Driver 2.1.1 with PreferredMode option none

Description Martin von Gagern 2007-09-24 01:46:48 UTC
This is a follow-up to bug 10860.

I have my X server configured and working with the i810 driver 1.7.4. Whenever I update to a 2.x version, be it 2.0.0 as originally reported (see URL as well) or 2.1.1 as I tried just now, I don't get my full configured resolution of 1600 x 1200, and the mode switching hotkeys (Ctrl + Alt + Plus or Minus) don't work any more, probably because there is only a single allowed mode. Removing my xorg.conf affects this behaviour, but changes it for worse. xrandr did work in previous tests, ahven't tried it this time.

Here is a short summary of my latest test runs:
Driver 1.7.4, xorg.conf unmodified: 1600 x 1200, hotkeys work
Driver 2.1.1, xorg.conf unmodified: 1152 x  768, no hotkeys
Driver 2.1.1, xorg.conf removed:    1024 x  768, no hotkeys
Driver 1.7.4, xorg.conf removed:     640 x  480, no hotkeys

I will attach my config and logs for these test runs.

Bug 11690 is another reason for me to stick with the 1.7.4 driver for now.
Comment 1 Martin von Gagern 2007-09-24 01:47:55 UTC
Created attachment 11710 [details]
xorg.conf
Comment 2 Martin von Gagern 2007-09-24 01:48:52 UTC
Created attachment 11711 [details]
Driver 1.7.4 with unmodified xorg.conf

This setup gave me a 1600 x 1200 resolution.
Comment 3 Martin von Gagern 2007-09-24 01:49:48 UTC
Created attachment 11712 [details]
Driver 2.1.1 with unmodified xorg.conf

This setup gave me a resolution of 1152 x 768.
Comment 4 Martin von Gagern 2007-09-24 01:50:55 UTC
Created attachment 11713 [details]
Driver 2.1.1 with xorg.conf removed

This setup gave me a resolution of 1024 x 768.
Comment 5 Hong Liu 2007-09-26 21:33:02 UTC
Would you please try add "option "PreferredMode" "1600x1200" " to monitor section to see if it helps?
Comment 6 Martin von Gagern 2007-09-27 00:57:12 UTC
Created attachment 11777 [details]
Driver 2.1.1 with PreferredMode option

(In reply to comment #5)
> Would you please try add "option "PreferredMode" "1600x1200" " to monitor
> section to see if it helps?

I tried this. The effect was that my screen went into power save mode and I found no way to get my display back except ssh from another machine and rebooting the system. Before that, I noticed that the Xorg process consumed all the CPU, didn't seem to execute any syscalls (according to strace), didn't respond to SIGTERM, but could be killed using SIGKILL. After killing it the monitor remained in powersave mode, but while shutting down it reactivated and displayed shutdown messages. In a second try, without killing the X server, no messages were displayed.

Perhaps I should repeat https://bugs.gentoo.org/show_bug.cgi?id=176501#c5 here and say that having a wrapper around my session startup script which calls xrandr works around the problem of the wrong resolution. Hotkeys still don't work in this setup, though.
Comment 7 Hong Liu 2007-09-27 20:02:54 UTC
(In reply to comment #6)
> Perhaps I should repeat https://bugs.gentoo.org/show_bug.cgi?id=176501#c5 here
> and say that having a wrapper around my session startup script which calls
> xrandr works around the problem of the wrong resolution. Hotkeys still don't
> work in this setup, though.
> 

What is the output of xrandr after you successfully change to 1600x1200?

Comment 8 Hong Liu 2007-09-28 01:05:01 UTC
Created attachment 11799 [details] [review]
patch to ry

Please try the attached patch which fixed the X dead problem and
retest with preferredmode option.
Comment 9 Martin von Gagern 2007-10-02 16:05:05 UTC
Created attachment 11870 [details]
Output of xrandr after resolution change

(In reply to comment #7)
> What is the output of xrandr after you successfully change to 1600x1200?

No output from the xrandr call that changes the resolution.
If I run xrandr without arguments afterwards, I get the attached list. Only difference to the list before is the selected current mode, which affects four lines of this output.
Comment 10 Martin von Gagern 2007-10-02 16:09:32 UTC
Created attachment 11871 [details]
Driver 2.1.1 with PreferredMode option

(In reply to comment #8)
> Please try the attached patch which fixed the X dead problem and
> retest with preferredmode option.

I get the correct resolution of 1600x1200, but at 60Hz instead of 85.
Setting "TargetRefresh" to "85" did not improve the situation.
Comment 11 Hong Liu 2007-10-07 19:36:27 UTC
(In reply to comment #10)
> I get the correct resolution of 1600x1200, but at 60Hz instead of 85.
> Setting "TargetRefresh" to "85" did not improve the situation.
> 

Yes, there is problem in X server for preferred mode selection. It just uses mode name (without refresh rate info) to select mode. I will try to add refresh rate info during mode selection.
Comment 12 Jesse Barnes 2007-10-31 15:28:10 UTC
The hotkey problem is being tracked in a separate bug:  #11546.  I've updated the summary for this one.
Comment 13 Michael Fu 2008-01-07 00:47:33 UTC
It seems now we've fixed the preferred mode option bug. Now move on to see the cause of "why 1600x1200 was not chosen by 2.x.x intel driver".

would you please post a xorg log file with ModeDebug option turns on?
Comment 14 Benjamin Close 2008-01-11 02:38:40 UTC
Bugzilla Upgrade Mass Bug Change

NEEDSINFO state was removed in Bugzilla 3.x, reopening any bugs previously listed as NEEDSINFO.

  - benjsc
    fd.o Wrangler
Comment 15 Michael Fu 2008-01-22 18:49:45 UTC
Martin, if still no response, we would have to reject this bug.. thanks...
Comment 16 Martin von Gagern 2008-01-23 02:43:35 UTC
(In reply to comment #15)
> Martin, if still no response, we would have to reject this bug.. thanks...

Sorry for not replying sooner. I've changed my setup here, especially my primary graphics card and monitor. The Intel chip is still on my mainboard, and the old monitor still around here somewhere, so I intend to test this nevertheless. Alas, my diploma thesis is nearing its completion, so I'd be glad if I could postpone this a month or so, as I'm pretty busy right now.
Comment 17 Michael Fu 2008-02-29 16:19:51 UTC
Martin, I'm gonna mark this bug as fixed as Keith has a patch checked in which fixed the same thing as hong did in comment# 8. Please reopen if you still meet this issue. thanks.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.