[G41][West Lake] Ubuntu 8.10, 9.04: System unable to detect display mode after S3 cycle and logout

[David Chen]

OS Version and Testing environment:
Ubuntu 8.10 - With pre-released kernel 2.6.27-14.33 (AGP fix) + Patched xserver-xorg-video-intel driver
Ubuntu 9.04 - Fresh installed Jaunty final release

Environment used for testing:
System with Intel G41 graphic controller

Summary:
After suspend/resume cycle, if user logs out, system will not be able to detect display mode, and prompts "Ubuntu is running at low-graphic mode".

Steps to Reproduce:
1. Power on the system and boot to the OS.
2. Enter correct username and password to login the system.
3. Put the system to sleep.
4. Resume the system.
5. Log current out - [System --> Log Out user --> Log Out].

Expected result:
Login window shows on the screen, waiting for user input.

Actual result:
Can't enter login prompt, but a message saying "Ubuntu is running at low-graphic mode".

Additional information:
* This bug is the continuation of LP#338598. Due to the lack of G41 support, system couldn't detect the supported resolution.
* Pre-release intrepid kernel and patched xorg driver has corrected the original issue.
* Both patches are in Jaunty final release by default.

[David Chen (david.chen)]

Please check attached Xorg log after the issue occurs.

[David Chen (david.chen)]

This is the original Xorg log for normal booted system.

[David Chen (david.chen)]

Followed Alberto's suggestion and added following options in xorg.conf under Device section:
Option "ForceSDVODetect" "on"
and
Option "ForceEnablePipeA" "True"

As it might be that for some reason when the computer resumes from standby it
connects to the wrong hardware pipe.

Result - Still failed

[minskey (chaohong-guo)]

Can you attach acpidump output ?

[Jerone Young (jerone)]

Can you do a vbios dump?

[minskey (chaohong-guo)]

Intel IGD uses some memory as communication channels between OS and VBIOS. We have seen some BIOS'es expect OSes to initialize some variables in that area, if OS failed to do that, some weird things might happen. So, we need the ACPI tables to check if this issue is caused by misunderstanding between OS and BIOS, is it possible for the reporter to attach acpidump output ? thanks

[David Chen (david.chen)]

After applying the following patch, this issue can no longer reproducible.

https://bugs.freedesktop.org/show_bug.cgi?id=21095#c10

[Leann Ogasawara (leannogasawara)]

Hi David,

The patch you referenced was committed as part of bug 365958 and it should be available for testing in the 2.6.28-12.43 kernel in jaunty-proposed.