acpi_video_flags Reset on Resume-From-RAM
Suspend-to-RAM used to (mostly) work on this machine, but there
seems to have been a regression between 2.6.22.1 and 2.6.22.5.
I have a ThinkPad Z61t, with an Intel Core 2 Duo and Intel 945GM
graphics. My kernel command line contains the argument, "acpi_sleep=s3_bios,s3_mode", which is the same as setting /proc/sys/kernel/acpi_video_flags to 3. This setting is suggested by ThinkWiki.org for getting the Intel graphics to recover properly after suspend-to-RAM.
Under 2.6.22.1, this arrangement worked fine. However, since I
upgraded the kernel to 2.6.22.5, the text console refuses to come back. When I inspect acpi_video_flags, the value has been reset to zero.
Sequence:
- Boot system.
- Inspect /proc/sys/kernel/acpi_video_flags -- it contains 3.
- Trigger suspend-to-RAM (done via 'hibernate-ram'). I have configured 'hibernate' to leave acpi_video_flags alone.
- Resume system. Text console does not come back. I can force X to recover the display by pressing CTRL-ALT-F1 (switch to console) followed by ALT-F7 (switch to X).
- Inspect /proc/sys/kernel/acpi_video_flags -- it contains 0.
- Force acpi_video_flags back to 3:
echo 3 > /proc/sys/kernel/acpi_video_flags
- Inspect /proc/sys/kernel/acpi_video_flags -- it contains 3.
- Trigger suspend-to-RAM.
- Resume system. Text console does not come back. acpi_video_flags contains zero again.
I'm running a custom-compiled kernel (config available upon
request). The highlights are SMP, PREEMPT, and NOHZ.
Note that nothing short of a full reboot will restore the console.
Switching to a virtual console results in a blank screen. Switching back to X so far always works. Since the console's busted, it's not immediately clear how I might diagnose this further. Suggestions?
Schwab
Quoted 30 lines · Leo L. Schwab
Suspend-to-RAM used to (mostly) work on this machine, but there
seems to have been a regression between 2.6.22.1 and 2.6.22.5.I have a ThinkPad Z61t, with an Intel Core 2 Duo and Intel 945GM
graphics. My kernel command line contains the argument, "acpi_sleep=s3_bios,s3_mode", which is the same as setting/proc/sys/kernel/acpi_video_flagsto 3. This setting is suggested by ThinkWiki.org for getting the Intel graphics to recover properly after suspend-to-RAM.Under 2.6.22.1, this arrangement worked fine. However, since I
upgraded the kernel to 2.6.22.5, the text console refuses to come back. When I inspectacpi_video_flags, the value has been reset to zero.Sequence:
- Boot system.
- Inspect/proc/sys/kernel/acpi_video_flags-- it contains 3.
- Trigger suspend-to-RAM (done via 'hibernate-ram'). I have configured 'hibernate' to leaveacpi_video_flagsalone.
- Resume system. Text console does not come back. I can force X to recover the display by pressing CTRL-ALT-F1 (switch to console) followed by ALT-F7 (switch to X).
- Inspect/proc/sys/kernel/acpi_video_flags-- it contains 0.
- Forceacpi_video_flagsback to 3:
echo 3 >/proc/sys/kernel/acpi_video_flags
- Inspect/proc/sys/kernel/acpi_video_flags-- it contains 3.
- Trigger suspend-to-RAM.
- Resume system. Text console does not come back.acpi_video_flagscontains zero again.
Quoted 7 lines
I'm running a custom-compiled kernel (config available upon
request). The highlights are SMP, PREEMPT, and NOHZ.Note that nothing short of a full reboot will restore the console.
Switching to a virtual console results in a blank screen. Switching back to X so far always works. Since the console's busted, it's not immediately clear how I might diagnose this further. Suggestions?
Try updating to 2.6.24-rc.... Or find out which patch causes the regression, and talk to its author.
Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html