Bug 42929

Summary:	GMA500: Regression: 3.3.0-rc7 - Display remains black after S3 (vga console)
Product:	Drivers	Reporter:	Dennis Jansen (dennis.jansen)
Component:	Video(DRI - Intel)	Assignee:	drivers_video-dri-intel (drivers_video-dri-intel)
Status:	RESOLVED CODE_FIX
Severity:	normal	CC:	alan, drivers_video-other, frank, lenb, rui.zhang
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	3.3.0-rc7	Subsystem:
Regression:	Yes	Bisected commit-id:
Bug Depends on:
Bug Blocks:	56331
Attachments:	dmesg after suspend lspci

Description Dennis Jansen 2012-03-14 13:08:24 UTC

It was last working in 3.2.1, in 3.3.0-rc7+ it's no longer working.

I will attach a dmesg and lspci. Let me know what else I can do to debug the issue.

PS. Responses might be a bit slow this time, sorry. Also, if I had time for a bisect, I wouldn't post this.

Comment 1 Dennis Jansen 2012-03-14 13:12:01 UTC

Important: The display stays off after trying to suspend. The system doesn't suspend, though. At least not noticeably. It is still responsive after suspend, though.

Comment 2 Dennis Jansen 2012-03-14 13:14:44 UTC

Created attachment 72582 [details]
dmesg after suspend

Comment 3 Dennis Jansen 2012-03-14 13:15:13 UTC

Created attachment 72583 [details]
lspci

Comment 4 Dennis Jansen 2012-03-14 17:17:07 UTC

Possibly related in some way to:
https://bugzilla.kernel.org/show_bug.cgi?id=42931
Where I loaded the gma500_gfx graphics driver module before suspend and suspend fails.

Comment 5 Zhang Rui 2012-03-19 08:47:56 UTC

IMO, it seems there are two issues here
1) S3 resumes immediately after suspend
2) black screen after suspend

issue 2) is probably caused by the broken gma500_gfx driver.

And issue 1) is probably caused by EHCI?

> [   17.524804] ehci_hcd 0000:00:1d.7: wake-up capability enabled by ACPI

could you please try to run 
"echo disabled > /sys/bus/pci/devices/0000:00:1d.0/power/wakeup" to disable the EHCI wakeup and see if the problem still exists?

Comment 6 Dennis Jansen 2012-03-19 09:30:55 UTC

Oh, yes, it's true. Sorry! I forgot the LID > /proc/acpi/wakeup.
So issue 1) is gone.

Comment 7 Len Brown 2012-03-20 02:23:06 UTC

re: LID > /proc/acpi/wakeup

Hmm, so the LID was causing an immediate wakeup?
Has this always been an issue and you've always used this workaround,
or is this a new problem?

re: black screen

ACPI is not responsible for resuming the screen -- the graphics driver is.
So this would be an issue with your graphics driver.
Can you describe the version that works and the version that fails?

Comment 8 Dennis Jansen 2012-03-20 07:10:15 UTC

No, this is an old problem. Since 2.6.34. See here:
https://bugzilla.kernel.org/show_bug.cgi?id=16512

Comment 9 Frank Seidel 2012-04-07 10:03:43 UTC

Having black screen issue also/still with released 3.3.0 kernel.
gma500_gfx is the gpu driver there (for the cedartrail gma3600).
Underneath / besides display not powering up all seems to work.
So to me this really could be related to #4291.

Comment 10 Alan 2012-08-31 17:06:26 UTC

Please test 3.5, that should have fixed most of these cases

Comment 11 Dennis Jansen 2012-09-16 18:08:05 UTC

It's fixed! Works great! Thanks!!!


Now the X display area is moved up about 20 pixels... opening another bug fot that. ;)