Bug 18872

Summary:	[RADEON:KMS:R100:SUSPEND] suspend to ram problems
Product:	Drivers	Reporter:	Maciej Rutecki (maciej.rutecki)
Component:	Video(DRI - non Intel)	Assignee:	drivers_video-dri
Status:	RESOLVED OBSOLETE
Severity:	normal	CC:	alan, alexdeucher, delist, florian, glisse, maciej.rutecki, mate.taixi, rjw
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.39	Subsystem:
Regression:	No	Bisected commit-id:
Bug Depends on:
Bug Blocks:	7216

Description Maciej Rutecki 2010-09-20 18:46:05 UTC

Subject    : Radeon KMS r100 and STR problems
Submitter  : Richard Mittendorfer <delist@gmx.net>
Date       : 2010-09-17 18:26
Message-ID : 20100917202647.aa4b6300.delist@gmx.net
References : http://marc.info/?l=linux-kernel&m=128474802809779&w=2

This entry is being used for tracking a regression from 2.6.35. Please don't
close it until the problem is fixed in the mainline.

Comment 1 Alex Deucher 2010-09-24 14:22:26 UTC

Any chance you can track down a working kernel and bisect it?

Comment 2 Richard Mittendorfer 2010-09-24 14:37:13 UTC

Sry, no. Since I switched to KMS I was unable to get the box out of STR alive.

Comment 3 Rafael J. Wysocki 2010-09-26 20:02:10 UTC

So that's not a regression, rather the driver not working correctly on your hardware.

Comment 4 Richard Mittendorfer 2010-09-26 22:38:45 UTC

True. Not a regression. I'm using KMS since .33(?), but hesitated to report as KMS is quite new - hoping it would fix itself.. :)

Comment 5 Rafael J. Wysocki 2011-01-16 22:39:28 UTC

Is the problem still present in 2.6.37?

Comment 6 Richard Mittendorfer 2011-01-16 22:59:27 UTC

Yep, still present. On -rc8 at least, switching to final .37 next days. Ideas highly welcome.

Comment 7 Rafael J. Wysocki 2011-01-16 23:06:06 UTC

Thanks for the information.

Does it work without X (ie. if you boot into the text console)?

Comment 8 Richard Mittendorfer 2011-01-17 00:06:05 UTC

Thanks for looking into this. It's the same behavior w/ and w/o Xorg. At the STR
process the backlight goes off shortly but remains lit during suspend. The screen
goes black and stays black when out of suspend. Some flickering on/off when
entering and leaving STR. There's short harddrive activity when coming out of
suspend. Also, suspend- and power-led blink/lit as they are supposed to. The box
is then locked up, not reacting to keyboard input or even sysrq. Hard power off
required.
I've seen it coming back out of suspend halfway sane (post on lkml) but that
hardly ever happens. The only known workaround for me is to go with nomodeset
(and loose 3d acceleration as well).

Comment 9 Rafael J. Wysocki 2011-01-17 00:22:37 UTC

Does the following:

# echo core > /sys/power/pm_test
# echo mem > /sys/power/state

return to the command prompt (after about 5-10 sec) or does it hang too?

Comment 10 Richard Mittendorfer 2011-01-17 01:32:03 UTC

On .37 now with PM_DEBUG and stuff there. Above method too hangs at quite the same position (when coming out of suspend after few seconds). Nothing in syslog, dmesg pm tracing (dmesg -s 1000000 | grep 'hash matches') as well, but didn't do much testing (won't be able to test/reply for the next hrs.)

Comment 11 Maté 2011-05-23 21:10:09 UTC

Same problem with vanilla 2.6.39 on debian testing.

"At the STR process the backlight goes off shortly but remains lit during suspend. The screen goes black and stays black when out of suspend. Some flickering on/off when entering and leaving STR. There's short harddrive activity when coming out of suspend [1]. Also, suspend- and power-led blink/lit as they are supposed to. The box is then locked up, not reacting to keyboard input or even sysrq. Hard power off required."

[1] not sure about this one

There's a switch in /etc/default/acpi-support:
#RADEON_LIGHT=true

Commented or uncommented doesn't change the behavior during STR.