[RS480] HP Dv8000 laptop suspends but does not resume

[markginter24]

HP Dv8000 - with Intrepid the xorg-ati driver does not allow suspend/resume but the fglrx one does. Trying Jaunty and - so far - the xorg-ati driver still does not allow for suspend/resume.

Turning off DRI does not help.

ProblemType: Bug
Architecture: i386
DistroRelease: Ubuntu 9.04
Lsusb:
 Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
 Bus 001 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
MachineType: Hewlett-Packard Pavilion dv8000 (EP404UA#ABA)
Package: linux-image-2.6.28-8-generic 2.6.28-8.24
ProcCmdLine: root=UUID=282af6ad-7313-487a-8c4d-eb65037455a1 ro quiet splash
ProcEnviron:
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.28-8.24-generic
SourcePackage: linux

[lspci]
00:00.0 Host bridge [0600]: ATI Technologies Inc RS480 Host Bridge [1002:5950] (rev 01)
     Subsystem: Hewlett-Packard Company Device [103c:309b]
01:05.0 VGA compatible controller [0300]: ATI Technologies Inc Radeon XPRESS 200M 5955 (PCIE) [1002:5955]
     Subsystem: Hewlett-Packard Company Device [103c:309b]

[Tormod Volden (tormodvolden)]

Please attach your Xorg.0.log

[markginter24 (mark-hanoverbeacon)]

Attached log file Xorg.0.log

[Tormod Volden (tormodvolden)]

I added a link to an upstream bug http://bugs.freedesktop.org/show_bug.cgi?id=19581 which might be the same issue, although the card is a bit different (RC410), they are both "XPRESS 200" cards.

I guess it resumes fine if DRI is disabled? (Option "DRI" "off")

[Matteo Z (matteozandi)]

I have a X700 card, suspend works fine with fglrx and intrepid. With jaunty, my card won't be supported anymore by fglrx, so if I have to switch to ati open driver.

With ati open source driver, the laptop actually suspends, but has problems with resume. It resumes, but video is all black, if I quickly move the mouse, I can see the arrow going around but no windows/menus are present.

Turning off DRI, seems to let the computer resume correctly.

[Matteo Z (matteozandi)]

marking as confirmed since Tormod's question was answered

[Tormod Volden (tormodvolden)]

Matteo, your chipset (RV410) is quite different, so please file a new bug. BTW, I have the same card as you and it works fine.

[Matteo Z (matteozandi)]

Sorry for the noise, I filed bug #351777 and would love to hear how to deal with suspend and ati open driver.

[Tormod Volden (tormodvolden)]

markginter24, does it work if DRI is disabled? (Option "DRI" "off" in xorg.conf)

[markginter24 (mark-hanoverbeacon)]

I tried two different configurations - both with DRI off:
1. I added the DRI off line to the xorg.conf file and allowed both the sideport and shared memory turned on in the bios
2. I left the DRI off line in the xorg.conf file and turned off the shared memory option in the bios

Neither one allowed me to resume from suspend.

[John Pye (jdpipe)]

Is this bug a duplicate of bug 367822?

[Jure Repinc (jlp)]

I have an HP Compaq nx6325 laptop with integrated ATI Technologies Inc RS482 [Radeon Xpress 200M]. Recently I've updated the driver from Git and started to notice that the laptop would freeze every time when waking up from suspend to RAM. It would also hang after some switches from console to X (I've been quickly pressing Ctrl+Alt+F1 and Ctrl+Alt+F7). When this happens the laptop would switch to X and the screen would remain black and the laptop was frozen.

So I've started to bisect the Git commits and it looks like that the commit whith which this bug appears is:
radeon: fix several dpms issues
http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=7d9f643ae3d07e51e644a5979ca90bc2c102bc89
After going back one commit al is working fine again.

[Jure Repinc (jlp)]

Created an attachment (id=26511)
Xorg.0.log

[Jure Repinc (jlp)]

Created an attachment (id=26512)
xorg.conf

[Koolfy (koolfy)]

Same chipset, same problem, same behaviour. (read this as "I confirm and I can always reproduce")

[agd5f (agd5f)]

*** Bug 22134 has been marked as a duplicate of this bug. ***

[Ken Bloom (kbloom)]

This is Debian bug #529830. (http://bugs.debian.org/529830) There they've bisected it to commit 421085949e195596000e37ea6693489db7c075b5 which is exactly the same commit as the one you've bisected it to here.

[agd5f (agd5f)]

What kernel or drm modules are you using?

[Ken Bloom (kbloom)]

[bloom@little-cat-a ~]$ uname -a
Linux little-cat-a 2.6.30-1-amd64 #1 SMP Sat Jul 18 12:55:06 UTC 2009 x86_64 GNU/Linux
[bloom@little-cat-a ~]$ lsmod
Module Size Used by
snd_atiixp 17220 0
radeon 141952 2
drm 172512 3 radeon
rfkill_input 6688 0
ppdev 7800 0
parport_pc 27080 0
lp 10612 0
parport 38224 3 ppdev,parport_pc,lp
dm_snapshot 22684 0
dm_mirror 14536 0
dm_region_hash 12704 1 dm_mirror
dm_log 9924 2 dm_mirror,dm_region_hash
dm_mod 59224 3 dm_snapshot,dm_mirror,dm_log
fuse 54576 1
powernow_k8 14052 0
ide_generic 2452 0 [permanent]
snd_atiixp_modem 13004 0
snd_ac97_codec 113960 2 snd_atiixp,snd_atiixp_modem
ac97_bus 2080 1 snd_ac97_codec
snd_pcm_oss 37200 0
snd_mixer_oss 15072 1 snd_pcm_oss
snd_pcm 78504 4 snd_atiixp,snd_atiixp_modem,snd_ac97_codec,snd_pcm_oss
snd_seq_midi 6976 0
snd_rawmidi 23008 1 snd_seq_midi
snd_seq_midi_event 7696 1 snd_seq_midi
arc4 2048 2
ecb 3072 2
snd_seq 51264 2 snd_seq_midi,snd_seq_midi_event
snd_timer 21824 2 snd_pcm,snd_seq
snd_seq_device 7476 3 snd_seq_midi,snd_rawmidi,snd_seq
joydev 10992 0
snd 63880 10 snd_atiixp,snd_atiixp_modem,snd_ac97_codec,snd_pcm_oss,snd_mixer_oss,snd_pcm,snd_rawmidi,snd_seq,snd_timer,snd_seq_device
soundcore 7984 1 snd
k8temp 5152 0
yenta_socket 24940 1
serio_raw 5844 0
tifm_7xx1 6176 0
b43 122136 0
shpchp 32588 0
rfkill 12084 3 rfkill_input,b43
rng_core 4872 1 b43
snd_page_alloc 9936 3 snd_atiixp,snd_atiixp_modem,snd_pcm
mac80211 160320 1 b43
cfg80211 65096 2 b43,mac80211
rsrc_nonstatic 10224 1 yenta_socket
input_polldev 4544 1 b43
pci_hotplug 29040 1 shpchp
tifm_core 9096 1 tifm_7xx1
i2c_piix4 11040 0
i2c_core 25408 2 drm,i2c_piix4
pcspkr 2800 0
wmi 7488 0
processor 40096 2 powernow_k8
battery 7840 0
psmouse 42140 0
evdev 10448 16
button 6512 0
ac 3872 0
ext3 122544 1
jbd 47976 1 ext3
mbcache 8804 1 ext3
ide_cd_mod 29016 0
cdrom 34360 1 ide_cd_mod
ide_gd_mod 23384 3
ide_pci_generic 4964 0
ata_generic 5924 0
libata 175260 1 ata_generic
scsi_mod 158560 1 libata
sdhci_pci 8112 0
sdhci 17540 1 sdhci_pci
ssb 43664 1 b43
atiixp 4292 2
ohci_hcd 22700 0
fi...

[László Vitos (vitos-laszlo)]

Same graphic chipset on Asus X51RL. I did not do a bisect, but for me version 6.9.0 of xf86-video-ati still works, while 6.10.0 doesn't.

[Ken Bloom (kbloom)]

Laszlo Vitos, if you're seeing this between 6.9 and 6.10, then you have some other bug with the same symptoms. This bug showed up between 6.12.1 and 6.12.2.

[Jure Repinc (jlp)]

# uname -a
Linux phoenix 2.6.31-desktop-0.rc4.1mnb #1 SMP Tue Jul 28 11:39:16 EDT 2009 x86_64 AMD Turion(tm) 64 X2 Mobile Technology TL-56 GNU/Linux

# lsmod
Module Size Used by
fuse 73920 2
af_packet 25184 2
ipv6 349040 16
arc4 2336 2
ecb 3680 2
snd_hda_intel 31528 0
radeon 375904 2
b43 176560 0
snd_hda_codec 94944 1 snd_hda_intel
rng_core 5640 1 b43
mac80211 210352 1 b43
snd_hwdep 10088 1 snd_hda_codec
drm 205728 3 radeon
snd_seq_dummy 3524 0
snd_seq_oss 38240 0
snd_seq_midi_event 9376 1 snd_seq_oss
snd_seq 67872 5 snd_seq_dummy,snd_seq_oss,snd_seq_midi_event
cfg80211 110520 2 b43,mac80211
snd_seq_device 9108 3 snd_seq_dummy,snd_seq_oss,snd_seq
rfkill 24680 1 cfg80211
snd_pcm_oss 53248 0
snd_pcm 104392 3 snd_hda_intel,snd_hda_codec,snd_pcm_oss
snd_timer 27600 2 snd_seq,snd_pcm
snd_mixer_oss 20064 1 snd_pcm_oss
snd 82984 10 snd_hda_intel,snd_hda_codec,snd_hwdep,snd_seq_oss,snd_seq,snd_seq_device,snd_pcm_oss,snd_pcm,snd_timer,snd_mixer_oss
ssb 55152 1 b43
tg3 124996 0
libphy 30496 1 tg3
i2c_piix4 15088 0
i2c_core 35296 2 drm,i2c_piix4
sr_mod 19300 0
soundcore 9408 1 snd
snd_page_alloc 11376 2 snd_hda_intel,snd_pcm
sdhci_pci 9920 0
sdhci 24068 1 sdhci_pci
led_class 5704 2 b43,sdhci
pcspkr 3392 0
serio_raw 7076 0
shpchp 39708 0
sg 36544 0
...

[Bryce Harrington (bryce)]

> Is this bug a duplicate of bug 367822?
Doubtful, the hardware is quite different.

[Ken Bloom (kbloom)]

This bug has gotten worse in verison 6.12.3. When I start the X
server, the whole machine freezes, and I never even see the GDM
greeter.

Reverting 421085949e195596000e37ea6693489db7c075b5 still fixes this
bug.

[Matthijs Kooijman (matthijskooijman)]

While trying to debug Bug 23666 (which seems to be a duplicate of this bug), I've also bisected to the same commit. I've found a very reliable way of reproducing this bug: If I unplug or plug my AC adapter and then (within a few seconds) switch from text to X, the machine freezes.

Also, it seems that the first radeon driver version you load determines the behaviour. That is, during bisecting I've had a few occasions where I booted using a bad version, installed a good version, restarted X and then it would still freeze (and also the other way around). Perhaps that helps?

[Matthijs Kooijman (matthijskooijman)]

*** Bug 23666 has been marked as a duplicate of this bug. ***

[Matthijs Kooijman (matthijskooijman)]

Since the bisected commit is still a bit of a big one, I've narrowed it down a bit more. While trying to understand the code, I think I found a bug in the commit (but unfortunately, fixing it didn't solve the problem of this report). I've created a fix for this small bug, which I put up here:

  http://git.stderr.nl/gitweb?p=matthijs/upstream/xf86-video-ati.git;a=commit;h=9222a02cf7

Anyway, back to the real bug. I've further narrowed the problem down to the change "- disable the crtcs in legacy crtc dpms". This branch shows my findings:

  http://git.stderr.nl/gitweb?p=matthijs/upstream/xf86-video-ati.git;a=shortlog;h=refs/heads/200m-debug

Two commits are significant here:
Working: http://git.stderr.nl/gitweb?p=matthijs/upstream/xf86-video-ati.git;a=commit;h=e59c01d
Broken: http://git.stderr.nl/gitweb?p=matthijs/upstream/xf86-video-ati.git;a=commit;h=3c4b0090

The first has most of the faulty commit applied, the second has the final part of the faulty commit applied as well. I've also included a some debug output in the working commit that shows the Legacy DPMS mode switches and register contents before the mode switch.

I'm attaching two Xorg logfiles, one made with the working version (which shows startup and a switch to text mode and back to X) and one with the broken version (which shows startup and a switch to text mode, nothing is logged after that due to the freeze).

[Matthijs Kooijman (matthijskooijman)]

Created an attachment (id=31702)
Xorg.log of a modeswitch with the broken code

[Matthijs Kooijman (matthijskooijman)]

Created an attachment (id=31703)
Xorg.log of a modeswitch with the working code

[agd5f (agd5f)]

I've gone ahead and committed the crtc2 dpms fix:
f082b1693d6f7f763ccf5a8436a89890ca2c6129

[agd5f (agd5f)]

Created an attachment (id=31711)
only mask/set what we need

Does this patch help (on top of git master or 6.12-branch)?

[agd5f (agd5f)]

(From update of attachment 31711)
nevermind.

[Ken Bloom (kbloom)]

Commit 3d59746808bc5f335104d27a8dce0fe94ab3cb78 works for me.

[agd5f (agd5f)]

*** Bug 19581 has been marked as a duplicate of this bug. ***

[agd5f (agd5f)]

Created an attachment (id=32086)
possible fix

Does this patch fix things?

[Ken Bloom (kbloom)]

I was confused, and 3d59746808bc5f335104d27a8dce0fe94ab3cb78 didn't actually fix things for me. I'm going to test with your patch, and so as not to confuse things and not embarrass myself, I'm going to keep quiet for a couple days before concluding that it works.

[agd5f (agd5f)]

No need to patch. I've pushed the patch to master and 6.12-branch.

[Matthijs Kooijman (matthijskooijman)]

Too late, I had just patched my 6.12-branch :-)

Anyway, I just rebooted to test and the problem seems gone for me. I used to be able to reproduce this problem everytime I did a VT switch while plugging my power in or out, but no freezes happened when I did so now. I'll report back if it freezes on me again, but it seems this fix works for me.

[Ken Bloom (kbloom)]

I've been testing this for a couple days now, and haven't had any freezes, and X behaves properly. I've noticed that when I try to switch to a VT, the VT doesn't show up properly (I get a blank black screen), but when I switch back to X, X shows up properly.

I'm using Xorg 1.6.5, Debian kernel 2.6.31-1-amd64 with a non-framebuffer console (/proc/fb is an empty file).

[Launchpad Janitor (janitor)]

This bug was fixed in the package xserver-xorg-video-ati - 1:6.12.99+git20100126.e5933fd7-0ubuntu1

---------------
xserver-xorg-video-ati (1:6.12.99+git20100126.e5933fd7-0ubuntu1) lucid; urgency=low

  * New upstream git snapshot 20100126 (master) up to commit
    e5933fd7, includes:
    + [3a30210d] RS4xx: fix 200M freezes on VT switch if CRTC is disabled
      (LP: #333377, #494672)
    + Speedups for r600
    + Fixes to various dpms / incorrect resolution issues
    + Fixes to low memory EXA; fix NoAccel to work with KMS
 -- Bryce Harrington <email address hidden> Wed, 27 Jan 2010 17:15:02 -0800