You are not logged in.

#1 2014-01-27 10:13:38

Andreaskem
Member
Registered: 2013-10-13
Posts: 67

Problems on 3.13 (AMD HDMI?)

Hi,

I wanted to test kernel 3.13 in [testing] yesterday and had problems. The first sign was that my system refused to shut down due to a hung kernel task. (Unfortunately, my kernel was tainted due to the r8168 module. I will probably try again with an untainted kernel tonight.)

Jan 27 00:05:05 thinklinux kernel: INFO: task kworker/u8:53:3998 blocked for more than 120 seconds.
Jan 27 00:05:05 thinklinux kernel:       Tainted: G           O 3.13.0-1-ARCH #1
Jan 27 00:05:05 thinklinux kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Jan 27 00:05:05 thinklinux kernel: kworker/u8:53   D 0000000000000246     0  3998      2 0x00000000
Jan 27 00:05:05 thinklinux kernel: Workqueue: hd-audio0 hdmi_repoll_eld [snd_hda_codec_hdmi]
Jan 27 00:05:05 thinklinux kernel:  ffff88000b6a9b68 0000000000000046 ffff88000b52d100 ffff88000b6a9fd8
Jan 27 00:05:05 thinklinux kernel:  0000000000014440 0000000000014440 ffff88000b52d100 ffff88000b6a9ac0
Jan 27 00:05:05 thinklinux kernel:  ffffffff812c0314 ffff8800db0c5000 0000000000000001 ffff8800db6b5a00
Jan 27 00:05:05 thinklinux kernel: Call Trace:
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff812c0314>] ? delay_tsc+0x64/0xc0
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff812c01ef>] ? __delay+0xf/0x20
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff812c022e>] ? __const_udelay+0x2e/0x30
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa058b031>] ? snd_hda_power_save+0x71/0x2f0 [snd_hda_codec]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa058b381>] ? codec_exec_verb+0xd1/0x290 [snd_hda_codec]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff81515ac9>] schedule+0x29/0x70
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff81515ec6>] schedule_preempt_disabled+0x16/0x20
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff8151831d>] __mutex_lock_slowpath+0x13d/0x390
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff81518582>] mutex_lock+0x12/0x22
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa0cd23a9>] hdmi_present_sense+0x49/0x360 [snd_hda_codec_hdmi]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff8151813e>] ? mutex_unlock+0xe/0x10
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa0cd27fa>] generic_hdmi_resume+0x5a/0x70 [snd_hda_codec_hdmi]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa058af19>] hda_call_codec_resume+0x179/0x1e0 [snd_hda_codec]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa058b1d8>] snd_hda_power_save+0x218/0x2f0 [snd_hda_codec]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa058b317>] codec_exec_verb+0x67/0x290 [snd_hda_codec]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa058b59f>] snd_hda_codec_read+0x5f/0x90 [snd_hda_codec]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa0cd3f29>] snd_hdmi_get_eld_ati+0x39/0x46b [snd_hda_codec_hdmi]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa0cd15cc>] atihdmi_pin_get_eld+0x3c/0x40 [snd_hda_codec_hdmi]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa0cd23ea>] hdmi_present_sense+0x8a/0x360 [snd_hda_codec_hdmi]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffffa0cd26ed>] hdmi_repoll_eld+0x2d/0x50 [snd_hda_codec_hdmi]
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff8107c587>] process_one_work+0x167/0x420
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff8107cf51>] worker_thread+0x121/0x3a0
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff8107ce30>] ? manage_workers.isra.23+0x2a0/0x2a0
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff81083612>] kthread+0xd2/0xf0
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff81083540>] ? kthread_create_on_node+0x180/0x180
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff815212fc>] ret_from_fork+0x7c/0xb0
Jan 27 00:05:05 thinklinux kernel:  [<ffffffff81083540>] ? kthread_create_on_node+0x180/0x180

This was repeated several times until I decided to do a hard shutdown. It seems that the AMD HDMI driver that got enabled by default in 3.13 does not work correctly for me (I'll probably try disabling it next). On a subsequent boot, ALSA complained

Jan 27 00:13:36 thinklinux systemd-udevd[121]: worker [126] /devices/platform/thinkpad_acpi/sound/card4/controlC4 timeout; kill it
Jan 27 00:13:36 thinklinux systemd-udevd[121]: seq 1553 '/devices/platform/thinkpad_acpi/sound/card4/controlC4' killed
Jan 27 00:13:36 thinklinux systemd-udevd[121]: worker [126] terminated by signal 9 (Killed)
Jan 27 00:13:38 thinklinux systemd-udevd[133]: timeout: killing '/usr/bin/alsactl restore ' [158]
Jan 27 00:13:38 thinklinux systemd-udevd[124]: timeout: killing '/usr/bin/alsactl restore ' [159]
Jan 27 00:13:39 thinklinux systemd-udevd[133]: timeout: killing '/usr/bin/alsactl restore ' [158]
Jan 27 00:13:39 thinklinux systemd-udevd[124]: timeout: killing '/usr/bin/alsactl restore ' [159]
Jan 27 00:13:39 thinklinux systemd-udevd[121]: worker [124] /devices/pci0000:00/0000:00:14.2/sound/card1/controlC1 timeout; kill it
Jan 27 00:13:39 thinklinux systemd-udevd[121]: seq 1614 '/devices/pci0000:00/0000:00:14.2/sound/card1/controlC1' killed
Jan 27 00:13:39 thinklinux systemd-udevd[121]: worker [133] /devices/pci0000:00/0000:00:01.1/sound/card0/controlC0 timeout; kill it
Jan 27 00:13:39 thinklinux systemd-udevd[121]: seq 1575 '/devices/pci0000:00/0000:00:01.1/sound/card0/controlC0' killed
Jan 27 00:13:39 thinklinux systemd-udevd[121]: worker [124] terminated by signal 9 (Killed)
Jan 27 00:13:39 thinklinux systemd-udevd[121]: worker [133] terminated by signal 9 (Killed)

I assume it's due to the same problem. I later noticed an error that happened before (and immediately after the graphics driver got initialized):

Jan 26 23:49:28 thinklinux kernel: pciehp 0000:00:07.0:pcie04: Device 0000:04:00.0 already exists at 0000:04:00, cannot hot-add
Jan 26 23:49:28 thinklinux kernel: pciehp 0000:00:07.0:pcie04: Cannot add device at 0000:04:00

Other errors in the journal that showed up after that (and didn't show up on 3.12):

Jan 27 10:31:39 thinklinux dhcpcd[214]: manage_link: No buffer space available
...
Jan 27 10:33:22 thinklinux systemd[1]: Failed unmounting Temporary Directory.
...
Jan 27 10:46:21 thinklinux systemd[362]: Failed to enqueue exit.target job: Unit systemd-exit.service failed to load: Host is down. See system logs and 'systemctl status systemd-exit.service' for details.
...

Everything was fine after I went back to 3.12. Maybe it is all related. I'll try to dig down deeper (with a clean kernel) as soon as I have the time.

Last edited by Andreaskem (2014-01-27 10:28:13)

Offline

#2 2014-01-27 10:29:38

frostschutz
Member
Registered: 2013-11-15
Posts: 429

Re: Problems on 3.13 (AMD HDMI?)

it also has a raid bug that makes it panic and crash

13 is unlucky

Offline

#3 2014-01-27 17:56:28

Andreaskem
Member
Registered: 2013-10-13
Posts: 67

Re: Problems on 3.13 (AMD HDMI?)

I tried again with a clean kernel and the ALSA message still showed up on 3.13. My machine also refused to shutdown showing the message about a hanging task until I hit the power button. It failed to unmount home and a temporary directory as before. The hotplug message did not show up but that might be due to devices showing up in a different order.

Everything seems to be fine with radeon.audio=0 on the kernel command line.

Offline

#4 2014-01-29 17:19:28

Andreaskem
Member
Registered: 2013-10-13
Posts: 67

Re: Problems on 3.13 (AMD HDMI?)

This is fixed by this commit in 3.13.1, meaning I can omit radeon.audio=0 and everything works fine.

Offline

#5 2014-01-30 09:33:21

frostschutz
Member
Registered: 2013-11-15
Posts: 429

Re: Problems on 3.13 (AMD HDMI?)

oooh, there's a 3.13.1, sweet. it also fixed the raid issue. hopefully.

mental note to self: in the future, wait for the .1 wink

Offline

#6 2014-02-05 23:13:06

HarlemSquirrel
Member
From: New York, NY
Registered: 2014-01-22
Posts: 95

Re: Problems on 3.13 (AMD HDMI?)

I'm glad to hear this is resolved! I haven't enabled the testing repo but am anxiously awaiting the supposed AMD graphics enhancements!  I am using an HD 6670 and the open source graphics drivers with GNOME.


"Melody reigns supreme!"
    -J. J. Johnson

Offline

Board footer

Powered by FluxBB