Bug 90111 - Intermittent hang when entering S3 suspend, started with kernel 3.16.2
Summary: Intermittent hang when entering S3 suspend, started with kernel 3.16.2
Status: CLOSED INSUFFICIENT_DATA
Alias: None
Product: ACPI
Classification: Unclassified
Component: Power-Sleep-Wake (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: acpi_power-sleep-wake
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-12-19 22:29 UTC by frederic
Modified: 2015-02-13 08:40 UTC (History)
2 users (show)

See Also:
Kernel Version: 3.16.2 and up
Subsystem:
Regression: No
Bisected commit-id:


Attachments
lscpu output (726 bytes, text/plain)
2014-12-19 22:30 UTC, frederic
Details
dmesg output (54.17 KB, text/plain)
2014-12-19 22:30 UTC, frederic
Details
lspci -vvv output (37.30 KB, text/plain)
2014-12-19 22:31 UTC, frederic
Details
lsmod output (3.62 KB, text/plain)
2014-12-19 22:31 UTC, frederic
Details
lsusb -vvv output (39.62 KB, text/plain)
2014-12-19 22:31 UTC, frederic
Details

Description frederic 2014-12-19 22:29:48 UTC
After upgrading from 3.16.1 to 3.16.2 suspend to ram started failing intermittently (typically takes about 5 suspend-resume cycles for it to happen). When it does, it hangs on a black screen with blinking cursor.

Distro is a pretty standard Arch Linux installation. The problem is still present in the currently packaged kernel (3.17.6).

Current hardware is a Gigabyte GA-990XA-UD3 rev 1.0 mainboard with an FX-8350 and an XFX R9 270X GPU. I am using the proprietary fglrx driver for my GPU, but switching to the radeon driver makes no difference, the hang happens intermittently with exactly the same symptoms.

Since first experiencing the bug, all hardware in this system has been replaced by other components except the mainboard (Gigabyte GA-990XA-UD3 rev 1.0) and the RAM. The bug still persists after changing all except the listed components. Original CPU was a Phenom 2 X4 970, GPU was an HD6870.
Comment 1 frederic 2014-12-19 22:30:29 UTC
Created attachment 161381 [details]
lscpu output
Comment 2 frederic 2014-12-19 22:30:49 UTC
Created attachment 161391 [details]
dmesg output
Comment 3 frederic 2014-12-19 22:31:02 UTC
Created attachment 161401 [details]
lspci -vvv output
Comment 4 frederic 2014-12-19 22:31:40 UTC
Created attachment 161411 [details]
lsmod output
Comment 5 frederic 2014-12-19 22:31:58 UTC
Created attachment 161421 [details]
lsusb -vvv output
Comment 6 frederic 2014-12-19 22:35:41 UTC
I have also played around with the /sys/power knobs:

Running through /sys/power/pm_test states, it appears that the hangs start occurring when pm_test is set to "processors".

Trying to use pm_trace did not point out a specific module.

Setting pm_async to zero has allowed me to do a fair number of suspend-resume cycles without experiencing a hang (yet), will update...
Comment 7 frederic 2014-12-29 18:17:24 UTC
Setting /sys/power/pm_async to 0 has allowed me to successfully suspend my system for more than a week now.

I'll try to do a bisect soon between 3.16.1 and 3.16.2.
Comment 8 Aaron Lu 2015-01-27 07:56:51 UTC
Any news on the bisect result?

Note You need to log in before you can comment on or make changes to this bug.