Bug 196927 - kworker hangs in mac80211 & cfg80211; wpa_supplicant survives SIGKILLs; power stays on after everything else shuts down
Summary: kworker hangs in mac80211 & cfg80211; wpa_supplicant survives SIGKILLs; power...
Status: RESOLVED CODE_FIX
Alias: None
Product: Networking
Classification: Unclassified
Component: Wireless (show other bugs)
Hardware: x86-64 Linux
: P1 normal
Assignee: networking_wireless@kernel-bugs.osdl.org
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-09-13 03:27 UTC by Kevin Dodd
Modified: 2017-10-05 16:22 UTC (History)
1 user (show)

See Also:
Kernel Version: 4.13.4
Subsystem:
Regression: Yes
Bisected commit-id:


Attachments
logs from boot through attempted-poweroff (111.62 KB, text/plain)
2017-09-13 03:27 UTC, Kevin Dodd
Details

Description Kevin Dodd 2017-09-13 03:27:38 UTC
Created attachment 258351 [details]
logs from boot through attempted-poweroff

This bug has a few symptoms, which I will describe in chronological
order (not in order of severity).

I get periodic complaints in my system logs about a kworker blocking
for more than 120 seconds. Each instance includes a stack trace. In
the attached logfile, this starts at timestamp 245.426296 on line
1136, and continues throughout.

wpa_supplicant refuses to die, even after root-level processes send
it SIGKILL multiple times. See timestamps 1209.706528, 1299.955722,
1390.206410, and 1480.455734.

After the journal stops and the system reports it has finished the
final step of shutdown, the computer still remains on. I left it like
this for several minutes with no change until I forced the computer
to power off by holding down the power button. (This means I cannot
reboot the computer remotely.)

Oddly enough, I can still use wi-fi just fine. Upload and download
speeds are not perceptibly affected, and neither is latency. It is
only when I try to STOP using wi-fi that the bug visibly affects
anything outside of system logs.

The bug occurs on Linux 4.13.1 but not 4.12.12, so I will be using
4.12.12 until I need to gather more info or until this is fixed.

Here's what "lspci -knn" says about my wi-fi card:

02:00.0 Network controller [0280]: Qualcomm Atheros QCA9377 802.11ac Wireless Network Adapter [168c:0042] (rev 31)
        Subsystem: Foxconn International, Inc. QCA9377 802.11ac Wireless Network Adapter [105b:e0a1]
        Kernel driver in use: ath10k_pci
        Kernel modules: ath10k_pci

Let me know if you need any more info.
Comment 1 Kevin Dodd 2017-09-13 03:54:47 UTC
I just remembered to mention: I have linux-firmware commit
a61ac5cf8374edbfe692d12f805a1b194f7fead2
Comment 3 Kevin Dodd 2017-09-14 14:44:07 UTC
(In reply to Johannes Berg from comment #2)
> This was already fixed in:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211.git/commit/
> ?h=master&id=bde59c475e0883e4c4294bcd9b9c7e08ae18c828

Thanks. Is that in 4.13.2 or will I have to wait for 4.13.3?
Comment 4 Kevin Dodd 2017-09-14 16:37:10 UTC
It looks like it is not in 4.13.2. I hope to see it in 4.13.3, then.
Comment 5 Kevin Dodd 2017-09-21 00:42:24 UTC
Is there any reason why 98e93e968e4947cd71c2eb69e323682daa453ee7 and bde59c475e0883e4c4294bcd9b9c7e08ae18c828 were not merged in 4.13.3?
Comment 6 Kevin Dodd 2017-09-28 20:35:26 UTC
Every time the stable kernel receives a "bugfix" release that lacks
this simple bugfix, I have to manually apply
patch 98e93e968e4947cd71c2eb69e323682daa453ee7 and
patch bde59c475e0883e4c4294bcd9b9c7e08ae18c828
and recompile the kernel myself. Is there a reason why you continue
to release "stable" kernels that freeze indefinitely during shutdown
and during wi-fi configuration, even though the fix is simple and has
already been written?
Comment 7 Kevin Dodd 2017-10-05 16:22:22 UTC
The fix has finally been merged in 4.13.5
so I am marking this "RESOLVED"

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