[BUG] 4.14 regression: Xorg hangs on exit

[BUG] 4.14 regression: Xorg hangs on exit

[Bob Tracy]

Apologies for the lack of detail, but the subject pretty much says it
all.  Xorg works fine with 4.13, but hangs on exit with 4.14.

Logging in remotely and applying the "kill -9" sledgehammer has no
effect.  System logs don't show anything unusual going on.  Rebooting
hangs because of the unkillable process: hitting the reset switch is
the only way forward.

Video driver is "radeon", reporting "ATOM BIOS: REDWOOD" at boot time.
Console is fb0 == radeondrmfb.

--Bob

Re: [BUG] 4.14 regression: Xorg hangs on exit

[Bob Tracy]

On Tue, Nov 21, 2017 at 11:12:35AM -0600, Bob Tracy wrote:
> Apologies for the lack of detail, but the subject pretty much says it
> all.  Xorg works fine with 4.13, but hangs on exit with 4.14.
> 
> Logging in remotely and applying the "kill -9" sledgehammer has no
> effect.  System logs don't show anything unusual going on.  Rebooting
> hangs because of the unkillable process: hitting the reset switch is
> the only way forward.
> 
> Video driver is "radeon", reporting "ATOM BIOS: REDWOOD" at boot time.
> Console is fb0 == radeondrmfb.

Only an "occasional" bad deal?  Lockup doesn't happen every time.
I'll make this a "watch" item for the time being.  I'd apologize for the
noise, but I never saw this happen prior to running the 4.14 kernel.
Maybe was just lucky up until now :-(.

Running a 32-bit SMP kernel per "uname -a" output below:

Linux gherkin 4.14.0 #1 SMP PREEMPT Thu Nov 16 23:34:38 CST 2017 i686 Intel(R) Core(TM) i7-2600S CPU @ 2.80GHz GenuineIntel GNU/Linux

The i7-2600S has four hyperthreaded cores (/proc/cpuinfo shows 8 CPUs).

--Bob

Re: [BUG] 4.14 regression: Xorg hangs on exit

[Tetsuo Handa]

On 2017/11/22 13:11, Bob Tracy wrote:
> On Tue, Nov 21, 2017 at 11:12:35AM -0600, Bob Tracy wrote:
>> Apologies for the lack of detail, but the subject pretty much says it
>> all.  Xorg works fine with 4.13, but hangs on exit with 4.14.
>>
>> Logging in remotely and applying the "kill -9" sledgehammer has no
>> effect.  System logs don't show anything unusual going on.  Rebooting
>> hangs because of the unkillable process: hitting the reset switch is
>> the only way forward.
>>
>> Video driver is "radeon", reporting "ATOM BIOS: REDWOOD" at boot time.
>> Console is fb0 == radeondrmfb.
> 
> Only an "occasional" bad deal?  Lockup doesn't happen every time.
> I'll make this a "watch" item for the time being.  I'd apologize for the
> noise, but I never saw this happen prior to running the 4.14 kernel.
> Maybe was just lucky up until now :-(.
> 
> Running a 32-bit SMP kernel per "uname -a" output below:
> 
> Linux gherkin 4.14.0 #1 SMP PREEMPT Thu Nov 16 23:34:38 CST 2017 i686 Intel(R) Core(TM) i7-2600S CPU @ 2.80GHz GenuineIntel GNU/Linux
> 
> The i7-2600S has four hyperthreaded cores (/proc/cpuinfo shows 8 CPUs).
> 
> --Bob
> 
When you hit that problem next time, please capture SysRq-t and SysRq-m output
after logging in remotely.

# dmesg -c > dmesg.txt
# echo t > /proc/sysrq-trigger
# echo m > /proc/sysrq-trigger
# dmesg -c >> dmesg.txt
# sleep 60
# echo t > /proc/sysrq-trigger
# echo m > /proc/sysrq-trigger
# dmesg -c >> dmesg.txt

Then, we will likely be able to know where the Xorg process got stuck at.

Re: [BUG] 4.14 regression: Xorg hangs on exit

[Bob Tracy]

On Wed, Nov 22, 2017 at 10:23:25PM +0900, Tetsuo Handa wrote:
> When you hit that problem next time, please capture SysRq-t and SysRq-m output
> after logging in remotely.
> 
> # dmesg -c > dmesg.txt
> # echo t > /proc/sysrq-trigger
> # echo m > /proc/sysrq-trigger
> # dmesg -c >> dmesg.txt
> # sleep 60
> # echo t > /proc/sysrq-trigger
> # echo m > /proc/sysrq-trigger
> # dmesg -c >> dmesg.txt

Thanks.  I figured there had to be a way to get a bit more insight into
what was happening.  System has been up for 12 hours without incident as
I type this: Xorg has been started and exited multiple times with no
issue.  Still going to keep an eye on it for a while.

--Bob