Re: [BUG RT] backtrace on v5.2.9-rt3

[Sean V Kelley]

> On Sep 13, 2019, at 6:42 AM, Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx> wrote:
> 
> On 2019-08-30 12:01:06 [-0400], Steven Rostedt wrote:
>> 
>> I'm triggering the following call splat on 5.2-rt.
>> 
>> [   63.099414] 006: BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:968
>> [   63.108423] 006: in_atomic(): 0, irqs_disabled(): 1, pid: 173, name: kworker/6:1
> â
>> [   63.141041] 006: Call Trace:
>> [   63.143916] 006:  dump_stack+0x5c/0x80
>> [   63.147654] 006:  ___might_sleep.cold.92+0x8d/0x9a
>> [   63.152423] 006:  rt_spin_lock+0x31/0x40
>> [   63.156332] 006:  intel_engine_breadcrumbs_irq+0x56/0x320 [i915]
>> [   63.162350] 006:  intel_engine_signal_breadcrumbs+0x11/0x20 [i915]
>> [   63.168540] 006:  i915_hangcheck_elapsed+0x199/0x420 [i915]
> â
>> [   63.309375] 006:  process_one_work+0x1e8/0x4c0
> â
>> Attached is the config and the full dmesg.
> 
> just for book keeping: Clark reported the same thing, posted patches and
> will try an alternative on Monday (as far as I'm been told). I will
> continue to reply in the other thread.
> 
> Sebastian

I was able to reproduce Clarkâs earlier splat report and also test the posted patches with no issues. 
This was also the approach that Chris Wilson suggested.  I agree Stevenâs issue looks to be the same, but with LPC did not have time to go back to it.

Sean