Comment # 5 on bug 1186332 from Coly Li

(In reply to Coly Li from comment #4)
> (In reply to Coly Li from comment #3)
> > (In reply to Petr Tesa������������k from comment #0)
> > > Created attachment 849559 [details]
> > > dmesg after lockup
> > > 
> > > I've just experienced a kernel stall on my Lenovo Thinkpad X270, attached.
> > > 
> > > At this point I have no idea how to proceed, and no, I cannot reproduce the
> > > hang at will.
> > 
> > The information is very limited. I understand there is no more stack trace,
> > and I can only guess what happened.
> > 
> > I see one fix is about using an encrypted swap device, do you encrypt the
> > swap partition ?
> 
> I see some suspicious and necessary fixes from dm-crypt deadlocking.
> 
> commit 528b16bfc3ae ("dm crypt: Avoid percpu_counter spinlock contention in
>  crypt_page_alloc()")
> commit c87a95dc28b1 ("dm crypt: defer decryption to a tasklet if interrupts
> disabled")
> commit a666e5c05e7c ("dm: fix deadlock when swapping to encrypted device")
> 
> Let me try to do the backport.

It turns out only the last patch is potential helpful to this workqueue
blocking condition.

I start to test this backport.

Coly Li