Re: Boot hang with 5.4-rc1, bisected to c5665868183f

[Catalin Marinas]

On Wed, Oct 02, 2019 at 07:26:51PM -0300, Marc Dionne wrote:
> 5.4-rc1 hangs early on boot for me; the stack trace that I could
> manage to see on screen suggested something kmemleak related, and it
> was fairly quick to bisect it down to commit c5665868183f ("mm:
> kmemleak: use the memory pool for early allocations").
> 
> Several call stacks periodically go by on screen, an interesting one
> is (manually transcribed):
> 
> RIP: 0010:_raw_write_unlock_irqrestore+0x15/0x20
>   find_and_remove_object+0x84/0x90
>   delete_object_full+0x10/0x20
>   __kmemleak_do_cleanup+0x2d/0x50
>   kmemleak_do_cleanup+0x59/0x60
>   process_one_work+0x1a4/0x3c0
>   worker_thread+0x50/0x3c0
> ...
> 
> Config is attached.

Thanks for the report. This was also mentioned here:

http://lkml.kernel.org/r/2ac37341-097e-17a2-fb6b-7912da9fa38e@xxxxxxxxx

I'll have a look. It seems to only happen on the kmemleak disabling
path (I managed to reproduce it as well).

-- 
Catalin