Re: [PATCH] mm: respect the __GFP_NOWARN flag when warning about stalls

[Michal Hocko]

On Wed 13-09-17 23:14:43, Tetsuo Handa wrote:
> Vlastimil Babka wrote:
> > On 09/13/2017 03:54 PM, Tetsuo Handa wrote:
> > > Michal Hocko wrote:
> > >> Let's see what others think about this.
> > > 
> > > Whether __GFP_NOWARN should warn about stalls is not a topic to discuss.
> > 
> > It is the topic of this thread, which tries to address a concrete
> > problem somebody has experienced. In that context, the rest of your
> > concerns seem to me not related to this problem, IMHO.
> 
> I suggested replacing warn_alloc() with safe/useful one rather than tweaking
> warn_alloc() about __GFP_NOWARN.

What you seem to ignore is that whatever method you use for reporting
stalling allocations you would still have to consider whether to dump
a stall information for __GFP_NOWARN ones. And as the current report
shows that might be a bad idea. So please stick to the topic and do not
move it towards _what_ is the proper way of stall detection.
-- 
Michal Hocko
SUSE Labs