OOM killer invoked while still one forth of mem is available

[Michal Hocko]

On Thu 26-04-18 22:29:17, LEROY Christophe wrote:
> Michal Hocko <mhocko at kernel.org> a écrit :
[...]
> > Yes, show_migration_types. But I do not see why unmovable pageblocks
> > should block the allocation. This is a GFP_KERNEL allocation request
> > essentially - thus unmovable itself. This smells like a bug. We are way
> > above reserves which could block the allocation.
> 
> Any suggestion on how to investigate that bug ? Anything to trace ?

try to enable allocator and vmscan tracepoints. Maybe it will tell
something.
-- 
Michal Hocko
SUSE Labs