Re: [PATCH v4] kmemleak: survive in a low-memory situation

[Catalin Marinas]

On Wed, Mar 27, 2019 at 11:21:58AM -0700, Matthew Wilcox wrote:
> On Wed, Mar 27, 2019 at 05:29:57PM +0000, Catalin Marinas wrote:
> > On Wed, Mar 27, 2019 at 09:44:32AM +0100, Michal Hocko wrote:
> > > As long as there is an implicit __GFP_NOFAIL then kmemleak is simply
> > > broken no matter what other gfp flags you play with. Has anybody looked
> > > at some sort of preallocation where gfpflags_allow_blocking context
> > > allocate objects into a pool that non-sleeping allocations can eat from?
> > 
> > Quick attempt below and it needs some more testing (pretty random pick
> > of the EMERGENCY_POOL_SIZE value). Also, with __GFP_NOFAIL removed, are
> > the other flags safe or we should trim them further?
> 
> Why not use mempool?

I had the wrong impression that it could sleep but it's only if
__GFP_DIRECT_RECLAIM is passed. See below for an updated patch.

> >  #define gfp_kmemleak_mask(gfp)	(((gfp) & (GFP_KERNEL | GFP_ATOMIC)) | \
> >  				 __GFP_NORETRY | __GFP_NOMEMALLOC | \
> > -				 __GFP_NOWARN | __GFP_NOFAIL)
> > +				 __GFP_NOWARN)
> 
> Why GFP_NORETRY?  And if I have specified one of the other retry policies
> in my gfp flags, you should presumably clear that off before setting
> GFP_NORETRY.

It only preserves GFP_KERNEL|GFP_ATOMIC from the original flags
while setting the NOWARN|NORETRY|NOMEMALLOC (the same flags seem to be
set by mempool_alloc()). Anyway, with the changes below, I'll let
mempool add the relevant flags while kmemleak only passes
GFP_KERNEL|GFP_ATOMIC from the original caller.

-----------------------8<------------------------------------------