Re: [PATCH 2/2] mm: drop PF_MEMALLOC_NORECLAIM

[Matthew Wilcox <willy@infradead.org>]

On Mon, Aug 26, 2024 at 06:51:55PM +0200, Michal Hocko wrote:
> On Mon 26-08-24 14:59:29, Matthew Wilcox wrote:
> > On Mon, Aug 26, 2024 at 10:47:13AM +0200, Michal Hocko wrote:
> > > From: Michal Hocko <mhocko@suse.com>
> > > 
> > > There is no existing user of the flag and the flag is dangerous because
> > > a nested allocation context can use GFP_NOFAIL which could cause
> > > unexpected failure. Such a code would be hard to maintain because it
> > > could be deeper in the call chain.
> > > 
> > > PF_MEMALLOC_NORECLAIM has been added even when it was pointed out [1]
> > > that such a allocation contex is inherently unsafe if the context
> > > doesn't fully control all allocations called from this context.
> > 
> > Wouldn't a straight-up revert of eab0af905bfc be cleaner?  Or is there
> > a reason to keep PF_MEMALLOC_NOWARN?
> 
> I wanted to make it PF_MEMALLOC_NORECLAIM specific. I do not have a
> strong case against PF_MEMALLOC_NOWARN TBH. It is a hack because the
> scope is claiming something about all allocations within the scope
> without necessarily knowing all of them (including potential future
> changes). But NOWARN is not really harmful so I do not care strongly.
> 
> If a plan revert is preferably, I will go with it.

There aren't any other users of PF_MEMALLOC_NOWARN and it definitely
seems like something you want at a callsite rather than blanket for every
allocation below this point.  We don't seem to have many PF_ flags left,
so let's not keep it around if there's no immediate plans for it.