[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CA+CK2bCyMWMXpz9xmPMRey0euS1AYuPEjmXo6OF9Hsv9x+f9GQ@mail.gmail.com>
Date: Fri, 4 Dec 2020 11:07:51 -0500
From: Pavel Tatashin <pasha.tatashin@...een.com>
To: Michal Hocko <mhocko@...e.com>
Cc: LKML <linux-kernel@...r.kernel.org>, linux-mm <linux-mm@...ck.org>,
Andrew Morton <akpm@...ux-foundation.org>,
Vlastimil Babka <vbabka@...e.cz>,
David Hildenbrand <david@...hat.com>,
Oscar Salvador <osalvador@...e.de>,
Dan Williams <dan.j.williams@...el.com>,
Sasha Levin <sashal@...nel.org>,
Tyler Hicks <tyhicks@...ux.microsoft.com>,
Joonsoo Kim <iamjoonsoo.kim@....com>, mike.kravetz@...cle.com,
Steven Rostedt <rostedt@...dmis.org>,
Ingo Molnar <mingo@...hat.com>, Jason Gunthorpe <jgg@...pe.ca>,
Peter Zijlstra <peterz@...radead.org>,
Mel Gorman <mgorman@...e.de>,
Matthew Wilcox <willy@...radead.org>,
David Rientjes <rientjes@...gle.com>,
John Hubbard <jhubbard@...dia.com>
Subject: Re: [PATCH 5/6] mm: honor PF_MEMALLOC_NOMOVABLE for all allocations
On Fri, Dec 4, 2020 at 3:54 AM Michal Hocko <mhocko@...e.com> wrote:
>
> On Fri 04-12-20 09:43:13, Michal Hocko wrote:
> > On Thu 03-12-20 10:15:41, Pavel Tatashin wrote:
> [...]
> > > Also, current_gfp_context() is used elsewhere, and in some
> > > places removing __GFP_MOVABLE from gfp_mask means that we will need to
> > > also change other things. For example [1], in try_to_free_pages() we
> > > call current_gfp_context(gfp_mask) which can reduce the maximum zone
> > > idx, yet we simply set it to: reclaim_idx = gfp_zone(gfp_mask), not to
> > > the newly determined gfp_mask.
> >
> > Yes and the direct reclaim should honor the movable zone restriction.
> > Why should we reclaim ZONE_MOVABLE when the allocation cannot really
> > allocate from it? Or have I misunderstood your concern?
>
> Btw. if we have gfp mask properly filtered for the fast path then we can
> remove the additional call to current_gfp_context from the direct
> reclaim path. Something for a separate patch.
Good point. I am thinking to make a preparation patch at the beginning
of the series where we move current_gfp_context() to the fast path,
and also address all other cases where this call is not going to be
needed anymore, or where the gfp_mask will needed to be set according
to what current_gfp_context() returned.
Thanks,
Pasha
Powered by blists - more mailing lists