Return-Path: Received: from mail-wm0-f41.google.com (mail-wm0-f41.google.com [74.125.82.41]) by kanga.kvack.org (Postfix) with ESMTP id ADDA482F86 for ; Wed, 18 Nov 2015 10:11:22 -0500 (EST) Received: by wmww144 with SMTP id w144so76412054wmw.0 for ; Wed, 18 Nov 2015 07:11:22 -0800 (PST) Received: from mail-wm0-f49.google.com (mail-wm0-f49.google.com. [74.125.82.49]) by mx.google.com with ESMTPS id m19si4585910wjr.103.2015.11.18.07.11.20 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Wed, 18 Nov 2015 07:11:20 -0800 (PST) Received: by wmww144 with SMTP id w144so201279500wmw.1 for ; Wed, 18 Nov 2015 07:11:20 -0800 (PST) Date: Wed, 18 Nov 2015 16:11:19 +0100 From: Michal Hocko Subject: Re: [PATCH 2/2] mm: do not loop over ALLOC_NO_WATERMARKS without triggering reclaim Message-ID: <20151118151119.GG19145@dhcp22.suse.cz> References: <1447680139-16484-1-git-send-email-mhocko@kernel.org> <1447680139-16484-3-git-send-email-mhocko@kernel.org> <564C91E9.8000904@suse.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <564C91E9.8000904@suse.cz> Sender: owner-linux-mm@kvack.org List-ID: To: Vlastimil Babka Cc: Andrew Morton , Mel Gorman , David Rientjes , linux-mm@kvack.org, LKML On Wed 18-11-15 15:57:45, Vlastimil Babka wrote: [...] > > --- a/mm/page_alloc.c > > +++ b/mm/page_alloc.c > > @@ -3046,32 +3046,36 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, > > * allocations are system rather than user orientated > > */ > > ac->zonelist = node_zonelist(numa_node_id(), gfp_mask); > > - do { > > - page = get_page_from_freelist(gfp_mask, order, > > - ALLOC_NO_WATERMARKS, ac); > > - if (page) > > - goto got_pg; > > - > > - if (gfp_mask & __GFP_NOFAIL) > > - wait_iff_congested(ac->preferred_zone, > > - BLK_RW_ASYNC, HZ/50); > > I've been thinking if the lack of unconditional wait_iff_congested() can affect > something negatively. I guess not? Considering that the wait_iff_congested is removed only for PF_MEMALLOC with __GFP_NOFAIL which should be non-existent in the kernel then I think the risk is really low. Even if there was a caller _and_ there was a congestion then the behavior wouldn't be much more worse than what we have currently. The system is out of memory hoplessly if ALLOC_NO_WATERMARKS allocation fails. -- Michal Hocko SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org