linux-kernel - Re: [PATCH 1/2] mm, memory_hotplug: do not fail offlining too early

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 5 Sep 2017 09:13:07 +0200
From:   Michal Hocko <mhocko@...nel.org>
To:     Anshuman Khandual <khandual@...ux.vnet.ibm.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
        Reza Arbab <arbab@...ux.vnet.ibm.com>,
        Yasuaki Ishimatsu <yasu.isimatu@...il.com>,
        qiuxishi@...wei.com, Igor Mammedov <imammedo@...hat.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>, linux-mm@...ck.org,
        LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/2] mm, memory_hotplug: do not fail offlining too early

On Tue 05-09-17 11:59:36, Anshuman Khandual wrote:
[...]
> > @@ -1634,43 +1634,25 @@ static int __ref __offline_pages(unsigned long start_pfn,
> >  
> >  	pfn = start_pfn;
> >  	expire = jiffies + timeout;
> > -	drain = 0;
> > -	retry_max = 5;
> >  repeat:
> >  	/* start memory hot removal */
> > -	ret = -EAGAIN;
> > +	ret = -EBUSY;
> >  	if (time_after(jiffies, expire))
> >  		goto failed_removal;
> >  	ret = -EINTR;
> >  	if (signal_pending(current))
> >  		goto failed_removal;
> > -	ret = 0;
> > -	if (drain) {
> > -		lru_add_drain_all_cpuslocked();
> > -		cond_resched();
> > -		drain_all_pages(zone);
> > -	}
> 
> Why we had this condition before that only when we fail in migration
> later in do_migrate_range function, drain the lru lists in the next
> attempt. Why not from the first attempt itself ? Just being curious.
 
I can only guess but draining used to invoke IPIs and that is really
costly so an optimistic attempt could try without draining and do that
only if the migration fails. Now that we have it all done in WQ context
there shouldn't be any reason to optimize for draining.
-- 
Michal Hocko
SUSE Labs