linux-kernel - Re: [PATCH -mm v2] mm: have order > 0 compaction start off where it left

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]

Message-ID: <4FF308CE.4070209@redhat.com>
Date:	Tue, 03 Jul 2012 10:59:26 -0400
From:	Rik van Riel <riel@...hat.com>
To:	Minchan Kim <minchan@...nel.org>
CC:	linux-mm@...ck.org, linux-kernel@...r.kernel.org,
	Mel Gorman <mel@....ul.ie>,
	Andrew Morton <akpm@...ux-foundation.org>, jaschut@...dia.gov,
	kamezawa.hiroyu@...fujitsu.com
Subject: Re: [PATCH -mm v2] mm: have order > 0 compaction start off where
 it left

On 06/28/2012 07:27 PM, Minchan Kim wrote:

>> index 7ea259d..2668b77 100644
>> --- a/mm/compaction.c
>> +++ b/mm/compaction.c
>> @@ -422,6 +422,17 @@ static void isolate_freepages(struct zone *zone,
>>   					pfn -= pageblock_nr_pages) {
>>   		unsigned long isolated;
>>
>> +		/*
>> +		 * Skip ahead if another thread is compacting in the area
>> +		 * simultaneously. If we wrapped around, we can only skip
>> +		 * ahead if zone->compact_cached_free_pfn also wrapped to
>> +		 * above our starting point.
>> +		 */
>> +		if (cc->order>  0&&  (!cc->wrapped ||
>
>
> So if (partial_compaction(cc)&&  ... ) or if (!full_compaction(cc)&&   ...

I am not sure that we want to abstract away what is happening
here.  We also are quite explicit with the meaning of cc->order
in compact_finished and other places in the compaction code.

>> +				      zone->compact_cached_free_pfn>
>> +				      cc->start_free_pfn))
>> +			pfn = min(pfn, zone->compact_cached_free_pfn);
>
>
> The pfn can be where migrate_pfn below?
> I mean we need this?
>
> if (pfn<= low_pfn)
> 	goto out;

That is a good point. I guess there is a small possibility that
another compaction thread is below us with cc->free_pfn and
cc->migrate_pfn, and we just inherited its cc->free_pfn via
zone->compact_cached_free_pfn, bringing us to below our own
cc->migrate_pfn.

Given that this was already possible with parallel compaction
in the past, I am not sure how important it is. It could result
in wasting a little bit of CPU, but your fix for it looks easy
enough.

Mel, any downside to compaction bailing (well, wrapping around)
a little earlier, like Minchan suggested?

>> @@ -463,6 +474,8 @@ static void isolate_freepages(struct zone *zone,
>>   		 */
>>   		if (isolated)
>>   			high_pfn = max(high_pfn, pfn);
>> +		if (cc->order>  0)
>> +			zone->compact_cached_free_pfn = high_pfn;
>
>
> Why do we cache high_pfn instead of pfn?

Reading the code, because we may not have isolated every
possible free page from this memory block.  The same reason
cc->free_pfn is set to high_pfn right before the function
exits.

> If we can't isolate any page, compact_cached_free_pfn would become low_pfn.
> I expect it's not what you want.

I guess we should only cache the value of high_pfn if
we isolated some pages?  In other words, this:

	if (isolated) {
		high_pfn = max(high_pfn, pfn);
		if (cc->order > 0)
			zone->compact_cached_free_pfn = high_pfn;
	}

-- 
All rights reversed
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/