LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Mike Kravetz <email@example.com>
To: Vlastimil Babka <firstname.lastname@example.org>,
Cc: Reinette Chatre <email@example.com>,
Michal Hocko <firstname.lastname@example.org>,
Christopher Lameter <email@example.com>,
Guy Shattah <firstname.lastname@example.org>,
Anshuman Khandual <email@example.com>,
Michal Nazarewicz <firstname.lastname@example.org>,
David Nellans <email@example.com>,
Laura Abbott <firstname.lastname@example.org>, Pavel Machek <email@example.com>,
Dave Hansen <firstname.lastname@example.org>,
Andrew Morton <email@example.com>
Subject: Re: [PATCH v2 0/4] Interface for higher order contiguous allocations
Date: Mon, 21 May 2018 17:15:06 -0700 [thread overview]
Message-ID: <firstname.lastname@example.org> (raw)
On 05/21/2018 05:00 AM, Vlastimil Babka wrote:
> On 05/04/2018 01:29 AM, Mike Kravetz wrote:
>> Vlastimil and Michal brought up the issue of allocation alignment. The
>> routine will currently align to 'nr_pages' (which is the requested size
>> argument). It does this by examining and trying to allocate the first
>> nr_pages aligned/nr_pages sized range. If this fails, it moves on to the
>> next nr_pages aligned/nr_pages sized range until success or all potential
>> ranges are exhausted.
> As I've noted in my patch 3/4 review, in fact nr_pages is first rounded
> up to an order, which makes this simpler, but suboptimal. I think we
> could perhaps assume that nr_pages that's a power of two should be
> aligned as such, and other values of nr_pages need no alignment? This
> should fit existing users, and can be extended to explicit alignment
> when such user appears?
I'm good with that. I do believe that minimum alignment will be
pageblock size alignment (for > MAX_ORDER allocations).
>> If we allow an alignment to be specified, we will
>> need to potentially check all alignment aligned/nr_pages sized ranges.
>> In the worst case where alignment = PAGE_SIZE, this could result in huge
>> increase in the number of ranges to check.
>> To help cut down on the number of ranges to check, we could identify the
>> first page that causes a range allocation failure and start the next
>> range at the next aligned boundary. I tried this, and we still end up
>> with a huge number of ranges and wasted CPU cycles.
> I think the wasted cycle issues is due to the current code structure,
> which is based on the CMA use-case, which assumes that the allocations
> will succeed, because the areas are reserved and may contain only
> movable allocations
> - performs only very basic pfn validity and belongs-to-zone checks
> for (pfn per pageblock) - the main cycle
> has_unmovable_pages() - cancel if yes
> move_freepages_block() - expensive!
> etc (not important)
> So I think the problem is that in the main cycle we might do a number of
> expensive move_freepages_block() operations, then hit a block where
> has_unmovable_pages() is true, cancel and do more expensive
> undo_isolate_page_range() operations.
> If we instead first scanned the range with has_unmovable_pages() and
> only start doing the expensive work when we find a large enough (aligned
> or not depending on caller) range, it should be much faster and there
> should be no algorithmic difference between aligned and non-aligned case.
Ok, I will give that a try.
Thanks again for looking at these.
prev parent reply other threads:[~2018-05-22 0:15 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-03 23:29 Mike Kravetz
2018-05-03 23:29 ` [PATCH v2 1/4] mm: change type of free_contig_range(nr_pages) to unsigned long Mike Kravetz
2018-05-18 9:12 ` Vlastimil Babka
2018-05-18 22:01 ` Mike Kravetz
2018-05-03 23:29 ` [PATCH v2 2/4] mm: check for proper migrate type during isolation Mike Kravetz
2018-05-18 10:32 ` Vlastimil Babka
2018-05-21 23:10 ` Mike Kravetz
2018-05-22 7:07 ` Vlastimil Babka
2018-05-03 23:29 ` [PATCH v2 3/4] mm: add find_alloc_contig_pages() interface Mike Kravetz
2018-05-21 8:54 ` Vlastimil Babka
2018-05-21 23:48 ` Mike Kravetz
2018-05-22 16:41 ` Reinette Chatre
2018-05-22 20:35 ` Mike Kravetz
2018-05-23 11:18 ` Vlastimil Babka
2018-05-23 18:07 ` Reinette Chatre
2018-05-28 13:12 ` Vlastimil Babka
2018-05-03 23:29 ` [PATCH v2 4/4] mm/hugetlb: use find_alloc_contig_pages() to allocate gigantic pages Mike Kravetz
2018-05-21 12:00 ` [PATCH v2 0/4] Interface for higher order contiguous allocations Vlastimil Babka
2018-05-22 0:15 ` Mike Kravetz [this message]
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--subject='Re: [PATCH v2 0/4] Interface for higher order contiguous allocations' \
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).