From: Johannes Weiner <hannes@cmpxchg.org>
To: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrew Morton <akpm@linux-foundation.org>,
Mel Gorman <mgorman@techsingularity.net>,
Miaohe Lin <linmiaohe@huawei.com>,
Kefeng Wang <wangkefeng.wang@huawei.com>, Zi Yan <ziy@nvidia.com>,
linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/6] mm: page_alloc: remove pcppage migratetype caching
Date: Tue, 12 Sep 2023 10:50:28 -0400 [thread overview]
Message-ID: <20230912145028.GA3228@cmpxchg.org> (raw)
In-Reply-To: <a389d846-c19a-42d3-6206-0a1c80e40b37@suse.cz>
On Tue, Sep 12, 2023 at 03:47:45PM +0200, Vlastimil Babka wrote:
> On 9/11/23 21:41, Johannes Weiner wrote:
> > The idea behind the cache is to save get_pageblock_migratetype()
> > lookups during bulk freeing. A microbenchmark suggests this isn't
> > helping, though. The pcp migratetype can get stale, which means that
> > bulk freeing has an extra branch to check if the pageblock was
> > isolated while on the pcp.
> >
> > While the variance overlaps, the cache write and the branch seem to
> > make this a net negative. The following test allocates and frees
> > batches of 10,000 pages (~3x the pcp high marks to trigger flushing):
> >
> > Before:
> > 8,668.48 msec task-clock # 99.735 CPUs utilized ( +- 2.90% )
> > 19 context-switches # 4.341 /sec ( +- 3.24% )
> > 0 cpu-migrations # 0.000 /sec
> > 17,440 page-faults # 3.984 K/sec ( +- 2.90% )
> > 41,758,692,473 cycles # 9.541 GHz ( +- 2.90% )
> > 126,201,294,231 instructions # 5.98 insn per cycle ( +- 2.90% )
> > 25,348,098,335 branches # 5.791 G/sec ( +- 2.90% )
> > 33,436,921 branch-misses # 0.26% of all branches ( +- 2.90% )
> >
> > 0.0869148 +- 0.0000302 seconds time elapsed ( +- 0.03% )
> >
> > After:
> > 8,444.81 msec task-clock # 99.726 CPUs utilized ( +- 2.90% )
> > 22 context-switches # 5.160 /sec ( +- 3.23% )
> > 0 cpu-migrations # 0.000 /sec
> > 17,443 page-faults # 4.091 K/sec ( +- 2.90% )
> > 40,616,738,355 cycles # 9.527 GHz ( +- 2.90% )
> > 126,383,351,792 instructions # 6.16 insn per cycle ( +- 2.90% )
> > 25,224,985,153 branches # 5.917 G/sec ( +- 2.90% )
> > 32,236,793 branch-misses # 0.25% of all branches ( +- 2.90% )
> >
> > 0.0846799 +- 0.0000412 seconds time elapsed ( +- 0.05% )
> >
> > A side effect is that this also ensures that pages whose pageblock
> > gets stolen while on the pcplist end up on the right freelist and we
> > don't perform potentially type-incompatible buddy merges (or skip
> > merges when we shouldn't), whis is likely beneficial to long-term
> > fragmentation management, although the effects would be harder to
> > measure. Settle for simpler and faster code as justification here.
>
> Makes sense to me, so
>
> > Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
>
> Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Thanks!
> > @@ -1577,7 +1556,6 @@ struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
> > continue;
> > del_page_from_free_list(page, zone, current_order);
> > expand(zone, page, order, current_order, migratetype);
> > - set_pcppage_migratetype(page, migratetype);
>
> Hm interesting, just noticed that __rmqueue_fallback() never did this
> AFAICS, sounds like a bug.
I don't quite follow. Which part?
Keep in mind that at this point __rmqueue_fallback() doesn't return a
page. It just moves pages to the desired freelist, and then
__rmqueue_smallest() gets called again. This changes in 5/6, but until
now at least all of the above would apply to fallback pages.
> > @@ -2145,7 +2123,7 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order,
> > * pages are ordered properly.
> > */
> > list_add_tail(&page->pcp_list, list);
> > - if (is_migrate_cma(get_pcppage_migratetype(page)))
> > + if (is_migrate_cma(get_pageblock_migratetype(page)))
> > __mod_zone_page_state(zone, NR_FREE_CMA_PAGES,
> > -(1 << order));
>
> This is potentially a source of overhead, I assume patch 6/6 might
> change that.
Yes, 6/6 removes it altogether.
But the test results in this patch's changelog are from this patch in
isolation, so it doesn't appear to be a concern even on its own.
> > @@ -2457,7 +2423,7 @@ void free_unref_page_list(struct list_head *list)
> > * Free isolated pages directly to the allocator, see
> > * comment in free_unref_page.
> > */
> > - migratetype = get_pcppage_migratetype(page);
> > + migratetype = get_pfnblock_migratetype(page, pfn);
> > if (unlikely(is_migrate_isolate(migratetype))) {
> > list_del(&page->lru);
> > free_one_page(page_zone(page), page, pfn, 0, migratetype, FPI_NONE);
>
> I think after this change we should move the isolated pages handling to
> the second loop below, so that we wouldn't have to call
> get_pfnblock_migratetype() twice per page. Dunno yet if some later patch
> does that. It would need to unlock pcp when necessary.
That sounds like a great idea. Something like the following?
Lightly tested. If you're good with it, I'll beat some more on it and
submit it as a follow-up.
---
From 429d13322819ab38b3ba2fad6d1495997819ccc2 Mon Sep 17 00:00:00 2001
From: Johannes Weiner <hannes@cmpxchg.org>
Date: Tue, 12 Sep 2023 10:16:10 -0400
Subject: [PATCH] mm: page_alloc: optimize free_unref_page_list()
Move direct freeing of isolated pages to the lock-breaking block in
the second loop. This saves an unnecessary migratetype reassessment.
Minor comment and local variable scoping cleanups.
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
mm/page_alloc.c | 49 +++++++++++++++++++++----------------------------
1 file changed, 21 insertions(+), 28 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index e3f1c777feed..9cad31de1bf5 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2408,48 +2408,41 @@ void free_unref_page_list(struct list_head *list)
struct per_cpu_pages *pcp = NULL;
struct zone *locked_zone = NULL;
int batch_count = 0;
- int migratetype;
-
- /* Prepare pages for freeing */
- list_for_each_entry_safe(page, next, list, lru) {
- unsigned long pfn = page_to_pfn(page);
- if (!free_pages_prepare(page, 0, FPI_NONE)) {
+ list_for_each_entry_safe(page, next, list, lru)
+ if (!free_pages_prepare(page, 0, FPI_NONE))
list_del(&page->lru);
- continue;
- }
-
- /*
- * Free isolated pages directly to the allocator, see
- * comment in free_unref_page.
- */
- migratetype = get_pfnblock_migratetype(page, pfn);
- if (unlikely(is_migrate_isolate(migratetype))) {
- list_del(&page->lru);
- free_one_page(page_zone(page), page, pfn, 0, migratetype, FPI_NONE);
- continue;
- }
- }
list_for_each_entry_safe(page, next, list, lru) {
unsigned long pfn = page_to_pfn(page);
struct zone *zone = page_zone(page);
+ int migratetype;
list_del(&page->lru);
migratetype = get_pfnblock_migratetype(page, pfn);
/*
- * Either different zone requiring a different pcp lock or
- * excessive lock hold times when freeing a large list of
- * pages.
+ * Zone switch, batch complete, or non-pcp freeing?
+ * Drop the pcp lock and evaluate.
*/
- if (zone != locked_zone || batch_count == SWAP_CLUSTER_MAX) {
+ if (unlikely(zone != locked_zone ||
+ batch_count == SWAP_CLUSTER_MAX ||
+ is_migrate_isolate(migratetype))) {
if (pcp) {
pcp_spin_unlock(pcp);
pcp_trylock_finish(UP_flags);
+ locked_zone = NULL;
}
- batch_count = 0;
+ /*
+ * Free isolated pages directly to the
+ * allocator, see comment in free_unref_page.
+ */
+ if (is_migrate_isolate(migratetype)) {
+ free_one_page(zone, page, pfn, 0,
+ migratetype, FPI_NONE);
+ continue;
+ }
/*
* trylock is necessary as pages may be getting freed
@@ -2459,12 +2452,12 @@ void free_unref_page_list(struct list_head *list)
pcp = pcp_spin_trylock(zone->per_cpu_pageset);
if (unlikely(!pcp)) {
pcp_trylock_finish(UP_flags);
- free_one_page(zone, page, pfn,
- 0, migratetype, FPI_NONE);
- locked_zone = NULL;
+ free_one_page(zone, page, pfn, 0,
+ migratetype, FPI_NONE);
continue;
}
locked_zone = zone;
+ batch_count = 0;
}
/*
--
2.42.0
next prev parent reply other threads:[~2023-09-12 14:50 UTC|newest]
Thread overview: 83+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-09-11 19:41 [PATCH V2 0/6] mm: page_alloc: freelist migratetype hygiene Johannes Weiner
2023-09-11 19:41 ` [PATCH 1/6] mm: page_alloc: remove pcppage migratetype caching Johannes Weiner
2023-09-11 19:59 ` Zi Yan
2023-09-11 21:09 ` Andrew Morton
2023-09-12 13:47 ` Vlastimil Babka
2023-09-12 14:50 ` Johannes Weiner [this message]
2023-09-13 9:33 ` Vlastimil Babka
2023-09-13 13:24 ` Johannes Weiner
2023-09-13 13:34 ` Vlastimil Babka
2023-09-12 15:03 ` Johannes Weiner
2023-09-14 7:29 ` Vlastimil Babka
2023-09-14 9:56 ` Mel Gorman
2023-09-27 5:42 ` Huang, Ying
2023-09-27 14:51 ` Johannes Weiner
2023-09-30 4:26 ` Huang, Ying
2023-10-02 14:58 ` Johannes Weiner
2023-09-11 19:41 ` [PATCH 2/6] mm: page_alloc: fix up block types when merging compatible blocks Johannes Weiner
2023-09-11 20:01 ` Zi Yan
2023-09-13 9:52 ` Vlastimil Babka
2023-09-14 10:00 ` Mel Gorman
2023-09-11 19:41 ` [PATCH 3/6] mm: page_alloc: move free pages when converting block during isolation Johannes Weiner
2023-09-11 20:17 ` Zi Yan
2023-09-11 20:47 ` Johannes Weiner
2023-09-11 20:50 ` Zi Yan
2023-09-13 14:31 ` Vlastimil Babka
2023-09-14 10:03 ` Mel Gorman
2023-09-11 19:41 ` [PATCH 4/6] mm: page_alloc: fix move_freepages_block() range error Johannes Weiner
2023-09-11 20:23 ` Zi Yan
2023-09-13 14:40 ` Vlastimil Babka
2023-09-14 13:37 ` Johannes Weiner
2023-09-14 10:03 ` Mel Gorman
2023-09-11 19:41 ` [PATCH 5/6] mm: page_alloc: fix freelist movement during block conversion Johannes Weiner
2023-09-13 19:52 ` Vlastimil Babka
2023-09-14 14:47 ` Johannes Weiner
2023-09-11 19:41 ` [PATCH 6/6] mm: page_alloc: consolidate free page accounting Johannes Weiner
2023-09-13 20:18 ` Vlastimil Babka
2023-09-14 4:11 ` Johannes Weiner
2023-09-14 23:52 ` [PATCH V2 0/6] mm: page_alloc: freelist migratetype hygiene Mike Kravetz
2023-09-15 14:16 ` Johannes Weiner
2023-09-15 15:05 ` Mike Kravetz
2023-09-16 19:57 ` Mike Kravetz
2023-09-16 20:13 ` Andrew Morton
2023-09-18 7:16 ` Vlastimil Babka
2023-09-18 14:52 ` Johannes Weiner
2023-09-18 17:40 ` Mike Kravetz
2023-09-19 6:49 ` Johannes Weiner
2023-09-19 12:37 ` Zi Yan
2023-09-19 15:22 ` Zi Yan
2023-09-19 18:47 ` Mike Kravetz
2023-09-19 20:57 ` Zi Yan
2023-09-20 0:32 ` Mike Kravetz
2023-09-20 1:38 ` Zi Yan
2023-09-20 6:07 ` Vlastimil Babka
2023-09-20 13:48 ` Johannes Weiner
2023-09-20 16:04 ` Johannes Weiner
2023-09-20 17:23 ` Zi Yan
2023-09-21 2:31 ` Zi Yan
2023-09-21 10:19 ` David Hildenbrand
2023-09-21 14:47 ` Zi Yan
2023-09-25 21:12 ` Zi Yan
2023-09-26 17:39 ` Johannes Weiner
2023-09-28 2:51 ` Zi Yan
2023-10-03 2:26 ` Zi Yan
2023-10-10 21:12 ` Johannes Weiner
2023-10-11 15:25 ` Johannes Weiner
2023-10-11 15:45 ` Johannes Weiner
2023-10-11 15:57 ` Zi Yan
2023-10-13 0:06 ` Zi Yan
2023-10-13 14:51 ` Zi Yan
2023-10-16 13:35 ` Zi Yan
2023-10-16 14:37 ` Johannes Weiner
2023-10-16 15:00 ` Zi Yan
2023-10-16 18:51 ` Johannes Weiner
2023-10-16 19:49 ` Zi Yan
2023-10-16 20:26 ` Johannes Weiner
2023-10-16 20:39 ` Johannes Weiner
2023-10-16 20:48 ` Zi Yan
2023-09-26 18:19 ` David Hildenbrand
2023-09-28 3:22 ` Zi Yan
2023-10-02 11:43 ` David Hildenbrand
2023-10-03 2:35 ` Zi Yan
2023-09-18 7:07 ` Vlastimil Babka
2023-09-18 14:09 ` Johannes Weiner
Reply instructions:
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:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230912145028.GA3228@cmpxchg.org \
--to=hannes@cmpxchg.org \
--cc=akpm@linux-foundation.org \
--cc=linmiaohe@huawei.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mgorman@techsingularity.net \
--cc=vbabka@suse.cz \
--cc=wangkefeng.wang@huawei.com \
--cc=ziy@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).