From: Jaegeuk Kim <jaegeuk@kernel.org>
To: linux-kernel@vger.kernel.org,
linux-f2fs-devel@lists.sourceforge.net, linux-mm@kvack.org,
Matthew Wilcox <willy@infradead.org>
Subject: Re: [PATCH 3/3 v2] mm/readahead: try to allocate high order pages for FADVISE_FAV_WILLNEED
Date: Wed, 3 Dec 2025 23:25:25 +0000 [thread overview]
Message-ID: <aTDG5RVirKFisCGT@google.com> (raw)
In-Reply-To: <20251202013212.964298-4-jaegeuk@kernel.org>
This patch assigns the max folio order for readahead. After applying this patch,
it starts with high-order page allocation successfully as shown in the below
traces.
Before:
f2fs_fadvise: dev = (252,16), ino = 14, i_size = 4294967296 offset:0, len:536870912, advise:3
page_cache_sync_ra: dev=252:16 ino=e index=0 req_count=2048 order=0 size=0 async_size=0 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=0 order=0 size=2048 async_size=1024 ra_pages=2048
page_cache_sync_ra: dev=252:16 ino=e index=2048 req_count=2048 order=0 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_unbounded: dev=252:16 ino=e index=2048 nr_to_read=2048 lookahead_size=0
page_cache_sync_ra: dev=252:16 ino=e index=4096 req_count=2048 order=0 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_unbounded: dev=252:16 ino=e index=4096 nr_to_read=2048 lookahead_size=0
page_cache_sync_ra: dev=252:16 ino=e index=6144 req_count=2048 order=0 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_unbounded: dev=252:16 ino=e index=6144 nr_to_read=2048 lookahead_size=0
...
page_cache_sync_ra: dev=252:16 ino=e index=129024 req_count=2048 order=0 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_unbounded: dev=252:16 ino=e index=129024 nr_to_read=2048 lookahead_size=0
After:
f2fs_fadvise: dev = (252,16), ino = 14, i_size = 4294967296 offset:0, len:536870912, advise:3
page_cache_sync_ra: dev=252:16 ino=e index=0 req_count=2048 order=0 size=0 async_size=0 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=0 order=2 size=2048 async_size=1024 ra_pages=2048
page_cache_sync_ra: dev=252:16 ino=e index=2048 req_count=2048 order=2 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=2048 order=4 size=2048 async_size=1024 ra_pages=2048
page_cache_sync_ra: dev=252:16 ino=e index=4096 req_count=2048 order=4 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=4096 order=6 size=2048 async_size=1024 ra_pages=2048
page_cache_sync_ra: dev=252:16 ino=e index=6144 req_count=2048 order=6 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=6144 order=8 size=2048 async_size=1024 ra_pages=2048
page_cache_sync_ra: dev=252:16 ino=e index=8192 req_count=2048 order=8 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=8192 order=10 size=2048 async_size=1024 ra_pages=2048
page_cache_sync_ra: dev=252:16 ino=e index=10240 req_count=2048 order=9 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=10240 order=11 size=2048 async_size=1024 ra_pages=2048
...
page_cache_ra_order: dev=252:16 ino=e index=126976 order=11 size=2048 async_size=1024 ra_pages=2048
page_cache_sync_ra: dev=252:16 ino=e index=129024 req_count=2048 order=9 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=129024 order=11 size=2048 async_size=1024 ra_pages=2048
page_cache_async_ra: dev=252:16 ino=e index=1024 req_count=2048 order=9 size=2048 async_size=1024 ra_pages=2048 mmap_miss=0 prev_pos=-1
For comparion, this is the trace of madvise(MADV_POPULATE_READ) which bumps up the order by 2.
page_cache_ra_order: dev=252:16 ino=e index=0 order=0 size=2048 async_size=512 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 0, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: MAJOR|RETRY
page_cache_async_ra: dev=252:16 ino=e index=1536 req_count=2048 order=0 size=2048 async_size=512 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=2048 order=2 size=2048 async_size=2048 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 1536, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
page_cache_async_ra: dev=252:16 ino=e index=2048 req_count=2048 order=2 size=2048 async_size=2048 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=4096 order=4 size=2048 async_size=2048 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 2048, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
page_cache_async_ra: dev=252:16 ino=e index=4096 req_count=2048 order=4 size=2048 async_size=2048 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=6144 order=6 size=2048 async_size=2048 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 4096, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
page_cache_async_ra: dev=252:16 ino=e index=6144 req_count=2048 order=6 size=2048 async_size=2048 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=8192 order=8 size=2048 async_size=2048 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 6144, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
page_cache_async_ra: dev=252:16 ino=e index=8192 req_count=2048 order=8 size=2048 async_size=2048 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=10240 order=10 size=2048 async_size=2048 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 8192, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
page_cache_async_ra: dev=252:16 ino=e index=10240 req_count=2048 order=9 size=2048 async_size=2048 ra_pages=2048 mmap_miss=0 prev_pos=-1
...
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 518144, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
page_cache_async_ra: dev=252:16 ino=e index=520192 req_count=2048 order=9 size=2048 async_size=2048 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=522240 order=11 size=2048 async_size=2048 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 520192, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
page_cache_async_ra: dev=252:16 ino=e index=522240 req_count=2048 order=9 size=2048 async_size=2048 ra_pages=2048 mmap_miss=0 prev_pos=-1
page_cache_ra_order: dev=252:16 ino=e index=524288 order=11 size=2048 async_size=2048 ra_pages=2048
f2fs_filemap_fault: dev = (252,16), ino = 14, index = 522240, flags: WRITE|KILLABLE|USER|REMOTE|0x8082000, ret: RETRY
Cc: linux-mm@kvack.org
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
Change log from v1:
- take the same madvise() behavior which bumps up ra->order by 2.
mm/readahead.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/mm/readahead.c b/mm/readahead.c
index 54c78f8276fe..61a469117209 100644
--- a/mm/readahead.c
+++ b/mm/readahead.c
@@ -593,7 +593,8 @@ void page_cache_sync_ra(struct readahead_control *ractl,
* trivial case: (index - prev_index) == 1
* unaligned reads: (index - prev_index) == 0
*/
- if (!index || req_count > max_pages || index - prev_index <= 1UL) {
+ if (!index || req_count > max_pages || index - prev_index <= 1UL ||
+ mapping_large_folio_support(ractl->mapping)) {
ra->start = index;
ra->size = get_init_ra_size(req_count, max_pages);
ra->async_size = ra->size > req_count ? ra->size - req_count :
@@ -627,7 +628,7 @@ void page_cache_sync_ra(struct readahead_control *ractl,
ra->size = min(contig_count + req_count, max_pages);
ra->async_size = 1;
readit:
- ra->order = 0;
+ ra->order += 2;
ractl->_index = ra->start;
page_cache_ra_order(ractl, ra);
}
--
2.52.0.223.gf5cc29aaa4-goog
prev parent reply other threads:[~2025-12-03 23:25 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-12-02 1:30 [PATCH 0/3 v2] improve fadvise(POSIX_FADV_WILLNEED) with large folio Jaegeuk Kim
2025-12-02 1:30 ` [PATCH 1/3] mm/readahead: fix the broken readahead for POSIX_FADV_WILLNEED Jaegeuk Kim
2025-12-02 1:30 ` [PATCH 2/3] mm/readahead: use page_cache_sync_ra for FADVISE_FAV_WILLNEED Jaegeuk Kim
2025-12-02 1:30 ` [PATCH 3/3] mm/readahead: try to allocate high order pages " Jaegeuk Kim
2025-12-02 22:56 ` Matthew Wilcox
2025-12-03 19:04 ` Jaegeuk Kim
2025-12-03 23:25 ` Jaegeuk Kim [this message]
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=aTDG5RVirKFisCGT@google.com \
--to=jaegeuk@kernel.org \
--cc=linux-f2fs-devel@lists.sourceforge.net \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=willy@infradead.org \
/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