linux-kernel - [PATCH] mm: remove an avoidable load of page refcount in page_ref_add

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-next>] [day] [month] [year] [list]

Message-ID: <20241207082931.1707465-1-mjguzik@gmail.com>
Date: Sat,  7 Dec 2024 09:29:31 +0100
From: Mateusz Guzik <mjguzik@...il.com>
To: yuzhao@...gle.com
Cc: akpm@...ux-foundation.org,
	willy@...radead.org,
	linux-kernel@...r.kernel.org,
	linux-mm@...ck.org,
	Mateusz Guzik <mjguzik@...il.com>
Subject: [PATCH] mm: remove an avoidable load of page refcount in page_ref_add_unless

Explicitly pre-checking the count adds nothing as atomic_add_unless
starts with doing the same thing. iow no functional changes.

disasm of stock filemap_get_read_batch from perf top while running
readseek2_processes -t 24:

  0.04 │ cb:   mov    0x34(%rbx),%eax           # first load
 73.11 │       test   %eax,%eax
       │     ↓ je     1bd
  0.09 │       mov    0x34(%rbx),%eax           # second load
  1.01 │ d9:   test   %eax,%eax
       │     ↓ je     1bd
  0.06 │       lea    0x1(%rax),%edx
  0.00 │       lea    0x34(%rbx),%r14
  0.00 │       lock   cmpxchg %edx,0x34(%rbx)
 14.06 │     ↑ jne    d9

Signed-off-by: Mateusz Guzik <mjguzik@...il.com>
---

I did not bother benchmarking, I don't think there is anything
warranting it for this one. fwiw it plausibly is worth few % in a
microbenchmark at higher core count.

 include/linux/page_ref.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/page_ref.h b/include/linux/page_ref.h
index 8c236c651d1d..fa203894876f 100644
--- a/include/linux/page_ref.h
+++ b/include/linux/page_ref.h
@@ -234,7 +234,7 @@ static inline bool page_ref_add_unless(struct page *page, int nr, int u)
 
 	rcu_read_lock();
 	/* avoid writing to the vmemmap area being remapped */
-	if (!page_is_fake_head(page) && page_ref_count(page) != u)
+	if (!page_is_fake_head(page))
 		ret = atomic_add_unless(&page->_refcount, nr, u);
 	rcu_read_unlock();
 
-- 
2.43.0