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>] [day] [month] [year] [list]
Date: Tue, 02 Jul 2024 15:50:09 +0100
From: David Howells <dhowells@...hat.com>
To: Steve French <sfrench@...ba.org>
cc: dhowells@...hat.com, Paulo Alcantara <pc@...guebit.com>,
    Jeff Layton <jlayton@...nel.org>,
    Christian Brauner <christian@...uner.io>,
    Matthew Wilcox <willy@...radead.org>, linux-cifs@...r.kernel.org,
    netfs@...ts.linux.dev, linux-fsdevel@...r.kernel.org,
    linux-mm@...ck.org, linux-kernel@...r.kernel.org
Subject: [PATCH] cifs: Fix read-performance regression by dropping readahead expansion

cifs_expand_read() is causing a performance regression of around 30% by
causing extra pagecache to be allocated for an inode in the readahead path
before we begin actually dispatching RPC requests, thereby delaying the
actual I/O.  The expansion is sized according to the rsize parameter, which
seems to be 4MiB on my test system; this is a big step up from the first
requests made by the fio test program.

Fix this by removing cifs_expand_readahead().  Readahead expansion is
mostly useful for when we're using the local cache if the local cache has a
block size greater than PAGE_SIZE, so we can dispense with it when not
caching.

Signed-off-by: David Howells <dhowells@...hat.com>
cc: Steve French <sfrench@...ba.org>
cc: Paulo Alcantara <pc@...guebit.com>
cc: Jeff Layton <jlayton@...nel.org>
cc: Matthew Wilcox <willy@...radead.org>
cc: linux-cifs@...r.kernel.org
cc: netfs@...ts.linux.dev
cc: linux-fsdevel@...r.kernel.org
cc: linux-mm@...ck.org
---
 fs/smb/client/file.c |   30 ------------------------------
 1 file changed, 30 deletions(-)

diff --git a/fs/smb/client/file.c b/fs/smb/client/file.c
index f1f2573bb18d..1374635e89fa 100644
--- a/fs/smb/client/file.c
+++ b/fs/smb/client/file.c
@@ -245,35 +245,6 @@ static int cifs_init_request(struct netfs_io_request *rreq, struct file *file)
 	return 0;
 }
 
-/*
- * Expand the size of a readahead to the size of the rsize, if at least as
- * large as a page, allowing for the possibility that rsize is not pow-2
- * aligned.
- */
-static void cifs_expand_readahead(struct netfs_io_request *rreq)
-{
-	unsigned int rsize = rreq->rsize;
-	loff_t misalignment, i_size = i_size_read(rreq->inode);
-
-	if (rsize < PAGE_SIZE)
-		return;
-
-	if (rsize < INT_MAX)
-		rsize = roundup_pow_of_two(rsize);
-	else
-		rsize = ((unsigned int)INT_MAX + 1) / 2;
-
-	misalignment = rreq->start & (rsize - 1);
-	if (misalignment) {
-		rreq->start -= misalignment;
-		rreq->len += misalignment;
-	}
-
-	rreq->len = round_up(rreq->len, rsize);
-	if (rreq->start < i_size && rreq->len > i_size - rreq->start)
-		rreq->len = i_size - rreq->start;
-}
-
 /*
  * Completion of a request operation.
  */
@@ -329,7 +300,6 @@ const struct netfs_request_ops cifs_req_ops = {
 	.init_request		= cifs_init_request,
 	.free_request		= cifs_free_request,
 	.free_subrequest	= cifs_free_subrequest,
-	.expand_readahead	= cifs_expand_readahead,
 	.clamp_length		= cifs_clamp_length,
 	.issue_read		= cifs_req_issue_read,
 	.done			= cifs_rreq_done,


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ