[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <2808175.1727601153@warthog.procyon.org.uk>
Date: Sun, 29 Sep 2024 10:12:33 +0100
From: David Howells <dhowells@...hat.com>
To: Leon Romanovsky <leon@...nel.org>
Cc: dhowells@...hat.com, Eduard Zingerman <eddyz87@...il.com>,
Christian Brauner <brauner@...nel.org>,
Manu Bretelle <chantr4@...il.com>, asmadeus@...ewreck.org,
ceph-devel@...r.kernel.org, christian@...uner.io, ericvh@...nel.org,
hsiangkao@...ux.alibaba.com, idryomov@...il.com, jlayton@...nel.org,
linux-afs@...ts.infradead.org, linux-cifs@...r.kernel.org,
linux-erofs@...ts.ozlabs.org, linux-fsdevel@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-mm@...ck.org,
linux-nfs@...r.kernel.org, marc.dionne@...istor.com,
netdev@...r.kernel.org, netfs@...ts.linux.dev, pc@...guebit.com,
smfrench@...il.com, sprasad@...rosoft.com, tom@...pey.com,
v9fs@...ts.linux.dev, willy@...radead.org
Subject: Re: [PATCH v2 19/25] netfs: Speed up buffered reading
Can you try the attached? I've also put it on my branch here:
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=netfs-fixes
David
---
9p: Don't revert the I/O iterator after reading
Don't revert the I/O iterator before returning from p9_client_read_once().
netfslib doesn't require the reversion and nor doed 9P directory reading.
Make p9_client_read() use a temporary iterator to call down into
p9_client_read_once(), and advance that by the amount read.
Reported-by: Manu Bretelle <chantr4@...il.com>
Reported-by: Eduard Zingerman <eddyz87@...il.com>
Reported-by: Leon Romanovsky <leon@...nel.org>
Signed-off-by: David Howells <dhowells@...hat.com>
cc: Eric Van Hensbergen <ericvh@...nel.org>
cc: Latchesar Ionkov <lucho@...kov.net>
cc: Dominique Martinet <asmadeus@...ewreck.org>
cc: Christian Schoenebeck <linux_oss@...debyte.com>
cc: v9fs@...ts.linux.dev
cc: netfs@...ts.linux.dev
cc: linux-fsdevel@...r.kernel.org
---
net/9p/client.c | 10 +++-------
1 file changed, 3 insertions(+), 7 deletions(-)
diff --git a/net/9p/client.c b/net/9p/client.c
index 5cd94721d974..be59b0a94eaf 100644
--- a/net/9p/client.c
+++ b/net/9p/client.c
@@ -1519,13 +1519,15 @@ p9_client_read(struct p9_fid *fid, u64 offset, struct iov_iter *to, int *err)
*err = 0;
while (iov_iter_count(to)) {
+ struct iov_iter tmp = *to;
int count;
- count = p9_client_read_once(fid, offset, to, err);
+ count = p9_client_read_once(fid, offset, &tmp, err);
if (!count || *err)
break;
offset += count;
total += count;
+ iov_iter_advance(to, count);
}
return total;
}
@@ -1567,16 +1569,12 @@ p9_client_read_once(struct p9_fid *fid, u64 offset, struct iov_iter *to,
}
if (IS_ERR(req)) {
*err = PTR_ERR(req);
- if (!non_zc)
- iov_iter_revert(to, count - iov_iter_count(to));
return 0;
}
*err = p9pdu_readf(&req->rc, clnt->proto_version,
"D", &received, &dataptr);
if (*err) {
- if (!non_zc)
- iov_iter_revert(to, count - iov_iter_count(to));
trace_9p_protocol_dump(clnt, &req->rc);
p9_req_put(clnt, req);
return 0;
@@ -1596,8 +1594,6 @@ p9_client_read_once(struct p9_fid *fid, u64 offset, struct iov_iter *to,
p9_req_put(clnt, req);
return n;
}
- } else {
- iov_iter_revert(to, count - received - iov_iter_count(to));
}
p9_req_put(clnt, req);
return received;
Powered by blists - more mailing lists