Thread (265 messages) 265 messages, 1 author, 2011-11-10

[095/264] Revert "NFS: Ensure that writeback_single_inode() calls write_inode() when syncing"

From: Greg KH <gregkh@suse.de>
Date: 2011-11-10 03:41:33
Also in: lkml

3.1-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Trond Myklebust <redacted>

commit 59b7c05fffba030e5d9e72324691e2f99aa69b79 upstream.

This reverts commit b80c3cb628f0ebc241b02e38dd028969fb8026a2.

The reverted commit was rendered obsolete by a VFS fix: commit
5547e8aac6f71505d621a612de2fca0dd988b439 (writeback: Update dirty flags in
two steps). We now no longer need to worry about writeback_single_inode()
missing our marking the inode for COMMIT in 'do_writepages()' call.

Reverting this patch, fixes a performance regression in which the inode
would continuously get queued to the dirty list, causing the writeback
code to unnecessarily try to send a COMMIT.

Signed-off-by: Trond Myklebust <redacted>
Tested-by: Simon Kirby <redacted>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

---
 fs/nfs/write.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -428,7 +428,6 @@ static void
 nfs_mark_request_dirty(struct nfs_page *req)
 {
 	__set_page_dirty_nobuffers(req->wb_page);
-	__mark_inode_dirty(req->wb_page->mapping->host, I_DIRTY_DATASYNC);
 }
 
 #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
@@ -762,6 +761,8 @@ int nfs_updatepage(struct file *file, st
 	status = nfs_writepage_setup(ctx, page, offset, count);
 	if (status < 0)
 		nfs_set_pageerror(page);
+	else
+		__set_page_dirty_nobuffers(page);
 
 	dprintk("NFS:       nfs_updatepage returns %d (isize %lld)\n",
 			status, (long long)i_size_read(inode));

Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help