From 2b57dc6cf9bf31edc0df430ea18dd1dbd3028975 Mon Sep 17 00:00:00 2001
From: Suresh Jayaraman <sjayaraman@suse.de>
Date: Wed, 11 Mar 2009 14:10:22 -0400
Subject: NFS: Minor __nfs_revalidate_inode cleanup

Remove redundant NFS_STALE() check, a leftover due to the commit
691beb13cdc88358334ef0ba867c080a247a760f

Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/inode.c | 3 ---
 1 file changed, 3 deletions(-)

(limited to 'fs/nfs/inode.c')

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 0c381686171..acaaa7c7efa 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -670,9 +670,6 @@ __nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
 	if (NFS_STALE(inode))
 		goto out;
 
-	if (NFS_STALE(inode))
-		goto out;
-
 	nfs_inc_stats(inode, NFSIOS_INODEREVALIDATE);
 	status = NFS_PROTO(inode)->getattr(server, NFS_FH(inode), &fattr);
 	if (status != 0) {
-- 
cgit v1.2.3-70-g09d2


From 37d9d76d8b3a2ac5817e1fa3263cfe0fdb439e51 Mon Sep 17 00:00:00 2001
From: NeilBrown <neilb@suse.de>
Date: Wed, 11 Mar 2009 14:10:23 -0400
Subject: NFS: flush cached directory information slightly more readily.

If cached directory contents becomes incorrect, there is no way to
flush the contents.  This contrasts with files where file locking is
the recommended way to ensure cache consistency between multiple
applications (a read-lock always flushes the cache).

Also while changes to files often change the size of the file (thus
triggering a cache flush), changes to directories often do not change
the apparent size (as the size is often rounded to a block size).

So it is particularly important with directories to avoid the
possibility of an incorrect cache wherever possible.

When the link count on a directory changes it implies a change in the
number of child directories, and so a change in the contents of this
directory.  So use that as a trigger to flush cached contents.

When the ctime changes but the mtime does not, there are two possible
reasons.
 1/ The owner/mode information has been changed.
 2/ utimes has been used to set the mtime backwards.

In the first case, a data-cache flush is not required.
In the second case it is.

So on the basis that correctness trumps performance, flush the
directory contents cache in this case also.

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/inode.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

(limited to 'fs/nfs/inode.c')

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index acaaa7c7efa..268ce3a4622 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -1113,8 +1113,16 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 				nfs_force_lookup_revalidate(inode);
 		}
 		/* If ctime has changed we should definitely clear access+acl caches */
-		if (!timespec_equal(&inode->i_ctime, &fattr->ctime))
+		if (!timespec_equal(&inode->i_ctime, &fattr->ctime)) {
 			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
+			/* and probably clear data for a directory too as utimes can cause
+			 * havoc with our cache.
+			 */
+			if (S_ISDIR(inode->i_mode)) {
+				invalid |= NFS_INO_INVALID_DATA;
+				nfs_force_lookup_revalidate(inode);
+			}
+		}
 	} else if (nfsi->change_attr != fattr->change_attr) {
 		dprintk("NFS: change_attr change on server for file %s/%ld\n",
 				inode->i_sb->s_id, inode->i_ino);
@@ -1148,8 +1156,11 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 	    inode->i_gid != fattr->gid)
 		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
 
-	if (inode->i_nlink != fattr->nlink)
+	if (inode->i_nlink != fattr->nlink) {
 		invalid |= NFS_INO_INVALID_ATTR;
+		if (S_ISDIR(inode->i_mode))
+			invalid |= NFS_INO_INVALID_DATA;
+	}
 
 	inode->i_mode = fattr->mode;
 	inode->i_nlink = fattr->nlink;
-- 
cgit v1.2.3-70-g09d2


From 9e6e70f8d8b6698e0017c56b86525aabe9c7cd4c Mon Sep 17 00:00:00 2001
From: Trond Myklebust <Trond.Myklebust@netapp.com>
Date: Wed, 11 Mar 2009 14:10:24 -0400
Subject: NFSv4: Support NFSv4 optional attributes in the struct nfs_fattr

Currently, filling struct nfs_fattr is more or less an all or nothing
operation, since NFSv2 and NFSv3 have only mandatory attributes.
In NFSv4, some attributes are optional, and so we may simply not be able to
fill in those fields. Furthermore, NFSv4 allows you to specify which
attributes you are interested in retrieving, thus permitting you to
optimise away retrieval of attributes that you know will no change...

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/inode.c          | 243 ++++++++++++++++++++++++++++++------------------
 fs/nfs/nfs2xdr.c        |   2 +-
 fs/nfs/nfs3xdr.c        |   6 +-
 fs/nfs/nfs4xdr.c        |   6 +-
 include/linux/nfs_xdr.h |  48 ++++++++--
 5 files changed, 202 insertions(+), 103 deletions(-)

(limited to 'fs/nfs/inode.c')

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 268ce3a4622..b7656bd3706 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -249,13 +249,10 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr)
 	struct inode *inode = ERR_PTR(-ENOENT);
 	unsigned long hash;
 
-	if ((fattr->valid & NFS_ATTR_FATTR) == 0)
+	if ((fattr->valid & NFS_ATTR_FATTR_FILEID) == 0)
 		goto out_no_inode;
-
-	if (!fattr->nlink) {
-		printk("NFS: Buggy server - nlink == 0!\n");
+	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) == 0)
 		goto out_no_inode;
-	}
 
 	hash = nfs_fattr_to_ino_t(fattr);
 
@@ -291,7 +288,8 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr)
 			    && fattr->size <= NFS_LIMIT_READDIRPLUS)
 				set_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
 			/* Deal with crossing mountpoints */
-			if (!nfs_fsid_equal(&NFS_SB(sb)->fsid, &fattr->fsid)) {
+			if ((fattr->valid & NFS_ATTR_FATTR_FSID)
+					&& !nfs_fsid_equal(&NFS_SB(sb)->fsid, &fattr->fsid)) {
 				if (fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL)
 					inode->i_op = &nfs_referral_inode_operations;
 				else
@@ -304,28 +302,45 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr)
 		else
 			init_special_inode(inode, inode->i_mode, fattr->rdev);
 
+		memset(&inode->i_atime, 0, sizeof(inode->i_atime));
+		memset(&inode->i_mtime, 0, sizeof(inode->i_mtime));
+		memset(&inode->i_ctime, 0, sizeof(inode->i_ctime));
+		nfsi->change_attr = 0;
+		inode->i_size = 0;
+		inode->i_nlink = 0;
+		inode->i_uid = -2;
+		inode->i_gid = -2;
+		inode->i_blocks = 0;
+		memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
+
 		nfsi->read_cache_jiffies = fattr->time_start;
 		nfsi->attr_gencount = fattr->gencount;
-		inode->i_atime = fattr->atime;
-		inode->i_mtime = fattr->mtime;
-		inode->i_ctime = fattr->ctime;
-		if (fattr->valid & NFS_ATTR_FATTR_V4)
+		if (fattr->valid & NFS_ATTR_FATTR_ATIME)
+			inode->i_atime = fattr->atime;
+		if (fattr->valid & NFS_ATTR_FATTR_MTIME)
+			inode->i_mtime = fattr->mtime;
+		if (fattr->valid & NFS_ATTR_FATTR_CTIME)
+			inode->i_ctime = fattr->ctime;
+		if (fattr->valid & NFS_ATTR_FATTR_CHANGE)
 			nfsi->change_attr = fattr->change_attr;
-		inode->i_size = nfs_size_to_loff_t(fattr->size);
-		inode->i_nlink = fattr->nlink;
-		inode->i_uid = fattr->uid;
-		inode->i_gid = fattr->gid;
-		if (fattr->valid & (NFS_ATTR_FATTR_V3 | NFS_ATTR_FATTR_V4)) {
+		if (fattr->valid & NFS_ATTR_FATTR_SIZE)
+			inode->i_size = nfs_size_to_loff_t(fattr->size);
+		if (fattr->valid & NFS_ATTR_FATTR_NLINK)
+			inode->i_nlink = fattr->nlink;
+		if (fattr->valid & NFS_ATTR_FATTR_OWNER)
+			inode->i_uid = fattr->uid;
+		if (fattr->valid & NFS_ATTR_FATTR_GROUP)
+			inode->i_gid = fattr->gid;
+		if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED)
+			inode->i_blocks = fattr->du.nfs2.blocks;
+		if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) {
 			/*
 			 * report the blocks in 512byte units
 			 */
 			inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
-		} else {
-			inode->i_blocks = fattr->du.nfs2.blocks;
 		}
 		nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
 		nfsi->attrtimeo_timestamp = now;
-		memset(nfsi->cookieverf, 0, sizeof(nfsi->cookieverf));
 		nfsi->access_cache = RB_ROOT;
 
 		unlock_new_inode(inode);
@@ -812,25 +827,31 @@ static void nfs_wcc_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 
-	if ((fattr->valid & NFS_ATTR_WCC_V4) != 0 &&
-			nfsi->change_attr == fattr->pre_change_attr) {
+	if ((fattr->valid & NFS_ATTR_FATTR_PRECHANGE)
+			&& (fattr->valid & NFS_ATTR_FATTR_CHANGE)
+			&& nfsi->change_attr == fattr->pre_change_attr) {
 		nfsi->change_attr = fattr->change_attr;
 		if (S_ISDIR(inode->i_mode))
 			nfsi->cache_validity |= NFS_INO_INVALID_DATA;
 	}
 	/* If we have atomic WCC data, we may update some attributes */
-	if ((fattr->valid & NFS_ATTR_WCC) != 0) {
-		if (timespec_equal(&inode->i_ctime, &fattr->pre_ctime))
+	if ((fattr->valid & NFS_ATTR_FATTR_PRECTIME)
+			&& (fattr->valid & NFS_ATTR_FATTR_CTIME)
+			&& timespec_equal(&inode->i_ctime, &fattr->pre_ctime))
 			memcpy(&inode->i_ctime, &fattr->ctime, sizeof(inode->i_ctime));
-		if (timespec_equal(&inode->i_mtime, &fattr->pre_mtime)) {
+
+	if ((fattr->valid & NFS_ATTR_FATTR_PREMTIME)
+			&& (fattr->valid & NFS_ATTR_FATTR_MTIME)
+			&& timespec_equal(&inode->i_mtime, &fattr->pre_mtime)) {
 			memcpy(&inode->i_mtime, &fattr->mtime, sizeof(inode->i_mtime));
 			if (S_ISDIR(inode->i_mode))
 				nfsi->cache_validity |= NFS_INO_INVALID_DATA;
-		}
-		if (i_size_read(inode) == nfs_size_to_loff_t(fattr->pre_size) &&
-		    nfsi->npages == 0)
-			i_size_write(inode, nfs_size_to_loff_t(fattr->size));
 	}
+	if ((fattr->valid & NFS_ATTR_FATTR_PRESIZE)
+			&& (fattr->valid & NFS_ATTR_FATTR_SIZE)
+			&& i_size_read(inode) == nfs_size_to_loff_t(fattr->pre_size)
+			&& nfsi->npages == 0)
+			i_size_write(inode, nfs_size_to_loff_t(fattr->size));
 }
 
 /**
@@ -850,35 +871,39 @@ static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fat
 
 
 	/* Has the inode gone and changed behind our back? */
-	if (nfsi->fileid != fattr->fileid
-			|| (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT)) {
+	if ((fattr->valid & NFS_ATTR_FATTR_FILEID) && nfsi->fileid != fattr->fileid)
+		return -EIO;
+	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT))
 		return -EIO;
-	}
 
-	if ((fattr->valid & NFS_ATTR_FATTR_V4) != 0 &&
+	if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 &&
 			nfsi->change_attr != fattr->change_attr)
 		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
 
 	/* Verify a few of the more important attributes */
-	if (!timespec_equal(&inode->i_mtime, &fattr->mtime))
+	if ((fattr->valid & NFS_ATTR_FATTR_MTIME) && !timespec_equal(&inode->i_mtime, &fattr->mtime))
 		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
 
-	cur_size = i_size_read(inode);
- 	new_isize = nfs_size_to_loff_t(fattr->size);
-	if (cur_size != new_isize && nfsi->npages == 0)
-		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
+	if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
+		cur_size = i_size_read(inode);
+		new_isize = nfs_size_to_loff_t(fattr->size);
+		if (cur_size != new_isize && nfsi->npages == 0)
+			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
+	}
 
 	/* Have any file permissions changed? */
-	if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)
-			|| inode->i_uid != fattr->uid
-			|| inode->i_gid != fattr->gid)
+	if ((fattr->valid & NFS_ATTR_FATTR_MODE) && (inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO))
+		invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
+	if ((fattr->valid & NFS_ATTR_FATTR_OWNER) && inode->i_uid != fattr->uid)
+		invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
+	if ((fattr->valid & NFS_ATTR_FATTR_GROUP) && inode->i_gid != fattr->gid)
 		invalid |= NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL;
 
 	/* Has the link count changed? */
-	if (inode->i_nlink != fattr->nlink)
+	if ((fattr->valid & NFS_ATTR_FATTR_NLINK) && inode->i_nlink != fattr->nlink)
 		invalid |= NFS_INO_INVALID_ATTR;
 
-	if (!timespec_equal(&inode->i_atime, &fattr->atime))
+	if ((fattr->valid & NFS_ATTR_FATTR_ATIME) && !timespec_equal(&inode->i_atime, &fattr->atime))
 		invalid |= NFS_INO_INVALID_ATIME;
 
 	if (invalid != 0)
@@ -890,11 +915,15 @@ static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fat
 
 static int nfs_ctime_need_update(const struct inode *inode, const struct nfs_fattr *fattr)
 {
+	if (!(fattr->valid & NFS_ATTR_FATTR_CTIME))
+		return 0;
 	return timespec_compare(&fattr->ctime, &inode->i_ctime) > 0;
 }
 
 static int nfs_size_need_update(const struct inode *inode, const struct nfs_fattr *fattr)
 {
+	if (!(fattr->valid & NFS_ATTR_FATTR_SIZE))
+		return 0;
 	return nfs_size_to_loff_t(fattr->size) > i_size_read(inode);
 }
 
@@ -1030,20 +1059,31 @@ int nfs_post_op_update_inode_force_wcc(struct inode *inode, struct nfs_fattr *fa
 	/* Don't do a WCC update if these attributes are already stale */
 	if ((fattr->valid & NFS_ATTR_FATTR) == 0 ||
 			!nfs_inode_attrs_need_update(inode, fattr)) {
-		fattr->valid &= ~(NFS_ATTR_WCC_V4|NFS_ATTR_WCC);
+		fattr->valid &= ~(NFS_ATTR_FATTR_PRECHANGE
+				| NFS_ATTR_FATTR_PRESIZE
+				| NFS_ATTR_FATTR_PREMTIME
+				| NFS_ATTR_FATTR_PRECTIME);
 		goto out_noforce;
 	}
-	if ((fattr->valid & NFS_ATTR_FATTR_V4) != 0 &&
-			(fattr->valid & NFS_ATTR_WCC_V4) == 0) {
+	if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 &&
+			(fattr->valid & NFS_ATTR_FATTR_PRECHANGE) == 0) {
 		fattr->pre_change_attr = NFS_I(inode)->change_attr;
-		fattr->valid |= NFS_ATTR_WCC_V4;
+		fattr->valid |= NFS_ATTR_FATTR_PRECHANGE;
 	}
-	if ((fattr->valid & NFS_ATTR_FATTR) != 0 &&
-			(fattr->valid & NFS_ATTR_WCC) == 0) {
+	if ((fattr->valid & NFS_ATTR_FATTR_CTIME) != 0 &&
+			(fattr->valid & NFS_ATTR_FATTR_PRECTIME) == 0) {
 		memcpy(&fattr->pre_ctime, &inode->i_ctime, sizeof(fattr->pre_ctime));
+		fattr->valid |= NFS_ATTR_FATTR_PRECTIME;
+	}
+	if ((fattr->valid & NFS_ATTR_FATTR_MTIME) != 0 &&
+			(fattr->valid & NFS_ATTR_FATTR_PREMTIME) == 0) {
 		memcpy(&fattr->pre_mtime, &inode->i_mtime, sizeof(fattr->pre_mtime));
+		fattr->valid |= NFS_ATTR_FATTR_PREMTIME;
+	}
+	if ((fattr->valid & NFS_ATTR_FATTR_SIZE) != 0 &&
+			(fattr->valid & NFS_ATTR_FATTR_PRESIZE) == 0) {
 		fattr->pre_size = i_size_read(inode);
-		fattr->valid |= NFS_ATTR_WCC;
+		fattr->valid |= NFS_ATTR_FATTR_PRESIZE;
 	}
 out_noforce:
 	status = nfs_post_op_update_inode_locked(inode, fattr);
@@ -1075,18 +1115,18 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 			__func__, inode->i_sb->s_id, inode->i_ino,
 			atomic_read(&inode->i_count), fattr->valid);
 
-	if (nfsi->fileid != fattr->fileid)
+	if ((fattr->valid & NFS_ATTR_FATTR_FILEID) && nfsi->fileid != fattr->fileid)
 		goto out_fileid;
 
 	/*
 	 * Make sure the inode's type hasn't changed.
 	 */
-	if ((inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT))
+	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT))
 		goto out_changed;
 
 	server = NFS_SERVER(inode);
 	/* Update the fsid? */
-	if (S_ISDIR(inode->i_mode) &&
+	if (S_ISDIR(inode->i_mode) && (fattr->valid & NFS_ATTR_FATTR_FSID) &&
 			!nfs_fsid_equal(&server->fsid, &fattr->fsid) &&
 			!test_bit(NFS_INO_MOUNTPOINT, &nfsi->flags))
 		server->fsid = fattr->fsid;
@@ -1096,14 +1136,27 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 	 */
 	nfsi->read_cache_jiffies = fattr->time_start;
 
-	nfsi->cache_validity &= ~(NFS_INO_INVALID_ATTR | NFS_INO_INVALID_ATIME
-			| NFS_INO_REVAL_PAGECACHE);
+	if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) || (fattr->valid & (NFS_ATTR_FATTR_MTIME|NFS_ATTR_FATTR_CTIME)))
+	    nfsi->cache_validity &= ~(NFS_INO_INVALID_ATTR
+		    | NFS_INO_INVALID_ATIME
+		    | NFS_INO_REVAL_PAGECACHE);
 
 	/* Do atomic weak cache consistency updates */
 	nfs_wcc_update_inode(inode, fattr);
 
 	/* More cache consistency checks */
-	if (!(fattr->valid & NFS_ATTR_FATTR_V4)) {
+	if (fattr->valid & NFS_ATTR_FATTR_CHANGE) {
+		if (nfsi->change_attr != fattr->change_attr) {
+			dprintk("NFS: change_attr change on server for file %s/%ld\n",
+					inode->i_sb->s_id, inode->i_ino);
+			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
+			if (S_ISDIR(inode->i_mode))
+				nfs_force_lookup_revalidate(inode);
+			nfsi->change_attr = fattr->change_attr;
+		}
+	}
+
+	if (fattr->valid & NFS_ATTR_FATTR_MTIME) {
 		/* NFSv2/v3: Check if the mtime agrees */
 		if (!timespec_equal(&inode->i_mtime, &fattr->mtime)) {
 			dprintk("NFS: mtime change on server for file %s/%ld\n",
@@ -1111,7 +1164,10 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
 			if (S_ISDIR(inode->i_mode))
 				nfs_force_lookup_revalidate(inode);
+			memcpy(&inode->i_mtime, &fattr->mtime, sizeof(inode->i_mtime));
 		}
+	}
+	if (fattr->valid & NFS_ATTR_FATTR_CTIME) {
 		/* If ctime has changed we should definitely clear access+acl caches */
 		if (!timespec_equal(&inode->i_ctime, &fattr->ctime)) {
 			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
@@ -1122,59 +1178,66 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 				invalid |= NFS_INO_INVALID_DATA;
 				nfs_force_lookup_revalidate(inode);
 			}
+			memcpy(&inode->i_ctime, &fattr->ctime, sizeof(inode->i_ctime));
 		}
-	} else if (nfsi->change_attr != fattr->change_attr) {
-		dprintk("NFS: change_attr change on server for file %s/%ld\n",
-				inode->i_sb->s_id, inode->i_ino);
-		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
-		if (S_ISDIR(inode->i_mode))
-			nfs_force_lookup_revalidate(inode);
 	}
 
 	/* Check if our cached file size is stale */
- 	new_isize = nfs_size_to_loff_t(fattr->size);
-	cur_isize = i_size_read(inode);
-	if (new_isize != cur_isize) {
-		/* Do we perhaps have any outstanding writes, or has
-		 * the file grown beyond our last write? */
-		if (nfsi->npages == 0 || new_isize > cur_isize) {
-			i_size_write(inode, new_isize);
-			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
+	if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
+		new_isize = nfs_size_to_loff_t(fattr->size);
+		cur_isize = i_size_read(inode);
+		if (new_isize != cur_isize) {
+			/* Do we perhaps have any outstanding writes, or has
+			 * the file grown beyond our last write? */
+			if (nfsi->npages == 0 || new_isize > cur_isize) {
+				i_size_write(inode, new_isize);
+				invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
+			}
+			dprintk("NFS: isize change on server for file %s/%ld\n",
+					inode->i_sb->s_id, inode->i_ino);
 		}
-		dprintk("NFS: isize change on server for file %s/%ld\n",
-				inode->i_sb->s_id, inode->i_ino);
 	}
 
 
-	memcpy(&inode->i_mtime, &fattr->mtime, sizeof(inode->i_mtime));
-	memcpy(&inode->i_ctime, &fattr->ctime, sizeof(inode->i_ctime));
-	memcpy(&inode->i_atime, &fattr->atime, sizeof(inode->i_atime));
-	nfsi->change_attr = fattr->change_attr;
+	if (fattr->valid & NFS_ATTR_FATTR_ATIME)
+		memcpy(&inode->i_atime, &fattr->atime, sizeof(inode->i_atime));
 
-	if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO) ||
-	    inode->i_uid != fattr->uid ||
-	    inode->i_gid != fattr->gid)
-		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
-
-	if (inode->i_nlink != fattr->nlink) {
-		invalid |= NFS_INO_INVALID_ATTR;
-		if (S_ISDIR(inode->i_mode))
-			invalid |= NFS_INO_INVALID_DATA;
+	if (fattr->valid & NFS_ATTR_FATTR_MODE) {
+		if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)) {
+			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
+			inode->i_mode = fattr->mode;
+		}
+	}
+	if (fattr->valid & NFS_ATTR_FATTR_OWNER) {
+		if (inode->i_uid != fattr->uid) {
+			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
+			inode->i_uid = fattr->uid;
+		}
+	}
+	if (fattr->valid & NFS_ATTR_FATTR_GROUP) {
+		if (inode->i_gid != fattr->gid) {
+			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL;
+			inode->i_gid = fattr->gid;
+		}
 	}
 
-	inode->i_mode = fattr->mode;
-	inode->i_nlink = fattr->nlink;
-	inode->i_uid = fattr->uid;
-	inode->i_gid = fattr->gid;
+	if (fattr->valid & NFS_ATTR_FATTR_NLINK) {
+		if (inode->i_nlink != fattr->nlink) {
+			invalid |= NFS_INO_INVALID_ATTR;
+			if (S_ISDIR(inode->i_mode))
+				invalid |= NFS_INO_INVALID_DATA;
+			inode->i_nlink = fattr->nlink;
+		}
+	}
 
-	if (fattr->valid & (NFS_ATTR_FATTR_V3 | NFS_ATTR_FATTR_V4)) {
+	if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) {
 		/*
 		 * report the blocks in 512byte units
 		 */
 		inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
- 	} else {
- 		inode->i_blocks = fattr->du.nfs2.blocks;
  	}
+	if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED)
+		inode->i_blocks = fattr->du.nfs2.blocks;
 
 	/* Update attrtimeo value if we're out of the unstable period */
 	if (invalid & NFS_INO_INVALID_ATTR) {
diff --git a/fs/nfs/nfs2xdr.c b/fs/nfs/nfs2xdr.c
index 28bab67d151..bea99992c30 100644
--- a/fs/nfs/nfs2xdr.c
+++ b/fs/nfs/nfs2xdr.c
@@ -136,7 +136,7 @@ xdr_decode_fattr(__be32 *p, struct nfs_fattr *fattr)
 	p = xdr_decode_time(p, &fattr->atime);
 	p = xdr_decode_time(p, &fattr->mtime);
 	p = xdr_decode_time(p, &fattr->ctime);
-	fattr->valid |= NFS_ATTR_FATTR;
+	fattr->valid |= NFS_ATTR_FATTR_V2;
 	fattr->rdev = new_decode_dev(rdev);
 	if (fattr->type == NFCHR && rdev == NFS2_FIFO_DEV) {
 		fattr->type = NFFIFO;
diff --git a/fs/nfs/nfs3xdr.c b/fs/nfs/nfs3xdr.c
index 6cdeacffde4..c0f7d02aced 100644
--- a/fs/nfs/nfs3xdr.c
+++ b/fs/nfs/nfs3xdr.c
@@ -177,7 +177,7 @@ xdr_decode_fattr(__be32 *p, struct nfs_fattr *fattr)
 	p = xdr_decode_time3(p, &fattr->ctime);
 
 	/* Update the mode bits */
-	fattr->valid |= (NFS_ATTR_FATTR | NFS_ATTR_FATTR_V3);
+	fattr->valid |= NFS_ATTR_FATTR_V3;
 	return p;
 }
 
@@ -233,7 +233,9 @@ xdr_decode_wcc_attr(__be32 *p, struct nfs_fattr *fattr)
 	p = xdr_decode_hyper(p, &fattr->pre_size);
 	p = xdr_decode_time3(p, &fattr->pre_mtime);
 	p = xdr_decode_time3(p, &fattr->pre_ctime);
-	fattr->valid |= NFS_ATTR_WCC;
+	fattr->valid |= NFS_ATTR_FATTR_PRESIZE
+		| NFS_ATTR_FATTR_PREMTIME
+		| NFS_ATTR_FATTR_PRECTIME;
 	return p;
 }
 
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index 5f0ee3e2bd8..7d220da3db3 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -3012,7 +3012,7 @@ static int decode_getfattr(struct xdr_stream *xdr, struct nfs_fattr *fattr, cons
 	if ((status = decode_attr_type(xdr, bitmap, &type)) != 0)
 		goto xdr_error;
 	fattr->type = nfs_type2fmt[type].nfs2type;
-	fmode = nfs_type2fmt[type].mode;
+	fattr->mode = nfs_type2fmt[type].mode;
 
 	if ((status = decode_attr_change(xdr, bitmap, &fattr->change_attr)) != 0)
 		goto xdr_error;
@@ -3026,7 +3026,7 @@ static int decode_getfattr(struct xdr_stream *xdr, struct nfs_fattr *fattr, cons
 						struct nfs4_fs_locations,
 						fattr))) != 0)
 		goto xdr_error;
-	if ((status = decode_attr_mode(xdr, bitmap, &fattr->mode)) != 0)
+	if ((status = decode_attr_mode(xdr, bitmap, &fmode)) != 0)
 		goto xdr_error;
 	fattr->mode |= fmode;
 	if ((status = decode_attr_nlink(xdr, bitmap, &fattr->nlink)) != 0)
@@ -3050,7 +3050,7 @@ static int decode_getfattr(struct xdr_stream *xdr, struct nfs_fattr *fattr, cons
 	if (fattr->fileid == 0 && fileid != 0)
 		fattr->fileid = fileid;
 	if ((status = verify_attr_len(xdr, savep, attrlen)) == 0)
-		fattr->valid = NFS_ATTR_FATTR | NFS_ATTR_FATTR_V3 | NFS_ATTR_FATTR_V4;
+		fattr->valid = NFS_ATTR_FATTR_V4;
 xdr_error:
 	dprintk("%s: xdr returned %d\n", __func__, -status);
 	return status;
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 2e5f00066af..b99295e07cd 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -27,7 +27,7 @@ static inline int nfs_fsid_equal(const struct nfs_fsid *a, const struct nfs_fsid
 }
 
 struct nfs_fattr {
-	unsigned short		valid;		/* which fields are valid */
+	unsigned int		valid;		/* which fields are valid */
 	__u64			pre_size;	/* pre_op_attr.size	  */
 	struct timespec		pre_mtime;	/* pre_op_attr.mtime	  */
 	struct timespec		pre_ctime;	/* pre_op_attr.ctime	  */
@@ -59,12 +59,46 @@ struct nfs_fattr {
 	unsigned long		gencount;
 };
 
-#define NFS_ATTR_WCC		0x0001		/* pre-op WCC data    */
-#define NFS_ATTR_FATTR		0x0002		/* post-op attributes */
-#define NFS_ATTR_FATTR_V3	0x0004		/* NFSv3 attributes */
-#define NFS_ATTR_FATTR_V4	0x0008		/* NFSv4 change attribute */
-#define NFS_ATTR_WCC_V4		0x0010		/* pre-op change attribute */
-#define NFS_ATTR_FATTR_V4_REFERRAL	0x0020		/* NFSv4 referral */
+#define NFS_ATTR_FATTR_TYPE		(1U << 0)
+#define NFS_ATTR_FATTR_MODE		(1U << 1)
+#define NFS_ATTR_FATTR_NLINK		(1U << 2)
+#define NFS_ATTR_FATTR_OWNER		(1U << 3)
+#define NFS_ATTR_FATTR_GROUP		(1U << 4)
+#define NFS_ATTR_FATTR_RDEV		(1U << 5)
+#define NFS_ATTR_FATTR_SIZE		(1U << 6)
+#define NFS_ATTR_FATTR_PRESIZE		(1U << 7)
+#define NFS_ATTR_FATTR_BLOCKS_USED	(1U << 8)
+#define NFS_ATTR_FATTR_SPACE_USED	(1U << 9)
+#define NFS_ATTR_FATTR_FSID		(1U << 10)
+#define NFS_ATTR_FATTR_FILEID		(1U << 11)
+#define NFS_ATTR_FATTR_ATIME		(1U << 12)
+#define NFS_ATTR_FATTR_MTIME		(1U << 13)
+#define NFS_ATTR_FATTR_CTIME		(1U << 14)
+#define NFS_ATTR_FATTR_PREMTIME		(1U << 15)
+#define NFS_ATTR_FATTR_PRECTIME		(1U << 16)
+#define NFS_ATTR_FATTR_CHANGE		(1U << 17)
+#define NFS_ATTR_FATTR_PRECHANGE	(1U << 18)
+#define NFS_ATTR_FATTR_V4_REFERRAL	(1U << 19)	/* NFSv4 referral */
+
+#define NFS_ATTR_FATTR (NFS_ATTR_FATTR_TYPE \
+		| NFS_ATTR_FATTR_MODE \
+		| NFS_ATTR_FATTR_NLINK \
+		| NFS_ATTR_FATTR_OWNER \
+		| NFS_ATTR_FATTR_GROUP \
+		| NFS_ATTR_FATTR_RDEV \
+		| NFS_ATTR_FATTR_SIZE \
+		| NFS_ATTR_FATTR_FSID \
+		| NFS_ATTR_FATTR_FILEID \
+		| NFS_ATTR_FATTR_ATIME \
+		| NFS_ATTR_FATTR_MTIME \
+		| NFS_ATTR_FATTR_CTIME)
+#define NFS_ATTR_FATTR_V2 (NFS_ATTR_FATTR \
+		| NFS_ATTR_FATTR_BLOCKS_USED)
+#define NFS_ATTR_FATTR_V3 (NFS_ATTR_FATTR \
+		| NFS_ATTR_FATTR_SPACE_USED)
+#define NFS_ATTR_FATTR_V4 (NFS_ATTR_FATTR \
+		| NFS_ATTR_FATTR_SPACE_USED \
+		| NFS_ATTR_FATTR_CHANGE)
 
 /*
  * Info on the file system
-- 
cgit v1.2.3-70-g09d2


From fb8a1f11b64e213d94dfa1cebb2a42a7b8c115c4 Mon Sep 17 00:00:00 2001
From: Trond Myklebust <Trond.Myklebust@netapp.com>
Date: Wed, 11 Mar 2009 14:10:29 -0400
Subject: NFS: cleanup - remove struct nfs_inode->ncommit

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/inode.c         |  1 -
 fs/nfs/write.c         | 25 ++++++++++++++++---------
 include/linux/nfs_fs.h |  3 +--
 3 files changed, 17 insertions(+), 12 deletions(-)

(limited to 'fs/nfs/inode.c')

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index b7656bd3706..00f116cdadc 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -1345,7 +1345,6 @@ static void init_once(void *foo)
 	INIT_LIST_HEAD(&nfsi->access_cache_entry_lru);
 	INIT_LIST_HEAD(&nfsi->access_cache_inode_lru);
 	INIT_RADIX_TREE(&nfsi->nfs_page_tree, GFP_ATOMIC);
-	nfsi->ncommit = 0;
 	nfsi->npages = 0;
 	atomic_set(&nfsi->silly_count, 1);
 	INIT_HLIST_HEAD(&nfsi->silly_list);
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 9f9845859fc..1a999939fed 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -404,7 +404,6 @@ nfs_mark_request_commit(struct nfs_page *req)
 	struct nfs_inode *nfsi = NFS_I(inode);
 
 	spin_lock(&inode->i_lock);
-	nfsi->ncommit++;
 	set_bit(PG_CLEAN, &(req)->wb_flags);
 	radix_tree_tag_set(&nfsi->nfs_page_tree,
 			req->wb_index,
@@ -523,6 +522,12 @@ static void nfs_cancel_commit_list(struct list_head *head)
 	}
 }
 
+static int
+nfs_need_commit(struct nfs_inode *nfsi)
+{
+	return radix_tree_tagged(&nfsi->nfs_page_tree, NFS_PAGE_TAG_COMMIT);
+}
+
 #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
 /*
  * nfs_scan_commit - Scan an inode for commit requests
@@ -538,16 +543,18 @@ static int
 nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
-	int res = 0;
 
-	if (nfsi->ncommit != 0) {
-		res = nfs_scan_list(nfsi, dst, idx_start, npages,
-				NFS_PAGE_TAG_COMMIT);
-		nfsi->ncommit -= res;
-	}
-	return res;
+	if (!nfs_need_commit(nfsi))
+		return 0;
+
+	return nfs_scan_list(nfsi, dst, idx_start, npages, NFS_PAGE_TAG_COMMIT);
 }
 #else
+static inline int nfs_need_commit(struct nfs_inode *nfsi)
+{
+	return 0;
+}
+
 static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
 {
 	return 0;
@@ -820,7 +827,7 @@ static int nfs_write_rpcsetup(struct nfs_page *req,
 	data->args.stable  = NFS_UNSTABLE;
 	if (how & FLUSH_STABLE) {
 		data->args.stable = NFS_DATA_SYNC;
-		if (!NFS_I(inode)->ncommit)
+		if (!nfs_need_commit(NFS_I(inode)))
 			data->args.stable = NFS_FILE_SYNC;
 	}
 
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index db867b04ac3..c9fecd3e8f0 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -166,8 +166,7 @@ struct nfs_inode {
 	 */
 	struct radix_tree_root	nfs_page_tree;
 
-	unsigned long		ncommit,
-				npages;
+	unsigned long		npages;
 
 	/* Open contexts for shared mmap writes */
 	struct list_head	open_files;
-- 
cgit v1.2.3-70-g09d2


From 72cb77f4a5ace37b12dcb47a0e8637a2c28ad881 Mon Sep 17 00:00:00 2001
From: Trond Myklebust <Trond.Myklebust@netapp.com>
Date: Wed, 11 Mar 2009 14:10:30 -0400
Subject: NFS: Throttle page dirtying while we're flushing to disk

The following patch is a combination of a patch by myself and Peter
Staubach.

Trond: If we allow other processes to dirty pages while a process is doing
a consistency sync to disk, we can end up never making progress.

Peter: Attached is a patch which addresses a continuing problem with
the NFS client generating out of order WRITE requests.  While
this is compliant with all of the current protocol
specifications, there are servers in the market which can not
handle out of order WRITE requests very well.  Also, this may
lead to sub-optimal block allocations in the underlying file
system on the server.  This may cause the read throughputs to
be reduced when reading the file from the server.

Peter: There has been a lot of work recently done to address out of
order issues on a systemic level.  However, the NFS client is
still susceptible to the problem.  Out of order WRITE
requests can occur when pdflush is in the middle of writing
out pages while the process dirtying the pages calls
generic_file_buffered_write which calls
generic_perform_write which calls
balance_dirty_pages_rate_limited which ends up calling
writeback_inodes which ends up calling back into the NFS
client to writes out dirty pages for the same file that
pdflush happens to be working with.

Signed-off-by: Peter Staubach <staubach@redhat.com>
[modification by Trond to merge the two similar patches]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/file.c          |  9 +++++++++
 fs/nfs/inode.c         | 12 ++++++++++++
 fs/nfs/internal.h      |  1 +
 fs/nfs/nfs4proc.c      | 10 +---------
 fs/nfs/pagelist.c      | 11 -----------
 fs/nfs/write.c         | 28 +++++++++++++++++++---------
 include/linux/nfs_fs.h |  1 +
 7 files changed, 43 insertions(+), 29 deletions(-)

(limited to 'fs/nfs/inode.c')

diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 90f292b520d..404c19c866a 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -354,6 +354,15 @@ static int nfs_write_begin(struct file *file, struct address_space *mapping,
 		file->f_path.dentry->d_name.name,
 		mapping->host->i_ino, len, (long long) pos);
 
+	/*
+	 * Prevent starvation issues if someone is doing a consistency
+	 * sync-to-disk
+	 */
+	ret = wait_on_bit(&NFS_I(mapping->host)->flags, NFS_INO_FLUSHING,
+			nfs_wait_bit_killable, TASK_KILLABLE);
+	if (ret)
+		return ret;
+
 	page = grab_cache_page_write_begin(mapping, index, flags);
 	if (!page)
 		return -ENOMEM;
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 00f116cdadc..c40adc5dd60 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -65,6 +65,18 @@ nfs_fattr_to_ino_t(struct nfs_fattr *fattr)
 	return nfs_fileid_to_ino_t(fattr->fileid);
 }
 
+/**
+ * nfs_wait_bit_killable - helper for functions that are sleeping on bit locks
+ * @word: long word containing the bit lock
+ */
+int nfs_wait_bit_killable(void *word)
+{
+	if (fatal_signal_pending(current))
+		return -ERESTARTSYS;
+	schedule();
+	return 0;
+}
+
 /**
  * nfs_compat_user_ino64 - returns the user-visible inode number
  * @fileid: 64-bit fileid
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index 340ede8f608..a55e69aa52e 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -165,6 +165,7 @@ extern void nfs_clear_inode(struct inode *);
 extern void nfs4_clear_inode(struct inode *);
 #endif
 void nfs_zap_acl_cache(struct inode *inode);
+extern int nfs_wait_bit_killable(void *word);
 
 /* super.c */
 void nfs_parse_ip_address(char *, size_t, struct sockaddr *, size_t *);
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 101f5f4c304..95f171e7e05 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -193,14 +193,6 @@ static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dent
 	kunmap_atomic(start, KM_USER0);
 }
 
-static int nfs4_wait_bit_killable(void *word)
-{
-	if (fatal_signal_pending(current))
-		return -ERESTARTSYS;
-	schedule();
-	return 0;
-}
-
 static int nfs4_wait_clnt_recover(struct nfs_client *clp)
 {
 	int res;
@@ -208,7 +200,7 @@ static int nfs4_wait_clnt_recover(struct nfs_client *clp)
 	might_sleep();
 
 	res = wait_on_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING,
-			nfs4_wait_bit_killable, TASK_KILLABLE);
+			nfs_wait_bit_killable, TASK_KILLABLE);
 	return res;
 }
 
diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
index 7f079209d70..e2975939126 100644
--- a/fs/nfs/pagelist.c
+++ b/fs/nfs/pagelist.c
@@ -176,17 +176,6 @@ void nfs_release_request(struct nfs_page *req)
 	kref_put(&req->wb_kref, nfs_free_request);
 }
 
-static int nfs_wait_bit_killable(void *word)
-{
-	int ret = 0;
-
-	if (fatal_signal_pending(current))
-		ret = -ERESTARTSYS;
-	else
-		schedule();
-	return ret;
-}
-
 /**
  * nfs_wait_on_request - Wait for a request to complete.
  * @req: request to wait upon.
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 1a999939fed..36fd35e0de8 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -313,19 +313,34 @@ static int nfs_writepages_callback(struct page *page, struct writeback_control *
 int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
 {
 	struct inode *inode = mapping->host;
+	unsigned long *bitlock = &NFS_I(inode)->flags;
 	struct nfs_pageio_descriptor pgio;
 	int err;
 
+	/* Stop dirtying of new pages while we sync */
+	err = wait_on_bit_lock(bitlock, NFS_INO_FLUSHING,
+			nfs_wait_bit_killable, TASK_KILLABLE);
+	if (err)
+		goto out_err;
+
 	nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGES);
 
 	nfs_pageio_init_write(&pgio, inode, wb_priority(wbc));
 	err = write_cache_pages(mapping, wbc, nfs_writepages_callback, &pgio);
 	nfs_pageio_complete(&pgio);
+
+	clear_bit_unlock(NFS_INO_FLUSHING, bitlock);
+	smp_mb__after_clear_bit();
+	wake_up_bit(bitlock, NFS_INO_FLUSHING);
+
 	if (err < 0)
-		return err;
-	if (pgio.pg_error < 0)
-		return pgio.pg_error;
+		goto out_err;
+	err = pgio.pg_error;
+	if (err < 0)
+		goto out_err;
 	return 0;
+out_err:
+	return err;
 }
 
 /*
@@ -1432,18 +1447,13 @@ static int nfs_write_mapping(struct address_space *mapping, int how)
 {
 	struct writeback_control wbc = {
 		.bdi = mapping->backing_dev_info,
-		.sync_mode = WB_SYNC_NONE,
+		.sync_mode = WB_SYNC_ALL,
 		.nr_to_write = LONG_MAX,
 		.range_start = 0,
 		.range_end = LLONG_MAX,
 		.for_writepages = 1,
 	};
-	int ret;
 
-	ret = __nfs_write_mapping(mapping, &wbc, how);
-	if (ret < 0)
-		return ret;
-	wbc.sync_mode = WB_SYNC_ALL;
 	return __nfs_write_mapping(mapping, &wbc, how);
 }
 
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index c9fecd3e8f0..933bc261c0d 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -206,6 +206,7 @@ struct nfs_inode {
 #define NFS_INO_STALE		(1)		/* possible stale inode */
 #define NFS_INO_ACL_LRU_SET	(2)		/* Inode is on the LRU list */
 #define NFS_INO_MOUNTPOINT	(3)		/* inode is remote mountpoint */
+#define NFS_INO_FLUSHING	(4)		/* inode is flushing out data */
 
 static inline struct nfs_inode *NFS_I(const struct inode *inode)
 {
-- 
cgit v1.2.3-70-g09d2


From 7fe5c398fc2186ed586db11106a6692d871d0d58 Mon Sep 17 00:00:00 2001
From: Trond Myklebust <Trond.Myklebust@netapp.com>
Date: Thu, 19 Mar 2009 15:35:50 -0400
Subject: NFS: Optimise NFS close()

Close-to-open cache consistency rules really only require us to flush out
writes on calls to close(), and require us to revalidate attributes on the
very last close of the file.

Currently we appear to be doing a lot of extra attribute revalidation
and cache flushes.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/file.c           | 11 ++---------
 fs/nfs/inode.c          | 41 +++++++++++++++++++++++++++++------------
 fs/nfs/internal.h       |  3 +++
 fs/nfs/nfs3proc.c       |  1 +
 fs/nfs/nfs4proc.c       | 10 ++++++++++
 fs/nfs/proc.c           |  1 +
 include/linux/nfs_xdr.h |  1 +
 7 files changed, 47 insertions(+), 21 deletions(-)

(limited to 'fs/nfs/inode.c')

diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 1eab9c9ad24..d451073c494 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -137,9 +137,6 @@ nfs_file_release(struct inode *inode, struct file *filp)
 			dentry->d_parent->d_name.name,
 			dentry->d_name.name);
 
-	/* Ensure that dirty pages are flushed out with the right creds */
-	if (filp->f_mode & FMODE_WRITE)
-		nfs_wb_all(dentry->d_inode);
 	nfs_inc_stats(inode, NFSIOS_VFSRELEASE);
 	return nfs_release(inode, filp);
 }
@@ -231,7 +228,6 @@ nfs_file_flush(struct file *file, fl_owner_t id)
 	struct nfs_open_context *ctx = nfs_file_open_context(file);
 	struct dentry	*dentry = file->f_path.dentry;
 	struct inode	*inode = dentry->d_inode;
-	int		status;
 
 	dprintk("NFS: flush(%s/%s)\n",
 			dentry->d_parent->d_name.name,
@@ -241,11 +237,8 @@ nfs_file_flush(struct file *file, fl_owner_t id)
 		return 0;
 	nfs_inc_stats(inode, NFSIOS_VFSFLUSH);
 
-	/* Ensure that data+attribute caches are up to date after close() */
-	status = nfs_do_fsync(ctx, inode);
-	if (!status)
-		nfs_revalidate_inode(NFS_SERVER(inode), inode);
-	return status;
+	/* Flush writes to the server and return any errors */
+	return nfs_do_fsync(ctx, inode);
 }
 
 static ssize_t
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index c40adc5dd60..a834d1d850b 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -541,6 +541,32 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
 	return err;
 }
 
+/**
+ * nfs_close_context - Common close_context() routine NFSv2/v3
+ * @ctx: pointer to context
+ * @is_sync: is this a synchronous close
+ *
+ * always ensure that the attributes are up to date if we're mounted
+ * with close-to-open semantics
+ */
+void nfs_close_context(struct nfs_open_context *ctx, int is_sync)
+{
+	struct inode *inode;
+	struct nfs_server *server;
+
+	if (!(ctx->mode & FMODE_WRITE))
+		return;
+	if (!is_sync)
+		return;
+	inode = ctx->path.dentry->d_inode;
+	if (!list_empty(&NFS_I(inode)->open_files))
+		return;
+	server = NFS_SERVER(inode);
+	if (server->flags & NFS_MOUNT_NOCTO)
+		return;
+	nfs_revalidate_inode(server, inode);
+}
+
 static struct nfs_open_context *alloc_nfs_open_context(struct vfsmount *mnt, struct dentry *dentry, struct rpc_cred *cred)
 {
 	struct nfs_open_context *ctx;
@@ -567,24 +593,15 @@ struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx)
 	return ctx;
 }
 
-static void __put_nfs_open_context(struct nfs_open_context *ctx, int wait)
+static void __put_nfs_open_context(struct nfs_open_context *ctx, int is_sync)
 {
-	struct inode *inode;
+	struct inode *inode = ctx->path.dentry->d_inode;
 
-	if (ctx == NULL)
-		return;
-
-	inode = ctx->path.dentry->d_inode;
 	if (!atomic_dec_and_lock(&ctx->count, &inode->i_lock))
 		return;
 	list_del(&ctx->list);
 	spin_unlock(&inode->i_lock);
-	if (ctx->state != NULL) {
-		if (wait)
-			nfs4_close_sync(&ctx->path, ctx->state, ctx->mode);
-		else
-			nfs4_close_state(&ctx->path, ctx->state, ctx->mode);
-	}
+	NFS_PROTO(inode)->close_context(ctx, is_sync);
 	if (ctx->cred != NULL)
 		put_rpccred(ctx->cred);
 	path_put(&ctx->path);
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index a55e69aa52e..2041f68ff1c 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -152,6 +152,9 @@ extern __be32 *nfs4_decode_dirent(__be32 *p, struct nfs_entry *entry, int plus);
 extern struct rpc_procinfo nfs4_procedures[];
 #endif
 
+/* proc.c */
+void nfs_close_context(struct nfs_open_context *ctx, int is_sync);
+
 /* dir.c */
 extern int nfs_access_cache_shrinker(int nr_to_scan, gfp_t gfp_mask);
 
diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c
index c55be7a7679..b82fe6847f1 100644
--- a/fs/nfs/nfs3proc.c
+++ b/fs/nfs/nfs3proc.c
@@ -834,4 +834,5 @@ const struct nfs_rpc_ops nfs_v3_clientops = {
 	.commit_done	= nfs3_commit_done,
 	.lock		= nfs3_proc_lock,
 	.clear_acl_cache = nfs3_forget_cached_acls,
+	.close_context	= nfs_close_context,
 };
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 95f171e7e05..97bacccff57 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -1572,6 +1572,15 @@ out_drop:
 	return 0;
 }
 
+void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
+{
+	if (ctx->state == NULL)
+		return;
+	if (is_sync)
+		nfs4_close_sync(&ctx->path, ctx->state, ctx->mode);
+	else
+		nfs4_close_state(&ctx->path, ctx->state, ctx->mode);
+}
 
 static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
 {
@@ -3776,6 +3785,7 @@ const struct nfs_rpc_ops nfs_v4_clientops = {
 	.commit_done	= nfs4_commit_done,
 	.lock		= nfs4_proc_lock,
 	.clear_acl_cache = nfs4_zap_acl_attr,
+	.close_context  = nfs4_close_context,
 };
 
 /*
diff --git a/fs/nfs/proc.c b/fs/nfs/proc.c
index 193465210d7..7be72d90d49 100644
--- a/fs/nfs/proc.c
+++ b/fs/nfs/proc.c
@@ -663,4 +663,5 @@ const struct nfs_rpc_ops nfs_v2_clientops = {
 	.commit_setup	= nfs_proc_commit_setup,
 	.lock		= nfs_proc_lock,
 	.lock_check_bounds = nfs_lock_check_bounds,
+	.close_context	= nfs_close_context,
 };
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 0691b9c188d..9708e78a4d4 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -868,6 +868,7 @@ struct nfs_rpc_ops {
 	int	(*lock)(struct file *, int, struct file_lock *);
 	int	(*lock_check_bounds)(const struct file_lock *);
 	void	(*clear_acl_cache)(struct inode *);
+	void	(*close_context)(struct nfs_open_context *ctx, int);
 };
 
 /*
-- 
cgit v1.2.3-70-g09d2