cluster/afr: Don't let NFS cache stat after writes

Problem:
Afr does post-ops after write but the stat buffer it unwinds is at the
time of write, so if nfs client caches this, it will see different
ctime when it does stat on it after post-op is done. From NFS client's
perspective it thinks the file is changed. Tar which depends on this
to be correct keeps giving 'file changed as we read it' warning.
If Afr instead has to choose to unwind after post-op, eager-lock,
delayed-post-op will have to be disabled which will lead to bad
performance for all write usecases.

Fix:
Don't let client cache stat after write.

Change-Id: Ic6062acc6e5cdd97a9c83c56bd529ec83cee8a23
BUG: 1302948
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Signed-off-by: Anuradha Talur <atalur@redhat.com>
Reviewed-on: http://review.gluster.org/13785
Smoke: Gluster Build System <jenkins@build.gluster.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Niels de Vos <ndevos@redhat.com>

7 files changed, 71 insertions, 30 deletions

diff --git a/xlators/cluster/afr/src/afr-dir-write.c b/xlators/cluster/afr/src/afr-dir-write.c
index 62ab54ae0ca..55aec7429a7 100644
--- a/xlators/cluster/afr/src/afr-dir-write.c
+++ b/xlators/cluster/afr/src/afr-dir-write.c
@@ -234,7 +234,9 @@ __afr_dir_write_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
         afr_local_t *local = NULL;
         int child_index = (long) cookie;
         int call_count = -1;
+        afr_private_t *priv = NULL;
 
+        priv  = this->private;
         local = frame->local;
 
 	LOCK (&frame->lock);
@@ -249,8 +251,13 @@ __afr_dir_write_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
         if (call_count == 0) {
 		__afr_dir_write_finalize (frame, this);
 
-		if (afr_txn_nothing_failed (frame, this))
-			local->transaction.unwind (frame, this);
+		if (afr_txn_nothing_failed (frame, this)) {
+                        /*if it did pre-op, it will do post-op changing ctime*/
+                        if (priv->consistent_metadata &&
+                            afr_needs_changelog_update (local))
+                                afr_zero_fill_stat (local);
+                        local->transaction.unwind (frame, this);
+                }
 
 		afr_mark_entry_pending_changelog (frame, this);
 
diff --git a/xlators/cluster/afr/src/afr-inode-write.c b/xlators/cluster/afr/src/afr-inode-write.c
index f240b5eec39..36889429657 100644
--- a/xlators/cluster/afr/src/afr-inode-write.c
+++ b/xlators/cluster/afr/src/afr-inode-write.c
@@ -182,7 +182,9 @@ __afr_inode_write_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
         afr_local_t *local = NULL;
         int child_index = (long) cookie;
         int call_count = -1;
+        afr_private_t *priv = NULL;
 
+        priv = this->private;
         local = frame->local;
 
         LOCK (&frame->lock);
@@ -198,8 +200,13 @@ __afr_inode_write_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
         if (call_count == 0) {
 		__afr_inode_write_finalize (frame, this);
 
-		if (afr_txn_nothing_failed (frame, this))
-			local->transaction.unwind (frame, this);
+		if (afr_txn_nothing_failed (frame, this)) {
+                        /*if it did pre-op, it will do post-op changing ctime*/
+                        if (priv->consistent_metadata &&
+                            afr_needs_changelog_update (local))
+                                afr_zero_fill_stat (local);
+                        local->transaction.unwind (frame, this);
+                }
 
                 local->transaction.resume (frame, this);
         }
@@ -230,8 +237,13 @@ void
 afr_writev_unwind (call_frame_t *frame, xlator_t *this)
 {
         afr_local_t *   local = NULL;
+        afr_private_t *priv = this->private;
+
         local = frame->local;
 
+       if (priv->consistent_metadata)
+               afr_zero_fill_stat (local);
+
         AFR_STACK_UNWIND (writev, frame,
                           local->op_ret, local->op_errno,
                           &local->cont.inode_wfop.prebuf,
diff --git a/xlators/cluster/afr/src/afr-transaction.c b/xlators/cluster/afr/src/afr-transaction.c
index 2acac027122..6fd44ce79f6 100644
--- a/xlators/cluster/afr/src/afr-transaction.c
+++ b/xlators/cluster/afr/src/afr-transaction.c
@@ -36,6 +36,37 @@ afr_changelog_do (call_frame_t *frame, xlator_t *this, dict_t *xattr,
 		  afr_changelog_resume_t changelog_resume,
                   afr_xattrop_type_t op);
 
+void
+afr_zero_fill_stat (afr_local_t *local)
+{
+        if (!local)
+                return;
+        if (local->transaction.type == AFR_DATA_TRANSACTION ||
+            local->transaction.type == AFR_METADATA_TRANSACTION) {
+                gf_zero_fill_stat (&local->cont.inode_wfop.prebuf);
+                gf_zero_fill_stat (&local->cont.inode_wfop.postbuf);
+        } else if (local->transaction.type == AFR_ENTRY_TRANSACTION ||
+                   local->transaction.type == AFR_ENTRY_RENAME_TRANSACTION) {
+                gf_zero_fill_stat (&local->cont.dir_fop.buf);
+                gf_zero_fill_stat (&local->cont.dir_fop.preparent);
+                gf_zero_fill_stat (&local->cont.dir_fop.postparent);
+                if (local->transaction.type == AFR_ENTRY_TRANSACTION)
+                        return;
+                gf_zero_fill_stat (&local->cont.dir_fop.prenewparent);
+                gf_zero_fill_stat (&local->cont.dir_fop.postnewparent);
+        }
+}
+
+gf_boolean_t
+afr_needs_changelog_update (afr_local_t *local)
+{
+        if (local->transaction.type == AFR_DATA_TRANSACTION)
+                return _gf_true;
+        if (!local->optimistic_change_log)
+                return _gf_true;
+        return _gf_false;
+}
+
 static int32_t
 afr_quorum_errno (afr_private_t *priv)
 {
@@ -85,9 +116,21 @@ int
 __afr_txn_write_done (call_frame_t *frame, xlator_t *this)
 {
         afr_local_t *local = NULL;
+        afr_private_t *priv = NULL;
+        gf_boolean_t unwind = _gf_false;
 
+        priv  = this->private;
         local = frame->local;
 
+        if (priv->consistent_metadata) {
+                LOCK (&frame->lock);
+                {
+                        unwind = (local->transaction.main_frame != NULL);
+                }
+                UNLOCK (&frame->lock);
+                if (unwind)/*It definitely did post-op*/
+                        afr_zero_fill_stat (local);
+        }
         local->transaction.unwind (frame, this);
 
         AFR_STACK_DESTROY (frame);
@@ -1232,8 +1275,7 @@ afr_changelog_pre_op (call_frame_t *frame, xlator_t *this)
 		goto err;
 	}
 
-	if ((local->transaction.type == AFR_DATA_TRANSACTION ||
-	     !local->optimistic_change_log)) {
+	if (afr_needs_changelog_update (local)) {
 
 		local->dirty[idx] = hton32(1);
 
diff --git a/xlators/cluster/afr/src/afr-transaction.h b/xlators/cluster/afr/src/afr-transaction.h
index 47d43d88991..c58531eff44 100644
--- a/xlators/cluster/afr/src/afr-transaction.h
+++ b/xlators/cluster/afr/src/afr-transaction.h
@@ -52,5 +52,7 @@ int __afr_txn_write_fop (call_frame_t *frame, xlator_t *this);
 int __afr_txn_write_done (call_frame_t *frame, xlator_t *this);
 call_frame_t *afr_transaction_detach_fop_frame (call_frame_t *frame);
 gf_boolean_t afr_has_quorum (unsigned char *subvols, xlator_t *this);
+gf_boolean_t afr_needs_changelog_update (afr_local_t *local);
+void afr_zero_fill_stat (afr_local_t *local);
 
 #endif /* __TRANSACTION_H__ */
diff --git a/xlators/nfs/server/src/nfs-common.c b/xlators/nfs/server/src/nfs-common.c
index 51a2b7e36f7..d9ea1e1ac47 100644
--- a/xlators/nfs/server/src/nfs-common.c
+++ b/xlators/nfs/server/src/nfs-common.c
@@ -106,25 +106,6 @@ nfs_mntpath_to_xlator (xlator_list_t *cl, char *path)
 }
 
 
-/* Returns 1 if the stat seems to be filled with zeroes. */
-int
-nfs_zero_filled_stat (struct iatt *buf)
-{
-        if (!buf)
-                return 1;
-
-        /* Do not use st_dev because it is transformed to store the xlator id
-         * in place of the device number. Do not use st_ino because by this time
-         * we've already mapped the root ino to 1 so it is not guaranteed to be
-         * 0.
-         */
-        if ((buf->ia_nlink == 0) && (buf->ia_ctime == 0))
-                return 1;
-
-        return 0;
-}
-
-
 void
 nfs_loc_wipe (loc_t *loc)
 {
diff --git a/xlators/nfs/server/src/nfs-common.h b/xlators/nfs/server/src/nfs-common.h
index fa7f4ebf212..77bdfb0bbf0 100644
--- a/xlators/nfs/server/src/nfs-common.h
+++ b/xlators/nfs/server/src/nfs-common.h
@@ -37,9 +37,6 @@ nfs_path_to_xlator (xlator_list_t *cl, char *path);
 extern xlator_t *
 nfs_mntpath_to_xlator (xlator_list_t *cl, char *path);
 
-extern int
-nfs_zero_filled_stat (struct iatt *buf);
-
 extern void
 nfs_loc_wipe (loc_t *loc);
 
diff --git a/xlators/nfs/server/src/nfs3-helpers.c b/xlators/nfs/server/src/nfs3-helpers.c
index 7eb491142f7..ad4c87e69d7 100644
--- a/xlators/nfs/server/src/nfs3-helpers.c
+++ b/xlators/nfs/server/src/nfs3-helpers.c
@@ -373,7 +373,7 @@ nfs3_stat_to_post_op_attr (struct iatt *buf)
          * returning these zeroed out attrs.
          */
         attr.attributes_follow = FALSE;
-        if (nfs_zero_filled_stat (buf))
+        if (gf_is_zero_filled_stat (buf))
                 goto out;
 
         nfs3_stat_to_fattr3 (buf, &(attr.post_op_attr_u.attributes));
@@ -394,7 +394,7 @@ nfs3_stat_to_pre_op_attr (struct iatt *pre)
          * returning these zeroed out attrs.
          */
         poa.attributes_follow = FALSE;
-        if (nfs_zero_filled_stat (pre))
+        if (gf_is_zero_filled_stat (pre))
                 goto out;
 
         poa.attributes_follow = TRUE;