glusterfs/0380-features-shard-Aggregate-size-block-count-in-iatt-be.patch

307 lines
10 KiB
Diff
Raw Normal View History

From 2cf22e54c8424949607c4a20df84887b838b2702 Mon Sep 17 00:00:00 2001
From: Krutika Dhananjay <kdhananj@redhat.com>
Date: Fri, 15 May 2020 11:29:36 +0530
Subject: [PATCH 380/382] features/shard: Aggregate size, block-count in iatt
before unwinding setxattr
Backport of:
> Upstream patch - https://review.gluster.org/c/glusterfs/+/24471
> Fixes: #1243
> Change-Id: I4da0eceb4235b91546df79270bcc0af8cd64e9ea
> Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Posix translator returns pre and postbufs in the dict in {F}SETXATTR fops.
These iatts are further cached at layers like md-cache.
Shard translator, in its current state, simply returns these values without
updating the aggregated file size and block-count.
This patch fixes this problem.
Change-Id: I4da0eceb4235b91546df79270bcc0af8cd64e9ea
BUG: 1823423
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/201135
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Xavi Hernandez Juan <xhernandez@redhat.com>
---
tests/bugs/shard/issue-1243.t | 31 ++++++
xlators/features/shard/src/shard.c | 218 +++++++++++++++++++++++++++++++++----
2 files changed, 225 insertions(+), 24 deletions(-)
create mode 100644 tests/bugs/shard/issue-1243.t
diff --git a/tests/bugs/shard/issue-1243.t b/tests/bugs/shard/issue-1243.t
new file mode 100644
index 0000000..b0c092c
--- /dev/null
+++ b/tests/bugs/shard/issue-1243.t
@@ -0,0 +1,31 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+
+cleanup;
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 $H0:$B0/${V0}{0,1}
+TEST $CLI volume set $V0 features.shard on
+TEST $CLI volume set $V0 features.shard-block-size 4MB
+TEST $CLI volume set $V0 performance.quick-read off
+TEST $CLI volume set $V0 performance.io-cache off
+TEST $CLI volume set $V0 performance.read-ahead off
+TEST $CLI volume set $V0 performance.strict-o-direct on
+TEST $CLI volume start $V0
+
+TEST $GFS --volfile-id=$V0 --volfile-server=$H0 $M0
+
+TEST $CLI volume set $V0 md-cache-timeout 10
+
+# Write data into a file such that its size crosses shard-block-size
+TEST dd if=/dev/zero of=$M0/foo bs=1048576 count=8 oflag=direct
+
+# Execute a setxattr on the file.
+TEST setfattr -n trusted.libvirt -v some-value $M0/foo
+
+# Size of the file should be the aggregated size, not the shard-block-size
+EXPECT '8388608' stat -c %s $M0/foo
+
+cleanup
diff --git a/xlators/features/shard/src/shard.c b/xlators/features/shard/src/shard.c
index ee38ed2..6ae4c41 100644
--- a/xlators/features/shard/src/shard.c
+++ b/xlators/features/shard/src/shard.c
@@ -5929,36 +5929,206 @@ out:
return 0;
}
-int32_t shard_fsetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd,
- dict_t *dict, int32_t flags, dict_t *xdata) {
- int op_errno = EINVAL;
+int32_t shard_common_set_xattr_cbk(call_frame_t *frame, void *cookie,
+ xlator_t *this, int32_t op_ret,
+ int32_t op_errno, dict_t *xdata) {
+ int ret = -1;
+ struct iatt *prebuf = NULL;
+ struct iatt *postbuf = NULL;
+ struct iatt *stbuf = NULL;
+ data_t *data = NULL;
+ shard_local_t *local = NULL;
- if (frame->root->pid != GF_CLIENT_PID_GSYNCD) {
- GF_IF_INTERNAL_XATTR_GOTO(SHARD_XATTR_PREFIX "*", dict, op_errno, out);
- }
+ local = frame->local;
- STACK_WIND_TAIL(frame, FIRST_CHILD(this), FIRST_CHILD(this)->fops->fsetxattr,
- fd, dict, flags, xdata);
- return 0;
-out:
- shard_common_failure_unwind(GF_FOP_FSETXATTR, frame, -1, op_errno);
- return 0;
+ if (op_ret < 0) {
+ local->op_ret = op_ret;
+ local->op_errno = op_errno;
+ goto err;
+ }
+
+ if (!xdata)
+ goto unwind;
+
+ data = dict_get(xdata, GF_PRESTAT);
+ if (data) {
+ stbuf = data_to_iatt(data, GF_PRESTAT);
+ prebuf = GF_MALLOC(sizeof(struct iatt), gf_common_mt_char);
+ if (prebuf == NULL) {
+ local->op_ret = -1;
+ local->op_errno = ENOMEM;
+ goto err;
+ }
+ *prebuf = *stbuf;
+ prebuf->ia_size = local->prebuf.ia_size;
+ prebuf->ia_blocks = local->prebuf.ia_blocks;
+ ret = dict_set_iatt(xdata, GF_PRESTAT, prebuf, false);
+ if (ret < 0) {
+ local->op_ret = -1;
+ local->op_errno = ENOMEM;
+ goto err;
+ }
+ prebuf = NULL;
+ }
+
+ data = dict_get(xdata, GF_POSTSTAT);
+ if (data) {
+ stbuf = data_to_iatt(data, GF_POSTSTAT);
+ postbuf = GF_MALLOC(sizeof(struct iatt), gf_common_mt_char);
+ if (postbuf == NULL) {
+ local->op_ret = -1;
+ local->op_errno = ENOMEM;
+ goto err;
+ }
+ *postbuf = *stbuf;
+ postbuf->ia_size = local->prebuf.ia_size;
+ postbuf->ia_blocks = local->prebuf.ia_blocks;
+ ret = dict_set_iatt(xdata, GF_POSTSTAT, postbuf, false);
+ if (ret < 0) {
+ local->op_ret = -1;
+ local->op_errno = ENOMEM;
+ goto err;
+ }
+ postbuf = NULL;
+ }
+
+unwind:
+ if (local->fd)
+ SHARD_STACK_UNWIND(fsetxattr, frame, local->op_ret, local->op_errno,
+ xdata);
+ else
+ SHARD_STACK_UNWIND(setxattr, frame, local->op_ret, local->op_errno,
+ xdata);
+ return 0;
+
+err:
+ GF_FREE(prebuf);
+ GF_FREE(postbuf);
+ shard_common_failure_unwind(local->fop, frame, local->op_ret,
+ local->op_errno);
+ return 0;
}
-int32_t shard_setxattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
- dict_t *dict, int32_t flags, dict_t *xdata) {
- int op_errno = EINVAL;
+int32_t shard_post_lookup_set_xattr_handler(call_frame_t *frame,
+ xlator_t *this) {
+ shard_local_t *local = NULL;
- if (frame->root->pid != GF_CLIENT_PID_GSYNCD) {
- GF_IF_INTERNAL_XATTR_GOTO(SHARD_XATTR_PREFIX "*", dict, op_errno, out);
- }
+ local = frame->local;
- STACK_WIND_TAIL(frame, FIRST_CHILD(this), FIRST_CHILD(this)->fops->setxattr,
- loc, dict, flags, xdata);
- return 0;
-out:
- shard_common_failure_unwind(GF_FOP_SETXATTR, frame, -1, op_errno);
- return 0;
+ if (local->op_ret < 0) {
+ shard_common_failure_unwind(local->fop, frame, local->op_ret,
+ local->op_errno);
+ return 0;
+ }
+
+ if (local->fd)
+ STACK_WIND(frame, shard_common_set_xattr_cbk, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->fsetxattr, local->fd,
+ local->xattr_req, local->flags, local->xattr_rsp);
+ else
+ STACK_WIND(frame, shard_common_set_xattr_cbk, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->setxattr, &local->loc,
+ local->xattr_req, local->flags, local->xattr_rsp);
+ return 0;
+}
+
+int32_t shard_common_set_xattr(call_frame_t *frame, xlator_t *this,
+ glusterfs_fop_t fop, loc_t *loc, fd_t *fd,
+ dict_t *dict, int32_t flags, dict_t *xdata) {
+ int ret = -1;
+ int op_errno = ENOMEM;
+ uint64_t block_size = 0;
+ shard_local_t *local = NULL;
+ inode_t *inode = loc ? loc->inode : fd->inode;
+
+ if ((IA_ISDIR(inode->ia_type)) || (IA_ISLNK(inode->ia_type))) {
+ if (loc)
+ STACK_WIND_TAIL(frame, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->setxattr, loc, dict, flags,
+ xdata);
+ else
+ STACK_WIND_TAIL(frame, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->fsetxattr, fd, dict, flags,
+ xdata);
+ return 0;
+ }
+
+ /* Sharded or not, if shard's special xattrs are attempted to be set,
+ * fail the fop with EPERM (except if the client is gsyncd.
+ */
+ if (frame->root->pid != GF_CLIENT_PID_GSYNCD) {
+ GF_IF_INTERNAL_XATTR_GOTO(SHARD_XATTR_PREFIX "*", dict, op_errno, err);
+ }
+
+ ret = shard_inode_ctx_get_block_size(inode, this, &block_size);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, 0, SHARD_MSG_INODE_CTX_GET_FAILED,
+ "Failed to get block size from inode ctx of %s",
+ uuid_utoa(inode->gfid));
+ goto err;
+ }
+
+ if (!block_size || frame->root->pid == GF_CLIENT_PID_GSYNCD) {
+ if (loc)
+ STACK_WIND_TAIL(frame, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->setxattr, loc, dict, flags,
+ xdata);
+ else
+ STACK_WIND_TAIL(frame, FIRST_CHILD(this),
+ FIRST_CHILD(this)->fops->fsetxattr, fd, dict, flags,
+ xdata);
+ return 0;
+ }
+
+ local = mem_get0(this->local_pool);
+ if (!local)
+ goto err;
+
+ frame->local = local;
+ local->fop = fop;
+ if (loc) {
+ if (loc_copy(&local->loc, loc) != 0)
+ goto err;
+ }
+
+ if (fd) {
+ local->fd = fd_ref(fd);
+ local->loc.inode = inode_ref(fd->inode);
+ gf_uuid_copy(local->loc.gfid, fd->inode->gfid);
+ }
+ local->flags = flags;
+ /* Reusing local->xattr_req and local->xattr_rsp to store the setxattr dict
+ * and the xdata dict
+ */
+ if (dict)
+ local->xattr_req = dict_ref(dict);
+ if (xdata)
+ local->xattr_rsp = dict_ref(xdata);
+
+ /* To-Do: Switch from LOOKUP which is path-based, to FSTAT if the fop is
+ * on an fd. This comes under a generic class of bugs in shard tracked by
+ * bz #1782428.
+ */
+ shard_lookup_base_file(frame, this, &local->loc,
+ shard_post_lookup_set_xattr_handler);
+ return 0;
+err:
+ shard_common_failure_unwind(fop, frame, -1, op_errno);
+ return 0;
+}
+
+int32_t shard_fsetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd,
+ dict_t *dict, int32_t flags, dict_t *xdata) {
+ shard_common_set_xattr(frame, this, GF_FOP_FSETXATTR, NULL, fd, dict, flags,
+ xdata);
+ return 0;
+}
+
+int32_t shard_setxattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
+ dict_t *dict, int32_t flags, dict_t *xdata) {
+ shard_common_set_xattr(frame, this, GF_FOP_SETXATTR, loc, NULL, dict, flags,
+ xdata);
+ return 0;
}
int shard_post_setattr_handler(call_frame_t *frame, xlator_t *this) {
--
1.8.3.1