rpms/glusterfs
5
0
Fork 0
Code Issues Pull Requests Releases Activity

autobuild v6.0-11

Browse Source 
Resolves: bz#1704562 bz#1733520 bz#1734423
Signed-off-by: Sunil Kumar Acharya <sheggodu@redhat.com>
This commit is contained in:
Sunil Kumar Acharya 2019-08-05 21:15:42 -04:00
parent ec44dc53fe
commit 4797c9e01f
4 changed files with 388 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

75
0263-tests-fix-ctime-related-tests.patch Normal file
View File

@ -0,0 +1,75 @@
From 427dab431f7e8c4c8a01e9e9ed0892708a3d22d2 Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Mon, 5 Aug 2019 08:33:29 +0530
Subject: [PATCH 263/265] tests: fix ctime related tests
With ctime being disabled by default, certain tests need to explicitly
turn this option off to sanitize the functionality
Label: DOWNSTREAM ONLY
Change-Id: Id70310b4b09e36bf66756fea447186bb073b5604
BUG: 1704562
Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/177704
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
tests/basic/ctime/ctime-noatime.t | 1 +
tests/basic/ctime/ctime-readdir.t | 1 +
tests/bugs/glusterd/bug-1696046.t | 8 +++++---
3 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/tests/basic/ctime/ctime-noatime.t b/tests/basic/ctime/ctime-noatime.t
index 609ccbd..a6c8d9c 100644
--- a/tests/basic/ctime/ctime-noatime.t
+++ b/tests/basic/ctime/ctime-noatime.t
@@ -20,6 +20,7 @@ function atime_compare {
TEST glusterd
TEST pidof glusterd
TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{0,1,2}
+TEST $CLI volume set $V0 ctime on
TEST $CLI volume set $V0 performance.stat-prefetch off
TEST $CLI volume set $V0 performance.read-ahead off
TEST $CLI volume set $V0 performance.quick-read off
diff --git a/tests/basic/ctime/ctime-readdir.t b/tests/basic/ctime/ctime-readdir.t
index 4564fc1..fa069b3 100644
--- a/tests/basic/ctime/ctime-readdir.t
+++ b/tests/basic/ctime/ctime-readdir.t
@@ -9,6 +9,7 @@ TEST glusterd
TEST $CLI volume create $V0 replica 3 ${H0}:$B0/brick{1,2,3};
TEST $CLI volume set $V0 performance.stat-prefetch on
+TEST $CLI volume set $V0 ctime on
TEST $CLI volume set $V0 performance.readdir-ahead off
TEST $CLI volume start $V0;
diff --git a/tests/bugs/glusterd/bug-1696046.t b/tests/bugs/glusterd/bug-1696046.t
index e1c1eb2..f7992f5 100644
--- a/tests/bugs/glusterd/bug-1696046.t
+++ b/tests/bugs/glusterd/bug-1696046.t
@@ -22,6 +22,8 @@ TEST pidof glusterd;
TEST $CLI volume set all cluster.brick-multiplex on
TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{1,2,3};
TEST $CLI volume create $V1 replica 3 $H0:$B0/${V1}{1,2,3};
+TEST $CLI volume set $V0 ctime on
+TEST $CLI volume set $V1 ctime on
## Start volume and verify
TEST $CLI volume start $V0;
@@ -64,9 +66,9 @@ TEST $CLI volume set $V0 diagnostics.brick-log-level DEBUG
# Do some operation
touch $M0/file1
-# Check debug message debug message should be exist only for V0
-# Server xlator is common in brick_mux so after enabling DEBUG log
-# some debug message should be available for other xlators like posix
+# Check debug message should exist only for V0 server xlator is common in
+# brick_mux so after enabling DEBUG log some debug message should be available
+# for other xlators like posix
brick_log_file=$logdir"/bricks/"`brick-log-file-name`
nofdlog=$(cat $brick_log_file | grep file1 | grep -v server | wc -l)
--
1.8.3.1

259
0264-gfapi-Fix-deadlock-while-processing-upcall.patch Normal file
View File

@ -0,0 +1,259 @@
From 52dc121c412de9c1cc3058d782b949dc7b25dc3e Mon Sep 17 00:00:00 2001
From: Soumya Koduri <skoduri@redhat.com>
Date: Thu, 25 Jul 2019 12:56:12 +0530
Subject: [PATCH 264/265] gfapi: Fix deadlock while processing upcall
As mentioned in bug1733166, there could be potential deadlock
while processing upcalls depending on how each xlator choose
to act on it. The right way of fixing such issues
is to change rpc callback communication process.
- https://github.com/gluster/glusterfs/issues/697
Till then, making changes in gfapi layer to avoid any I/O
processing.
This is backport of below mainline patch
> https://review.gluster.org/#/c/glusterfs/+/23108/
> bz#1733166
> https://review.gluster.org/#/c/glusterfs/+/23107/ (release-6)
Change-Id: I2079e95339e5d761d5060707f4555cfacab95c83
fixes: bz#1733520
Signed-off-by: Soumya Koduri <skoduri@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/177675
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
api/src/glfs-fops.c | 164 +++++++++++++++++++++++++++++++++++++++++-----------
1 file changed, 131 insertions(+), 33 deletions(-)
diff --git a/api/src/glfs-fops.c b/api/src/glfs-fops.c
index 396f18c..e6adea5 100644
--- a/api/src/glfs-fops.c
+++ b/api/src/glfs-fops.c
@@ -34,7 +34,7 @@
struct upcall_syncop_args {
struct glfs *fs;
- struct glfs_upcall *up_arg;
+ struct gf_upcall upcall_data;
};
#define READDIRBUF_SIZE (sizeof(struct dirent) + GF_NAME_MAX + 1)
@@ -5716,8 +5716,28 @@ out:
static int
upcall_syncop_args_free(struct upcall_syncop_args *args)
{
- if (args && args->up_arg)
- GLFS_FREE(args->up_arg);
+ dict_t *dict = NULL;
+ struct gf_upcall *upcall_data = NULL;
+
+ if (args) {
+ upcall_data = &args->upcall_data;
+ switch (upcall_data->event_type) {
+ case GF_UPCALL_CACHE_INVALIDATION:
+ dict = ((struct gf_upcall_cache_invalidation *)(upcall_data
+ ->data))
+ ->dict;
+ break;
+ case GF_UPCALL_RECALL_LEASE:
+ dict = ((struct gf_upcall_recall_lease *)(upcall_data->data))
+ ->dict;
+ break;
+ }
+ if (dict)
+ dict_unref(dict);
+
+ GF_FREE(upcall_data->client_uid);
+ GF_FREE(upcall_data->data);
+ }
GF_FREE(args);
return 0;
}
@@ -5727,14 +5747,7 @@ glfs_upcall_syncop_cbk(int ret, call_frame_t *frame, void *opaque)
{
struct upcall_syncop_args *args = opaque;
- /* Here we not using upcall_syncop_args_free as application
- * will be cleaning up the args->up_arg using glfs_free
- * post processing upcall.
- */
- if (ret) {
- upcall_syncop_args_free(args);
- } else
- GF_FREE(args);
+ (void)upcall_syncop_args_free(args);
return 0;
}
@@ -5743,29 +5756,17 @@ static int
glfs_cbk_upcall_syncop(void *opaque)
{
struct upcall_syncop_args *args = opaque;
+ struct gf_upcall *upcall_data = NULL;
struct glfs_upcall *up_arg = NULL;
struct glfs *fs;
+ int ret = -1;
fs = args->fs;
- up_arg = args->up_arg;
-
- if (fs->up_cbk && up_arg) {
- (fs->up_cbk)(up_arg, fs->up_data);
- return 0;
- }
-
- return -1;
-}
+ upcall_data = &args->upcall_data;
-static struct upcall_syncop_args *
-upcall_syncop_args_init(struct glfs *fs, struct gf_upcall *upcall_data)
-{
- struct upcall_syncop_args *args = NULL;
- int ret = -1;
- struct glfs_upcall *up_arg = NULL;
-
- if (!fs || !upcall_data)
+ if (!upcall_data) {
goto out;
+ }
up_arg = GLFS_CALLOC(1, sizeof(struct gf_upcall), glfs_release_upcall,
glfs_mt_upcall_entry_t);
@@ -5795,6 +5796,8 @@ upcall_syncop_args_init(struct glfs *fs, struct gf_upcall *upcall_data)
if (up_arg->reason == GLFS_UPCALL_EVENT_NULL) {
gf_msg(THIS->name, GF_LOG_DEBUG, errno, API_MSG_INVALID_ENTRY,
"Upcall_EVENT_NULL received. Skipping it.");
+ ret = 0;
+ GLFS_FREE(up_arg);
goto out;
} else if (ret) {
gf_msg(THIS->name, GF_LOG_ERROR, errno, API_MSG_INVALID_ENTRY,
@@ -5802,6 +5805,85 @@ upcall_syncop_args_init(struct glfs *fs, struct gf_upcall *upcall_data)
goto out;
}
+ if (fs->up_cbk && up_arg)
+ (fs->up_cbk)(up_arg, fs->up_data);
+
+ /* application takes care of calling glfs_free on up_arg post
+ * their processing */
+
+out:
+ return ret;
+}
+
+static struct gf_upcall_cache_invalidation *
+gf_copy_cache_invalidation(struct gf_upcall_cache_invalidation *src)
+{
+ struct gf_upcall_cache_invalidation *dst = NULL;
+
+ if (!src)
+ goto out;
+
+ dst = GF_CALLOC(1, sizeof(struct gf_upcall_cache_invalidation),
+ glfs_mt_upcall_entry_t);
+
+ if (!dst) {
+ gf_msg(THIS->name, GF_LOG_ERROR, ENOMEM, API_MSG_ALLOC_FAILED,
+ "Upcall entry allocation failed.");
+ goto out;
+ }
+
+ dst->flags = src->flags;
+ dst->expire_time_attr = src->expire_time_attr;
+ dst->stat = src->stat;
+ dst->p_stat = src->p_stat;
+ dst->oldp_stat = src->oldp_stat;
+
+ if (src->dict)
+ dst->dict = dict_copy_with_ref(src->dict, NULL);
+
+ return dst;
+out:
+ return NULL;
+}
+
+static struct gf_upcall_recall_lease *
+gf_copy_recall_lease(struct gf_upcall_recall_lease *src)
+{
+ struct gf_upcall_recall_lease *dst = NULL;
+
+ if (!src)
+ goto out;
+
+ dst = GF_CALLOC(1, sizeof(struct gf_upcall_recall_lease),
+ glfs_mt_upcall_entry_t);
+
+ if (!dst) {
+ gf_msg(THIS->name, GF_LOG_ERROR, ENOMEM, API_MSG_ALLOC_FAILED,
+ "Upcall entry allocation failed.");
+ goto out;
+ }
+
+ dst->lease_type = src->lease_type;
+ memcpy(dst->tid, src->tid, 16);
+
+ if (src->dict)
+ dst->dict = dict_copy_with_ref(src->dict, NULL);
+
+ return dst;
+out:
+ return NULL;
+}
+
+static struct upcall_syncop_args *
+upcall_syncop_args_init(struct glfs *fs, struct gf_upcall *upcall_data)
+{
+ struct upcall_syncop_args *args = NULL;
+ int ret = -1;
+ struct gf_upcall *t_data = NULL;
+
+ if (!fs || !upcall_data)
+ goto out;
+
args = GF_CALLOC(1, sizeof(struct upcall_syncop_args),
glfs_mt_upcall_entry_t);
if (!args) {
@@ -5819,15 +5901,31 @@ upcall_syncop_args_init(struct glfs *fs, struct gf_upcall *upcall_data)
* notification without taking any lock/ref.
*/
args->fs = fs;
- args->up_arg = up_arg;
+ t_data = &(args->upcall_data);
+ t_data->client_uid = gf_strdup(upcall_data->client_uid);
- /* application takes care of calling glfs_free on up_arg post
- * their processing */
+ gf_uuid_copy(t_data->gfid, upcall_data->gfid);
+ t_data->event_type = upcall_data->event_type;
+
+ switch (t_data->event_type) {
+ case GF_UPCALL_CACHE_INVALIDATION:
+ t_data->data = gf_copy_cache_invalidation(
+ (struct gf_upcall_cache_invalidation *)upcall_data->data);
+ break;
+ case GF_UPCALL_RECALL_LEASE:
+ t_data->data = gf_copy_recall_lease(
+ (struct gf_upcall_recall_lease *)upcall_data->data);
+ break;
+ }
+
+ if (!t_data->data)
+ goto out;
return args;
out:
- if (up_arg) {
- GLFS_FREE(up_arg);
+ if (ret) {
+ GF_FREE(args->upcall_data.client_uid);
+ GF_FREE(args);
}
return NULL;
--
1.8.3.1

47
0265-fuse-add-missing-GF_FREE-to-fuse_interrupt.patch Normal file
View File

@ -0,0 +1,47 @@
From 7455900798446681fea1a2693fac9b423ad9722a Mon Sep 17 00:00:00 2001
From: Csaba Henk <csaba@redhat.com>
Date: Tue, 9 Jul 2019 11:06:49 +0200
Subject: [PATCH 265/265] fuse: add missing GF_FREE to fuse_interrupt
Upstream:
(Reviewed on https://review.gluster.org/c/glusterfs/+/23016)
> Change-Id: Id7e003e4a53d0a0057c1c84e1cd704c80a6cb015
> Fixes: bz#1728047
> Signed-off-by: Csaba Henk <csaba@redhat.com>
BUG: 1734423
Change-Id: I50640bf9b56349ab9b07140bdce8a45a7d07ba7a
Signed-off-by: Csaba Henk <csaba@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/177298
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
xlators/mount/fuse/src/fuse-bridge.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/xlators/mount/fuse/src/fuse-bridge.c b/xlators/mount/fuse/src/fuse-bridge.c
index c05866b..1c946a2 100644
--- a/xlators/mount/fuse/src/fuse-bridge.c
+++ b/xlators/mount/fuse/src/fuse-bridge.c
@@ -661,7 +661,7 @@ fuse_interrupt(xlator_t *this, fuse_in_header_t *finh, void *msg,
" failed to allocate timed message",
finh->unique, fii->unique);
- return;
+ goto out;
}
dmsg->fuse_out_header.unique = finh->unique;
@@ -673,6 +673,9 @@ fuse_interrupt(xlator_t *this, fuse_in_header_t *finh, void *msg,
send_fuse_timed(this, dmsg);
}
+
+out:
+ GF_FREE(finh);
}
/*
--
1.8.3.1

8
glusterfs.spec
View File

@ -231,7 +231,7 @@ Release: 0.1%{?prereltag:.%{prereltag}}%{?dist}
%else
Name: glusterfs
Version: 6.0
Release: 10%{?dist}
Release: 11%{?dist}
ExcludeArch: i686
%endif
License: GPLv2 or LGPLv3+
@ -571,6 +571,9 @@ Patch0259: 0259-gluster-block-tuning-perf-options.patch
Patch0260: 0260-ctime-Set-mdata-xattr-on-legacy-files.patch
Patch0261: 0261-features-utime-Fix-mem_put-crash.patch
Patch0262: 0262-glusterd-ctime-Disable-ctime-by-default.patch
Patch0263: 0263-tests-fix-ctime-related-tests.patch
Patch0264: 0264-gfapi-Fix-deadlock-while-processing-upcall.patch
Patch0265: 0265-fuse-add-missing-GF_FREE-to-fuse_interrupt.patch
%description
GlusterFS is a distributed file-system capable of scaling to several
@ -2279,6 +2282,9 @@ fi
%endif
%changelog
* Tue Aug 06 2019 Sunil Kumar Acharya <sheggodu@redhat.com> - 6.0-11
- fixes bugs bz#1733520 bz#1734423
* Fri Aug 02 2019 Sunil Kumar Acharya <sheggodu@redhat.com> - 6.0-10
- fixes bugs bz#1713890