diff options
author | Shehjar Tikoo <shehjart@gluster.com> | 2009-06-26 13:05:00 +0000 |
---|---|---|
committer | Anand V. Avati <avati@dev.gluster.com> | 2009-06-29 10:51:16 -0700 |
commit | 304e4274ca9b0339539581c5413e3339078c1182 (patch) | |
tree | 95313a38ebb829e22e9256ab65f86fac69f6c3ae /libglusterfsclient | |
parent | d6ce087395220ea0b6e4bff4d3e561de624447a3 (diff) |
libglusterfsclient: Fix glusterfs_mount-vmp search deadlock
It is possible that the only translator in the libglusterfsclient
tree is the posix. In that case, inside gluster_init, the graph
init routines will need to call lstat on the posix subdirectory.
Since even the glusterfs stack is running over booster, those
calls will also first require vmp searching. BUT, the vmp lock
is the same as the mount lock that was already taken when we entered
glusterfs_mount, so a deadlock occurs.
Signed-off-by: Anand V. Avati <avati@dev.gluster.com>
Diffstat (limited to 'libglusterfsclient')
-rwxr-xr-x | libglusterfsclient/src/libglusterfsclient.c | 31 |
1 files changed, 20 insertions, 11 deletions
diff --git a/libglusterfsclient/src/libglusterfsclient.c b/libglusterfsclient/src/libglusterfsclient.c index 89a7dff8a..576fef7f1 100755 --- a/libglusterfsclient/src/libglusterfsclient.c +++ b/libglusterfsclient/src/libglusterfsclient.c @@ -73,7 +73,16 @@ struct { int entries; }vmplist; -pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER; + +/* Protects the VMP list above. */ +pthread_mutex_t vmplock = PTHREAD_MUTEX_INITIALIZER; + +/* Ensures only one thread is ever calling glusterfs_mount. + * Since that function internally calls routines which + * use the yacc parser code using global vars, this process + * needs to be syncronised. + */ +pthread_mutex_t mountlock = PTHREAD_MUTEX_INITIALIZER; char * libgf_vmp_virtual_path(struct vmp_entry *entry, const char *path) @@ -1319,11 +1328,11 @@ libgf_vmp_search_entry (char *path) if (!path) goto out; - pthread_mutex_lock (&lock); + pthread_mutex_lock (&vmplock); { entry = _libgf_vmp_search_entry (path); } - pthread_mutex_unlock (&lock); + pthread_mutex_unlock (&vmplock); out: if (entry) @@ -1390,7 +1399,7 @@ glusterfs_mount (char *vmp, glusterfs_init_params_t *ipars) if (!vmp_resolved) goto out; - pthread_mutex_lock (&lock); + pthread_mutex_lock (&mountlock); { vmp_entry = _libgf_vmp_search_entry (vmp); if (vmp_entry) { @@ -1410,7 +1419,7 @@ glusterfs_mount (char *vmp, glusterfs_init_params_t *ipars) } } unlock: - pthread_mutex_unlock (&lock); + pthread_mutex_unlock (&mountlock); out: if (vmp_resolved) @@ -1454,11 +1463,11 @@ libgf_umount (char *vmp) { int ret = -1; - pthread_mutex_lock (&lock); + pthread_mutex_lock (&vmplock); { ret = _libgf_umount (vmp); } - pthread_mutex_unlock (&lock); + pthread_mutex_unlock (&vmplock); return ret; } @@ -1489,7 +1498,7 @@ glusterfs_umount_all (void) { struct vmp_entry *entry = NULL, *tmp = NULL; - pthread_mutex_lock (&lock); + pthread_mutex_lock (&vmplock); { list_for_each_entry_safe (entry, tmp, &vmplist.list, list) { /* even if there are errors, continue with other @@ -1498,7 +1507,7 @@ glusterfs_umount_all (void) _libgf_umount (entry->vmp); } } - pthread_mutex_unlock (&lock); + pthread_mutex_unlock (&vmplock); return 0; } @@ -1509,8 +1518,8 @@ glusterfs_reset (void) INIT_LIST_HEAD (&vmplist.list); vmplist.entries = 0; - memset (&lock, 0, sizeof (lock)); - pthread_mutex_init (&lock, NULL); + memset (&vmplock, 0, sizeof (vmplock)); + pthread_mutex_init (&vmplock, NULL); first_fini = first_init = 1; } |