Solaris: do not call VFS_RELE with GLOCK

VFS_RELE can call afs_freevfs, which grabs GLOCK. Thus, don't call
VFS_RELE with GLOCK held or we can try to recursively acquire GLOCK
and panic.  This is currently unlikely to occur (sans vfs refcount
unbalances) without support for forced unmounts, since the last vfs
ref will usually come from the caller of afs_unmount(). But it still
may be possible.

Change-Id: I152bd13abc2e90f258f98194833c50c3a2846904
Reviewed-on: http://gerrit.openafs.org/1966
Tested-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

diff --git a/src/afs/SOLARIS/osi_vnodeops.c b/src/afs/SOLARIS/osi_vnodeops.c
index 0ec2e09..4c88297 100644 (file)
--- a/src/afs/SOLARIS/osi_vnodeops.c
+++ b/src/afs/SOLARIS/osi_vnodeops.c
@@ -1817,7 +1817,11 @@ afs_inactive(struct vcache *avc, afs_ucred_t *acred)
     afs_InactiveVCache(avc, acred);
 
 #ifdef AFS_SUN58_ENV
+    AFS_GUNLOCK();
+    /* VFS_RELE must be called outside of GLOCK, since it can potentially
+     * call afs_freevfs, which acquires GLOCK */
     VFS_RELE(afs_globalVFS);
+    AFS_GLOCK();
 #endif
 
     return 0;