From 263ae23c5fd69a76e83561e1210e434cc73cb73d Mon Sep 17 00:00:00 2001 From: Jean-Pierre André Date: Wed, 22 Aug 2012 10:00:22 +0200 Subject: Released the locks before deleting inodes --- kernel/fuse_vnops.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/kernel/fuse_vnops.c b/kernel/fuse_vnops.c index a6e8526..237283f 100644 --- a/kernel/fuse_vnops.c +++ b/kernel/fuse_vnops.c @@ -2221,6 +2221,7 @@ sendmsg: * XXX: This call releases vnode with vn_free, can this * be an issue? Or does VN_RELE do the required job for us? */ + VN_RELE(vp); fuse_vnode_free(vp, sep); vp = NULL; } @@ -2997,6 +2998,7 @@ fuse_remove(vnode_t *dvp, char *name, cred_t *credp, caller_context_t *ct, * XXX: This call releases vnode with vn_free, can this * be an issue? Or does VN_RELE do the required job for us? */ + VN_RELE(vp); fuse_vnode_free(vp, sep); cleanup: fuse_free_msg(msgp); @@ -3031,6 +3033,7 @@ fuse_vnode_free(struct vnode *vp, fuse_session_t *sep) } vp->v_data = NULL; + VFS_RELE(vp->v_vfsp); if (!vn_has_cached_data(vp)) /* be safe : better leak than corruption */ vn_free(vp); } @@ -3074,6 +3077,7 @@ void fuse_destroy_cache(fuse_session_t *sep) fuse_vnode_cache_remove(vp, sep); fuse_free_vdata(vp); vp->v_data = NULL; + VFS_RELE(vp->v_vfsp); if (!vn_has_cached_data(vp)) vn_free(vp); item = (fuse_avl_cache_node_t*) -- cgit v1.2.3