diff options
author | Trond Myklebust <Trond.Myklebust@netapp.com> | 2011-03-25 18:15:11 (GMT) |
---|---|---|
committer | Trond Myklebust <Trond.Myklebust@netapp.com> | 2011-03-27 15:48:07 (GMT) |
commit | 4d65c520fb4abed970069d18c119cfe85624f46d (patch) | |
tree | 74671eb263380317f1e4a958a6a7f73749e98eec /fs/nfs/write.c | |
parent | 16c29dafcc86024048f1dbb8349d31cb22c7c55a (diff) | |
download | linux-fsl-qoriq-4d65c520fb4abed970069d18c119cfe85624f46d.tar.xz |
NFS: Fix a hang in the writeback path
Now that the inode scalability patches have been merged, it is no longer
safe to call igrab() under the inode->i_lock.
Now that we no longer call nfs_clear_request() until the nfs_page is
being freed, we know that we are always holding a reference to the
nfs_open_context, which again holds a reference to the path, and so
the inode cannot be freed until the last nfs_page has been removed
from the radix tree and freed.
We can therefore skip the igrab()/iput() altogether.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Diffstat (limited to 'fs/nfs/write.c')
-rw-r--r-- | fs/nfs/write.c | 13 |
1 files changed, 3 insertions, 10 deletions
diff --git a/fs/nfs/write.c b/fs/nfs/write.c index 85d7525..af0c627 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -389,11 +389,8 @@ static int nfs_inode_add_request(struct inode *inode, struct nfs_page *req) spin_lock(&inode->i_lock); error = radix_tree_insert(&nfsi->nfs_page_tree, req->wb_index, req); BUG_ON(error); - if (!nfsi->npages) { - igrab(inode); - if (nfs_have_delegation(inode, FMODE_WRITE)) - nfsi->change_attr++; - } + if (!nfsi->npages && nfs_have_delegation(inode, FMODE_WRITE)) + nfsi->change_attr++; set_bit(PG_MAPPED, &req->wb_flags); SetPagePrivate(req->wb_page); set_page_private(req->wb_page, (unsigned long)req); @@ -423,11 +420,7 @@ static void nfs_inode_remove_request(struct nfs_page *req) clear_bit(PG_MAPPED, &req->wb_flags); radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index); nfsi->npages--; - if (!nfsi->npages) { - spin_unlock(&inode->i_lock); - iput(inode); - } else - spin_unlock(&inode->i_lock); + spin_unlock(&inode->i_lock); nfs_release_request(req); } |