Re: [PATCH] nfs: fix bdi_unregister() before sb kill

From: Jens Axboe
Date: Thu Sep 17 2009 - 15:40:37 EST


On Thu, Sep 17 2009, Jens Axboe wrote:
> Hi,
>
> This can cause a hang on NFS umount, since the bdi gets pruned before we
> flush any pending dirty IO. Peter, can you check whether it fixes your
> issue?

There's another problem with NFS && backing devices. NFS may call
bdi_destroy() on a bdi without ever called bdi_init(). This was always a
bad idea, now it's an issue.

So, Trond, can I safely add a server->flags private flag to indicate
that we have called bdi_init()? Then nfs_free_server() knows when to
call bdi_destroy(). Seems like the safest fix, since error handling is
currently 'just call nfs_free_server()'.

diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index e350bd6..8ffee33 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -936,6 +936,7 @@ static int nfs_probe_fsinfo(struct nfs_server *server, struct nfs_fh *mntfh, str
error = bdi_init(&server->backing_dev_info);
if (error)
goto out_error;
+ server->flags |= NFS_MOUNT_BDI_INITED;


/* Get some general file system info */
@@ -1021,7 +1022,8 @@ void nfs_free_server(struct nfs_server *server)
nfs_put_client(server->nfs_client);

nfs_free_iostats(server->io_stats);
- bdi_destroy(&server->backing_dev_info);
+ if (server->flags & NFS_MOUNT_BDI_INITED)
+ bdi_destroy(&server->backing_dev_info);
kfree(server);
nfs_release_automount_timer();
dprintk("<-- nfs_free_server()\n");
diff --git a/include/linux/nfs_mount.h b/include/linux/nfs_mount.h
index 4499016..842ef74 100644
--- a/include/linux/nfs_mount.h
+++ b/include/linux/nfs_mount.h
@@ -69,5 +69,6 @@ struct nfs_mount_data {
#define NFS_MOUNT_LOOKUP_CACHE_NONEG 0x10000
#define NFS_MOUNT_LOOKUP_CACHE_NONE 0x20000
#define NFS_MOUNT_NORESVPORT 0x40000
+#define NFS_MOUNT_BDI_INITED 0x80000

#endif

--
Jens Axboe

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/