[Cluster-devel] [GFS2 Patch] nfsd readdirplus assertion failure
Steven Whitehouse
swhiteho at redhat.com
Tue Feb 6 13:28:32 UTC 2007
Hi,
Now added to the GFS2 -nmw git tree. Thanks,
Steve.
On Tue, 2007-02-06 at 03:52 -0500, Wendy Cheng wrote:
> Glock assertion failure found in '07 NFS connectathon. One of the NFSDs
> is doing a "readdirplus" procedure call. It passes the logic into
> gfs2_readdir() where it obtains its directory inode glock. This is then
> followed by filehandle construction that invokes lookup code. It hits
> the assertion failure while trying to obtain the inode glock again
> inside gfs2_drevalidate().
>
> This patch bypasses the recursive glock call if caller already holds the
> lock.
>
> Signed-off-by: S. Wendy Cheng <wcheng at redhat.com>
>
> ops_dentry.c | 16 +++++++++++-----
> 1 files changed, 11 insertions(+), 5 deletions(-)
>
> --- linux-git/fs/gfs2/ops_dentry.c 2007-02-06 01:07:26.000000000 -0500
> +++ linux/fs/gfs2/ops_dentry.c 2007-02-06 01:05:35.000000000 -0500
> @@ -46,6 +46,7 @@ static int gfs2_drevalidate(struct dentr
> struct gfs2_inum_host inum;
> unsigned int type;
> int error;
> + int had_lock=0;
>
> if (inode && is_bad_inode(inode))
> goto invalid;
> @@ -53,9 +54,12 @@ static int gfs2_drevalidate(struct dentr
> if (sdp->sd_args.ar_localcaching)
> goto valid;
>
> - error = gfs2_glock_nq_init(dip->i_gl, LM_ST_SHARED, 0, &d_gh);
> - if (error)
> - goto fail;
> + had_lock = gfs2_glock_is_locked_by_me(dip->i_gl);
> + if (!had_lock) {
> + error = gfs2_glock_nq_init(dip->i_gl, LM_ST_SHARED, 0, &d_gh);
> + if (error)
> + goto fail;
> + }
>
> error = gfs2_dir_search(parent->d_inode, &dentry->d_name, &inum,
> &type);
> switch (error) {
> @@ -82,13 +86,15 @@ static int gfs2_drevalidate(struct dentr
> }
>
> valid_gunlock:
> - gfs2_glock_dq_uninit(&d_gh);
> + if (!had_lock)
> + gfs2_glock_dq_uninit(&d_gh);
> valid:
> dput(parent);
> return 1;
>
> invalid_gunlock:
> - gfs2_glock_dq_uninit(&d_gh);
> + if (!had_lock)
> + gfs2_glock_dq_uninit(&d_gh);
> invalid:
> if (inode && S_ISDIR(inode->i_mode)) {
> if (have_submounts(dentry))
>
>
More information about the Cluster-devel
mailing list