Hi Olga
On Mon, 2019-12-16 at 17:13 -0500, Olga Kornievskaia wrote:
> From: Olga Kornievskaia <kolga@xxxxxxxxxx>
>
> Ever since the commit 0e0cb35b417f, it's possible to lose an open
> stateid
> while retrying a CLOSE due to ERR_OLD_STATEID. Once that happens,
> operations that require openstateid fail with EAGAIN which is
> propagated
> to the application then tests like generic/446 and generic/168 fail
> with
> "Resource temporarily unavailable".
>
> Instead of returning this error, initiate state recovery when
> possible to
> recover the open stateid and then try calling
> nfs4_select_rw_stateid()
> again.
>
> Fixes: 0e0cb35b417f ("NFSv4: Handle NFS4ERR_OLD_STATEID in
> CLOSE/OPEN_DOWNGRADE")
> Signed-off-by: Olga Kornievskaia <kolga@xxxxxxxxxx>
> ---
> fs/nfs/nfs4proc.c | 3 +++
> fs/nfs/nfs4state.c | 2 +-
> fs/nfs/pnfs.c | 2 +-
> 3 files changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> index 76d37161409a..66f9631ba012 100644
> --- a/fs/nfs/nfs4proc.c
> +++ b/fs/nfs/nfs4proc.c
> @@ -3239,6 +3239,8 @@ static int _nfs4_do_setattr(struct inode
> *inode,
> nfs_put_lock_context(l_ctx);
> if (status == -EIO)
> return -EBADF;
> + else if (status)
> + goto out;
> } else {
> zero_stateid:
> nfs4_stateid_copy(&arg->stateid, &zero_stateid);
> @@ -3251,6 +3253,7 @@ static int _nfs4_do_setattr(struct inode
> *inode,
> put_cred(delegation_cred);
> if (status == 0 && ctx != NULL)
> renew_lease(server, timestamp);
> +out:
> trace_nfs4_setattr(inode, &arg->stateid, status);
> return status;
> }
> diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
> index 34552329233d..8686451893a6 100644
> --- a/fs/nfs/nfs4state.c
> +++ b/fs/nfs/nfs4state.c
> @@ -1064,7 +1064,7 @@ int nfs4_select_rw_stateid(struct nfs4_state
> *state,
> * choose to use.
> */
> goto out;
> - ret = nfs4_copy_open_stateid(dst, state) ? 0 : -EAGAIN;
> + ret = nfs4_copy_open_stateid(dst, state) ? 0 :
> -NFS4ERR_BAD_STATEID;
This is not acceptable. We can't return NFSv4 error values to functions
that expect POSIX errors.
For instance pnfs_update_layout() tries to apply ERR_PTR() to this
return value, which breaks badly for non-POSIX errors (it returns an
invalid pointer that fails the IS_ERR() test).
> out:
> if (nfs_server_capable(state->inode, NFS_CAP_STATEID_NFSV41))
> dst->seqid = 0;
> diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
> index cec3070ab577..11d646bbd2cb 100644
> --- a/fs/nfs/pnfs.c
> +++ b/fs/nfs/pnfs.c
> @@ -1998,7 +1998,7 @@ pnfs_update_layout(struct inode *ino,
> trace_pnfs_update_layout(ino, pos, count,
> iomode, lo, lseg,
> PNFS_UPDATE_LAYOUT_INVALID_OPEN
> );
> - if (status != -EAGAIN)
> + if (status != -EAGAIN && status !=
> -NFS4ERR_BAD_STATEID)
> goto out_unlock;
> spin_unlock(&ino->i_lock);
> nfs4_schedule_stateid_recovery(server, ctx-
> >state);
--
Trond Myklebust
Linux NFS client maintainer, Hammerspace
trond.myklebust@xxxxxxxxxxxxxxx
