Return-Path: linux-nfs-owner@vger.kernel.org Received: from casper.infradead.org ([85.118.1.10]:51183 "EHLO casper.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751694AbaHUQHa (ORCPT ); Thu, 21 Aug 2014 12:07:30 -0400 Received: from ip-64-134-168-64.public.wayport.net ([64.134.168.64] helo=localhost) by casper.infradead.org with esmtpsa (Exim 4.80.1 #2 (Red Hat Linux)) id 1XKUtd-00009V-0B for linux-nfs@vger.kernel.org; Thu, 21 Aug 2014 16:07:29 +0000 From: Christoph Hellwig To: linux-nfs@vger.kernel.org Subject: [PATCH 06/19] pnfs: avoid using stale stateids after layoutreturn Date: Thu, 21 Aug 2014 11:09:22 -0500 Message-Id: <1408637375-11343-7-git-send-email-hch@lst.de> In-Reply-To: <1408637375-11343-1-git-send-email-hch@lst.de> References: <1408637375-11343-1-git-send-email-hch@lst.de> Sender: linux-nfs-owner@vger.kernel.org List-ID: After we issued a layoutreturn operations the may free the layout stateid and will thus cause bad stateid error when the client uses it again. We currently try to avoid this case by chosing the open stateid if not lsegs are present for this inode. But various places can hold refererence on lsegs and thus cause the list not to be empty shortly after a layout return. Add an explicit flag to mark the current layout stateid invalid and force usage of the openstateid after we did a full file layoutreturn. Signed-off-by: Christoph Hellwig --- fs/nfs/pnfs.c | 7 ++++++- fs/nfs/pnfs.h | 1 + 2 files changed, 7 insertions(+), 1 deletion(-) diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 32a14a6..bce7f1b 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -741,7 +741,8 @@ pnfs_choose_layoutget_stateid(nfs4_stateid *dst, struct pnfs_layout_hdr *lo, status = -EAGAIN; } else if (!nfs4_valid_open_stateid(open_state)) { status = -EBADF; - } else if (list_empty(&lo->plh_segs)) { + } else if (list_empty(&lo->plh_segs) || + test_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags)) { int seq; do { @@ -863,6 +864,8 @@ _pnfs_return_layout(struct inode *ino) dprintk("NFS: %s no layout segments to return\n", __func__); goto out; } + + set_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags); lo->plh_block_lgets++; spin_unlock(&ino->i_lock); pnfs_free_lseg_list(&tmp_list); @@ -1383,6 +1386,8 @@ pnfs_layout_process(struct nfs4_layoutget *lgp) lo->plh_barrier = be32_to_cpu(res->stateid.seqid); } + clear_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags); + pnfs_get_lseg(lseg); pnfs_layout_insert_lseg(lo, lseg); diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h index aca3dff..16cd14f 100644 --- a/fs/nfs/pnfs.h +++ b/fs/nfs/pnfs.h @@ -65,6 +65,7 @@ enum { NFS_LAYOUT_BULK_RECALL, /* bulk recall affecting layout */ NFS_LAYOUT_ROC, /* some lseg had roc bit set */ NFS_LAYOUT_RETURN, /* Return this layout ASAP */ + NFS_LAYOUT_INVALID_STID, /* layout stateid id is invalid */ }; enum layoutdriver_policy_flags { -- 1.9.1