Return-Path: linux-nfs-owner@vger.kernel.org Received: from casper.infradead.org ([85.118.1.10]:45767 "EHLO casper.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753975AbaHGHVd (ORCPT ); Thu, 7 Aug 2014 03:21:33 -0400 From: Christoph Hellwig To: Trond Myklebust Cc: linux-nfs@vger.kernel.org Subject: [PATCH 04/17] pnfs: don't check sequence on new stateids in layoutget Date: Thu, 7 Aug 2014 09:23:36 +0200 Message-Id: <1407396229-4785-5-git-send-email-hch@lst.de> In-Reply-To: <1407396229-4785-1-git-send-email-hch@lst.de> References: <1407396229-4785-1-git-send-email-hch@lst.de> Sender: linux-nfs-owner@vger.kernel.org List-ID: When layoutget returns an entirely new layout stateid it should not check the generation counter as the new stateid will start with a new counter entirely unrelated to old one. The current behavior causes constant layoutget failures against a block server which allocates a new stateid after an recall that removed all outstanding layouts. Signed-off-by: Christoph Hellwig --- fs/nfs/pnfs.c | 36 +++++++++++++++++++----------------- 1 file changed, 19 insertions(+), 17 deletions(-) diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index c72376c..22183fc 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -685,17 +685,6 @@ static bool pnfs_seqid_is_newer(u32 s1, u32 s2) return (s32)(s1 - s2) > 0; } -static void -pnfs_verify_layout_stateid(struct pnfs_layout_hdr *lo, - const nfs4_stateid *new, - struct list_head *free_me_list) -{ - if (nfs4_stateid_match_other(&lo->plh_stateid, new)) - return; - /* Layout is new! Kill existing layout segments */ - pnfs_mark_matching_lsegs_invalid(lo, free_me_list, NULL); -} - /* update lo->plh_stateid with new if is more recent */ void pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new, @@ -1370,16 +1359,29 @@ pnfs_layout_process(struct nfs4_layoutget *lgp) goto out_forget_reply; } - if (pnfs_layoutgets_blocked(lo, 1) || - pnfs_layout_stateid_blocked(lo, &res->stateid)) { + if (pnfs_layoutgets_blocked(lo, 1)) { dprintk("%s forget reply due to state\n", __func__); goto out_forget_reply; } - /* Check that the new stateid matches the old stateid */ - pnfs_verify_layout_stateid(lo, &res->stateid, &free_me); - /* Done processing layoutget. Set the layout stateid */ - pnfs_set_layout_stateid(lo, &res->stateid, false); + if (nfs4_stateid_match_other(&lo->plh_stateid, &res->stateid)) { + /* existing state ID, make sure the sequence number matches. */ + if (pnfs_layout_stateid_blocked(lo, &res->stateid)) { + dprintk("%s forget reply due to sequence\n", __func__); + goto out_forget_reply; + } + pnfs_set_layout_stateid(lo, &res->stateid, false); + } else { + /* + * We got an entirely new state ID. Mark all segments for the + * inode invalid, and don't bother validating the stateid + * sequence number. + */ + pnfs_mark_matching_lsegs_invalid(lo, &free_me, NULL); + + nfs4_stateid_copy(&lo->plh_stateid, &res->stateid); + lo->plh_barrier = be32_to_cpu(res->stateid.seqid); + } pnfs_get_lseg(lseg); pnfs_layout_insert_lseg(lo, lseg); -- 1.9.1