From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from cuda.sgi.com (cuda1.sgi.com [192.48.157.11]) by oss.sgi.com (8.14.3/8.14.3/SuSE Linux 0.8) with ESMTP id o2BMf5Cq194424 for ; Thu, 11 Mar 2010 16:41:06 -0600 Received: from mail.internode.on.net (localhost [127.0.0.1]) by cuda.sgi.com (Spam Firewall) with ESMTP id 99705141F781 for ; Thu, 11 Mar 2010 14:42:37 -0800 (PST) Received: from mail.internode.on.net (bld-mail13.adl6.internode.on.net [150.101.137.98]) by cuda.sgi.com with ESMTP id FDPBHvJgOvrWPkla for ; Thu, 11 Mar 2010 14:42:37 -0800 (PST) From: Dave Chinner Subject: [PATCH 09/19] xfs: Ensure we force all busy extents in range to disk Date: Fri, 12 Mar 2010 09:42:07 +1100 Message-Id: <1268347337-7160-10-git-send-email-david@fromorbit.com> In-Reply-To: <1268347337-7160-1-git-send-email-david@fromorbit.com> References: <1268347337-7160-1-git-send-email-david@fromorbit.com> List-Id: XFS Filesystem from SGI List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: xfs-bounces@oss.sgi.com Errors-To: xfs-bounces@oss.sgi.com To: stable@kernel.org Cc: xfs@oss.sgi.com >>From fd45e4784164d1017521086524e3442318c67370 Date: Sat, 2 Jan 2010 02:38:56 +0000 When we search for and find a busy extent during allocation we force the log out to ensure the extent free transaction is on disk before the allocation transaction. The current implementation has a subtle bug in it--it does not handle multiple overlapping ranges. That is, if we free lots of little extents into a single contiguous extent, then allocate the contiguous extent, the busy search code stops searching at the first extent it finds that overlaps the allocated range. It then uses the commit LSN of the transaction to force the log out to. Unfortunately, the other busy ranges might have more recent commit LSNs than the first busy extent that is found, and this results in xfs_alloc_search_busy() returning before all the extent free transactions are on disk for the range being allocated. This can lead to potential metadata corruption or stale data exposure after a crash because log replay won't replay all the extent free transactions that cover the allocation range. Modified-by: Alex Elder (Dropped the "found" argument from the xfs_alloc_busysearch trace event.) Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Signed-off-by: Alex Elder --- fs/xfs/xfs_alloc.c | 52 +++++++++++++++++++++------------------------------- 1 files changed, 21 insertions(+), 31 deletions(-) diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c index 2cf944e..4cd1c23 100644 --- a/fs/xfs/xfs_alloc.c +++ b/fs/xfs/xfs_alloc.c @@ -2703,45 +2703,35 @@ xfs_alloc_search_busy(xfs_trans_t *tp, xfs_mount_t *mp; xfs_perag_busy_t *bsy; xfs_agblock_t uend, bend; - xfs_lsn_t lsn; + xfs_lsn_t lsn = 0; int cnt; mp = tp->t_mountp; spin_lock(&mp->m_perag[agno].pagb_lock); - cnt = mp->m_perag[agno].pagb_count; - uend = bno + len - 1; - /* search pagb_list for this slot, skipping open slots */ - for (bsy = mp->m_perag[agno].pagb_list; cnt; bsy++) { - - /* - * (start1,length1) within (start2, length2) - */ - if (bsy->busy_tp != NULL) { - bend = bsy->busy_start + bsy->busy_length - 1; - if ((bno > bend) || (uend < bsy->busy_start)) { - cnt--; - } else { - TRACE_BUSYSEARCH("xfs_alloc_search_busy", - "found1", agno, bno, len, tp); - break; - } - } - } - /* - * If a block was found, force the log through the LSN of the - * transaction that freed the block + * search pagb_list for this slot, skipping open slots. We have to + * search the entire array as there may be multiple overlaps and + * we have to get the most recent LSN for the log force to push out + * all the transactions that span the range. */ - if (cnt) { - TRACE_BUSYSEARCH("xfs_alloc_search_busy", "found", agno, bno, len, tp); - lsn = bsy->busy_tp->t_commit_lsn; - spin_unlock(&mp->m_perag[agno].pagb_lock); - xfs_log_force(mp, lsn, XFS_LOG_FORCE|XFS_LOG_SYNC); - } else { - TRACE_BUSYSEARCH("xfs_alloc_search_busy", "not-found", agno, bno, len, tp); - spin_unlock(&mp->m_perag[agno].pagb_lock); + for (cnt = 0; cnt < mp->m_perag[agno].pagb_count; cnt++) { + bsy = &mp->m_perag[agno].pagb_list[cnt]; + if (!bsy->busy_tp) + continue; + bend = bsy->busy_start + bsy->busy_length - 1; + if (bno > bend || uend < bsy->busy_start) + continue; + + /* (start1,length1) within (start2, length2) */ + if (XFS_LSN_CMP(bsy->busy_tp->t_commit_lsn, lsn) > 0) + lsn = bsy->busy_tp->t_commit_lsn; } + spin_unlock(&mp->m_perag[agno].pagb_lock); + TRACE_BUSYSEARCH("xfs_alloc_search_busy", lsn ? "found" : "not-found", + agno, bno, len, tp); + if (lsn) + xfs_log_force(mp, lsn, XFS_LOG_FORCE|XFS_LOG_SYNC); } -- 1.6.5 _______________________________________________ xfs mailing list xfs@oss.sgi.com http://oss.sgi.com/mailman/listinfo/xfs