Date: Tue, 1 Jul 2014 06:59:24 +0000 (UTC) From: Scott Long <scottl@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-10@freebsd.org Subject: svn commit: r268077 - stable/10/sys/ufs/ffs Message-ID: <201407010659.s616xOfc096242@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: scottl Date: Tue Jul 1 06:59:23 2014 New Revision: 268077 URL: http://svnweb.freebsd.org/changeset/base/268077 Log: Merge r265463: Due to reasons unknown at this time, the system can be forced to write a journal block even when there are no journal entries to be written. Until the root cause is found, handle this case by ensuring that a valid journal segment is always written. Second, the data buffer used for writing journal entries was never being scrubbed of old data. Fix this. Submitted by: Takehara Mikihito Obtained from: Netflix, Inc. Modified: stable/10/sys/ufs/ffs/ffs_softdep.c Modified: stable/10/sys/ufs/ffs/ffs_softdep.c ============================================================================== --- stable/10/sys/ufs/ffs/ffs_softdep.c Tue Jul 1 06:50:35 2014 (r268076) +++ stable/10/sys/ufs/ffs/ffs_softdep.c Tue Jul 1 06:59:23 2014 (r268077) @@ -1275,6 +1275,7 @@ static int stat_cleanup_blkrequests; /* static int stat_cleanup_inorequests; /* Number of inode cleanup requests */ static int stat_cleanup_retries; /* Number of cleanups that needed to flush */ static int stat_cleanup_failures; /* Number of cleanup requests that failed */ +static int stat_emptyjblocks; /* Number of potentially empty journal blocks */ SYSCTL_INT(_debug_softdep, OID_AUTO, max_softdeps, CTLFLAG_RW, &max_softdeps, 0, ""); @@ -1334,6 +1335,8 @@ SYSCTL_INT(_debug_softdep, OID_AUTO, cle &stat_cleanup_failures, 0, ""); SYSCTL_INT(_debug_softdep, OID_AUTO, flushcache, CTLFLAG_RW, &softdep_flushcache, 0, ""); +SYSCTL_INT(_debug_softdep, OID_AUTO, emptyjblocks, CTLFLAG_RD, + &stat_emptyjblocks, 0, ""); SYSCTL_DECL(_vfs_ffs); @@ -3328,6 +3331,24 @@ softdep_process_journal(mp, needwk, flag */ data = bp->b_data; off = 0; + + /* + * Always put a header on the first block. + * XXX As with below, there might not be a chance to get + * into the loop. Ensure that something valid is written. + */ + jseg_write(ump, jseg, data); + off += JREC_SIZE; + data = bp->b_data + off; + + /* + * XXX Something is wrong here. There's no work to do, + * but we need to perform and I/O and allow it to complete + * anyways. + */ + if (LIST_EMPTY(&ump->softdep_journal_pending)) + stat_emptyjblocks++; + while ((wk = LIST_FIRST(&ump->softdep_journal_pending)) != NULL) { if (cnt == 0) @@ -3377,6 +3398,11 @@ softdep_process_journal(mp, needwk, flag data = bp->b_data + off; cnt--; } + + /* Clear any remaining space so we don't leak kernel data */ + if (size > off) + bzero(data, size - off); + /* * Write this one buffer and continue. */
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201407010659.s616xOfc096242>