Hello, On further testing with more iterations with nilfs 2.1 with a only a date rewind (across reboots also) the old dates do not get cleaned up and the new checkpoints with the rewound dates are not cleaned up (my previous testing with 2.1 daemon was on a loopback mount and involved 2 iterations - all testing now was on live systems) - though the 2.1 daemon does not crash it is a do nothing process. A few times we had nilfs_cleanerd 2.1 crashes (reboots fixed that - no crashes, but no checkpoint cleanup after the reboot). Here is the stacktrace from the crash for a 3.0.4 kernel:
Dec 2 13:53:22 kernel: Pid: 1717, comm: nilfs_cleanerd Not tainted 3.0.4 #4 Dec 2 13:53:22 kernel: Call Trace: Dec 2 13:53:22 kernel: [<c043e1d0>] ? warn_slowpath_common+0x65/0x7a Dec 2 13:53:22 kernel: [<c043e1f9>] ? warn_slowpath_null+0x14/0x18 Dec 2 13:53:22 kernel: [<f84c000d>] ? nilfs_ioctl_move_blocks+0x11d/0x199 [nilfs2] Dec 2 13:53:22 kernel: [<f84c02cd>] ? nilfs_ioctl_clean_segments+0x236/0x2b6 [nilfs2] Dec 2 13:53:22 kernel: [<f84bfd23>] ? nilfs_ioctl_get_bdescs+0x68/0x7b [nilfs2] Dec 2 13:53:22 kernel: [<f84c06f2>] ? nilfs_ioctl+0x192/0x1bb [nilfs2] Dec 2 13:53:22 kernel: [<f84c0560>] ? nilfs_ioctl_set_alloc_range+0x12b/0x12b [nilfs2] Dec 2 13:53:22 kernel: [<c04fffdc>] ? vfs_ioctl+0x1e/0x38 Dec 2 13:53:22 kernel: [<c050061c>] ? do_vfs_ioctl+0x164/0x16b Dec 2 13:53:22 kernel: [<c0500668>] ? sys_ioctl+0x45/0x5c Dec 2 13:53:22 kernel: [<c07311df>] ? sysenter_do_call+0x12/0x28 Dec 2 13:53:22 kernel: [<c0720000>] ? ab8500_regulator_probe+0x147/0x1af Dec 2 13:53:22 kernel: ---[ end trace 34bfcccc859adad2 ]--- Dec 2 13:53:22 kernel: NILFS: GC failed during preparation: cannot read source blocks: err=-17 Dec 2 13:53:22 nilfs_cleanerd[1717]: cannot clean segments: File exists Dec 2 13:53:22 nilfs_cleanerd[1717]: shutdown Dec 2 14:06:10 nilfs_cleanerd[15310]: start Dec 2 14:06:12 kernel: ------------[ cut here ]------------ Dec 2 14:06:12 kernel: WARNING: at fs/nilfs2/ioctl.c:449 nilfs_ioctl_move_blocks+0x11d/0x199 [nilfs2]() Dec 2 14:06:12 kernel: Pid: 15310, comm: nilfs_cleanerd Tainted: G W 3.0.4 #4 Dec 2 14:06:12 kernel: Call Trace: Dec 2 14:06:12 kernel: [<c043e1d0>] ? warn_slowpath_common+0x65/0x7a Dec 2 14:06:12 kernel: [<c043e1f9>] ? warn_slowpath_null+0x14/0x18 Dec 2 14:06:12 kernel: [<f84c000d>] ? nilfs_ioctl_move_blocks+0x11d/0x199 [nilfs2] Dec 2 14:06:12 kernel: [<f84c02cd>] ? nilfs_ioctl_clean_segments+0x236/0x2b6 [nilfs2] Dec 2 14:06:12 kernel: [<f84bfd23>] ? nilfs_ioctl_get_bdescs+0x68/0x7b [nilfs2] Dec 2 14:06:12 kernel: [<f84c06f2>] ? nilfs_ioctl+0x192/0x1bb [nilfs2] Dec 2 14:06:12 kernel: [<f84c0560>] ? nilfs_ioctl_set_alloc_range+0x12b/0x12b [nilfs2] Dec 2 14:06:12 kernel: [<c04fffdc>] ? vfs_ioctl+0x1e/0x38 Dec 2 14:06:12 kernel: [<c050061c>] ? do_vfs_ioctl+0x164/0x16b Dec 2 14:06:12 kernel: [<c0500668>] ? sys_ioctl+0x45/0x5c Dec 2 14:06:12 kernel: [<c07311df>] ? sysenter_do_call+0x12/0x28 Dec 2 14:06:12 kernel: ---[ end trace 34bfcccc859adad3 ]--- Dec 2 14:06:12 kernel: NILFS: GC failed during preparation: cannot read source blocks: err=-17 Dec 2 14:06:12 nilfs_cleanerd[15310]: cannot clean segments: File exists Dec 2 14:06:12 nilfs_cleanerd[15310]: shutdown Dec 2 14:08:11 nilfs_cleanerd[15574]: start Dec 2 14:08:13 kernel: ------------[ cut here ]------------ Dec 2 14:08:13 kernel: WARNING: at fs/nilfs2/ioctl.c:449 nilfs_ioctl_move_blocks+0x11d/0x199 [nilfs2]() Note, that moving the date forward from to the most forward checkpoint in the future cleans all checkpoints in both 2.0 & 2.1 daemons. Zahid -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Zahid Chowdhury Sent: Monday, December 05, 2011 2:16 PM To: Ryusuke Konishi Cc: [email protected]; dexen deVries Subject: RE: nilfs_cleanerd from nilfs-utils shutdown on version 2.0 and 2.1 does not fail but says nothing and does not clean the old checkpoints nor newer (actually older) ones. Hello Ryusuke, I have successfully run the nilfs utils 2.1 with a Centos 5.5 kernel with a nilfs module builtin and cleaned up all checkpoints with no issues whatsoever. Thus, no kernel bug is caused even in the old 2.6.18 kernel from the time rewind. I suggest everybody upgrade to nilfs utils 2.1 wherever possible. Thanks everybody for your help. Zahid -----Original Message----- From: Ryusuke Konishi [mailto:[email protected]] Sent: Sunday, December 04, 2011 6:57 AM To: Zahid Chowdhury Cc: [email protected]; dexen deVries Subject: Re: nilfs_cleanerd from nilfs-utils shutdown on version 2.0 and 2.1 does not fail but says nothing and does not clean the old checkpoints nor newer (actually older) ones. Hi, On Fri, 2 Dec 2011 16:33:09 -0800, Zahid Chowdhury wrote: > Hello, > If I move the system date forward, have some checkpoints created and then > move the date backward a 2.0 cleanerd daemon fails on this error: > Nov 30 14:39:37 nilfs_cleanerd[5789]: start > Nov 30 14:39:38 kernel: nilfs_ioctl_move_inode_block: conflicting data > > buffer: ino=4, cno=0, offset=0, blocknr=665655, vblocknr=566462 > Nov 30 14:39:38 kernel: NILFS: GC failed during preparation: cannot read > source blocks: err=-17 > Nov 30 14:39:38 nilfs_cleanerd[5789]: cannot clean segments: File exists > Nov 30 14:39:38 nilfs_cleanerd[5789]: shutdown > > I cannot ever start up the daemon. If I move to a 2.1 daemon, then it logs no > errors, but it cleans no old or newer (really older) checkpoints - it just > sits in a do-nothing mode (strace(1) shows he is hung on a mq_timedreceive > syscall). Hmm, this error seems to be caused by a known bug which was already fixed on nilfs-utils 2.1 with the following patch. It might be an actual corruption by the kernel code of nilfs2 if you were using old kernels, but it's most likely due to the bug. I will backport the fix to nilfs-utils 2.0 series and make another release of it. Regards, Ryusuke Konishi --- From: Ryusuke Konishi <[email protected]> nilfs_cleanerd: fix move block errors with cpfile and sufile This fixes the following gc error related to cpfile and sufile: nilfs_ioctl_move_inode_block: conflicting data buffer: ino=4, cno=0, offset=0, blocknr=78648, vblocknr=62283 Blocks of cpfile and sufile should be judged live only if they are latest, and should not depends on the protection period. Signed-off-by: Ryusuke Konishi <[email protected]> --- sbin/cleanerd/cleanerd.c | 10 ++++++++++ 1 files changed, 10 insertions(+), 0 deletions(-) diff --git a/sbin/cleanerd/cleanerd.c b/sbin/cleanerd/cleanerd.c index 45a0be0..138a444 100644 --- a/sbin/cleanerd/cleanerd.c +++ b/sbin/cleanerd/cleanerd.c @@ -748,6 +748,16 @@ static int nilfs_vdesc_is_live(const struct nilfs_vdesc *vdesc, long low, high, index; int s; + if (vdesc->vd_cno == 0) { + /* + * live/dead judge for sufile and cpfile should not + * depend on protection period and snapshots. Without + * this check, gc will cause buffer conflict error + * because their checkpoint number is always zero. + */ + return vdesc->vd_period.p_end == NILFS_CNO_MAX; + } + if (vdesc->vd_period.p_end == NILFS_CNO_MAX || vdesc->vd_period.p_end > protect) return 1; -- 1.7.7.4 -- To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
