* [FIXED?] reiser4: problem with key inconsistency
@ 2012-10-28 21:56 Edward Shishkin
2012-10-31 17:24 ` Ivan Shapovalov
0 siblings, 1 reply; 8+ messages in thread
From: Edward Shishkin @ 2012-10-28 21:56 UTC (permalink / raw)
To: reiserfs-devel
Cc: Luciano, I put the Who? in Mishehu, Jonáš Vidra,
dimas000, Bringfried Stecklum, Dušan Čolić,
Alexey Shvetsov
[-- Attachment #1: Type: text/plain, Size: 689 bytes --]
Hi all.
It seems I do have the long-awaited fix.
The symptoms of the problem: kernel starts to issue complaints
about key inconsistency with the suggestions to check the partition
with fsck. Fsck sometimes finds corruptions, but sometimes reports
that file system is consistent.
Please apply the attached patch to reiser4 stuff for 3.6.2, or
download reiser4-for-3.6.4 (which already contains this patch), and
let me know if the problem has not gone:
http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
Please, don't rush to deploy production systems on reiser4: it can
take up to 80 hours to reproduce the problem with high workload.
Thank you for your help.
Edward.
[-- Attachment #2: reiser4-update-dkeys-in-squeeze_right_twig.patch --]
[-- Type: text/x-patch, Size: 1451 bytes --]
Update dkeys of the pair of nodes, which took participation
in the "implicit shift" during squalloc on the twig level in
flush time.
Signed-off-by: Edward Shishkin <edward.shishkin@gmail.com>
---
fs/reiser4/flush.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)
--- linux-3.6.2_.orig/fs/reiser4/flush.c
+++ linux-3.6.2_/fs/reiser4/flush.c
@@ -450,6 +450,7 @@ static int znode_check_flushprepped(znod
{
return jnode_check_flushprepped(ZJNODE(node));
}
+static void update_znode_dkeys(znode * left, znode * right);
/* Flush position functions */
static void pos_init(flush_pos_t *pos);
@@ -1525,6 +1526,7 @@ static int squeeze_right_twig(znode * le
int ret = SUBTREE_MOVED;
coord_t coord; /* used to iterate over items */
reiser4_key stop_key;
+ reiser4_tree *tree;
assert("jmacd-2008", !node_is_empty(right));
coord_init_first_unit(&coord, right);
@@ -1552,6 +1554,18 @@ static int squeeze_right_twig(znode * le
ON_DEBUG(shift_check(vp, left, coord.node));
}
+ /*
+ * @left and @right nodes participated in the
+ * implicit shift, determined by the pair of
+ * functions:
+ * . squalloc_extent() - append units to the @left
+ * . squalloc_right_twig_cut() - cut the units from @right
+ * so update their delimiting keys
+ */
+ tree = znode_get_tree(left);
+ write_lock_dk(tree);
+ update_znode_dkeys(left, right);
+ write_unlock_dk(tree);
if (node_is_empty(coord.node))
ret = SQUEEZE_SOURCE_EMPTY;
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [FIXED?] reiser4: problem with key inconsistency
2012-10-28 21:56 [FIXED?] reiser4: problem with key inconsistency Edward Shishkin
@ 2012-10-31 17:24 ` Ivan Shapovalov
2012-10-31 18:43 ` Edward Shishkin
2012-11-03 16:02 ` Edward Shishkin
0 siblings, 2 replies; 8+ messages in thread
From: Ivan Shapovalov @ 2012-10-31 17:24 UTC (permalink / raw)
To: edward.shishkin; +Cc: reiserfs-devel
On 28 октября 2012 22:56:14 Edward Shishkin wrote:
> Hi all.
> It seems I do have the long-awaited fix.
>
> The symptoms of the problem: kernel starts to issue complaints
> about key inconsistency with the suggestions to check the partition
> with fsck. Fsck sometimes finds corruptions, but sometimes reports
> that file system is consistent.
>
> Please apply the attached patch to reiser4 stuff for 3.6.2, or
> download reiser4-for-3.6.4 (which already contains this patch), and
> let me know if the problem has not gone:
> http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
>
> Please, don't rush to deploy production systems on reiser4: it can
> take up to 80 hours to reproduce the problem with high workload.
>
> Thank you for your help.
> Edward.
Alas, it hasn't gone... Or maybe it's a different problem, but today (on a
debugless build) I've got a massive crash (mostly -2 and -5 errors for
different VFS operations - e.g.
nikita-717/edward-1611/edward-430/nikita-2219/nikita-2221, many vs-3533 and
edward-1460/edward-1608/edward-156 warnings).
Fsck said SIGSEGV, so I had to restore from a daily backup. Don't know if that
is related to allocator, but anyway.
And I don't have much time these days to set up VM with a clone of my system
for kgdb, sorry... Though I hope I'll do that on next week (on vacations).
BTW, changing RAM didn't help, and old RAM passes memtest - so here it's truly
a bug somewhere in allocator.
Regards,
Ivan.
--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [FIXED?] reiser4: problem with key inconsistency
2012-10-31 17:24 ` Ivan Shapovalov
@ 2012-10-31 18:43 ` Edward Shishkin
2012-11-04 12:14 ` Ivan Shapovalov
2012-11-03 16:02 ` Edward Shishkin
1 sibling, 1 reply; 8+ messages in thread
From: Edward Shishkin @ 2012-10-31 18:43 UTC (permalink / raw)
To: Ivan Shapovalov; +Cc: reiserfs-devel
On 10/31/2012 06:24 PM, Ivan Shapovalov wrote:
> On 28 октября 2012 22:56:14 Edward Shishkin wrote:
>> Hi all.
>> It seems I do have the long-awaited fix.
>>
>> The symptoms of the problem: kernel starts to issue complaints
>> about key inconsistency with the suggestions to check the partition
>> with fsck. Fsck sometimes finds corruptions, but sometimes reports
>> that file system is consistent.
>>
>> Please apply the attached patch to reiser4 stuff for 3.6.2, or
>> download reiser4-for-3.6.4 (which already contains this patch), and
>> let me know if the problem has not gone:
>> http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
>>
>> Please, don't rush to deploy production systems on reiser4: it can
>> take up to 80 hours to reproduce the problem with high workload.
>>
>> Thank you for your help.
>> Edward.
>
> Alas, it hasn't gone... Or maybe it's a different problem, but today (on a
> debugless build) I've got a massive crash (mostly -2 and -5 errors for
> different VFS operations - e.g.
> nikita-717/edward-1611/edward-430/nikita-2219/nikita-2221, many vs-3533 and
> edward-1460/edward-1608/edward-156 warnings).
Hmm, strange..
It has gone for me.. At least, my machine works already
110 hours without complaints (before fixing I encountered
this usually in 30-50 h). Well, I'll continue stressing.
Anybody else has the problems with the patched stuff?
> Fsck said SIGSEGV, so I had to restore from a daily backup. Don't know if that
> is related to allocator, but anyway.
>
> And I don't have much time these days to set up VM with a clone of my system
> for kgdb, sorry... Though I hope I'll do that on next week (on vacations).
>
> BTW, changing RAM didn't help, and old RAM passes memtest - so here it's truly
> a bug somewhere in allocator.
May I ask you also to test your partition with badblock program?
Thanks,
Edward.
--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [FIXED?] reiser4: problem with key inconsistency
2012-10-31 18:43 ` Edward Shishkin
@ 2012-11-04 12:14 ` Ivan Shapovalov
0 siblings, 0 replies; 8+ messages in thread
From: Ivan Shapovalov @ 2012-11-04 12:14 UTC (permalink / raw)
To: Edward Shishkin; +Cc: reiserfs-devel
On 31 Nov 2012 19:43:06 Edward Shishkin wrote:
> On 10/31/2012 06:24 PM, Ivan Shapovalov wrote:
> > On 28 октября 2012 22:56:14 Edward Shishkin wrote:
> >> Hi all.
> >> It seems I do have the long-awaited fix.
> >>
> >> The symptoms of the problem: kernel starts to issue complaints
> >> about key inconsistency with the suggestions to check the partition
> >> with fsck. Fsck sometimes finds corruptions, but sometimes reports
> >> that file system is consistent.
> >>
> >> Please apply the attached patch to reiser4 stuff for 3.6.2, or
> >> download reiser4-for-3.6.4 (which already contains this patch), and
> >> let me know if the problem has not gone:
> >> http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
> >>
> >> Please, don't rush to deploy production systems on reiser4: it can
> >> take up to 80 hours to reproduce the problem with high workload.
> >>
> >> Thank you for your help.
> >> Edward.
> >
> > Alas, it hasn't gone... Or maybe it's a different problem, but today (on a
> > debugless build) I've got a massive crash (mostly -2 and -5 errors for
> > different VFS operations - e.g.
> > nikita-717/edward-1611/edward-430/nikita-2219/nikita-2221, many vs-3533
> > and
> > edward-1460/edward-1608/edward-156 warnings).
>
> Hmm, strange..
> It has gone for me.. At least, my machine works already
> 110 hours without complaints (before fixing I encountered
> this usually in 30-50 h). Well, I'll continue stressing.
>
> Anybody else has the problems with the patched stuff?
>
> > Fsck said SIGSEGV, so I had to restore from a daily backup. Don't know if
> > that is related to allocator, but anyway.
> >
> > And I don't have much time these days to set up VM with a clone of my
> > system for kgdb, sorry... Though I hope I'll do that on next week (on
> > vacations).
> >
> > BTW, changing RAM didn't help, and old RAM passes memtest - so here it's
> > truly a bug somewhere in allocator.
>
> May I ask you also to test your partition with badblock program?
3 passes (non-destructive R/W test, random pattern on each pass), no errors.
--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [FIXED?] reiser4: problem with key inconsistency
2012-10-31 17:24 ` Ivan Shapovalov
2012-10-31 18:43 ` Edward Shishkin
@ 2012-11-03 16:02 ` Edward Shishkin
2012-11-04 12:09 ` Ivan Shapovalov
2013-01-07 11:05 ` Dušan Čolić
1 sibling, 2 replies; 8+ messages in thread
From: Edward Shishkin @ 2012-11-03 16:02 UTC (permalink / raw)
To: Ivan Shapovalov; +Cc: reiserfs-devel
On 10/31/2012 06:24 PM, Ivan Shapovalov wrote:
> On 28 октября 2012 22:56:14 Edward Shishkin wrote:
>> Hi all.
>> It seems I do have the long-awaited fix.
>>
>> The symptoms of the problem: kernel starts to issue complaints
>> about key inconsistency with the suggestions to check the partition
>> with fsck. Fsck sometimes finds corruptions, but sometimes reports
>> that file system is consistent.
>>
>> Please apply the attached patch to reiser4 stuff for 3.6.2, or
>> download reiser4-for-3.6.4 (which already contains this patch), and
>> let me know if the problem has not gone:
>> http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
>>
>> Please, don't rush to deploy production systems on reiser4: it can
>> take up to 80 hours to reproduce the problem with high workload.
>>
>> Thank you for your help.
>> Edward.
>
> Alas, it hasn't gone... Or maybe it's a different problem, but today (on a
> debugless build) I've got a massive crash (mostly -2 and -5 errors for
> different VFS operations - e.g.
> nikita-717/edward-1611/edward-430/nikita-2219/nikita-2221, many vs-3533 and
> edward-1460/edward-1608/edward-156 warnings).
> Fsck said SIGSEGV,
There are pending patches for reiser4progs, please don't forget to apply
them when checking your partitions:
http://marc.info/?l=reiserfs-devel&m=129703129519745&w=2
http://marc.info/?l=reiserfs-devel&m=127504471326353&w=2
Thanks,
Edward.
so I had to restore from a daily backup. Don't know if that
> is related to allocator, but anyway.
>
> And I don't have much time these days to set up VM with a clone of my system
> for kgdb, sorry... Though I hope I'll do that on next week (on vacations).
>
> BTW, changing RAM didn't help, and old RAM passes memtest - so here it's truly
> a bug somewhere in allocator.
>
> Regards,
> Ivan.
--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [FIXED?] reiser4: problem with key inconsistency
2012-11-03 16:02 ` Edward Shishkin
@ 2012-11-04 12:09 ` Ivan Shapovalov
2013-01-07 11:05 ` Dušan Čolić
1 sibling, 0 replies; 8+ messages in thread
From: Ivan Shapovalov @ 2012-11-04 12:09 UTC (permalink / raw)
To: Edward Shishkin; +Cc: reiserfs-devel
On 03 Nov 2012 17:02:10 Edward Shishkin wrote:
> On 10/31/2012 06:24 PM, Ivan Shapovalov wrote:
> > On 28 октября 2012 22:56:14 Edward Shishkin wrote:
> >> Hi all.
> >> It seems I do have the long-awaited fix.
> >>
> >> The symptoms of the problem: kernel starts to issue complaints
> >> about key inconsistency with the suggestions to check the partition
> >> with fsck. Fsck sometimes finds corruptions, but sometimes reports
> >> that file system is consistent.
> >>
> >> Please apply the attached patch to reiser4 stuff for 3.6.2, or
> >> download reiser4-for-3.6.4 (which already contains this patch), and
> >> let me know if the problem has not gone:
> >> http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
> >>
> >> Please, don't rush to deploy production systems on reiser4: it can
> >> take up to 80 hours to reproduce the problem with high workload.
> >>
> >> Thank you for your help.
> >> Edward.
> >
> > Alas, it hasn't gone... Or maybe it's a different problem, but today (on a
> > debugless build) I've got a massive crash (mostly -2 and -5 errors for
> > different VFS operations - e.g.
> > nikita-717/edward-1611/edward-430/nikita-2219/nikita-2221, many vs-3533
> > and
> > edward-1460/edward-1608/edward-156 warnings).
> > Fsck said SIGSEGV,
>
> There are pending patches for reiser4progs, please don't forget to apply
> them when checking your partitions:
>
> http://marc.info/?l=reiserfs-devel&m=129703129519745&w=2
> http://marc.info/?l=reiserfs-devel&m=127504471326353&w=2
Ah, these were not applied. Anyway, I don't have the faulty partition image,
so I'll probably wait for the next crash (with unpatched fsck) and check if
the problem is gone after patching.
Thanks,
Ivan.
>
>
> > so I had to restore from a daily backup. Don't know if that
> > is related to allocator, but anyway.
> >
> > And I don't have much time these days to set up VM with a clone of my
> > system for kgdb, sorry... Though I hope I'll do that on next week (on
> > vacations).
> >
> > BTW, changing RAM didn't help, and old RAM passes memtest - so here it's
> > truly a bug somewhere in allocator.
> >
> > Regards,
> > Ivan.
--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [FIXED?] reiser4: problem with key inconsistency
2012-11-03 16:02 ` Edward Shishkin
2012-11-04 12:09 ` Ivan Shapovalov
@ 2013-01-07 11:05 ` Dušan Čolić
2013-01-17 18:40 ` Edward Shishkin
1 sibling, 1 reply; 8+ messages in thread
From: Dušan Čolić @ 2013-01-07 11:05 UTC (permalink / raw)
To: reiserfs-devel
On Sat, Nov 3, 2012 at 5:02 PM, Edward Shishkin
<edward.shishkin@gmail.com> wrote:
> On 10/31/2012 06:24 PM, Ivan Shapovalov wrote:
>>
>> On 28 октября 2012 22:56:14 Edward Shishkin wrote:
>>>
>>> Hi all.
>>> It seems I do have the long-awaited fix.
>>>
>>> The symptoms of the problem: kernel starts to issue complaints
>>> about key inconsistency with the suggestions to check the partition
>>> with fsck. Fsck sometimes finds corruptions, but sometimes reports
>>> that file system is consistent.
>>>
>>> Please apply the attached patch to reiser4 stuff for 3.6.2, or
>>> download reiser4-for-3.6.4 (which already contains this patch), and
>>> let me know if the problem has not gone:
>>> http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
>>>
>>> Please, don't rush to deploy production systems on reiser4: it can
>>> take up to 80 hours to reproduce the problem with high workload.
>>>
>>> Thank you for your help.
>>> Edward.
>>
>>
>> Alas, it hasn't gone... Or maybe it's a different problem, but today (on a
>> debugless build) I've got a massive crash (mostly -2 and -5 errors for
>> different VFS operations - e.g.
>> nikita-717/edward-1611/edward-430/nikita-2219/nikita-2221, many vs-3533
>> and
>> edward-1460/edward-1608/edward-156 warnings).
>> Fsck said SIGSEGV,
>
>
>
> There are pending patches for reiser4progs, please don't forget to apply
> them when checking your partitions:
>
> http://marc.info/?l=reiserfs-devel&m=129703129519745&w=2
> http://marc.info/?l=reiserfs-devel&m=127504471326353&w=2
>
Is this bug fixed in 3.7?
What is blocking the release of reiser4progs-1.0.8 with those pending patches?
Are there some other open bugs?
Thanks
Dušan
PS. Sorry for sending this again, vger.kernel.org rejected my first mail.
> Thanks,
> Edward.
>
>
>
> so I had to restore from a daily backup. Don't know if that
>>
>> is related to allocator, but anyway.
>>
>> And I don't have much time these days to set up VM with a clone of my
>> system
>> for kgdb, sorry... Though I hope I'll do that on next week (on vacations).
>>
>> BTW, changing RAM didn't help, and old RAM passes memtest - so here it's
>> truly
>> a bug somewhere in allocator.
>>
>> Regards,
>> Ivan.
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [FIXED?] reiser4: problem with key inconsistency
2013-01-07 11:05 ` Dušan Čolić
@ 2013-01-17 18:40 ` Edward Shishkin
0 siblings, 0 replies; 8+ messages in thread
From: Edward Shishkin @ 2013-01-17 18:40 UTC (permalink / raw)
To: Dušan Čolić; +Cc: reiserfs-devel
On 01/07/2013 12:05 PM, Dušan Čolić wrote:
> On Sat, Nov 3, 2012 at 5:02 PM, Edward Shishkin
> <edward.shishkin@gmail.com> wrote:
>> On 10/31/2012 06:24 PM, Ivan Shapovalov wrote:
>>> On 28 октября 2012 22:56:14 Edward Shishkin wrote:
>>>> Hi all.
>>>> It seems I do have the long-awaited fix.
>>>>
>>>> The symptoms of the problem: kernel starts to issue complaints
>>>> about key inconsistency with the suggestions to check the partition
>>>> with fsck. Fsck sometimes finds corruptions, but sometimes reports
>>>> that file system is consistent.
>>>>
>>>> Please apply the attached patch to reiser4 stuff for 3.6.2, or
>>>> download reiser4-for-3.6.4 (which already contains this patch), and
>>>> let me know if the problem has not gone:
>>>> http://sourceforge.net/projects/reiser4/files/reiser4-for-linux-3.x/
>>>>
>>>> Please, don't rush to deploy production systems on reiser4: it can
>>>> take up to 80 hours to reproduce the problem with high workload.
>>>>
>>>> Thank you for your help.
>>>> Edward.
>>>
>>> Alas, it hasn't gone... Or maybe it's a different problem, but today (on a
>>> debugless build) I've got a massive crash (mostly -2 and -5 errors for
>>> different VFS operations - e.g.
>>> nikita-717/edward-1611/edward-430/nikita-2219/nikita-2221, many vs-3533
>>> and
>>> edward-1460/edward-1608/edward-156 warnings).
>>> Fsck said SIGSEGV,
>>
>>
>> There are pending patches for reiser4progs, please don't forget to apply
>> them when checking your partitions:
>>
>> http://marc.info/?l=reiserfs-devel&m=129703129519745&w=2
>> http://marc.info/?l=reiserfs-devel&m=127504471326353&w=2
>>
> Is this bug fixed in 3.7?
So, we have found 2 problems, which led to this kind of corruption
(key inconsistency):
1) nobody updated dkeys (fixed in 3.6):
http://marc.info/?l=reiserfs-devel&m=135146138331012&w=2
2) default ->migratepage() didn't work for reiser4 (isolated in 3.7).
>
> What is blocking the release of reiser4progs-1.0.8 with those pending patches?
>
> Are there some other open bugs?
Since we are not in mainline tree, we'll periodically encounter various
oopses in new releases. However, I think it will be easy to fix.
Edward.
--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2013-01-17 18:40 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-10-28 21:56 [FIXED?] reiser4: problem with key inconsistency Edward Shishkin
2012-10-31 17:24 ` Ivan Shapovalov
2012-10-31 18:43 ` Edward Shishkin
2012-11-04 12:14 ` Ivan Shapovalov
2012-11-03 16:02 ` Edward Shishkin
2012-11-04 12:09 ` Ivan Shapovalov
2013-01-07 11:05 ` Dušan Čolić
2013-01-17 18:40 ` Edward Shishkin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).