Just finished copying 135 gigs of files into my dom0 system over
the network using smbclient. No errors occured during the copy.
Immediately afterwards i tried to run a find command, but it''s
segfaulting on me:
<1>Unable to handle kernel NULL pointer dereference at virtual address
00000000
printing eip:
c0037d86
*pde=00000000(55555000)
Oops: 0000
CPU: 0
EIP: 0819:[<c0037d86>] Not tainted
EFLAGS: 00211213
eax: 00000000 ebx: fffffff0 ecx: 0000000d edx: 00196f45
esi: 00000000 edi: c07fffa0 ebp: c07fff00 esp: c07ffee0
ds: 0821 es: 0821 ss: 0821
Process find (pid: 975, stackpage=c07ff000)<1>
Stack: c07fff50 00000000 c07fffa0 00000000 c1092758 c24bd000 00196f45 00000003
c07fff14 c002f571 c1d3aa40 c07fff50 c07fff50 c07fff5c c002fd40 c1d3aa40
c07fff50 00000000 c07fffa0 c24bd000 00000000 c278f8c0 0000000e c12c8974
Call Trace: [<c002f571>] [<c002fd40>] [<c002ffed>]
[<c003015e>] [<c00303cd>]
[<c002cf9a>] [<c009c08b>]
I''ve tried running find a few times, and the only differences each time
are in edi/ebp/esp/stackpage/stack/pid
I''m currently running rev 40f41ae00utn5d2f3tlNLcvG_QhiBA from
2004-07-13 17:24:48
This also happened to me on an older rev, but i was messing about with the vmmon
stuff at the time, and i also wanted to move up to a newer rev to see if that
fixed things before saying anything. Now, last time the find worked again after
a reboot, but that''s obviously no good. For the moment this time
i''ve just left
things as-is right now.
--
Jody Belka
knew (at) pimb (dot) org
-------------------------------------------------------
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xen-devel
Can you point us at the vmlinux file for this backtrace? Also, I can''t see the repository revision you''ve described -- what is the output of ''bk sccslog | head'' on your repository? Thanks, Keir> Just finished copying 135 gigs of files into my dom0 system over > the network using smbclient. No errors occured during the copy. > Immediately afterwards i tried to run a find command, but it''s > segfaulting on me: > > <1>Unable to handle kernel NULL pointer dereference at virtual address 00000000 > printing eip: > c0037d86 > *pde=00000000(55555000) > Oops: 0000 > CPU: 0 > EIP: 0819:[<c0037d86>] Not tainted > EFLAGS: 00211213 > eax: 00000000 ebx: fffffff0 ecx: 0000000d edx: 00196f45 > esi: 00000000 edi: c07fffa0 ebp: c07fff00 esp: c07ffee0 > ds: 0821 es: 0821 ss: 0821 > Process find (pid: 975, stackpage=c07ff000)<1> > Stack: c07fff50 00000000 c07fffa0 00000000 c1092758 c24bd000 00196f45 00000003 > c07fff14 c002f571 c1d3aa40 c07fff50 c07fff50 c07fff5c c002fd40 c1d3aa40 > c07fff50 00000000 c07fffa0 c24bd000 00000000 c278f8c0 0000000e c12c8974 > Call Trace: [<c002f571>] [<c002fd40>] [<c002ffed>] [<c003015e>] [<c00303cd>] > [<c002cf9a>] [<c009c08b>] > > > I''ve tried running find a few times, and the only differences each time > are in edi/ebp/esp/stackpage/stack/pid > > I''m currently running rev 40f41ae00utn5d2f3tlNLcvG_QhiBA from 2004-07-13 17:24:48 > > This also happened to me on an older rev, but i was messing about with the vmmon > stuff at the time, and i also wanted to move up to a newer rev to see if that > fixed things before saying anything. Now, last time the find worked again after > a reboot, but that''s obviously no good. For the moment this time i''ve just left > things as-is right now. > > -- > Jody Belka > knew (at) pimb (dot) org > > > ------------------------------------------------------- > This SF.Net email is sponsored by BEA Weblogic Workshop > FREE Java Enterprise J2EE developer tools! > Get your free copy of BEA WebLogic Workshop 8.1 today. > http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/xen-devel------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
Also, to clarify: there was no other domain than DOM0 involved in this transfer -- the transfer was from a remote machine? -- Keir> > Can you point us at the vmlinux file for this backtrace? Also, I can''t > see the repository revision you''ve described -- what is the output of > ''bk sccslog | head'' on your repository? > > Thanks, > Keir > > > Just finished copying 135 gigs of files into my dom0 system over > > the network using smbclient. No errors occured during the copy. > > Immediately afterwards i tried to run a find command, but it''s > > segfaulting on me: > > > > <1>Unable to handle kernel NULL pointer dereference at virtual address 00000000 > > printing eip: > > c0037d86 > > *pde=00000000(55555000) > > Oops: 0000 > > CPU: 0 > > EIP: 0819:[<c0037d86>] Not tainted > > EFLAGS: 00211213 > > eax: 00000000 ebx: fffffff0 ecx: 0000000d edx: 00196f45 > > esi: 00000000 edi: c07fffa0 ebp: c07fff00 esp: c07ffee0 > > ds: 0821 es: 0821 ss: 0821 > > Process find (pid: 975, stackpage=c07ff000)<1> > > Stack: c07fff50 00000000 c07fffa0 00000000 c1092758 c24bd000 00196f45 00000003 > > c07fff14 c002f571 c1d3aa40 c07fff50 c07fff50 c07fff5c c002fd40 c1d3aa40 > > c07fff50 00000000 c07fffa0 c24bd000 00000000 c278f8c0 0000000e c12c8974 > > Call Trace: [<c002f571>] [<c002fd40>] [<c002ffed>] [<c003015e>] [<c00303cd>] > > [<c002cf9a>] [<c009c08b>] > > > > > > I''ve tried running find a few times, and the only differences each time > > are in edi/ebp/esp/stackpage/stack/pid > > > > I''m currently running rev 40f41ae00utn5d2f3tlNLcvG_QhiBA from 2004-07-13 17:24:48 > > > > This also happened to me on an older rev, but i was messing about with the vmmon > > stuff at the time, and i also wanted to move up to a newer rev to see if that > > fixed things before saying anything. Now, last time the find worked again after > > a reboot, but that''s obviously no good. For the moment this time i''ve just left > > things as-is right now. > > > > -- > > Jody Belka > > knew (at) pimb (dot) org > > > > > > ------------------------------------------------------- > > This SF.Net email is sponsored by BEA Weblogic Workshop > > FREE Java Enterprise J2EE developer tools! > > Get your free copy of BEA WebLogic Workshop 8.1 today. > > http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click > > _______________________________________________ > > Xen-devel mailing list > > Xen-devel@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/xen-devel > > > > ------------------------------------------------------- > This SF.Net email is sponsored by BEA Weblogic Workshop > FREE Java Enterprise J2EE developer tools! > Get your free copy of BEA WebLogic Workshop 8.1 today. > http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/xen-devel------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
> > > Just finished copying 135 gigs of files into my dom0 system over > > > the network using smbclient. No errors occured during the copy. > > > Immediately afterwards i tried to run a find command, but it''s > > > segfaulting on me:> Also, to clarify: there was no other domain than DOM0 involved in this > transfer -- the transfer was from a remote machine?Further, did dom0 have swap configured? Ian ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
On Wed, 14 Jul, 2004 at 23:54:57 +0000, Jody Belka wrote:> Just finished copying 135 gigs of files into my dom0 system over > the network using smbclient. No errors occured during the copy. > Immediately afterwards i tried to run a find command, but it''s > segfaulting on me:On Thu, Jul 15th, 2004 at 09:00 (CEST), Keir Fraser wrote:> Can you point us at the vmlinux file for this backtrace? Also, I can''t > see the repository revision you''ve described -- what is the output of > ''bk sccslog | head'' on your repository?I''ve uploaded the vmlinuz, System.map and config files to http://test.pimb.org/xen/20040713172448.40f41ae00utn5d2f3tlNLcvG_QhiBA/ The output of the bk sccslog command is as follows: -------- ChangeSet 1.1076 04/07/13 18:24:48 kaf24@scramble.cl.cam.ac.uk +1 -0 Fairly major fixes to the network frontend driver. Much saner now. -------- I generally prefer to give out the MD5KEY names for the revisions, since the changeset numbers can, and often do, change whenever they feel like it. You can use the MD5KEY names anywhere you''d use a changeset number anyway, after all. On Thu, Jul 15th, 2004 at 09:10 (CEST), Keir Fraser wrote:> Also, to clarify: there was no other domain than DOM0 involved in this > transfer -- the transfer was from a remote machine?That''s right. there /is/ one other domain running, but it had nothing to do with the transfer at all, being purely a dns cache and server. On Thu, Jul 15, 2004 at 08:30:52AM +0100, Ian Pratt wrote:> Further, did dom0 have swap configured?Yes. I allocated 48megs of physical memory to dom0, and added an additional an extra 125megs of swap to it. The output of free as we speak (still haven''t rebooted the system) is: total used free shared buffers cached Mem: 44 41 2 0 1 8 -/+ buffers/cache: 32 12 Swap: 125 5 120 -- Jody Belka knew (at) pimb (dot) org ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
The backtrace is (innermost to outermost): d_lookup cached_lookup link_path_walk path_walk path_lookup __user_walk sys_lstat64 system_call The crash in d_lookup is due to a corrupted directory-entry hash-table list. Doesn''t look very Xen-related. This area of code isn''t modified for Xen, and no Xen drivers are involved in your data transfer. I wonder if this is reproducible in native-x86 Linux, if you run it with the same configuration? -- Keir> Just finished copying 135 gigs of files into my dom0 system over > the network using smbclient. No errors occured during the copy. > Immediately afterwards i tried to run a find command, but it''s > segfaulting on me: > > <1>Unable to handle kernel NULL pointer dereference at virtual address 00000000 > printing eip: > c0037d86 > *pde=00000000(55555000) > Oops: 0000 > CPU: 0 > EIP: 0819:[<c0037d86>] Not tainted > EFLAGS: 00211213 > eax: 00000000 ebx: fffffff0 ecx: 0000000d edx: 00196f45 > esi: 00000000 edi: c07fffa0 ebp: c07fff00 esp: c07ffee0 > ds: 0821 es: 0821 ss: 0821 > Process find (pid: 975, stackpage=c07ff000)<1> > Stack: c07fff50 00000000 c07fffa0 00000000 c1092758 c24bd000 00196f45 00000003 > c07fff14 c002f571 c1d3aa40 c07fff50 c07fff50 c07fff5c c002fd40 c1d3aa40 > c07fff50 00000000 c07fffa0 c24bd000 00000000 c278f8c0 0000000e c12c8974 > Call Trace: [<c002f571>] [<c002fd40>] [<c002ffed>] [<c003015e>] [<c00303cd>] > [<c002cf9a>] [<c009c08b>] > > > I''ve tried running find a few times, and the only differences each time > are in edi/ebp/esp/stackpage/stack/pid > > I''m currently running rev 40f41ae00utn5d2f3tlNLcvG_QhiBA from 2004-07-13 17:24:48 > > This also happened to me on an older rev, but i was messing about with the vmmon > stuff at the time, and i also wanted to move up to a newer rev to see if that > fixed things before saying anything. Now, last time the find worked again after > a reboot, but that''s obviously no good. For the moment this time i''ve just left > things as-is right now. > > -- > Jody Belka > knew (at) pimb (dot) org > > > ------------------------------------------------------- > This SF.Net email is sponsored by BEA Weblogic Workshop > FREE Java Enterprise J2EE developer tools! > Get your free copy of BEA WebLogic Workshop 8.1 today. > http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/xen-devel------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
A few months ago (before the 1.2 release) I had a similar problem: I''d do lots of file IO (big copies) and lots of swapping and sometimes processes like find would segfault due to a kernel null pointer dereference. Perhaps this is the same problem again. At the time, we decided it was probably a latent Linux bug and not Xen-specific but I haven''t managed to provoke it on native Linux. Mark ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
A bit offtopic, but maybe this will help somebody: On Thu, Jul 15, 2004 at 01:20:05PM +0000, Jody Belka wrote:> On Thu, Jul 15th, 2004 at 09:00 (CEST), Keir Fraser wrote: > > Can you point us at the vmlinux file for this backtrace? Also, I can''t > > see the repository revision you''ve described -- what is the output of > > ''bk sccslog | head'' on your repository?"bk changes -r+" is a more concise equivalent... but as Jody points out, you can just "bk changes -r40f41ae00utn5d2f3tlNLcvG_QhiBA" to find out what changeset that is.> I generally prefer to give out the MD5KEY names for the > revisions, since the changeset numbers can, and often do, > change whenever they feel like it. You can use the MD5KEY names > anywhere you''d use a changeset number anyway, after all.That''s perfectly true. In fact that cset has a different version number in my local tree than it does in Jody''s output. MD5KEYs and KEYs are eternal, everything else is temporary. (Well, I suppose the contents are also eternal. But comments and revision numbers, in particular, are mutable.) And on a slightly more xen-related topic:> On Thu, Jul 15, 2004 at 08:30:52AM +0100, Ian Pratt wrote: > > Further, did dom0 have swap configured? > > Yes. I allocated 48megs of physical memory to dom0, and added > an additional an extra 125megs of swap to it. The output of free > as we speak (still haven''t rebooted the system) is: > > total used free shared buffers cached > Mem: 44 41 2 0 1 8 > -/+ buffers/cache: 32 12 > Swap: 125 5 120Hmm, looks like your free(1) has -m as the default... I''ve always seen -k as the default. Or else you got it to boot with 11 pages of memory, which would be quite a feat. -andy ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel
Jody Belka
2004-Jul-16 20:57 UTC
Re: a few bk hints (was Re: [Xen-devel] segfault in dom0)
On Thu, Jul 15, 2004 at 08:33:41PM -0500, Andy Isaacson wrote:> And on a slightly more xen-related topic: > > > On Thu, Jul 15, 2004 at 08:30:52AM +0100, Ian Pratt wrote: > > > Further, did dom0 have swap configured? > > > > Yes. I allocated 48megs of physical memory to dom0, and added > > an additional an extra 125megs of swap to it. The output of free > > as we speak (still haven''t rebooted the system) is: > > > > total used free shared buffers cached > > Mem: 44 41 2 0 1 8 > > -/+ buffers/cache: 32 12 > > Swap: 125 5 120 > > Hmm, looks like your free(1) has -m as the default... I''ve always seen > -k as the default. > > Or else you got it to boot with 11 pages of memory, which would be > quite a feat.Nope, just a shell alias; i generally prefer to see it in megs, so i made it easier for me. -- Jody Belka knew (at) pimb (dot) org ------------------------------------------------------- This SF.Net email is sponsored by BEA Weblogic Workshop FREE Java Enterprise J2EE developer tools! Get your free copy of BEA WebLogic Workshop 8.1 today. http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click _______________________________________________ Xen-devel mailing list Xen-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xen-devel