https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #41 from prochazka 2011-04-21
12:39:11 ---
Hi,
I never test #38 from avi kivity suggest,
and now i suppres all bridge ( brctl) and netfilter configuration from my
server,
i'm using openvswitch without this problem.
Regards,
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #40 from Avi Kivity 2011-04-21 12:13:33 ---
(In reply to comment #39)
> Is this issue resolved?
> I'm seeing almost the same issue with 2.6.38 + qemu-kvm-0.14.0 (on NEC
> Express5800/110Ge server), except that my backtrace is call
https://bugzilla.kernel.org/show_bug.cgi?id=27052
Taisuke Yamada changed:
What|Removed |Added
CC||tai.ker...@cc.rakugaki.org
--- Comme
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #38 from Avi Kivity 2011-02-14 16:02:06 ---
Please revert ca44ac386181ba7 and try again (with netfilter enabled).
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are receiving this mail beca
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #37 from prochazka 2011-02-14
15:51:14 ---
Sorry I do not read your post to netfilter devel list.
Nicolas
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are receiving this mail because: --
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #36 from prochazka 2011-02-14
15:50:02 ---
I agree with you but i must resend this problem to kernel module bridge or
netfilter, I must isolate the issue .
Regards,
Nicolas Prochazka.
--
Configure bugmail: https://bugzilla.ke
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #35 from Avi Kivity 2011-02-14 14:52:24 ---
Even if it isn't a kvm bug, it needs to be fixed.
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #34 from prochazka 2011-02-13
20:13:53 ---
When I said it worked with 2.6.36 , in fact, I not used brigde and netfilter
with this test ,
so now as avi says, perhaps it's more brctl or netfilter bug and not kvm.
If you think it
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #33 from Marcelo Tosatti 2011-02-13 19:39:09
---
So something else appears to be corrupting memory (i verified and its unlikely
to be KVM). Perhaps go back to 2.6.36.3 in an attempt to isolate the issue?
--
Configure bugmail: h
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #32 from prochazka 2011-02-13
16:38:03 ---
Can you explain your idea ?
In fact, we are using DNAT and SNAT rules to redirect and access VM Guest ( VNC
and RDP ) . And to reproduce bug, connection seems to be important :
1 - Run
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #31 from Avi Kivity 2011-02-13 16:11:53 ---
What happens if you disable (rmmod) all netfilter modules?
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are receiving this mail because: --
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #30 from prochazka 2011-02-12
17:06:02 ---
hi marcelo,
I enable CONFIG_SLUB_DEBUG=y and boot with slub_debug=FZPU,
but now I cannot reproduce bug
I have a lot of : BUG kmalloc-1024: Object padding overwritten
example :
Feb 1
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #29 from Marcelo Tosatti 2011-02-11 21:33:17
---
gfn=dead01ff idx=511 lvl=1 d=0 spte=0 old_spte=f001
The last spte is 0xf001, it should be 0xf001 (two bytes at
end of pgtable zeroed). Something is scr
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #28 from prochazka 2011-02-11
15:43:22 ---
Marcelo,
debug info
Regards,
Nicolas
general protection fault: [#1] SMP
last sysfs file: /sys/devices/system/cpu/cpu7/cache/index2/shared_cpu_map
CPU 0
Modules linked in: kvm_i
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #27 from Marcelo Tosatti 2011-02-10 16:57:59
---
Created an attachment (id=47152)
--> (https://bugzilla.kernel.org/attachment.cgi?id=47152)
kvm-debug-spte-gfn-2.patch
--
Configure bugmail: https://bugzilla.kernel.org/userprefs
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #26 from Marcelo Tosatti 2011-02-10 16:57:17
---
Nicolas,
New debug patch attached. Please try it on top of clean 2.6.37.
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are receiving this
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #25 from prochazka 2011-02-10
14:16:51 ---
cmdline
/usr/local/bin/qemu -name Soins_003 -vga std -net
tap,vlan=0,name=interne,ifname=vmtap5 -net
nic,vlan=0,macaddr=ac:de:48:1d:e8:2c,model=e1000 -cpu host -localtime -usb
-usbdevic
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #24 from prochazka 2011-02-10
14:14:25 ---
I can now reproduce it under this circonstance on different server
- Windows XP guest SP2 : guest OS seems to be important, other XP sp3 works
fine
- connect with vnc to this guest an
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #23 from Marcelo Tosatti 2011-02-10 13:50:08
---
Nicolas,
On comment #2 you mention the bug could not be reproduced, but in comment #3
you report it without hugepages enabled. So, were you using hugepages or not,
in the reports
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #22 from Marcelo Tosatti 2011-02-10 13:36:25
---
Problem description:
Present spte is dropped while syncing 32-bit level 1 shadow page. But
sp->gfns[index] contains uninitialized value (0 or f001), so
gfn->rmap conve
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #21 from prochazka 2011-02-10
08:25:35 ---
Dear all,
Do you have more information about this bug submit ?
Regards,
Nicolas Prochazka.
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are r
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #20 from prochazka 2011-02-08
10:50:13 ---
gfn=f001 idx=511 lvl=1 d=0 spte=0
sp->gfns[511] = f001
sp->gfns[511] = f001
sp->gfns[511] = f001
sp->gfns[510] = f001
sp->gfns
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #19 from prochazka 2011-02-08
09:46:04 ---
An other log without patch #18 / #17 ( as #16 )
I try with #18
Feb 8 00:16:37 bergson18494 rmap_remove: gfn=0 index=511 level=1
Feb 8 00:16:37 bergson18494 rmap_remove: 88027ed4
https://bugzilla.kernel.org/show_bug.cgi?id=27052
Marcelo Tosatti changed:
What|Removed |Added
Attachment #45162|0 |1
is obsolete|
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #17 from Marcelo Tosatti 2011-02-04 21:11:55
---
Nicolas,
Please try kvm-debug-spte-gfn.patch on top of plain 2.6.37 (without the
previous two patches).
Thanks
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?t
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #16 from prochazka 2011-01-28
06:58:03 ---
Et voilà :
Jan 28 01:28:18 bergson25412 rmap_remove: 88011ce3fff8 1->BUG
Jan 28 01:28:18 bergson25412 [ cut here ]
Jan 28 01:28:18 bergson25412 kernel BUG
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #15 from prochazka 2011-01-25
17:29:00 ---
previous test is with hugepages.
I try to test but i can reproduce only to one server, and this server is not
mine.
Regards,
Nicolas.
--
Configure bugmail: https://bugzilla.kernel.or
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #14 from Marcelo Tosatti 2011-01-25 17:08:42
---
Created an attachment (id=45162)
--> (https://bugzilla.kernel.org/attachment.cgi?id=45162)
debug rmap remove
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #13 from Marcelo Tosatti 2011-01-25 17:07:58
---
Nicolas,
Can you please back out the first patch and apply this debug one instead?
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are rece
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #12 from Marcelo Tosatti 2011-01-25 13:55:05
---
It gets inconsistent if mmu_set_spte updates an already present spte with a
different gfn. For example:
- unsync page spte N gfn A
- guest updates gpte that spte N shadows with g
https://bugzilla.kernel.org/show_bug.cgi?id=27052
Avi Kivity changed:
What|Removed |Added
CC||a...@redhat.com
--- Comment #11 from Avi
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #10 from Marcelo Tosatti 2011-01-25 13:09:07
---
Nicolas,
It would be good if you can test with hugepages enabled again.
Thanks for the detailed report.
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=emai
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #9 from prochazka 2011-01-25
08:49:48 ---
seems to be closed,
but our tests are not exhaustive
Regards,
NP.
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are receiving this mail because
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #8 from Marcelo Tosatti 2011-01-21 03:29:36
---
Created an attachment (id=44552)
--> (https://bugzilla.kernel.org/attachment.cgi?id=44552)
update sp->gfns on pte update path
--
Configure bugmail: https://bugzilla.kernel.org/us
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #7 from Marcelo Tosatti 2011-01-21 03:27:36
---
Nicolas,
My bad. Can you please try the following patch.
--
Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email
--- You are receiving this mail because: --
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #6 from prochazka 2011-01-20
19:45:49 ---
hello,
I do not understand, patch seems to be already apply on 2.6.37 kernel tree,
and my test are based on this release.
NP.
--
Configure bugmail: https://bugzilla.kernel.org/userpr
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #5 from Marcelo Tosatti 2011-01-20 17:30:38
---
Created an attachment (id=44522)
--> (https://bugzilla.kernel.org/attachment.cgi?id=44522)
KVM: MMU: fix rmap_remove on non present sptes
KVM: MMU: fix rmap_remove on non present
https://bugzilla.kernel.org/show_bug.cgi?id=27052
Marcelo Tosatti changed:
What|Removed |Added
CC||mtosa...@redhat.com
--- Comment #4
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #3 from prochazka 2011-01-19
11:34:21 ---
Sorry,
witout hugepage, bug is alway here :
rmap_remove: 8802455bfff8 0->BUG
[ cut here ]
kernel BUG at arch/x86/kvm/mmu.c:695!
invalid opcode: [#
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #2 from prochazka 2011-01-19
11:26:16 ---
it seems without hugepage, i can not reproduce this bugs
( -mem-prealloc -mem-path /hugepages )
/usr/local/bin/qemu -name R005 -vga std -net
tap,vlan=0,name=interne,ifname=vmtap5 -net
https://bugzilla.kernel.org/show_bug.cgi?id=27052
--- Comment #1 from prochazka 2011-01-19
11:05:05 ---
cpuinfo and cmdline :
/usr/local/bin/qemu -name R005 -vga std -net
tap,vlan=0,name=interne,ifname=vmtap5 -net
nic,vlan=0,macaddr=ac:de:48:3f:74:73,model=rtl8139 -localtime -usb -usbdev
41 matches
Mail list logo
