kernel BUG look familiar?

kernel BUG look familiar?

[Matt Ayres]

This was reported by one of my users and it shows on every boot for 
them.  I wanted to know if this was a known issue before filing a bug 
report.

Pausing... Pausing... Pausing... Pausing... <3>BUG: soft lockup detected 
on CPU#0!

Pid: 1, comm: init
EIP: 0061:[<c0106b0d>] CPU: 0
EIP is at delay_tsc+0xd/0x15
EFLAGS: 00000297 Not tainted (2.6.16-xen #2)
EAX: 51d9f61d EBX: 0022d869 ECX: 51cc64c4 EDX: 00000101
ESI: 00000002 EDI: c010eeb0 EBP: bf955008 DS: 007b ES: 007b
CR0: 8005003b CR2: b7f17e57 CR3: 07903000 CR4: 00000660
[<c010efb2>] do_fixup_4gb_segment+0x102/0x128
[<c0110a18>] do_page_fault+0x1c5/0x7dd
[<c01049d3>] error_code+0x2b/0x30
Pausing... Continuing...


Host info..

xen_major              : 3
xen_minor              : 0
xen_extra              : .2-1
xen_caps               : xen-3.0-x86_32p
platform_params        : virt_start=0xf5800000
xen_changeset          : Tue Apr 11 22:14:24 2006 +0100 9612:1f2520faecda

Re: kernel BUG look familiar?

[David F. Barrera]

There's already a bug for a similar issue:

http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=543

-- 
Regards,

David F Barrera
Linux Technology Center
Systems and Technology Group, IBM

"The wisest men follow their own direction. "
                                                       Euripides




Matt Ayres wrote:

> This was reported by one of my users and it shows on every boot for 
> them.  I wanted to know if this was a known issue before filing a bug 
> report.
>
> Pausing... Pausing... Pausing... Pausing... <3>BUG: soft lockup 
> detected on CPU#0!
>
> Pid: 1, comm: init
> EIP: 0061:[<c0106b0d>] CPU: 0
> EIP is at delay_tsc+0xd/0x15
> EFLAGS: 00000297 Not tainted (2.6.16-xen #2)
> EAX: 51d9f61d EBX: 0022d869 ECX: 51cc64c4 EDX: 00000101
> ESI: 00000002 EDI: c010eeb0 EBP: bf955008 DS: 007b ES: 007b
> CR0: 8005003b CR2: b7f17e57 CR3: 07903000 CR4: 00000660
> [<c010efb2>] do_fixup_4gb_segment+0x102/0x128
> [<c0110a18>] do_page_fault+0x1c5/0x7dd
> [<c01049d3>] error_code+0x2b/0x30
> Pausing... Continuing...
>
>
> Host info..
>
> xen_major              : 3
> xen_minor              : 0
> xen_extra              : .2-1
> xen_caps               : xen-3.0-x86_32p
> platform_params        : virt_start=0xf5800000
> xen_changeset          : Tue Apr 11 22:14:24 2006 +0100 9612:1f2520faecda
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel

Re: kernel BUG look familiar?

[Matt Ayres]

David F. Barrera wrote:
> There's already a bug for a similar issue:
> 
> http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=543
> 

Keir has already committed a fix for it.

# HG changeset patch
# User kaf24@firebug.cl.cam.ac.uk
# Node ID 1eec4d22bd8f5aba974cfb60325da95d826dc9e0
# Parent  765250390f84c7439d679ffadd07c9454baff778
Touch softlockup watchdog while warning user about use of
tls libraries.

Signed-off-by: Keir Fraser <keir@xensource.com>

RE: kernel BUG look familiar?

[Ian Pratt]

> This was reported by one of my users and it shows on every boot for 
> them.  I wanted to know if this was a known issue before filing a bug 
> report.

I've never seen this. It looks fairly harmless, but worth understanding.

Please can you post a full boot log, state what distro is being used in
the domU, and whether tls has been disabled or whether 'echo "hwcap 0
nosegneg" > /etc/ld.so.conf.d/xen' has been executed.

Thanks,
Ian 
 

> Pausing... Pausing... Pausing... Pausing... <3>BUG: soft 
> lockup detected 
> on CPU#0!
> 
> Pid: 1, comm: init
> EIP: 0061:[<c0106b0d>] CPU: 0
> EIP is at delay_tsc+0xd/0x15
> EFLAGS: 00000297 Not tainted (2.6.16-xen #2)
> EAX: 51d9f61d EBX: 0022d869 ECX: 51cc64c4 EDX: 00000101
> ESI: 00000002 EDI: c010eeb0 EBP: bf955008 DS: 007b ES: 007b
> CR0: 8005003b CR2: b7f17e57 CR3: 07903000 CR4: 00000660
> [<c010efb2>] do_fixup_4gb_segment+0x102/0x128
> [<c0110a18>] do_page_fault+0x1c5/0x7dd
> [<c01049d3>] error_code+0x2b/0x30
> Pausing... Continuing...
> 
> 
> Host info..
> 
> xen_major              : 3
> xen_minor              : 0
> xen_extra              : .2-1
> xen_caps               : xen-3.0-x86_32p
> platform_params        : virt_start=0xf5800000
> xen_changeset          : Tue Apr 11 22:14:24 2006 +0100 
> 9612:1f2520faecda
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
> 

Re: kernel BUG look familiar?

[Keir Fraser]

> > This was reported by one of my users and it shows on every boot for 
> > them.  I wanted to know if this was a known issue before filing a bug 
> > report.
> 
> I've never seen this. It looks fairly harmless, but worth understanding.
> 
> Please can you post a full boot log, state what distro is being used in
> the domU, and whether tls has been disabled or whether 'echo "hwcap 0
> nosegneg" > /etc/ld.so.conf.d/xen' has been executed.

There's a 5-second pause the first time negative segment offsets are
used, to provide a clear warning to the user. The softlockup watchdog
is supposed to kick in after 10 seconds, so something here is a bit
confused. It's probably that mdelay() is delaying more than a
second. It's a software delay loop which we really ought to replace
with something better (HYPERVISOR_poll is the obvious candidate).

The issue can be worked around by calling touch_softlockup_watchdog()
in the 5-second delay loop in arch/i386/kernel/fixup.c.

 -- Keir

Re: kernel BUG look familiar?

[Matt Ayres]

Keir Fraser wrote:

> 
> There's a 5-second pause the first time negative segment offsets are
> used, to provide a clear warning to the user. The softlockup watchdog
> is supposed to kick in after 10 seconds, so something here is a bit
> confused. It's probably that mdelay() is delaying more than a
> second. It's a software delay loop which we really ought to replace
> with something better (HYPERVISOR_poll is the obvious candidate).
> 
> The issue can be worked around by calling touch_softlockup_watchdog()
> in the 5-second delay loop in arch/i386/kernel/fixup.c.
> 

Thanks for the quick fix Keir!