Timer ISR/0: Timer went backwards error...

Timer ISR/0: Timer went backwards error...

[Sandesh]

Hi all,
  I am getting lots of Timer ISR/0: Timer went backwards error...
due to which my dom0 is crashing. I tried hpet_force=1 argument for xen
command line even then the problem presists. Other than that i could not
find any solution on the net. 
  Has this problem been resolved??

Iam using AMD Athlon(tm) 64bit processor, installed with FC6 &
xen-3.2.0.

Thanks,
  Sandesh


Please do not print this email unless it is absolutely necessary. 

The information contained in this electronic message and any attachments to this message are intended for the exclusive use of the addressee(s) and may contain proprietary, confidential or privileged information. If you are not the intended recipient, you should not disseminate, distribute or copy this e-mail. Please notify the sender immediately and destroy all copies of this message and any attachments. 

WARNING: Computer viruses can be transmitted via email. The recipient should check this email and any attachments for the presence of viruses. The company accepts no liability for any damage caused by any virus transmitted by this email. 

www.wipro.com

Re: Timer ISR/0: Timer went backwards error...

[Keir Fraser]

hpet_force=1 means nothing to Xen. To see what platform timer is being used,
look for a line 'Platform timer is <foo>' in Xen's boot output. To change
it, specify for example 'clocksource=hpet' as a Xen boot parameter.

Could you also send a few of the 'went backwards' messages, so we can see
how badly time is getting skewed. Perhaps we can work out where the issue
arises from too -- it seems that a significant proprotion of those who have
issues with these messages are seeing them on AMD boxes. I haven't confirmed
the correlation though.

 -- Keir

On 3/7/08 12:44, "Sandesh" <sandesh.ahiremath@wipro.com> wrote:

> Hi all,
>   I am getting lots of Timer ISR/0: Timer went backwards error...
> due to which my dom0 is crashing. I tried hpet_force=1 argument for xen
> command line even then the problem presists. Other than that i could not
> find any solution on the net.
>   Has this problem been resolved??
> 
> Iam using AMD Athlon(tm) 64bit processor, installed with FC6 &
> xen-3.2.0.
> 
> Thanks,
>   Sandesh
> 
> 
> Please do not print this email unless it is absolutely necessary.
> 
> The information contained in this electronic message and any attachments to
> this message are intended for the exclusive use of the addressee(s) and may
> contain proprietary, confidential or privileged information. If you are not
> the intended recipient, you should not disseminate, distribute or copy this
> e-mail. Please notify the sender immediately and destroy all copies of this
> message and any attachments.
> 
> WARNING: Computer viruses can be transmitted via email. The recipient should
> check this email and any attachments for the presence of viruses. The company
> accepts no liability for any damage caused by any virus transmitted by this
> email. 
> 
> www.wipro.com
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel

Re: Timer ISR/0: Timer went backwards error...

[Sandesh]

On Thu, 2008-07-03 at 13:29 +0100, Keir Fraser wrote:
> 
> Could you also send a few of the 'went backwards' messages, so we can
> see
> how badly time is getting skewed. Perhaps we can work out where the
> issue
> arises from too -- it seems that a significant proprotion of those who
> have
> issues with these messages are seeing them on AMD boxes. I haven't
> confirmed
> the correlation though.
here is the log...repeated many times...   

 Timer ISR/0: Time went backwards: delta=-9201298748592621662 delta_cpu=-9201298748582621662 shadow=73000238797 off=-9201298747702859870 processed=73890000000 cpu_processed=73880000000
   0: 73880000000
 Timer ISR/0: Time went backwards: delta=-9201298748575066948 delta_cpu=-9201298748565066948 shadow=73000238797 off=-9201298747685305644 processed=73890000000 cpu_processed=73880000000
   0: 73880000000
 Timer ISR/0: Time went backwards: delta=-9201298748557613457 delta_cpu=-9201298748547613457 shadow=73000238797 off=-9201298747667852175 processed=73890000000 cpu_processed=73880000000
   0: 73880000000
 Timer ISR/0: Time went backwards: delta=-9201298748540109642 delta_cpu=-9201298748530109642 shadow=73000238797 off=-9201298747650348375 processed=73890000000 cpu_processed=73880000000
   0: 73880000000

Thanks,
 Sandesh



Please do not print this email unless it is absolutely necessary. 

The information contained in this electronic message and any attachments to this message are intended for the exclusive use of the addressee(s) and may contain proprietary, confidential or privileged information. If you are not the intended recipient, you should not disseminate, distribute or copy this e-mail. Please notify the sender immediately and destroy all copies of this message and any attachments. 

WARNING: Computer viruses can be transmitted via email. The recipient should check this email and any attachments for the presence of viruses. The company accepts no liability for any damage caused by any virus transmitted by this email. 

www.wipro.com

RE: Timer ISR/0: Timer went backwards error...

[Dan Magenheimer]

Sandesh --

hpet_force was a Xen parameter in Dave Winchell's last hpet
patch that hasn't yet been accepted into the tree.  However
I think it would have only worked on hvm guests.

Xen system time's dependency on tsc is most susceptible on
pre-Barcelona SMP AMD boxes (where tsc isn't synchronized
between processors), though yours seems to have a very
bad skew.  As Keir suggests "clocksource=hpet" as
a Xen command line parameter will hopefully
work for those boxes.

Dan

> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com
> [mailto:xen-devel-bounces@lists.xensource.com]On Behalf Of Keir Fraser
> Sent: Thursday, July 03, 2008 6:30 AM
> To: Sandesh; xen-devel@lists.xensource.com
> Subject: Re: [Xen-devel] Timer ISR/0: Timer went backwards error...
> 
> 
> hpet_force=1 means nothing to Xen. To see what platform timer 
> is being used,
> look for a line 'Platform timer is <foo>' in Xen's boot 
> output. To change
> it, specify for example 'clocksource=hpet' as a Xen boot parameter.
> 
> Could you also send a few of the 'went backwards' messages, 
> so we can see
> how badly time is getting skewed. Perhaps we can work out 
> where the issue
> arises from too -- it seems that a significant proprotion of 
> those who have
> issues with these messages are seeing them on AMD boxes. I 
> haven't confirmed
> the correlation though.
> 
>  -- Keir
> 
> On 3/7/08 12:44, "Sandesh" <sandesh.ahiremath@wipro.com> wrote:
> 
> > Hi all,
> >   I am getting lots of Timer ISR/0: Timer went backwards error...
> > due to which my dom0 is crashing. I tried hpet_force=1 
> argument for xen
> > command line even then the problem presists. Other than 
> that i could not
> > find any solution on the net.
> >   Has this problem been resolved??
> >
> > Iam using AMD Athlon(tm) 64bit processor, installed with FC6 &
> > xen-3.2.0.
> >
> > Thanks,
> >   Sandesh
> >
> >
> > Please do not print this email unless it is absolutely necessary.
> >
> > The information contained in this electronic message and 
> any attachments to
> > this message are intended for the exclusive use of the 
> addressee(s) and may
> > contain proprietary, confidential or privileged 
> information. If you are not
> > the intended recipient, you should not disseminate, 
> distribute or copy this
> > e-mail. Please notify the sender immediately and destroy 
> all copies of this
> > message and any attachments.
> >
> > WARNING: Computer viruses can be transmitted via email. The 
> recipient should
> > check this email and any attachments for the presence of 
> viruses. The company
> > accepts no liability for any damage caused by any virus 
> transmitted by this
> > email.
> >
> > www.wipro.com
> >
> > _______________________________________________
> > Xen-devel mailing list
> > Xen-devel@lists.xensource.com
> > http://lists.xensource.com/xen-devel
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>

RE: Timer ISR/0: Timer went backwards error...

[Sandesh]

>   As Keir suggests "clocksource=hpet" as
> a Xen command line parameter will hopefully
> work for those boxes.
 I tired "clocksource=hpet and clocksource=cyclone", and both of them
failed to initialize respective timers, so by default APIC timer is
getting installed.
However with "clocksource=pit", init_pit seemed to have initialized PIT
timer but it did not solve the problem. Am still getting the Time went
backwards...

To be more specific about the problem, it happens only when i try to
return from the hypervisor kdb. But since many others have/had the same
problem without the use of hypervisor kdb, i dont think its due to the
kdb. I had never got this error before, so suerly kdb must be triggering
something here.

Thanks,
  Sandesh



Please do not print this email unless it is absolutely necessary. 

The information contained in this electronic message and any attachments to this message are intended for the exclusive use of the addressee(s) and may contain proprietary, confidential or privileged information. If you are not the intended recipient, you should not disseminate, distribute or copy this e-mail. Please notify the sender immediately and destroy all copies of this message and any attachments. 

WARNING: Computer viruses can be transmitted via email. The recipient should check this email and any attachments for the presence of viruses. The company accepts no liability for any damage caused by any virus transmitted by this email. 

www.wipro.com

Re: Timer ISR/0: Timer went backwards error...

[Mukesh Rathor]

> To be more specific about the problem, it happens only when i try to
> return from the hypervisor kdb. But since many others have/had the same

That's a known issue in kdb. Like I mentioned in my kdb presentation, I'm still
trying to get the timer stuff right, hopefully, in next release I'll have it
figured out and fixed. 
I am sure that is not the cause of your dom0 crash. Like I had mentioned in our
offline emails, kdb is not properly tested on 32bit hyp, so the crash could well
be something kdb related that needs to be investigated.

Thanks,
Mukesh


Sandesh <sandesh.ahiremath <at> wipro.com> writes:

> 
> 
> >   As Keir suggests "clocksource=hpet" as
> > a Xen command line parameter will hopefully
> > work for those boxes.
>  I tired "clocksource=hpet and clocksource=cyclone", and both of them
> failed to initialize respective timers, so by default APIC timer is
> getting installed.
> However with "clocksource=pit", init_pit seemed to have initialized PIT
> timer but it did not solve the problem. Am still getting the Time went
> backwards...
> 
> To be more specific about the problem, it happens only when i try to
> return from the hypervisor kdb. But since many others have/had the same
> problem without the use of hypervisor kdb, i dont think its due to the
> kdb. I had never got this error before, so suerly kdb must be triggering
> something here.
> 
> Thanks,
>   Sandesh
> 
> Please do not print this email unless it is absolutely necessary. 
> 
> The information contained in this electronic message and any attachments to
this message are intended for
> the exclusive use of the addressee(s) and may contain proprietary,
confidential or privileged
> information. If you are not the intended recipient, you should not
disseminate, distribute or copy this
> e-mail. Please notify the sender immediately and destroy all copies of this
message and any attachments. 
> 
> WARNING: Computer viruses can be transmitted via email. The recipient should
check this email and any
> attachments for the presence of viruses. The company accepts no liability for
any damage caused by any
> virus transmitted by this email. 
> 
> www.wipro.com
> 

Re: Timer ISR/0: Timer went backwards error...

[Keir Fraser]

On 3/7/08 17:43, "Sandesh" <sandesh.ahiremath@wipro.com> wrote:

> To be more specific about the problem, it happens only when i try to
> return from the hypervisor kdb. But since many others have/had the same
> problem without the use of hypervisor kdb, i dont think its due to the
> kdb. I had never got this error before, so suerly kdb must be triggering
> something here.

Quite a relevant bit of information I'd say. It's certainly related, since
time structures in Xen will not be updated and kept synchronised while kdb
has the system locked down. Indeed the time deltas you see are huge, so the
timeout in kdb has sent time handling haywire.

 -- Keir