Lost Ticks

Lost Ticks

[Marc Perkel]

Got a dual core Athlon 64 X2 on an Asus board using NVidia chipset and 
getting lost ticks. The software clock of course is totally messed up. 
I've scanned google for a solution and see others complaining about bad 
code in the SMM BIOS. I have the latest bios and whatever they need to 
fix - isn't.

So - what do I do to make it work?

Yes - I compiled the kernel 2.6.13.1 and used a speed of 100 - the 
lowest setting - and that did help some. But - the problem needs to go away.

If I compile a kermel without any of the power management at all - will 
that fix the problem? I need a work around.

Thanks in advance.

-- 
Marc Perkel - marc@perkel.com

Spam Filter: http://www.junkemailfilter.com
    My Blog: http://marc.perkel.com

Re: Lost Ticks

[Frank van Maarseveen]

On Sun, Sep 18, 2005 at 09:23:40PM -0700, Marc Perkel wrote:
> Got a dual core Athlon 64 X2 on an Asus board using NVidia chipset and 
> getting lost ticks. The software clock of course is totally messed up. 
> I've scanned google for a solution and see others complaining about bad 
> code in the SMM BIOS. I have the latest bios and whatever they need to 
> fix - isn't.
> 
> So - what do I do to make it work?

See http://bugzilla.kernel.org/show_bug.cgi?id=5105

On the kernel command-line:

x86_64:	try "notsc"
i386:	try "clock=pit"

"nosmp" works but isn't fun.

-- 
Frank

Re: Lost Ticks

[Scott Lampert]

I have the exact same problem on a ASUS A8N-SLI Premium board and 
Athlon64 4800+ X2 with every BIOS up to 1008-01.  Running with notsc is 
the only way to get it to work.

As an aside BIOS version 1008-003 is available for this board however 
this one seems to be WAY worse as the board won't even boot.  It gets 
panics before the boot messages unless you boot with noapic and after 
that it gets checksum errors on the RSDP.  I'm afraid to see what the 
next official BIOS version does. :/
    -Scott

Frank van Maarseveen wrote:

>On Sun, Sep 18, 2005 at 09:23:40PM -0700, Marc Perkel wrote:
>  
>
>>Got a dual core Athlon 64 X2 on an Asus board using NVidia chipset and 
>>getting lost ticks. The software clock of course is totally messed up. 
>>I've scanned google for a solution and see others complaining about bad 
>>code in the SMM BIOS. I have the latest bios and whatever they need to 
>>fix - isn't.
>>
>>So - what do I do to make it work?
>>    
>>
>
>See http://bugzilla.kernel.org/show_bug.cgi?id=5105
>
>On the kernel command-line:
>
>x86_64:	try "notsc"
>i386:	try "clock=pit"
>
>"nosmp" works but isn't fun.
>
>  
>

Re: Lost Ticks

[Marc Perkel]

Yeah - there's a patch for that here:

http://bugzilla.kernel.org/show_bug.cgi?id=5105

I haven't tried it yet but will later when I go to the data center. You 
might want to try it and let me know if it actually fixed the problem.

diff --git a/arch/x86_64/kernel/time.c b/arch/x86_64/kernel/time.c
--- a/arch/x86_64/kernel/time.c
+++ b/arch/x86_64/kernel/time.c
@@ -959,9 +959,6 @@ static __init int unsynchronized_tsc(voi
  	   are handled in the OEM check above. */
  	if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL)
  		return 0;
- 	/* All in a single socket - should be synchronized */
- 	if (cpus_weight(cpu_core_map[0]) == num_online_cpus())
- 		return 0;
 #endif
  	/* Assume multi socket systems are not synchronized */
  	return num_online_cpus() > 1;


Mizery loves company. I'm glad I'm not the only one with this problem.


Scott Lampert wrote:

> I have the exact same problem on a ASUS A8N-SLI Premium board and 
> Athlon64 4800+ X2 with every BIOS up to 1008-01.  Running with notsc 
> is the only way to get it to work.
>
> As an aside BIOS version 1008-003 is available for this board however 
> this one seems to be WAY worse as the board won't even boot.  It gets 
> panics before the boot messages unless you boot with noapic and after 
> that it gets checksum errors on the RSDP.  I'm afraid to see what the 
> next official BIOS version does. :/
>    -Scott
>
> Frank van Maarseveen wrote:
>
>> On Sun, Sep 18, 2005 at 09:23:40PM -0700, Marc Perkel wrote:
>>  
>>
>>> Got a dual core Athlon 64 X2 on an Asus board using NVidia chipset 
>>> and getting lost ticks. The software clock of course is totally 
>>> messed up. I've scanned google for a solution and see others 
>>> complaining about bad code in the SMM BIOS. I have the latest bios 
>>> and whatever they need to fix - isn't.
>>>
>>> So - what do I do to make it work?
>>>   
>>
>>
>> See http://bugzilla.kernel.org/show_bug.cgi?id=5105
>>
>> On the kernel command-line:
>>
>> x86_64:    try "notsc"
>> i386:    try "clock=pit"
>>
>> "nosmp" works but isn't fun.
>>
>>  
>>
> -
> To unsubscribe from this list: send the line "unsubscribe 
> linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/


-- 
Marc Perkel - marc@perkel.com

Spam Filter: http://www.junkemailfilter.com
    My Blog: http://marc.perkel.com

Re: Lost Ticks

[Scott Lampert]

Yeah I tried that patch.  It simply emulates adding notsc to the command 
line.  However there was a thread on LKML between Andi Kleen and the 
provider of this patch and Andi indicated he wasn't convinced that this 
fixes the bug, but simply masks the real bug.  For now I'm sticking with 
the notsc option until they figure out more details.
    -Scott

Marc Perkel wrote:

> Yeah - there's a patch for that here:
>
> http://bugzilla.kernel.org/show_bug.cgi?id=5105
>
> I haven't tried it yet but will later when I go to the data center. 
> You might want to try it and let me know if it actually fixed the 
> problem.
>
> diff --git a/arch/x86_64/kernel/time.c b/arch/x86_64/kernel/time.c
> --- a/arch/x86_64/kernel/time.c
> +++ b/arch/x86_64/kernel/time.c
> @@ -959,9 +959,6 @@ static __init int unsynchronized_tsc(voi
>         are handled in the OEM check above. */
>      if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL)
>          return 0;
> -     /* All in a single socket - should be synchronized */
> -     if (cpus_weight(cpu_core_map[0]) == num_online_cpus())
> -         return 0;
> #endif
>      /* Assume multi socket systems are not synchronized */
>      return num_online_cpus() > 1;
>
>
> Mizery loves company. I'm glad I'm not the only one with this problem.
>
>
> Scott Lampert wrote:
>
>> I have the exact same problem on a ASUS A8N-SLI Premium board and 
>> Athlon64 4800+ X2 with every BIOS up to 1008-01.  Running with notsc 
>> is the only way to get it to work.
>>
>> As an aside BIOS version 1008-003 is available for this board however 
>> this one seems to be WAY worse as the board won't even boot.  It gets 
>> panics before the boot messages unless you boot with noapic and after 
>> that it gets checksum errors on the RSDP.  I'm afraid to see what the 
>> next official BIOS version does. :/
>>    -Scott
>>
>> Frank van Maarseveen wrote:
>>
>>> On Sun, Sep 18, 2005 at 09:23:40PM -0700, Marc Perkel wrote:
>>>  
>>>
>>>> Got a dual core Athlon 64 X2 on an Asus board using NVidia chipset 
>>>> and getting lost ticks. The software clock of course is totally 
>>>> messed up. I've scanned google for a solution and see others 
>>>> complaining about bad code in the SMM BIOS. I have the latest bios 
>>>> and whatever they need to fix - isn't.
>>>>
>>>> So - what do I do to make it work?
>>>>   
>>>
>>>
>>>
>>> See http://bugzilla.kernel.org/show_bug.cgi?id=5105
>>>
>>> On the kernel command-line:
>>>
>>> x86_64:    try "notsc"
>>> i386:    try "clock=pit"
>>>
>>> "nosmp" works but isn't fun.
>>>
>>>  
>>>
>> -
>> To unsubscribe from this list: send the line "unsubscribe 
>> linux-kernel" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at  http://www.tux.org/lkml/
>
>
>