Native preemption is the culprit [was Re: today's CURRENT lockups]

[Jon Noack]

On 07/09/04 13:31, Kenneth Culver wrote:
> Quoting Jon Noack <noackjr at alumni.rice.edu>:
>> On 07/09/04 13:06, Steve Kargl wrote:
>>> On Fri, Jul 09, 2004 at 12:53:53PM -0500, Jon Noack wrote:
>>>> On 07/09/04 12:25, Steve Kargl wrote:
>>>>> On Fri, Jul 09, 2004 at 12:19:09PM -0500, Jon Noack wrote:
>>>>>> On 07/09/04 12:15, John Baldwin wrote:
>>>>>>> My test machine is not a true SMP machine either, just
>>>>>>> HTT. It has been running a -j 256 worldloop overnight
>>>>>>> with no problems, so I committed a slightly modified
>>>>>>> version of the patch yesterday.
>>>>>> 
>>>>>> Did you test with a UP kernel?  After your latest commit I
>>>>>> have been experiencing regular hard locks on my pre-HTT P4
>>>>>> machine. Backing out rev. 1.114 of sched_ule.c fixes it.
>>>>>> See my previous message (Re: FreeBSD keeps hanging......):
>>>>>>  http://docs.freebsd.org/cgi/mid.cgi?40EECC49.3070501
>>>>> 
>>>>> I've tested it on a UP kernel (HTT enabled, ACPI disabled,
>>>>> APIC enabled, ULE).   Appears to work fine.
>>>> 
>>>> Perhaps it's just extraneous information, but if it's really a
>>>> UP kernel (as in, no 'options SMP') then whether you have HTT
>>>> enabled doesn't matter at all -- the kernel won't use it.
>>> 
>>> Yes, I know. :-)  I'm just reporting what dmesg tells me about
>>> the CPU.
>>> 
>>> CPU: Mobile Intel(R) Pentium(R) 4 - M CPU 2.00GHz (1994.12-MHz 
>>> 686-class CPU) Origin = "GenuineIntel"  Id = 0xf27  Stepping = 7
>>>  
>>> Features=0xbfebf9ff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CLFLUSH,DTS,ACPI,MMX,FXSR,SSE,SSE2,SS,HTT,TM,PBE>
>> 
>> Mine also lists HTT, but doesn't actually support it: CPU: Intel(R)
>> Pentium(R) 4 CPU 2.53GHz (2539.10-MHz 686-class CPU) Origin =
>> "GenuineIntel"  Id = 0xf24  Stepping = 4
>> 
>> Features=0x3febfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CLFLUSH,DTS,ACPI,MMX,FXSR,SSE,SSE2,SS,HTT,TM>
>> 
>> I remember a discussion about this a long time ago, but don't
>> remember the details.  Regardless, the flag doesn't actually mean
>> the processor supports HTT.
> 
> I was under the impression that all P4's support HTT, but a lot of 
> bioses don't allow users to enable it.

Nope, they didn't support HTT until the 3.06 GHz (533 MHz bus).  Models 
introduced after that support HTT; if they use an 800 MHz bus then they 
do support HTT, regardless of speed.  That's why you can get a 2.4 GHz 
(800 MHz bus) w/ HTT but an older 2.8 GHz (533 MHz bus) won't support it.

In any case, the flag just means we can check if the processor supports HTT:
http://lists.freebsd.org/pipermail/freebsd-current/2004-January/017920.html

Jon