Re: [PATCH] - x86_64-add-ioapic-nmi-support-fix-3

[Randy Dunlap <randy.dunlap@oracle.com>]

John Keller wrote:
>> [adding Andi Kleen]
>>
>> John Keller wrote:
>>> Place all the IOACPI NMI support code under CONFIG_ACPI
>>> to clear up build errors with certain configs.
>>>
>>> Signed-off-by: John Keller <jpk@sgi.com>
>>> ---
>> Is there some architectural reason that IO APIC NMI support should
>> require ACPI?
> 
> OK, I guess standing alone this description was a bit misleading.
> The code referred to here is the new code supporting the ACPI NMI SRC
> structure that can be specified in the MADT. Without ACPI support this
> code is not relevant. All the code touched by this patch was introduced
> by eariler versions of this patchset.
> 
> This patch makes no changes to the workings of nmi_watchdog.

OK, thanks for the clarification.

> John
> 
> 
>> Is this a new requirement?  It seems like a step backwards to me.
>>
>>
>> Documentation/nmi_watchdog.txt doesn't say anything about ACPI being
>> needed.  It does say:
>>
>> "For x86-64, the needed APIC is always compiled in, and the NMI watchdog is
>> always enabled with I/O-APIC mode (nmi_watchdog=1). Currently, local APIC
>> mode (nmi_watchdog=2) does not work on x86-64.
>>
>> Using local APIC (nmi_watchdog=2) needs the first performance register, so
>> you can't use it for other purposes (such as high precision performance
>> profiling.) However, at least oprofile and the perfctr driver disable the
>> local APIC NMI watchdog automatically."


-- 
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***