Re: [PATCH] powerpc/mm: Add trace point for tracking hash pte fault

["Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>]

Anton Blanchard <anton@samba.org> writes:

> Hi Aneesh,
>
>> yes. We do use jump label. I also verified that looking at .s
>> 
>> #APP
>>  # 23 "./arch/powerpc/include/asm/jump_label.h" 1
>>         1:
>>         nop
>>         .pushsection __jump_table,  "aw"
>>         .llong 1b, .L201, __tracepoint_hash_fault+8      #,
>>         .popsection
>> 
>>  # 0 "" 2
>
> So we insert a single nop, and the slow path is in another section. I'd
> be surprised if we could measure this, unless the nop causes a branch
> target alignment issue the slow path caused some hot path icache layout
> issues.
>
>> Without patch
>> sys: 0m11.2425
>> 
>> With patch:
>> sys: 0m11.3258
>>
>> ie, a -0.7% impact 
>> 
>> If that impact is high we could possibly put that tracepoint within
>> #ifdef CONFIG_DEBUG_VM ?
>
> Did the real time change? I'd be careful about comparing based on
> system time.


Yes it did. 
11.8769 with patch
11.7924 without patch

I will look at the perf stat data difference between the both runs and update.

-aneesh