Re: [PATCH] CHECK_IRQ_PER_CPU() to avoid dead code in __do_IRQ()

[Zachary Amsden <zach@vmware.com>]

Karsten Wiese wrote:

>Am Montag, 8. August 2005 13:19 schrieb Alexander Nyberg:
>  
>
>>There are many places where one could replace run-time tests with 
>>#ifdef's but it makes reading more difficult (and in longer terms
>>maintainence). Have you benchmarked any workload that benefits 
>>from this?
>>    
>>
>
>Performance gain is small, but measurable: 0,02%
>Tested on an Atlon64 running at 1000MHz.
>I took this value from 9 runs (each with/without the patch) of 
>	$ time lame x.wav
>where each took about 1 minute.
>3000 Interrupts/s were generated at the time by running
>	$ jackd -R -dalsa -p64 -n2
>
>0,02% really isn't that much....but Athlon64 is better than P4
>with branch predictions, I think.
>
>Erm... ok, I won't insist on having this patch applied ;-) 
>
>   Karsten
>  
>

Removing dead code is always good - 0.02% is small, but if 100 kernel 
developers all did the same, that adds up to 2% rather quickly, and that 
is nothing to sneeze at.  I like your patch, but you should add some 
comments for maintainability about what CHECK_IRQ_PER_CPU does - see 
include/asm-generic/pgtable.h for similar styling.  If also probably 
doesn't hurt to leave IRQ_PER_CPU defined even when 
ARCH_HAS_CHECK_IRQ_PER_CPU is not, since it looks cleaner and prevents 
future collisions with bits defined inside of an #ifdef.

Zach