Re: Performance of function calls of SH?(Fw: [RFC][selinux] Tuning SELinux permission check

[Paul M. <le...@li...>]

On Mon, Sep 10, 2007 at 01:35:14PM +0900, Yuichi Nakamura wrote:
> > 3) Result for SH
> >                 Base   SELinux  Overhead(%)
> > Simple read     2.6781 3.6538  36.43(before 141.5)
> > Simple write    2.0781 3.321   59.80(before 155.9)
> In other archs, the effect of inlining was not so big,
> but in SH it was big.
> 
> I could not find the reason.
> Does anyone know why?
> 
> Are function calls on SH heavy?
> Or gcc optimization does not work well?(I used gcc 3.4.5)
> 
I suspect you're mostly being bitten by your compiler version, it would
be nice to see the numbers without the manual inlining with a more
up-to-date compiler (4.1.x, 4.2.x, etc.).

> 1) Result for x86(Pentium 4 2.6Ghz), kernel 2.6.22
> 2) Result for ARM(Intel XScale (PXA270 416MHz)), kernel 2.6.17
> 3) Result for SH(SH4, SH7751R), kernel 2.6.22

What versions of the compiler did you have for x86 and ARM?

It's probably worth trying to do the manual inling for those old
compilers (especially as there don't seem to be any regressions for any
of the other platforms -- though you didn't list before and after code
size), though I would be surprised if it's still problematic in current
compilers. If you don't notice an improvement in more recent compilers,
you may want to roll together a simple testcase and shove it in to the
GCC bugzilla.