Re: Some functions are not inlined by gcc 3.2, resulting code is ugly
Denis Vlasenko (vda@port.imtp.ilyichevsk.odessa.ua)
Sun, 3 Nov 2002 22:20:25 -0200
On 3 November 2002 14:21, Alan Cox wrote:
> On Sun, 2002-11-03 at 15:37, Jakub Jelinek wrote:
> > On Sun, Nov 03, 2002 at 04:14:26PM -0200, Denis Vlasenko wrote:
> > > Here is the cure: force_inline will guarantee inlining.
> > >
> > > To use _only_ with functions which meant to be almost
> > > optimized away to nothing but are large and gcc might decide
> > > they are _too_ large for inlining.
> >
> > Well, you can as well bump -finline-limit, like
> > -finline-limit=2000. The default is too low for kernel code (and
> > glibc too).
>
> I would venture the reverse interpretation for modern processors, the
> kernel inlines far far too much
I agree that there are far too many large inlines. But.
__constant_c_and_count_memset *has to* be inlined.
There is large switch statement which meant to be optimized out.
It does optimize out *only if* count is compile-time constant.
--
vda
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/