Re: [RFC,PATCH] optimize fixed size kmalloc calls

[Andrew Morton]

Brian Gerst <[e-mail]> wrote:
>
> > What do you think about the attached patch? It's not pretty, but it 
> > should work will all gcc versions. Any other ideas?
> > Just FYI: The function that forced me to use use switch/case instead of 
> > if is interrupts_open in fs/proc/proc_misc.c.
> 
> How about this?  GCC 3.2.2 is able to optimize it away properly from the 
> tests that I've run.

It doesn't compile with older gcc's - see the `extern' decl in the middle
of the function.

The compiler should have been given a way of disabling this.  It'd going to
bite again and again.

I'll fix it up.

void *kmalloc(size_t size, int flags)
{
        if (__builtin_constant_p(size)) {
                int i = 0;

# 1 "include/linux/kmalloc_sizes.h" 1

        if (size <= 32) goto found; else i++;

        if (size <= 64) goto found; else i++;

        if (size <= 128) goto found; else i++;

        if (size <= 256) goto found; else i++;
        if (size <= 512) goto found; else i++;
        if (size <= 1024) goto found; else i++;
        if (size <= 2048) goto found; else i++;
        if (size <= 4096) goto found; else i++;
        if (size <= 8192) goto found; else i++;
        if (size <= 16384) goto found; else i++;
        if (size <= 32768) goto found; else i++;
        if (size <= 65536) goto found; else i++;
        if (size <= 131072) goto found; else i++;
# 84 "include/linux/slab.h" 2

                extern void __you_cannot_kmalloc_that_much(void);
                __you_cannot_kmalloc_that_much();
found:
                return kmem_cache_alloc((flags & 0x01) ?
                        malloc_sizes[i].cs_dmacachep :
                        malloc_sizes[i].cs_cachep, flags);
        }
        return __kmalloc(size, flags);
}