Re: [RHEL5 PATCH 1/4] Provide fallback full 64-bit divide/modulus ops for gcc

[David Howells]

Andi Kleen <ak@xxxxxxx> wrote:

> At least Linus' traditional argument against this is that it's better
> to open code these (do_div) so that it's clear to the coder that they
> are really costly.

do_div() is not a full replacement for __udivdi3(), __umoddi3() or
__udivmoddi4(), though I suspect we don't need divisor >= 2^32 anywhere atm.

There are places where the compiler emits these that aren't entirely obvious,
one of which IIRC is in ext2 inode allocation.

David
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/