Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

[PATCH] small sha512 cleanup

0 views
Skip to first unread message

Denis Vlasenko

unread,
Oct 1, 2004, 3:37:25 PM10/1/04
to jmo...@redhat.com, da...@davemloft.net, linux-...@vger.kernel.org
Looks like open-coded be_to_cpu.
GCC produces rather poor code for this.
be_to_cpu produces asm()s which are ~4 times shorter.

Compile-tested only.

I am not sure whether input can be 64bit-unaligned.
If it indeed can be, replace:

((u64*)(input))[I] -> get_unaligned( ((u64*)(input))+I )
--
vda

sha512.c.diff
0 new messages