This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: epoll_pwait broken?
On 1/23/07, Davin McCall <davmac@davmac.org> wrote:
Ugh, I'm a little braindead today. Ok, you might be able to optimize the constant load in that you can keep the same constant value on the stack for successive calls. However, in most cases, you have still turned a single "load constant into register" operation (on the glibc side) into "push constant onto stack" (on the user side) together with "load register from the stack" (on the glibc side, in epoll_pwait).
Yes, you're right for IA32. On other architectures with enough
function call argument registers this is faster. Unfortunately at six
arguments you are getting to the limit of most architectures.
c.