> Just to see what access_ok() was responsible for (in terms of size)
> I built two identical kernels with access_ok() defined to zero in one
> of them. The sizes were as follows:
>
> ------------------------------------------------------------------------
> Memory: 6884k/8192k available (488k kernel code, 384k reserved, 436k data)
>
> Memory: 6800k/8192k available (564k kernel code, 384k reserved, 444k data)
> ------------------------------------------------------------------------
>
> The config was for math-emu, wd80x3, NFS, hd, ext2, proc and serial.
> The resulting kernel wouldn't get past the execve of init, but at least
> I got the above Mem line out of it. :)
#define it 1 and you get past that line too ;)
would be nice to see what speed difference it makes. I have the feeling
that P5s see lots of pipeline stalls in access_ok().
-- mingo