On Saturday 10 February 2001  4:55 am, Jason R Thorpe wrote:
> On Fri, Feb 09, 2001 at 11:46:20PM +0000, Chris Gilbert wrote:
>  > Oh updated make configure time is, which is half that it is when UVMHIST
>  > is enabled (doh):
>  > 	106.59 real        20.35 user        62.93 sys
>  > I'm going to try it with the new bootloader in a mo.
>  >
>  > Anyway either give push the process creation and destruction.
>
> So, this will make the umapping of the entire address space happen
> while the process is not curproc.  In fact, the unmap in the non-curproc
> path was already happening, but there was a redundant unmap in exit1().
>
> It made things ever so slightly faster on my 700MHz P-III -- I wasn't
> expecting to see much improvement on that system :-)  Anyway, please
> try it on your ARM systems and tell me what improvement you see.
>
> (I'm considering just committing it anyhow, since the unmap in exit1()
> is clearly redundant...)

Certainly faster, the difference is probably even bigger for the slower 
risc-pc's.

The fork/exit loop comes in with:
8.43 real         0.00 user         2.07 sys
compared with previous to this patch
9.17 real         0.00 user         2.06 sys

The gmake configure test comes in at:
99.07 real        20.96 user        54.08 sys

Which is about 10% off the sys time.

Thanks,
Chris