Re: corrupt modversions.h built in 2.2.13

Ron Flory (ron.flory@adtran.com)
Fri, 22 Oct 1999 08:54:27 -0500


Michael Elizabeth Chastain wrote:
>
> David Woodhouse writes:
>
> > Non-recursive make, anyone?
>
> I can't get my mind around that vision (and I've tried). I am simply
> too familiar with the "one make per directory" world.

<rant>

There's no polite way to say it, but this is how msdos/windoze people
have been conditioned. We don't have the same built-in restrictions.

Its not so much the recursive make, as the ability to spawn concurrent
compilations that are independent [as stated in the makefile rules] of
one another. The modversions.h builder apparently circumvents 'makes'
built-in control mechanisms.

The use of 'make -j<n>' is -very- common in the *nix world, where one
compiler instance can utilize the CPU(s) while another is blocking for
disk I/O, even if you have just one CPU (remember, I'm looking at this
mostly from an SMP approach). There are other performance benefits as
well. Such concurrent compilations result in a significant reduction in
compile time for sizable projects such as our kernel.

Under 2.0.3x, 'make -j25' really hauled ass (just over a minute).
2.2.x seems to have restructured the makefiles so the kernel build
process leaves most of my CPU's idle, and seems to take forever. When
compiling, any time the 'percent-idle' is not 0, is wasted time.

thanks for listening...

</rant>

bye-

ron

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