Hmmm...I don't think that that is a fair assesment of the linux
kernel...From the standpoint of the desktop (where esentually the
kernel is really targeted) -O2 makes much more sence...therefore hand
optimaisation<sp> that are only valid with -O2, have been added on
numerous occations...Why? Moores Law...Memory is cheap...disk space is
even cheaper...Speed is more critical then size...
Working on Embedded Linux myself, I understand the frustration...I
often find it disconserting that I can configure a kernel with almost
nothing enabled and it still come out 400k...
Jeremy
MontaVista Software
On Tue, Jun 12, 2001 at 09:16:38PM +0100, Philip Blundell wrote:
> >Note that your kernel will probably break when you compile it with -Os,
> >kernels have to be compiled with -O2.
>
> Maybe, but that's due to brain damage in the kernel. Compiling everything
> with -Os is a worthy goal, and if the kernel can't cope with that it ought to
> be fixed. (It already builds itself with -Os on 26-bit platforms,
> incidentally.)
>
> p.
>
>
Received on Tue Jun 12 13:46:39 2001
This archive was generated by hypermail 2.1.8 : Tue May 04 2004 - 09:44:01 EDT