Re: Referring what kernel-images to build to the technical committee?

[David Schleef <ds@schleef.org>]

On Thu, Apr 26, 2001 at 10:13:01PM +0200, Russell Coker wrote:
> So a 386 compiled kernel can still support MMX, 3DNow! and MTRR?  In that 
> case we only need a 386 kernel, but it might be nice to have a PentiumMMX 
> compiled kernel as well (that should give better performance on all brands of 
> CPU that are better than 486).

Except that the PentiumMMX kernel won't run on a Pentium -- it
uses MMX instructions for memcpy().  (I now ramble into things
I only slightly remember, so please check before believing...)
I think the general consensus is that the original Pentium
instruction ordering is best all-around, whereas PentiumPro
is the worst.

> I am thinking of the various IDE options for dealing with broken drives and 
> controllers, PCI Quirks, etc.  These things will break some machines when set 
> one way and break other machines on the other setting.

I haven't heard of many problems like this recently.  But I'm sure
there will be such bugs during the lifetime of 2.4.

> So we ship half the kernel as binary and compile the other half after 
> installation?  Sounds terrible.  Why not just compile custom kernels for 
> every user?

(Yes, that doesn sound terrible.)  No, I meant to ship and install
one standard complete kernel.  If the user wants to run the
automatic kernel optimisation script and compile a new kernel,
cool.  But the idea is to make it as simple as "Do you want an
optimized kernel?"

> It wouldn't be that difficult to write a script that asks a dozen easy 
> questions, checks the /proc data, and then compiles an optimised kernel.

I tend to think that asking fewer questions is better -- a script
will better know how to optimise the kernel for someone that is
unprepared.  And the people that want more configuration options
should be compiling their own kernels anyway.




dave...