Hi Wolfgang!

On Mon, Jul 26, 2004 at 07:16:03PM +0200, Wolfgang Denk wrote:

In message [ref] you wrote:

quoted

Deployment if you're already almost ready to ship is sane enough, I
suppose -- but to actually put more effort into 2.4 wouldn't make much
sense.

It depends on your requirements.

If you need a kernel that is
	(1) stable and working [2.6 is still pretty much useless for  example
	    on  8xx  systems],
	(2) small [2.6 is usually >= 20% bigger than 2.4],
and/or

Even with CONFIG_EMBEDDED or Matt Mackall's -tiny tree?

	(3) fast [2.6 is usually >= 10% slower than 2.4],
then 2.4 may be the better choice.

Where are you you seeing this slowdown?

v2.6 is faster in almost all workloads (it often consumes less memory, VM/FS/core API's
are faster, etc).

rkload, and we are talking about embedded setups/workloads here.

So, I'm sure we should go hunt the slowdowns... At least make the kernel development
community known about them. For sure we should make efforts to fix any performance
regressions in v2.6.

quoted

Because nobody's really that interested in it. For what it's worth, I've
abandoned all pretence of continuing to support 2.4 in the MTD/JFFS2 CVS

David, you can do what you want. But it is not up to  you  to  decide
what  other people are doing. I declare that there ARE people who are
interested in 2.4 kernel - both for maintenance and development work.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/