[cgl_discussion] About Preemptive and O1 scheduler patch

george anzinger george at mvista.com
Tue Sep 24 10:53:59 PDT 2002

First, I am confused.  Is this the right list for this
discussion?  Or should it be on the
<developer at carrierlinux.org> list?  At first, I thought that
this list would replace that one, but this list is now open
to the world and seems to be concentrating on higher level
issues, while the other seems to still be involved with the
day to day CVS issues.  Any one know which we should use and
for what?

"Zou, Yixiong" wrote:
> The CVS repository for the preempt-kernel
> http://cvs.developer.osdl.org/viewcvs/viewcvs.cgi/kernel/linux-2.4.18/patche
> s/preempt-kernel/ contains 6 files. Does anybody know which one is really
> applied to the CGLE kernel? Is there an easy way to reverse the
> preempt-kernel patch in CGLE kernel?

It is a mess.  Robert did not understand that the complete
virgin kernel patch was to be kept here and submitted fix
patches along the way which were to the CVS tree, not the
virgin kernel.  I continued that with the latest fixes.  We
are trying to get a "proper" virgin kernel patch together as
we speak, uh, write.

As to the question of reversing, none of these patches
should be of much help.  They were never intended to be as
the target was supposed to be the virgin kernel.  Your best
bet is the tags, but we were not asked to use (in the old
TLT tree) a "fixed" pre commitment tag, so this may not work
very well.  In this regard, you might try finding the post
tags, extracting the date/ time and using a pre tag of that
date less 30 mins or so.  (We are remote and these things
take time :)

Also note that the lock-break patch and the preempt-stats
patch both depend on the preempt patch so they would need to
be reversed as well.

This, of course, brings up the question of why.  The preempt
code is designed to disappear from the code if you turn off
the CONFIG_PREEMPT option.  What more do you want/need?

George Anzinger   george at mvista.com
Preemption patch:

More information about the cgl_discussion mailing list