From: Robin Gilks <hidden> Date: 2003-06-17 04:06:29
Greetings
With the recent release of 2.4.21 I see many patches have been swept up
and incorporated. No sign of the low latency & preemption patches though
(not really suprising in a stable series) but does anyone have a set of
patches for this new tree for ppc only?
Many thanks.
--
Robin Gilks
Senior Design Engineer Phone: (+64)(3) 357 1569
Tait Electronics Fax : (+64)(3) 359 4632
PO Box 1645 Christchurch Email : robin.gilks@tait.co.nz
New Zealand
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
From: Marius Groeger <hidden> Date: 2003-06-17 05:34:21
On Tue, 17 Jun 2003, Robin Gilks wrote:
With the recent release of 2.4.21 I see many patches have been swept up
and incorporated. No sign of the low latency & preemption patches though
(not really suprising in a stable series) but does anyone have a set of
patches for this new tree for ppc only?
The low latency patch is now part of the kpreempt patch. A kpreempt
patch for 2.4.21-pre1 can be found here:
http://www.kernel.org/pub/linux/kernel/people/rml/preempt-kernel/v2.4/
Some tweaks may be required to apply it on the current development
kernel.
Regards,
Marius
--
Marius Groeger [off-list ref]
Software Engineering
SYSGO Real-Time Solutions AG | Embedded and Real-Time Software
Am Pfaffenstein 14
55270 Klein-Winternheim, Germany
Voice: +49-6136-9948-0 | FAX: +49-6136-9948-10
www.sysgo.de | www.elinos.com | www.osek.de
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
With the recent release of 2.4.21 I see many patches have been swept up
and incorporated. No sign of the low latency & preemption patches though
(not really suprising in a stable series) but does anyone have a set of
patches for this new tree for ppc only?
Yes, of course, lock breaking is also needed. I always use both
kpreempt and lock-break. Wrong assumption on my part.
Anyway, with this combination you don't need the low-latency patches
anymore (according to www.tech9.net/rml/linux):
"With the preemptible kernel, the need for explicit scheduling points, like
in the low-latency patches, are no more. However, since we can not preempt
while locks are held, we can take a similar model as low-latency and
"break" (drop and immediately reacquire) locks to improve system response.
The trick is finding when and where we can safely break the locks (periods
of quiescence) and how to safely recover. The majority of the lock
breaking is in the VM and VFS code. This patch is for users with strong
system response requirements affected by the worst-case latencies caused
by long-held locks."
Regards,
Marius
--
Marius Groeger [off-list ref]
Software Engineering
SYSGO Real-Time Solutions AG | Embedded and Real-Time Software
Am Pfaffenstein 14
55270 Klein-Winternheim, Germany
Voice: +49-6136-9948-0 | FAX: +49-6136-9948-10
www.sysgo.de | www.elinos.com | www.osek.de
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
Yes, of course, lock breaking is also needed. I always use both
kpreempt and lock-break. Wrong assumption on my part.
OK :)
I also use preempt + lock-break.
Preempt patch alone is not good enough for our applications.
Anyway, with this combination you don't need the low-latency patches
anymore (according to www.tech9.net/rml/linux):
<snip>
He is probably right.
Although lock-break patch wasn't updated for some time.
Andrew Morton's low latency patch seems to be more up to date ...
I'm thinking about trying it instead of lock-break or maybe just full patch
set from (http://members.optusnet.com.au/ckolivas/kernel/). It probably
will require some PPC mods though...
Eugene.
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/