Updated preempt-kernel patches
Updated preempt-kernel patches for 2.4.10, 2.4.12, 2.4.13-pre1, and (phew!) 2.4.12-ac1 are available at:
Note the patches for 2.4.10 are updated over previously releases.
These patches allow create a preemptible kernel -- increasing system response time. Everyone is encouraged to give it a whirl and report back.
Major changes include code to prevent CPU changes on preempt (should prevent a class of race conditions), an SMP compile fix, and various cleanups.
Full change log:
20011011
- remove pgalloc.h preemption disable statements -- not needed with the new CPU switch prevention
- fix compile on SMP in some configurations (ac tree only)
20011005
- rearrange sched.c so we can patch cleanly against rtsched (or vice versa)
20011004
- prevent the case of preemption causing a CPU switch by locking tasks to the current CPU in preempt_schedule
- revert to _raw_spin_xxx from _spin_xxx since PPC uses that convention.
20011003
- push pcmcia-cs tools fix into their next release
20011001
- fix spin_lock_prefetch optimization causing (harmless) compile warnings
Robert Love
Quoted 3 lines · Robert Love
These patches allow create a preemptible kernel -- increasing system response time. Everyone is encouraged to give it a whirl and report back.
If you are lucky, maybe it will _decrease_ system response time, too. Whoops.
Robert Love
Quoted 1 line · Robert Love
- fix compile on SMP in some configurations (ac tree only)
Looks like I forgot to merge that one. Fix follows below (its needed by some ac-tree users who also compile SMP).
include/asm-i386/spinlock.h +1 -1
@@ -97,7 +97,7 @@ :"=q" (oldval), "=m" (lock->lock) \ :"0" (1) : "memory" -static inline void spin_unlock(spinlock_t *lock) +static inline void _raw_spin_unlock(spinlock_t *lock) {char oldval;#if SPINLOCK_DEBUG
Robert Love
Quoted 6 lines · Robert Love
Quoted 1 line · Robert Love
- fix compile on SMP in some configurations (ac tree only)
Looks like I forgot to merge that one. Fix follows below (its needed by some ac-tree users who also compile SMP).
Again? ;)
Any idea how it would work on any smp systems using the -ac tree?
Quoted 9 lines · Mike Fedyk
Quoted 6 lines · Robert Love
Quoted 1 line · Robert Love
- fix compile on SMP in some configurations (ac tree only)
Looks like I forgot to merge that one. Fix follows below (its needed by some ac-tree users who also compile SMP).
Again? ;)
Yes, again :)
The 2.4.12-ac1 patch at http://tech9.net/rml/linux has been updated so you don't need the previous patch.
Quoted 1 line
Any idea how it would work on any smp systems using the -ac tree?
It should run fine, give it a whirl.
Robert Love
Quoted 5 lines · Robert Love
Quoted 4 lines · Mike Fedyk
Quoted 3 lines · Robert Love
Quoted 1 line · Robert Love
- fix compile on SMP in some configurations (ac tree only)
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Looks like I forgot to merge that one. Fix follows below (its needed by some ac-tree users who also compile SMP).
Again? ;)
Yes, again :)
;)
Quoted 7 lines
The 2.4.12-ac1 patch at http://tech9.net/rml/linux has been updated so you don't need the previous patch.
Quoted 1 line
Any idea how it would work on any smp systems using the -ac tree?
It should run fine, give it a whirl.
Don't worry, I'm just bugging... ;)
SMP patch is working great:
Now : 1 day(s), 23:14:50 running Linux 2.4.10-ac11+smp+preempt+vm_hogstop One : 6 day(s), 01:51:56 running Linux 2.4.9-ac10_ext3_0.9.9_dir-speed_interactivity, ended Wed Sep 26 14:34:33 2001 Two : 4 day(s), 03:37:47 running Linux 2.4.9-ac16-vfree, ended Mon Oct 1 18:51:31 2001
Three: 3 day(s), 00:59:12 running Linux 2.4.10-ac4, ended Mon Oct 8 19:53:54 2001
I reset ud after switching to 2.4-ac from 2.2...
Quoted 2 lines
Robert Love
Mike
Great.
Let me know if you test 2.4.12-ac1 and how that goes.
Robert Love