PERFORCE change 67916 for review
John Baldwin
jhb at FreeBSD.org
Thu Dec 30 11:22:51 PST 2004
http://perforce.freebsd.org/chv.cgi?CH=67916
Change 67916 by jhb at jhb_slimer on 2004/12/30 19:22:08
IPIs no longer suck, so turn this back on.
Affected files ...
.. //depot/projects/smpng/sys/kern/sched_4bsd.c#44 edit
.. //depot/projects/smpng/sys/notes#27 edit
Differences ...
==== //depot/projects/smpng/sys/kern/sched_4bsd.c#44 (text+ko) ====
@@ -229,7 +229,7 @@
/* Enable forwarding of wakeups to all other cpus */
SYSCTL_NODE(_kern_sched, OID_AUTO, ipiwakeup, CTLFLAG_RD, NULL, "Kernel SMP");
-static int forward_wakeup_enabled = 0;
+static int forward_wakeup_enabled = 1;
SYSCTL_INT(_kern_sched_ipiwakeup, OID_AUTO, enabled, CTLFLAG_RW,
&forward_wakeup_enabled, 0,
"Forwarding of wakeup to idle CPUs");
==== //depot/projects/smpng/sys/notes#27 (text+ko) ====
@@ -35,8 +35,6 @@
- Untested
- Change amd64 to use [ls]fence instructions for memory barriers.
- Untested (and no hardware, maybe peter can test)
-- Turn off the ipiwakeups in 4BSD since the currently implementation can
- lead to IPI deadlocks
- Add a kproc API that does what kthread does right now
- Add a real kthread API that creates just another thread inside of a kproc
- Figure out what needs to be done to make a new kthread.. should each
More information about the p4-projects
mailing list