Date: Sat, 13 May 2006 17:22:19 +0200 From: Ivan Voras <ivoras@fer.hr> To: Sam Leffler <sam@errno.com> Cc: stable@freebsd.org Subject: Re: pagezero again Message-ID: <4465F9AB.8010107@fer.hr> In-Reply-To: <4465F0E4.8020205@errno.com> References: <4464FA8D.5060309@fer.hr> <44652D88.1070408@FreeBSD.org> <4465CA8B.7000509@fer.hr> <4465EE4D.2090701@FreeBSD.org> <4465F0E4.8020205@errno.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Sam Leffler wrote:
> Maybe this is the same problem jhb recently fixed in head with threads
> getting assigned the wrong priority under certain conditions (it showed
> up most easily with taskq threads).
Is it this one:
jhb 2006-04-17 18:20:38 UTC
FreeBSD src repository
Modified files:
sys/dev/hwpmc hwpmc_mod.c
sys/dev/random randomdev_soft.c
sys/kern kern_intr.c kern_poll.c kern_synch.c
kern_thr.c kern_umtx.c sched_4bsd.c
subr_taskqueue.c uipc_mqueue.c
sys/vm vm_zeroidle.c
Log:
Change msleep() and tsleep() to not alter the calling thread's priority
if the specified priority is zero. This avoids a race where the calling
thread could read a snapshot of it's current priority, then a different
thread could change the first thread's priority, then the original thread
would call sched_prio() inside msleep() undoing the change made by the
second thread. I used a priority of zero as no thread that calls
msleep()
or tsleep() should be specifying a priority of zero anyway.
The various places that passed 'curthread->td_priority' or some variant
as the priority now pass 0.
Revision Changes Path
1.24 +1 -8 src/sys/dev/hwpmc/hwpmc_mod.c
1.13 +2 -4 src/sys/dev/random/randomdev_soft.c
1.132 +3 -5 src/sys/kern/kern_intr.c
1.26 +1 -3 src/sys/kern/kern_poll.c
1.279 +5 -3 src/sys/kern/kern_synch.c
1.43 +2 -2 src/sys/kern/kern_thr.c
1.36 +4 -6 src/sys/kern/kern_umtx.c
1.78 +1 -1 src/sys/kern/sched_4bsd.c
1.36 +1 -1 src/sys/kern/subr_taskqueue.c
1.15 +2 -2 src/sys/kern/uipc_mqueue.c
1.37 +1 -3 src/sys/vm/vm_zeroidle.c
Would it be enough just to apply this change to my 6.1-R tree:
http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/vm/vm_zeroidle.c.diff?r1=1.36&r2=1.37
?
It's probably not, so: any chance of MFC-ing the commit? :)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4465F9AB.8010107>