Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 02 Oct 2000 16:33:03 -0700 (PDT)
From:      John Baldwin <jhb@FreeBSD.org>
To:        Doug Rabson <dfr@nlsystems.com>
Cc:        alpha@freebsd.org, cp@bsdi.com, smp@freebsd.org
Subject:   Re: Status update
Message-ID:  <XFMail.001002163303.jhb@FreeBSD.org>
In-Reply-To: <Pine.BSF.4.21.0009260948210.35016-100000@salmon.nlsystems.com>

next in thread | previous in thread | raw e-mail | index | archive | help

On 26-Sep-00 Doug Rabson wrote:
> On Tue, 26 Sep 2000, Doug Rabson wrote:
> 
>> On Sat, 23 Sep 2000, John Baldwin wrote:
>> > 
>> > There are also a few more weirdism's on the alpha.  In a few places in
>> > sys/kern, we call spl0() instead of splx().  I've added some debugging
>> > code to
>> > do a printf() if we aren't actually at IPL_0 (what spl0 used to do) after
>> > the
>> > mtx_exit().  It does trigger in several cases during /etc/rc at least, but
>> > the
>> > machine seems to be running stable regardless (I'll be running a
>> > buildworld -j
>> > 8 tonight to stress test it).  My question is: is it ok for the code to
>> > run
>> > with some interrupts disabled or do we need to replace the calls to spl0()
>> > with enable_intr()?
>> 
>> I'm testing this now and I'm seeing a flood of diagnostic messages like:
>> 
>>      ../../kern/kern_fork.c:537:fork1() spl0 needs fixing 
>> 
>> I think these are all due to the fact that sched_lock is held at that
>> point. The preceding mtx_exit() just decrements the recursion count.
>> I haven't worked out exactly why sched_lock is held here because it
>> shouldn't be. Possibly its because we don't clear pcb_schednest in
>> cpu_fork().
> 
> As I suspected, clearing pcb_schednest makes this lot go away. Try this
> version of vm_machdep.c:
> 
> Index: vm_machdep.c
> ===================================================================
> RCS file: /home/ncvs/src/sys/alpha/alpha/vm_machdep.c,v
> retrieving revision 1.33
> diff -u -r1.33 vm_machdep.c
> --- vm_machdep.c      2000/09/07 01:32:39     1.33
> +++ vm_machdep.c      2000/09/26 08:38:55
> @@ -210,6 +210,13 @@
>               up->u_pcb.pcb_context[2] = (u_long) p2; /* s2: a0 */
>               up->u_pcb.pcb_context[7] =
>                   (u_int64_t)switch_trampoline;       /* ra: assembly magic */
> +
> +             /*
> +              * Clear the saved recursion count for sched_lock
> +              * since the child needs only one count which is
> +              * released in switch_trampoline.
> +              */
> +             up->u_pcb.pcb_schednest = 0;
>       }
>  }

The x86 code does this, so the alpha code should as well.  Go ahead
and commit this, it shouldnt' break anything. :-P  I'll remove the
extra cruft from the MTX_EXIT() assembly in mutex.h afterwards.

> -- 
> Doug Rabson                           Mail:  dfr@nlsystems.com

-- 

John Baldwin <jhb@FreeBSD.org> -- http://www.FreeBSD.org/~jhb/
PGP Key: http://www.baldwin.cx/~john/pgpkey.asc
"Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.001002163303.jhb>