Re: [SMP patch] hard-lockup debugging, NMI-watchdog, 2.1.97
Bill Broadhurst (bbroad@CX492564-a.dt1.sdca.home.com)
Sun, 19 Apr 1998 17:03:59 -0700
On Mon, Apr 20, 1998 at 01:02:27AM +0200, MOLNAR Ingo wrote:
>
> On Mon, 20 Apr 1998, Keith Owens wrote:
>
> > On Mon, 20 Apr 1998 00:41:15 +0200 (CEST),
> > MOLNAR Ingo <mingo@chiara.csoma.elte.hu> wrote:
> > >+extern spinlock_t console_lock;
> > >+ * since NMIs dont listen to _any_ locks, we have to be extremely
> > >+ * careful not to rely on unsafe variables. The printk might lock
> > >+ * up though, so we have to break up console_lock first ...
> >
> > But it does not break console_lock.
>
> yes, i've been thinking about other locks as well. We need some kind of
> runtime registering for static spinlocks and we have to break them all.
> Or, like earlier NMI-watchdog patches did, we can halt the whole system
> and print out the first oops. (so one has a chance to write it down from
> screen)
>
> Ingo
OK I loaded the patch, re-compiled, and ran the SCSI backup which was
always locking solid. Here's what I got.
Kernel Panic: Attempted to kill idle PID
In Swapper task - Not Synching.
But, at least, the SysRq keys worked - mostly. The info keys worked
but the Unmount and Sync were useless - they just froze after the
sign-on for the key.
Does this tell you anything?
-bb
--
----------------------------------------------------------------------
Dr. Bill Broadhurst | Independent contract Engineer.
(619)296-3710 | BIOS, Firmware, & Diagnostics.
bbroad@CX492564-a.dt1.sdca.home.com | Finger for PGP 5.0 public key.
----------------------------------------------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu