Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 2 Dec 1998 06:12:10 -0600 (CST)
From:      Daniel Ortmann <ortmann@sparc.isl.net>
To:        bde@zeta.org.au (Bruce Evans), gurney_j@efn.org (John-Mark Gurney)
Cc:        bde@zeta.org.au, kato@ganko.eps.nagoya-u.ac.jp, andyf@speednet.com.au, freebsd-current@FreeBSD.ORG, geoffb@demon.net, johan@granlund.nu, mike@smith.net.au, ortmann@sparc.isl.net
Subject:   a problem tracked down (was "Re: sio breakage")
Message-ID:  <199812021212.GAA01094@pyrl.eye>
In-Reply-To: <199812010450.PAA27337@godzilla.zeta.org.au> from Bruce Evans at "Dec 1, 1998  3:50:12 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
> >> Fixes:
> >> kernel: send a fatal signal to applications that do this.
> >
> >How to detect this?
> 
> Near the beginning of trap():
> 
> 	if ((frame.tf_eflags & PSL_I) == 0)
> 		fatal(...);
> 
> but this is too strong, since the application may be in the middle
> of changing hardware state.  Returning to the application probably
> won't make the problem (the application's problem of changing state
> atomically) worse, so we should enable interrupts and print an error
> message.
> 
> PSI_I must be checked even in traps from kernel mode because the trap
> may be nested.  In kernel mode, just enable interrupts and continue.

I found the problem with my slow-running clock!

Here is the fix:

1) disable power management in the BIOS
2) remove apm from my kernel's config file

Perhaps merely one of the above is sufficient.  I did both 1 & 2.

I recall that my sio silo overflow messages began to appear suddenly
after a period of time ... as if a switch were thrown.  Perhaps the
above steps will also make those messages go away; this is probable
since they seemed to appear at the same time.

I will log a pr right away.

-- 
Daniel Ortmann       IBM Circuit Technology
2414 30 av NW, #D    E315, bldg 040-2
Rochester, MN 55901  3605 Hwy 52 N
507.288.7732 (h)     507.253.6795 (w)
ortmann@isl.net      ortmann@us.ibm.com
--
"The answers are so simple and we all know where to look,
but it's easier just to avoid the question." -- Kansas

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



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199812021212.GAA01094>