* Power Management/PCMCIA conflict causes system freeze
@ 2003-07-22 18:24 Larry LeBlanc
2003-07-22 18:34 ` Russell King
0 siblings, 1 reply; 3+ messages in thread
From: Larry LeBlanc @ 2003-07-22 18:24 UTC (permalink / raw)
To: linux-kernel
Hello,
I hinted at this issue when I asked about PCI IRQs yesterday but I've
received some feedback that indicates IRQs aren't my problem so I'll
post the basic problem I'm trying to solve and see if anyone has any ideas.
I am running Linux (RH 8.0, upgraded to kernel 2.4.20) on a Kontron
Envoy mobile data server - essentially a ruggedized PIII-700 MHz PC
designed to be installed in the trunk of a car. It comes with a 5 minute
UPS that issues power state change notifications so that it is possible
to gracefully shutdown the system when the car power is turned off. If
the power is restored within 30 seconds, the shutdown should abort and
the system should continue running (or the shutdown should complete and
then the system will auto-restart).
My problem is that as soon as the state change notification arrives, the
system freezes. If the power remains off for more than 30 seconds, the
UPS turns off and the system goes down hard. I'm using ext3 and I don't
do a lot of writing so this hasn't caused me any trouble - yet. The big
problem is that if the car power is restored within 30 seconds, the
system remains frozen until you once again turn the car off for more
than 30 seconds. Since the system is mounted in the trunk, it may not be
obvious that this problem has occurred.
This problem also manifests itself when I attempt to do a shutdown (-h
or -r). Everything shuts down properly but then hangs after the "Power
down" or "Restarting system" message is printed on the console. I don't
expect to have to reboot this system very often since it will naturally
reboot every time the car turns off and back on again, but there may be
occasions where I want to trigger a reboot and I won't be able to.
So how does PCMCIA relate to all of this? What we discovered after
gradually stripping things out of the system is that everything works
fine as long as the pcmcia modules (pcmcia_core, yenta_socket) are not
loaded into the kernel. Once they are loaded, I see the above behaviour,
regardless of whether any cards are in the CardBus adapter. Even if I
subsequently unload the pcmcia modules the problem persists.
Unfortunately I need PCMCIA support for my application, so leaving the
modules unloaded is not an option.
My system is using a Texas Instruments PCI1450 CardBus adapter, if that
helps. On boot yenta assigns IRQs 9 and 15 for the two sockets. 9 is
also used by my VGA adapter and 15 is used by my audio controller, but I
understand (now) that PCI is made to handle this kind of sharing and it
should not, in and of itself, be a problem. In fact /proc/interrupts
does not show the VGA and audio controller assignments - I found those
via lspci -vv. I believe that means that the drivers have not actually
requested them yet (not surprising since I don't really use either the
VGA or the audio controller).
Any ideas? I've tried turning on debug in apm but all that happens is
the debug messages freeze with the rest of the system. Is there any
other location I should turn on debug messages to try to get an idea of
where the system is hanging?
Thanks,
Larry
P.S. The system ships by default with Windows XP and everything works
fine including PCMCIA. So from a hardware perspective I think everything
is OK. I refuse to believe that I have to move to Windows in order to
make this work...there has to be a Linux answer...
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Power Management/PCMCIA conflict causes system freeze
2003-07-22 18:24 Power Management/PCMCIA conflict causes system freeze Larry LeBlanc
@ 2003-07-22 18:34 ` Russell King
[not found] ` <3F1D91AB.3030707@inmotiontechnology.com>
0 siblings, 1 reply; 3+ messages in thread
From: Russell King @ 2003-07-22 18:34 UTC (permalink / raw)
To: Larry LeBlanc; +Cc: linux-kernel
On Tue, Jul 22, 2003 at 11:24:25AM -0700, Larry LeBlanc wrote:
> Any ideas? I've tried turning on debug in apm but all that happens is
> the debug messages freeze with the rest of the system. Is there any
> other location I should turn on debug messages to try to get an idea of
> where the system is hanging?
You need to enable sysrq, cause a hang, and then issue a sysrq-t and
sysrq-p to find out where you're hanging.
Note that if you're using a serial console, you need to hold the serial
port open for sysrq requests to be processed.
--
Russell King (rmk@arm.linux.org.uk) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Power Management/PCMCIA conflict causes system freeze
[not found] ` <3F1D91AB.3030707@inmotiontechnology.com>
@ 2003-07-22 19:39 ` Larry LeBlanc
0 siblings, 0 replies; 3+ messages in thread
From: Larry LeBlanc @ 2003-07-22 19:39 UTC (permalink / raw)
To: Larry LeBlanc; +Cc: Russell King, linux-kernel
Larry LeBlanc wrote:
> Hi Russell,
>
> No luck...I enabled sysrq and verified it was working by issuing
> sysrq-t and sysrq-p while the system was healthy. But when I cause the
> hang sysrq-t and sysrq-p no longer do anything. One source I looked up
> indicated that "Since it [Magic SysRq] is implemented as a part of the
> keyboard driver, it is guaranteed to work most of the time, unless the
> kernel itself is dead." So my conclusion is that my kernel is dead.
>
> Note: I tried sysrq-b on my system when it is healthy and it causes
> the same hang.
>
> Larry
>
> Russell King wrote:
>
>>On Tue, Jul 22, 2003 at 11:24:25AM -0700, Larry LeBlanc wrote:
>>
>>
>>>Any ideas? I've tried turning on debug in apm but all that happens is
>>>the debug messages freeze with the rest of the system. Is there any
>>>other location I should turn on debug messages to try to get an idea of
>>>where the system is hanging?
>>>
>>>
>>
>>You need to enable sysrq, cause a hang, and then issue a sysrq-t and
>>sysrq-p to find out where you're hanging.
>>
>>Note that if you're using a serial console, you need to hold the serial
>>port open for sysrq requests to be processed.
>>
>>
>>
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2003-07-22 19:24 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-07-22 18:24 Power Management/PCMCIA conflict causes system freeze Larry LeBlanc
2003-07-22 18:34 ` Russell King
[not found] ` <3F1D91AB.3030707@inmotiontechnology.com>
2003-07-22 19:39 ` Larry LeBlanc
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).