Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

pccard kernel config for OmniBook 500

0 views
Skip to first unread message

Najib Ninaba

unread,
Nov 19, 2001, 9:48:02 AM11/19/01
to
Hello,

Is there anybody on the list who has a working pccard configuration for
4.4-STABLE on HP OmniBook 500? I realised that in my kernel config file,
it has (I'm typing from memory here..):

device pcic0 at isa? irq 0 port ... # continues on

What irq should I assign for this particular notebook? And what irqs
should be on /etc/pccard.conf? From dmesg and vmstat -i, seems like the
only available irqs seems to be 9 and 11.

Thanks in advance.

Regards,
--
Najib Ninaba
"Life's unfair - but having the root password helps!"
--
PGP fingerprint
B338 7EC2 B737 3BAA 03B1 38C7 0832 B6CF 160F 496E
--
To verify the signature, get my PGP public key at
http://www.keyserver.net/en/ KeyID:0x160F496E

Brooks Davis

unread,
Nov 19, 2001, 1:46:44 PM11/19/01
to
On Mon, Nov 19, 2001 at 10:47:21PM +0800, Najib Ninaba wrote:
> Hello,
>
> Is there anybody on the list who has a working pccard configuration for
> 4.4-STABLE on HP OmniBook 500? I realised that in my kernel config file,
> it has (I'm typing from memory here..):
>
> device pcic0 at isa? irq 0 port ... # continues on
>
> What irq should I assign for this particular notebook? And what irqs
> should be on /etc/pccard.conf? From dmesg and vmstat -i, seems like the
> only available irqs seems to be 9 and 11.

I believe the default entry with the following line in /boot/loader.conf
will work:

hw.pcic.init_routing=1

It works for me on -current and the support is pretty near identical at
this point.

-- Brooks

--
Any statement of the form "X is the one, true Y" is FALSE.
PGP fingerprint 655D 519C 26A7 82E7 2529 9BF0 5D8E 8BE9 F238 1AD4

Najib Ninaba

unread,
Nov 22, 2001, 5:36:45 AM11/22/01
to
Hello,

I've tried the below-mentioned line without any success. It boots up
normally but still freezes at the point when I plug in my aironet 350
card or ibm modem card. Any other leads?

Regards,
Najib Ninaba

On Mon, Nov 19, 2001 at 10:46:34AM -0800, Brooks Davis wrote:
> I believe the default entry with the following line in /boot/loader.conf
> will work:
>
> hw.pcic.init_routing=1
>
> It works for me on -current and the support is pretty near identical at
> this point.
>
> -- Brooks
>
> --
> Any statement of the form "X is the one, true Y" is FALSE.
> PGP fingerprint 655D 519C 26A7 82E7 2529 9BF0 5D8E 8BE9 F238 1AD4

--

Warner Losh

unread,
Nov 22, 2001, 3:13:54 PM11/22/01
to
: I've tried the below-mentioned line without any success. It boots up

: normally but still freezes at the point when I plug in my aironet 350
: card or ibm modem card. Any other leads?

Try -stable, if you aren't already, otherwise no. If I had acess to
one of these beasts, then maybe I could figure out the problem...

Warner

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

Brooks Davis

unread,
Nov 22, 2001, 3:54:39 PM11/22/01
to
On Thu, Nov 22, 2001 at 06:35:33PM +0800, Najib Ninaba wrote:
> I've tried the below-mentioned line without any success. It boots up
> normally but still freezes at the point when I plug in my aironet 350
> card or ibm modem card. Any other leads?

Hmm, odd. It's definatly working here in -current. I was using an
Aironet 340 quite a bit at a confrence last week. Here's what my hint's
file says about the pcic (this translates to the kernel config line
in stable.)

hint.pcic.0.at="isa"
# hint.pcic.0.irq="10" # Default to polling
hint.pcic.0.port="0x3e0"
hint.pcic.0.maddr="0xd0000"

Guido van Rooij

unread,
Nov 25, 2001, 6:57:08 AM11/25/01
to
On Thu, Nov 22, 2001 at 01:13:44PM -0700, Warner Losh wrote:
> : I've tried the below-mentioned line without any success. It boots up

> : normally but still freezes at the point when I plug in my aironet 350
> : card or ibm modem card. Any other leads?
>
> Try -stable, if you aren't already, otherwise no. If I had acess to
> one of these beasts, then maybe I could figure out the problem...

I have access to one now. I have the exact same symptoms with a freshly
installed 4.4-release. It will not even mount a disk when not
using it in polling mode. init_routing=1 also does not seem to make
an effect. The very moment a card is inserted, the machine freezes.
But not solidly! I can go to DDB.
Problem is that I do not have a docking station and thus no serial port
so I cannot use remote gdb on it.

a trace gives that it was in doreti() and eip=0xc029dc70 which is the entry
point of doreti()

What is really funny is that my notebook automatically suspended to disk
(hibernate) during this debugging due to a low battery. When it
came back, the hang was over!?

In fact, after that I could insert other cards as well.
I tried some more and when the system hangs, pressing the suspend
key (either normal suspend or hibernate) does not work anymore.
I also tried rebooting a machine and do a normal suspend after which
I inserted the card. That does not work either.

If the card was already inserted at boot time the system does not
hang, and the card is correctly seen. However, in that case, removal
of the card hangs the machine.

If I hybernate the system and afterwards insert a card it also hangs.

If I could help in anyway debugging this problem I'd surely like to.

-Guido

Guido van Rooij

unread,
Nov 25, 2001, 7:24:01 AM11/25/01
to
It gets weirder: If the machine hangs and I go to DDB and press 'n'
a number of times, the system 'unhangs' itsself and card insertion and removal
are working correctly.

Looking at vmstat -i aftrewards, I see a ridiculous amount of interrupts
on irq 10 (assigned to the builtin xl0).

If I set hw.pcic.irq=10 the booting failes with a panic (cant route ISA CSC
interrupt followed by panic: resource_list_release: can't find resource).

I tried again rebooting with hw.pcic.init_routing=1 and hw.pcic.intr_path=2
It hangs as before. The funny thing is the if I go to the debugger,
I see all input I typed. It thus looks like the screen is no longer
updated. Yet that is not the complete story. What I tried next was
to press return to get a single user /bin/sh and type echo ctrl-G.
I got a beep but the beep never stopped. Using ddb again to use the 'n'
trick resulted after a couple of time in a crash of the sh !?

Guido van Rooij

unread,
Nov 26, 2001, 4:16:21 PM11/26/01
to
On Sun, Nov 25, 2001 at 07:27:07AM +0800, Najib Ninaba wrote:
> Hello,
>
> After checking through the mailing-list for some clues, I just realised
> that the my kernel did not actually have any irq assigned to the pcic
> devices. Most of the dmesgs posted here have an irq after in their dmesg
> pcic line but not in mine. Could that be the problem of my notebook
> freezing whenever I plug in my card? The lines pci_cfgintr_linked and
> pci_cfgintr were also missing in my dmesg. Attached is my dmesg.
>

I had an irc session with Warner Losh and a patch has been made to address
the problem. After the patch, you should no longer use hw.pcic.intr_path=1

The patch has just been committed. (you need version 1.1.2.6
of sys/i386/isa/pci_cfgreg.c)

Najib Ninaba

unread,
Nov 27, 2001, 8:12:23 AM11/27/01
to
Hello,

It worked! Thanks, Warner for solving it and Guido for chasing the bug
;) Finally I can use this notebook for some real work. Thanks again, guys.

Regards,
Najib Ninaba

On Mon, Nov 26, 2001 at 10:16:11PM +0100, Guido van Rooij wrote:
> On Sun, Nov 25, 2001 at 07:27:07AM +0800, Najib Ninaba wrote:
> > Hello,
> >
> > After checking through the mailing-list for some clues, I just realised
> > that the my kernel did not actually have any irq assigned to the pcic
> > devices. Most of the dmesgs posted here have an irq after in their dmesg
> > pcic line but not in mine. Could that be the problem of my notebook
> > freezing whenever I plug in my card? The lines pci_cfgintr_linked and
> > pci_cfgintr were also missing in my dmesg. Attached is my dmesg.
> >
>
> I had an irc session with Warner Losh and a patch has been made to address
> the problem. After the patch, you should no longer use hw.pcic.intr_path=1
>
> The patch has just been committed. (you need version 1.1.2.6
> of sys/i386/isa/pci_cfgreg.c)
>
> -Guido

--

0 new messages