3com 3crwb6096b crash my system..

[nic]

Hi

After I got a new Toshiba r100 laptop (from a Dell latitude) my 3com bluetooth pcmcia card startet killing my laptop. It uses the serial_cs driver and worked well before, but now when I add the card, the system freezes and I have to reboot.

System infos:
Gentoo Linux
Kernel 2.6.7 (own kernel)
pcmcia-cs version 3.2.7
Toshiba R100 Laptop (yento_socket)

[David Hinds]

Do you have any other PCMCIA cards, or is this the only one you can try?

It could be a few things.  Maybe the IO port range and/or interrupt being assigned to the card conflicts with some other device.  Or maybe the memory probe that happens when a card is first inserted is causing the problem.

-- Dave

[nic]

Hi

I'm really sorry about the delay.

This is what I've found out so far. When I plug in the card or start pcmcia with the card in the slot, my computer hangs. If I remove it fast enough, the computer comes back to life.
This is the dmegs output:
Yenta: CardBus bridge found at 0000:02:0b.0 [1179:0001]
Yenta: ISA IRQ mask 0x04b8, PCI irq 11
Socket status: 30000007
cs: memory probe 0x0c0000-0x0fffff: excluding 0xc0000-0xcbfff 0xe0000-0xfffff
Serial: 8250/16550 driver $Revision: 1.90 $ 48 ports, IRQ sharing enabled
Serial: 8250/16550 driver $Revision: 1.90 $ 48 ports, IRQ sharing enabled
ttyS16 at I/O 0x100 (irq = 11) is a 16C950/954
ttyS17 at I/O 0x108 (irq = 11) is a 8250

and the /var/log/messages:
Oct 27 00:44:15 nicto cardmgr[23884]: socket 0: Serial or Modem
Oct 27 00:44:15 nicto ttyS16 at I/O 0x100 (irq = 11) is a 16C950/954
Oct 27 00:44:15 nicto ttyS17 at I/O 0x108 (irq = 11) is a 8250
Oct 27 00:44:19 nicto cardmgr[23884]: executing: './serial start ttyS17 2>&1'
Oct 27 00:44:19 nicto cardmgr[23884]: + Cannot get serial info: Input/output error
Oct 27 00:44:19 nicto cardmgr[23884]: + Missing argument for irq
Oct 27 00:44:19 nicto cardmgr[23884]: executing: './serial stop ttyS17 2>&1'
Oct 27 00:44:19 nicto cardmgr[23884]: + /dev/modem: No such file or directory

I just upgraded to kernel 2.6.9.

[David Hinds]

Well, the first thing I'd try would be to add:

  exclude port 0x100-0x10f

to /etc/pcmcia/config.opts on the theory that it might be an IO port conflict.

I'm assuming that re-inserting the card after you've rescued it also results in a lockup?

-- Dave

[nic]

Thanks for the reply

First I have tried a few other cards (network, externel cdrom drive) and they both work just fine.

Second, I tried 'exclude port 0x100-0x10f' in config.opts, but it didn't help.

No, if i reinsert the card afterwards it still freezes.
Could it be a conflict on irq 11?

/nicholai

[David Hinds]

I don't think it can be an interrupt conflict, if your other cards work ok.  I think the problem is that the kernel serial_cs driver is misdetecting the card as a dual port device.

Can you temporarily comment out the device definition for the serial_cs driver in /etc/pcmcia/config (so that your system doesn't lock up), and then post the output of 'dump_cis' with this card inserted?

-- Dave

[nic]

Okay. I commented out device "serial_cs" ... in /etc/pcmcia/config

Here is my dump_cis:

Socket 0:
  dev_info
    NULL 0ns, 512b
  manfid 0x0101, 0x0041
  funcid serial_port [post]
  serial_interface
    uart 16550 [space/mark/odd/even] [5/6/7/8] [1/1.5/2]
  vers_1 7.1, "3Com", "3CRWB6096B Bluetooth PC Card", "", ""
  config base 0x00f8 mask 0x000f last_index 0x02
  cftable_entry 0x01 [default]
    [rdybsy]
    Vcc Vnom 3300mV Iavg 70mA
    io 0x0000-0x000f [lines=4] [8bit] [range]
    irq mask 0xffff [level] [shared]
  cftable_entry 0x02 [default]
    [rdybsy]
    Vcc Vnom 5V Iavg 70mA
    io 0x0000-0x000f [lines=4] [8bit] [range]
    irq mask 0xffff [level] [shared]

I did notice in my kernel config some options under
Character Devices ->
        Serial Drivers ->
              PCMCIA device support
Some stuff like 'Support for sharing serial intetrupts'  and more.

Is that related?

Thanks,

Nic