Lockups when booting with CardBus NIC inserted

Lockups when booting with CardBus NIC inserted

[Felipe Alfaro Solana]

Hi,

I tried jumping from 2.6.0-test4-mm6 to 2.6.0-test4-bk9. However, with
2.6.0-test4-bk9, I have noticed that I can no longer boot if my CardBus
NIC is inserted into one of the CardBus sockets. Doing so causes the
kernel to lock up hard when checking the cardbus socket. The kernel is
able to boot if the card is not inserted into the slot.

This is very similar to a previous experience I had when Russell King
was implementing a new state machine to handle the PCMCIA/CArdBus events
(and if my memory serves me well, making the number of pccardd kernel
threads equal to the number of CardBus slots).

Is anyone experiencing this?

I notified Russell King about this, but haven't heard of him since then.
I assume he is working on a fix.

Thanks!

PS: This also happens with 2.6.0-test5 and 2.6.0-test5-mm1.

Re: Lockups when booting with CardBus NIC inserted

[Miffe]

Felipe Alfaro Solana wrote:
> Is anyone experiencing this?

Yes, in 2.6.0-test5.
It stops after

PCI: Enabling device 0000:00:03.0 (0000 -> 0002)
Yenta: CardBus bridge found at 0000:00:03.0 [1028:00bb]
Yenta: Using CSCINT to route CSC interrupts to PCI
Yenta: Routing CardBus interrupts to PCI
Yenta: ISA IRQ list 0098, PCI irq10
Socket status: 30000020

Re: Lockups when booting with CardBus NIC inserted

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 1888 bytes --]

On Tue, 09 Sep 2003 21:18:01 +0200, Felipe Alfaro Solana <felipe_alfaro@linuxmail.org>  said:
> Hi,
> 
> I tried jumping from 2.6.0-test4-mm6 to 2.6.0-test4-bk9. However, with
> 2.6.0-test4-bk9, I have noticed that I can no longer boot if my CardBus
> NIC is inserted into one of the CardBus sockets. Doing so causes the
> kernel to lock up hard when checking the cardbus socket. The kernel is
> able to boot if the card is not inserted into the slot.

Seeing the same thing here on a Dell Latitude C840.

.  -test4-mm6 was OK, -test5-mm1 was NOT

The card in question is a Xircom 10/100Ethernet/56K modem card.

Under -test4-mm6, it does:

Sep  9 09:41:18 turing-police kernel: Console: switching to colour frame buffer device 160x64
Sep  9 09:41:18 turing-police kernel: PCI: Enabling device 0000:02:01.0 (0000 -> 0002)
Sep  9 09:41:18 turing-police kernel: Yenta: CardBus bridge found at 0000:02:01.0 [1028:00d5]
Sep  9 09:41:18 turing-police kernel: Yenta: Using CSCINT to route CSC interrupts to PCI
Sep  9 09:41:18 turing-police kernel: Yenta: Routing CardBus interrupts to PCI
Sep  9 09:41:18 turing-police kernel: Yenta: ISA IRQ list 00b8, PCI irq9
Sep  9 09:41:18 turing-police kernel: Socket status: 30000006
Sep  9 09:41:18 turing-police kernel: PCI: Enabling device 0000:02:01.1 (0000 -> 0002)
Sep  9 09:41:18 turing-police kernel: Yenta: CardBus bridge found at 0000:02:01.1 [1028:00d5]
Sep  9 09:41:18 turing-police kernel: Yenta: Using CSCINT to route CSC interrupts to PCI
Sep  9 09:41:18 turing-police kernel: Yenta: Routing CardBus interrupts to PCI
Sep  9 09:41:18 turing-police kernel: Yenta: ISA IRQ list 00b8, PCI irq9
Sep  9 09:41:18 turing-police kernel: Socket status: 30000006


Under -test5-mm1 it says "Socket status: 30000020" after the *first* one (02:01.0)
and locks up hard, not even a SysRq.  Is it perhaps upset about the fact it's
a multifunction card?

[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]