2.5.67 + i82365 -> oops

2.5.67 + i82365 -> oops

[Jean Tourrilhes]

	Hi,

	I've got troubles with 2.5.67 and an ISA Pcmcia bridge
(i82365). Half of the time, after boot, the card is recognised as
memory_cs (which is not compiled), where it should be airo_cs.
	When trying to restart the Pcmcia package (using the init
script), I got the nice oops below...

	Have a nice day...

	Jean

Apr 17 14:50:55 lagaffe cardmgr[307]: executing: 'modprobe -r memory_cs'
Apr 17 14:50:55 lagaffe cardmgr[307]: + FATAL: Module memory_cs not found.
Apr 17 14:50:55 lagaffe cardmgr[307]: modprobe exited with status 1
Apr 17 14:50:55 lagaffe cardmgr[307]: exiting
Apr 17 14:50:57 cs: IO port probe 0x0c00-0x0cff:lagaffe cardmgr[ clean.
1918]: watching cs: IO port probe 0x0800-0x08ff:2 sockets
 clean.
cs: IO port probe 0x0100-0x04ff: excluding 0x200-0x207 0x220-0x22f 0x330-0x337 0x388-0x39f 0x3c0-0x3e7 0x3f0-0x3ff 0x480-0x48f 0x4d0-0x4d7
cs: IO port probe 0x0a00-0x0aff: clean.
Apr 17 14:50:57 lagaffe cardmgr[1919]: starting, version is 3.2.3
Apr 17 14:50:57 lagaffe cardmgr[1919]: socket 0: Anonymous Memory
Apr 17 14:50:57 lagaffe cardmgr[1919]: executing: 'modprobe memory_cs'
Apr 17 14:50:57 lagaffe cardmgr[1919]: + FATAL: Module memory_cs not found.
Apr 17 14:50:57 lagaffe cardmgr[1919]: modprobe exited with status 1
Apr 17 14:50:57 lagaffe cardmgr[1919]: module /lib/modules/2.5.67/pcmcia/memory_cs.o not available
Apr 17 14:50:57 lagaffe cardmgr[1919]: bind 'memory_cs' to socket 0 failed: Invalid argument
Unable to handle kernel NULL pointer dereference at virtual address 00000014
 printing eip:
c0180fae
*pde = 00000000
Oops: 0000 [#1]
CPU:    0
EIP:    0060:[<c0180fae>]    Not tainted
EFLAGS: 00010256
EIP is at pci_remove_behind_bridge+0x6/0x38
eax: 00000000   ebx: cb60d000   ecx: cc838dc0   edx: 00000000
esi: 00000000   edi: cc838de4   ebp: 00000000   esp: cbf93f28
ds: 007b   es: 007b   ss: 0068
Process events/0 (pid: 6, threadinfo=cbf92000 task=cbfa0670)
Stack: cb60d000 cb60d000 cc85bde9 00000000 cc859376 cb60d000 cb60d000 cb60d000 
       cb60d00c cb60d000 00000080 cc838de4 00000000 cc8596a7 cb60d000 00000003 
       cb60d000 cc8596e6 cb60d000 00000080 00000000 00000141 cc83646d cb60d000 
Call Trace:
 [<cc85bde9>] cb_free+0xd/0x14 [pcmcia_core]
 [<cc859376>] shutdown_socket+0x7a/0xe8 [pcmcia_core]
 [<cc838de4>] socket+0x24/0x1a0 [i82365]
 [<cc8596a7>] do_shutdown+0x57/0x5c [pcmcia_core]
 [<cc8596e6>] parse_events+0x3a/0xd8 [pcmcia_core]
 [<cc83646d>] pcic_bh+0x5d/0x74 [i82365]
 [<cc838fe4>] pcic_task+0x4/0x40 [i82365]
 [<c0126f04>] worker_thread+0x1bc/0x294
 [<c0126d48>] worker_thread+0x0/0x294
 [<cc836410>] pcic_bh+0x0/0x74 [i82365]
 [<c0116d8c>] default_wake_function+0x0/0x1c
 [<c0116d8c>] default_wake_function+0x0/0x1c
 [<c0106f51>] kernel_thread_helper+0x5/0xc

Code: 8b 46 14 85 c0 74 26 8b 50 14 83 c0 14 8b 1a 39 c2 74 1a 8d 

Re: 2.5.67 + i82365 -> oops

[Russell King]

On Thu, Apr 17, 2003 at 03:00:18PM -0700, Jean Tourrilhes wrote:
> 	I've got troubles with 2.5.67 and an ISA Pcmcia bridge
> (i82365). Half of the time, after boot, the card is recognised as
> memory_cs (which is not compiled), where it should be airo_cs.

Well, the oops can be solved by application of this patch.  As for the
random detection (or non-detection) of airo_cs, can you say that any
kernel has reliably detected it (and if so, which kernel?)  Basically
I'd like to know if the problem exists with:

	2.4
	2.5 before Dominik/my changes
	2.5 after changes

diff -Nru a/drivers/pcmcia/cardbus.c b/drivers/pcmcia/cardbus.c
--- a/drivers/pcmcia/cardbus.c	Fri Apr 18 00:55:08 2003
+++ b/drivers/pcmcia/cardbus.c	Fri Apr 18 00:55:08 2003
@@ -270,5 +270,6 @@
 {
 	struct pci_dev *bridge = s->cap.cb_dev;
 
-	pci_remove_behind_bridge(bridge);
+	if (bridge)
+		pci_remove_behind_bridge(bridge);
 }


-- 
Russell King (rmk@arm.linux.org.uk)                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html