Re: autoloading module for pcmcia network card
Thomas Friedrichsmeier [Donnerstag, 3. April 2003 11:03]:
>| Apr 2 19:03:30 flyer cardmgr[2086]: starting, version is 3.1.33
>| Apr 2 19:03:31 flyer cardmgr[2086]: watching 2 sockets
>| Apr 2 19:03:31 flyer cardmgr[2086]: Card Services release does not match
>| Apr 2 19:03:31 flyer cardmgr[2086]: socket 0: CardBus hotplug device
>| where that thing about release not matching does indeed look a bit
>| suspicious. I don't know, however, whether that's the result of my messing
>| with the config-files in /etc/pcmcia, or what it is all about.
Assuming you installed pcmcia-cs (to get the daemon and the tools ) and
pcmcia-modules-2.4.xx (your kernel-version), for kernel-pcmcia.
Then it's not unusual the tools (cardmgr) won't be the same version as the
kernel-pcmcia-support (i.e., the modules). Which is said to be no problem,
somewhere (maybe it was the Howto ?)
All i can say is, that after I got the package pcmcia-source and recompiled
both modules and tools against you actual kernel, this error message didn't
occur no more. And the config was accepted. I didn't need to install hotplug,
for i've got only 16 bit cardbus cards.
>| Now, with hotplug installed, it is:
>| Apr 3 10:30:10 flyer cardmgr[329]: starting, version is 3.1.33
>| Apr 3 10:30:10 flyer cardmgr[329]: watching 2 sockets
>| Apr 3 10:30:10 flyer cardmgr[329]: Card Services release does not match
>| Apr 3 10:30:11 flyer cardmgr[329]: initializing socket 0
Now this is weird. Same circumstances ?
>| Apr 3 10:30:11 flyer cardmgr[329]: socket 0: CardBus hotplug device
>| Apr 3 10:30:13 flyer insmod:
>| /lib/modules/2.4.20-686/kernel/drivers/net/8139cp.o: init_module: No such
>| device
The driver doesn't recognize his targeted device. Seems as you can't use the
cp. It's for 8193C+ chips, maybe not your's.
Now what about the ('gerneric') too?
>| Note, that before insmod fails on 8139cp, 8139too has already been loaded.
ok.
>| /var/log/messages:
>| Apr 3 10:30:09 flyer kernel: Linux Kernel Card Services 3.1.22
See what i mean. ------------------------------------------------------^^^
>| Apr 3 10:30:09 flyer kernel: options: [pci] [cardbus] [pm]
>| Apr 3 10:30:09 flyer kernel: PCI: Guessed IRQ 9 for device 00:13.0
>| Apr 3 10:30:09 flyer kernel: PCI: Sharing IRQ 9 with 00:13.1
>| Apr 3 10:30:09 flyer kernel: PCI: Guessed IRQ 9 for device 00:13.1
>| Apr 3 10:30:09 flyer kernel: PCI: Sharing IRQ 9 with 00:13.0
>| Apr 3 10:30:09 flyer kernel: Yenta IRQ list 08b8, PCI irq9
Did i mention the standalone package uses the PCIC=i82365 socket ?
According to the manpage, there should be some options possible to pass,
like pc_debug=1 ... didn't try that, though.
>| Apr 3 10:30:09 flyer kernel: Socket status: 30000827
>| Apr 3 10:30:09 flyer kernel: Yenta IRQ list 08b8, PCI irq9
>| Apr 3 10:30:09 flyer kernel: Socket status: 30000007
>| Apr 3 10:30:10 flyer kernel: cs: cb_alloc(bus 2): vendor 0x10ec, device
>| 0x8139
>| Apr 3 10:30:11 flyer kernel: 8139too Fast Ethernet driver 0.9.26
>| Apr 3 10:30:11 flyer kernel: eth0: RealTek RTL8139 Fast Ethernet at
fine good ol' kernel :-)
>| 0xca8fa000, 00:40:f5:4f:10:a1, IRQ 9
>| Apr 3 10:30:12 flyer kernel: 8139cp: 10/100 PCI Ethernet driver v0.3.0
>| (Sep 29, 2002)
hmm. Why does it load the cp as well ? One driver's enough.
>| Apr 3 10:30:13 flyer pci.agent: ... can't load module 8139cp
>| Apr 3 10:30:13 flyer pci.agent: missing kernel or user mode driver 8139cp
jesus. get rid of hotplug, i'd say. It's quite confusing to deal with two
agents + the kernel-pci at once ... Make it more easy !
Seems you didn't use hotplug 'pcimodules' (package pciutils ?) to coldplug
the module, btw.
--> /usr/share/doc/hotplug/README.gz,
search for 'pcmcia' and you directly get the point.
good luck
micha.
Reply to: