Can't hotplug two cardbus devices -- a 3c575 ethernet adapter and a USB host-controller.

Can't hotplug two cardbus devices -- a 3c575 ethernet adapter and a USB host-controller.

[Miles Lane]

First off, I have successfully hotplugged my USB mouse on
this machine, so I don't think my configuration is wrong.

My config:

	kernel 2.4.0-pre9
	modutils 2.4.0 + David Brownell's depmod patch.
	The latest hotplug package from the linux-hotplug site.
	usbutils is not installed.

I just tested and got this:

Jan 22 02:20:35 agate kernel: cs: socket c4ffa000 timed out during 
reset.  Try increasing setup_delay.
Jan 22 02:20:36 agate kernel: cs: cb_alloc(bus 5): vendor 0x1045, device 
0xc861
Jan 22 02:20:36 agate kernel:   got res[11000000:11000fff] for resource 
0 of PCI device 1045:c861
Jan 22 02:20:36 agate kernel: PCI: Enabling device 05:00.0 (0000 -> 0002)
Jan 22 02:20:36 agate kernel: PCI: No IRQ known for interrupt pin A of 
device 05:00.0.
Jan 22 02:20:36 agate cardmgr[437]: initializing socket 1
Jan 22 02:20:36 agate cardmgr[437]: unsupported card in socket 1
Jan 22 02:20:36 agate /sbin/hotplug: arguments (pci) env 
(PCI_CLASSÀ310 PCI_ID\x1045:C861 ACTION­d TERM=dumb HOSTTYPE=i386 
PATH=/bin:/sbin:/usr/sbin:/usr/bin HOME=/ SHELL=/bin/bash DEBUG=yes 
PCI_SUBSYS_ID\x1045:C861 OSTYPE=Linux PCI_SLOT_NAME\x05:00.0 SHLVL=1 
_=/usr/bin/env)
Jan 22 02:20:36 agate cardmgr[437]:   no product info available

# pci module         vendor     device     subvendor  subdevice  class 
     class_mask driver_data
ohci1394             0xffffffff 0xffffffff 0xffffffff 0xffffffff 
0x000c0010 0x00ffffff 0x00000000

It looks to me like something is hosed in creating this entry.
We don't want the vendor/device IDs to be "0xffffffff 0xffffffff",
do we?  How are those values getting into the modules.pcimap file?

Also interesting is the fact that during the boot process,
my 3c575 cardbus ethernet PC Card was noticed, but no hotplug
events happened for it:

Jan 22 02:12:50 agate kernel: Yenta IRQ list 0698, PCI irq11
Jan 22 02:12:51 agate kernel: Socket status: 30000068
Jan 22 02:12:51 agate kernel: Yenta IRQ list 0698, PCI irq11
Jan 22 02:12:51 agate kernel: Socket status: 30000006
Jan 22 02:12:51 agate kernel: devfs: v0.102 (20000622) Richard Gooch 
(rgooch@atnf.csiro.au)
Jan 22 02:12:51 agate kernel: devfs: devfs_debug: 0x0
Jan 22 02:12:51 agate kernel: devfs: boot_options: 0x2
Jan 22 02:12:51 agate kernel: cs: cb_alloc(bus 1): vendor 0x10b7, device 
0x5157
Jan 22 02:12:52 agate kernel:   got res[1000:107f] for resource 0 of PCI 
device 10b7:5157
Jan 22 02:12:52 agate kernel:   got res[10800000:1080007f] for resource 
1 of PCI device 10b7:5157
Jan 22 02:12:52 agate kernel:   got res[10800080:108000ff] for resource 
2 of PCI device 10b7:5157
Jan 22 02:12:52 agate kernel:   got res[10400000:1041ffff] for resource 
6 of PCI device 10b7:5157
Jan 22 02:12:52 agate kernel: PCI: Enabling device 01:00.0 (0000 -> 0003)
Jan 22 02:12:52 agate kernel: PCI: No IRQ known for interrupt pin A of 
device 01:00.0.

Instead, when I loaded the driver by hand, the hotplug event got
triggered:

Jan 22 02:22:23 agate kernel: 3c59x.c:LK1.1.12 06 Jan 2000  Donald 
Becker and others. http://www.scyld.com/network/vortex.html $Revision: 
1.102.2.46 $
Jan 22 02:22:23 agate kernel: See Documentation/networking/vortex.txt
Jan 22 02:22:23 agate kernel: eth0: 3Com PCI 3CCFE575BT Cyclone CardBus 
at 0x1000, PCI: Setting latency timer of device 01:00.0 to 64
Jan 22 02:22:23 agate kernel:  00:10:4b:7c:9d:9d, IRQ 11
Jan 22 02:22:23 agate kernel:   product code 'NV' rev 07.1 date 06-09-98
Jan 22 02:22:23 agate kernel: eth0: CardBus functions mapped 
10800080->c5a53080
Jan 22 02:22:23 agate kernel:   8K byte-wide RAM 5:3 Rx:Tx split, MII 
interface.
Jan 22 02:22:23 agate kernel:   MII transceiver found at address 0, 
status 782d.
Jan 22 02:22:23 agate kernel:   Enabling bus-master transmits and 
whole-frame receives.
Jan 22 02:22:23 agate /sbin/hotplug: arguments (net) env 
(ACTION=register TERM=dumb HOSTTYPE=i386 
PATH=/bin:/sbin:/usr/sbin:/usr/bin HOME=/ SHELL=/bin/bash DEBUG=yes 
INTERFACE=eth0 OSTYPE=Linux SHLVL=1 _=/usr/bin/env)
Jan 22 02:22:23 agate kernel: eth0: using default media MII

Here is the modules.pcimap line that shows that there should have been a 
match:

3c59x                0x000010b7 0x00005157 0xffffffff 0xffffffff 
0x00000000 0x00000000 0x0000001a

Lastly, I am getting a SCSI hotplug error early in the boot cycle.

    Jan 22 02:12:47 agate kernel: request_module[scsi_hostadapter]: Root 
fs not mounted

I thought this was fixed?

	Miles


_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Can't hotplug two cardbus devices -- a 3c575 ethernet adapter and a

[Andrew Morton]

Miles Lane wrote:
> 
> First off, I have successfully hotplugged my USB mouse on
> this machine, so I don't think my configuration is wrong.
> 
> My config:
> 
>         kernel 2.4.0-pre9
>         modutils 2.4.0 + David Brownell's depmod patch.
>         The latest hotplug package from the linux-hotplug site.
>         usbutils is not installed.
> 
> I just tested and got this:
> 
> Jan 22 02:20:35 agate kernel: cs: socket c4ffa000 timed out during
> reset.  Try increasing setup_delay.

This is where things start to come unstuck.  The NIC isn't
reset properly and its PCI space subsequently seems to read
as mostly 0xff's.

I'd roll out my standard answer: "turn off ACPI", but I don't
think you're using ACPI?

Something's screwy in the cardbus initialisation and I don't
know what.  I assume everything works OK if you eject
and reseat the card?

> ...
> 
> Lastly, I am getting a SCSI hotplug error early in the boot cycle.
> 
>     Jan 22 02:12:47 agate kernel: request_module[scsi_hostadapter]: Root
> fs not mounted
> 
> I thought this was fixed?

Mostly.

Perhaps you could stick a `show_trace(0)' into request_module()
where this print happens, feed the dmesg output through ksymoops?

_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Can't hotplug two cardbus devices -- a 3c575 ethernet adapter and a

[Miles Lane]

BTW:  What action is this hotplug call supposed to accomplish?

Jan 22 10:40:45 agate /sbin/hotplug: arguments (net) env 
(ACTION=register TERM=dumb HOSTTYPE=i386 
PATH=/bin:/sbin:/usr/sbin:/usr/bin HOME=/ SHELL=/bin/bash DEBUG=yes 
INTERFACE=eth0 OSTYPE=Linux SHLVL=1 _=/usr/bin/env)

When I load the 3c59x driver by hand (the hotplug
"add" hotplug event isn't loading the driver),
this "register" hotplug process kicks off.  However,
it doesn't do anything I can determine, like
run "ifup eth0".


_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Can't hotplug two cardbus devices -- a 3c575 ethernet adapter

[David Brownell]

> 
> When I insert my 3c575, I get:
> 
> ...
> Jan 22 12:55:30 agate cardmgr[444]: initializing socket 0
> Jan 22 12:55:30 agate cardmgr[444]: socket 0: 3Com 3CCFE575B/3CXFE575B 
> Fast EtherLink XL
> Jan 22 12:55:30 agate /sbin/hotplug: arguments (pci) env 
> (PCI_CLASS 000 PCI_ID\x10B7:5157 ACTION­d TERM=dumb HOSTTYPE=i386 
> PATH=/bin:/sbin:/usr/sbin:/usr/bin HOME=/ SHELL=/bin/bash DEBUG=yes 
> PCI_SUBSYS_ID\x10B7:5B57 OSTYPE=Linux PCI_SLOT_NAME\x01:00.0 SHLVL=1 
> _=/usr/bin/env)

You need to apply that patch to /etc/hotplug/pci.agent that was sent
by at the beginning of last week (from Dan Zink @ COmpaq).  The PCI
ID syntax changed again, it uses colons not "/" ... one hopes it stays
stable now!


> Jan 22 12:55:30 agate cardmgr[444]: module 
> /lib/modules/2.4.1-pre9/pcmcia/cb_enabler.o not available
> Jan 22 12:55:30 agate cardmgr[444]: module 
> /lib/modules/2.4.1-pre9/pcmcia/3c575_cb.o not available
> Jan 22 12:55:31 agate cardmgr[444]: get dev info on socket 0 failed: 
> Resource temporarily unavailable
> 
> I see no sign of "modprobe 3c59x" occuring.

Likely you just need that patch.


> When I run "modprobe 3c59x" by hand, I get:
> 
> Jan 22 12:59:38 agate kernel: 3c59x.c:LK1.1.12 06 Jan 2000  Donald 
> Becker and others. http://www.scyld.com/network/vortex.html $Revision: 
> 1.102.2.46 $
> Jan 22 12:59:38 agate kernel: See Documentation/networking/vortex.txt
> Jan 22 12:59:38 agate kernel: eth0: 3Com PCI 3CCFE575BT Cyclone CardBus 
> at 0x1000, PCI: Setting latency timer of device 01:00.0 to 64
> Jan 22 12:59:38 agate kernel:  00:10:4b:7c:9d:9d, IRQ 11
> Jan 22 12:59:38 agate kernel:   product code 'NV' rev 07.1 date 06-09-98
> Jan 22 12:59:38 agate kernel: eth0: CardBus functions mapped 
> 10800080->c5a50080
> Jan 22 12:59:38 agate kernel:   8K byte-wide RAM 5:3 Rx:Tx split, MII 
> interface.
> Jan 22 12:59:38 agate kernel:   MII transceiver found at address 0, 
> status 782d.
> Jan 22 12:59:38 agate kernel:   Enabling bus-master transmits and 
> whole-frame receives.
> Jan 22 12:59:38 agate /sbin/hotplug: arguments (net) env 
> (ACTION=register TERM=dumb HOSTTYPE=i386 
> PATH=/bin:/sbin:/usr/sbin:/usr/bin HOME=/ SHELL=/bin/bash DEBUG=yes 
> INTERFACE=eth0 OSTYPE=Linux SHLVL=1 _=/usr/bin/env)
> Jan 22 12:59:39 agate kernel: eth0: using default media MII
> 
> And, I see no sign of "ifup eth0" being run.

What would syslog show though?  I know that MII messages come
out of pegasus when the interface is brought up.

You should check whether "ifconfig eth0" shows that it's up.

- Dave



_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Can't hotplug two cardbus devices -- a 3c575 ethernet adapter

[Miles Lane]

David Brownell wrote:

<snip>

> You need to apply that patch to /etc/hotplug/pci.agent that was sent
> by at the beginning of last week (from Dan Zink @ COmpaq).  The PCI
> ID syntax changed again, it uses colons not "/" ... one hopes it stays
> stable now!

So, has CVS been updated with these changes?  Also, isn't a new
release set needed in the download section of
http://sourceforge.net/project/showfiles.php?group_id\x17679 ?

>> When I run "modprobe 3c59x" by hand, I get:
<snip>

>> And, I see no sign of "ifup eth0" being run.

> You should check whether "ifconfig eth0" shows that it's up.

I'll give that a try, thanks!

	Miles


_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel