Jump to content

updating system caused PCMCIA to fail


jdion81
 Share

Recommended Posts

I updated my system (running 10.0 CE or is it now OE?) with a mirror listed on the urpmi config site. I assume I am now running official after the update. I could not get the last few updates for a day or two. When I finally got the updates I shut down the computer and the next time I started it the pcmcia card was not found, I have tried 2 other pc cards and they are not found either. I also do not have power management. (Klaptop is not starting, I dont know how else to check power management, but the laptop is getting *really* hot and the fan runs for about 30 seconds every 30 seconds.)

 

I updated using urpmi, is there a way to tell what packages were installed or updated as I feel this is what has cause this issue. I do know that one of the packages was a kernel source package (dont remember the kernel version number though) I already had the kernel source installed for my system.

 

Mod note: Moved from software by phunni

Edited by phunni
Link to comment
Share on other sites

urpmi maintains a log file: /var/log/urpmi.log

 

You'll probably want to download the binary kernel update to go with the new kernel-source package (unless you want to compile the kernel yourself). I don't know if urpmi will allow you to install the binary kernel; it won't do it as part of an update since the kernel must be installed (rpm -i) not updated (rpm -U) to prevent removal of the currently running kernel's files. Unless you want to experiment with this, use a browser or ftp program to download the kernel. The most recent kernel is 2.6.3-8mdk, but depending on when you updated your system you might have picked up the 2.6.3-7mdk so check to see which source package you have first.

 

To install a new kernel package use rpm -ivh kernel-2.6.3.....rpm and reboot. The install script adds an entry to you lilo menu for your old kernel so you can boot that if the new one doesn't work.

Link to comment
Share on other sites

Hum.. Did you update your kernel? If so have you tried rebooting the last working verion? Have you checked /etc/lilo.conf to see if the kernel is being booted with the same options as the one that worked? If the kernel hasn't changed then I'd start looking for clues in the kernel boot messages (/var/log/messages) to see what hardware is being detected and how its being configured. There should be some output from the relevant modules as they load.

Link to comment
Share on other sites

I just went into services and told it to stop then start pcmcia. When I started pcmcia I got a message that said " Starting PCMCIA services: cardmgr[2820]: no sockets found! done." I know the card works and I know that the pcmcia slot works as I have booted off a knoppix cd and I am able to use the card there.

 

The only thing that has changed on mandrake since the card has worked last is I installed a few updates using uprmi at the cli. The updates installed were:

 

libqt3-3.2.3-17.1.100mdk.i586.rpm

qt3-common-3.2.3-17.1.100mdk.i586.rpm

kernel-source-2.6.3-8mdk.i586

Link to comment
Share on other sites

Wow this is weird, I just went to System > Configuration > KDE > Information > PCMCIA

 

It says:

Card 0: D-Link DWL-650

Card 1:

 

Which is correct I do have the DWL 650 in the card slot.

 

I am assuming that KDE is reading this information from a file somewhere that is not current or is not being updated.

Link to comment
Share on other sites

cardmgr can't see any pcmcia sockets so the problem is likely further up-stream. I'd check the following:

  • run lsmod to see if the pcmcia modules are loaded
  • run MCC->Hardware->Hardware - see if your socket(s) are detected
  • scan through the output of dmesg for pcmcia related messages

The boot messages ( run dmesg or less /var/log/messages) are still the best place to look for problems in hardware detection, module loading and intialization and service startup.

Link to comment
Share on other sites

:wall: well just for the heck of it (and to get windows to run again :screwy: and resize my partitions) I reinstalled xp and mdk 10. When I did everything worked wonderfully. This includes winxp, pcmcia, wireless, acpi. Then while compiling dri to get dri enabled for my video card the computer halted. I checked the system logs and it said that the computer had hit critical temp and was being halted (92'C) After bringing the computer back up and checking the logs the wireless was once again not working and kde > information > PCMCIA said that the card was inserted (this did not change when the card was removed and the computer rebooted.

 

I am running kernel 2.6.3-7mdk (came with official)

 

no change if booting when booting with linux acpi=off

 

This is really getting to me as I rely heavily on the wireless connection.

 

I checked harddrake and it says that the PCMCIA card controller is loaded and working (OZ6912) and I checked the syslogs for pcmcia and wireless card messages (orinoco_cs) and found:

 

pcmcia                 Starting PCMCIA services
pcmcia                 cardmgr[736]: no sockets found!
cardmgr[736]       no sockets found!
pcmcia                 Starting pcmcia: succeeded

 

I may be reverting to mdk 9.2 if this keeps up. (seems I cannot get pcmcia / wireless to work again unless I reinstall) :furious3:

Link to comment
Share on other sites

You probably ran into file system recheck problems after your machine crashed. Did you go through a log recovery on the reboot? Just guessing, but it may be that cardmgr depends on state information that was either destroyed/trashed when your machine went down, or is normally cleared when when cardmgr shuts down.

 

I would be interesting to look at /var/lib/pcmcia/stab - it should contain your current configuration information. You could try running cardmgr directly from the command line, adding the -v option to see if you get more useful messages;-- kill any currently running cardmgr before you do this. Also, take a look at /etc/sysconfig/pcmcia to see what options are passed to cardmgr (and the kernel modules) when /etc/init.d/pcmcia is run during the startup process.

 

Does your hardware regularly overheat? This might be the real problem. Unclean shutdowns are bad for any linux/unix system.

 

I should add that I'm not a laptop user so all I can offer you is debugging strategies. I hope it helps.

Link to comment
Share on other sites

I did go through the file system check on the next boot and nothing unusual was reported.

 

This was the first time that the computer was halted due to an overheat. (I had the machine on my lap ontop of a pillow -- blocking the exhaust on the bottom)

 

I appreciate your help with this.

Link to comment
Share on other sites

Could you post your /etc/sysconfig/pcmcia file here, plus /var/lib/pcmcia/stab (if it exists and I suspect it does)? Also, check for a /var/lib/pcmcia/sheme file and post the contents (it may be empty). Finally, post your /etc/pcmcia/config.opts file.

 

I suspect that removing the files /var/lib/pcmcia/stab, /var/lib/pcmcia/scheme and /var/run/cardmrg.pid MAY fix your pcmcia problems, but I haven't yet established a proof. It's a very tangled nest of programs calling scripts calling programs and not easy to trace (but very interesting).

 

I have established that the normal mode of operation is for the stab and cardmgr.pid files to be removed when the PCMCIA subsystem is shut down. When you system shutdown abnormally this would not have happened.

 

Finally, run lsmod and tell me if the following modules are loaded: pcmcia_core, ds and whatever the value (if any) of PCIC is in the /etc/sysconfig/pcmcia file.

Link to comment
Share on other sites

/etc/sysconfig/pcmcia:

 

PCIC=yenta_socket
PCMCIA=yes

 

/var/lib/pcmcia/sheme is empty (I have deleted and rebooted and there is no change)

 

/var/lib/pcmcia/stab doesnt exist

 

any ideas?

 

I ran pcic_probe and it says:

PCI bridge probe: O2 Micro 6912 foudn, 2 sockets.

Link to comment
Share on other sites

When cardmgr issues the message "no sockets found!" it does so because it failed to open a device file for major pcmcia minor socket-0. This happens only if no sockets have been registered by yenta_socket. Note that cardmgr itself creates the device files in /var/lib/pcmcia and immediately unlinks them so they can't be seen by any other process. The device-major number for pcmcia is dynamically assigned.

 

When pcmcia services are started, the /etc/init.d/pcmcia script first loads three kernel modules: pcmcia_core (core cardbus services), yenta_socket (cardbus controller chip interface) and ds (cardbus driver services). You would only get the "no sockets found!" message if pcmcia_core is loaded and registered: cardmgr has to look up the major number for its device file in /proc/devices (key "pcmcia") and you would have got a different message if the lookup had failed.

 

So the question is: did yenta_socket find and register any sockets? If it did you should see a /var/log/message in the form of "Yenta: CardBus bridge found at ..." followed by one or more "Socket status: ..." . If it failed, then you should see "No cardbus resource!".

 

yenta_socket should be seeing your hardware since pcic_probe, which is independent of the pcmcia kernel drivers can see it. Also, yenta_socket is the correct driver for your socket controller hardware (unless pcic_probe misidentified it which is not likely).

 

If you do find messages that indicate that yenta does see your sockets, one possible reason for cardmgr failing would be a race-condition -- cardmgr trying to open a node for a socket that hasn't yet been registered by yenta. One way to check for this would be to run cardmgr manually (as root) after system startup. If it works at that time, it would confirm the race-condition hypothesis. If not there is a most likely a bug in the pcmcia kernel modules implementation.

 

Good luck.

Link to comment
Share on other sites

OK this is wierd.

 

The current Install that I have is a MDK 9.2 fivestar (installed from CDs) then upgraded to 10.0 OE using a network boot floppy and installing from a mirror. So I still have the 2.4 kernel installed now. I have been booting with the 2.6 Kernel and cannot get the pc card to work after it just stopped working. Just now for the heck of it I booted from the 2.4 kernel and it gets to the finding new hardware part of the boot sequence and "beep... beep" the card is found. I dont know if maybe booting from the old kernel cleared a bad file out or what but I seem to be up for now.

 

 

Thanks ALOT Kjel

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
 Share

×
×
  • Create New...