-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Tuesday, August 19, 2003, 1:16:22 PM, Ognen wrote: > Can you elaborate on this statement? (not that I do not agree, I have a > bunch of pcmcia related problems myself...)
Certainly! Here are copies of the main messages I've posted looking for help. The first describes how I cannot upgrade my ThinkPad 770ED running RH 8.0 from the 2.4.18 kernel that works to the current crop of 2.4.20, due to PCMCIA problems. The second describes how I have failed in my efforts at building a 2.4.21 kernel from scratch that is capable of talking PCMCIA. The third message is a continuation of the 2.4.21 effort. At first, all I wanted was to be able to continue to upgrade to the latest RH supplied kernel. I'll still accept that, but now I want more! Now, I want to be able to build my own kernel. I *did* manage to build a kernel that boots on the 770ED and seems to access all of the hardware that the 2.4.18 kernel was, EXCEPT that it won't talk to the PCMCIA card. Since that's where my Ethernet and Modem are, it's kinda important to me. :-) Ron. -----BEGIN PGP SIGNATURE----- Version: PGP Personal Privacy 6.5.8 Comment: Until recently, the last PGP with full source disclosure. iQA/AwUBP0KZM28pw+2/9pUJEQL/tQCfWmUnl38cYMrYZB8W3HP9ntxAwsYAoMt0 /szP3S5s3RusNix9BKKl8W9o =0Cpx -----END PGP SIGNATURE----- ===== Message that began my quest: ===== -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello linux-thinkpad, I'm running RH 8.0 on my 770ED. The system sits on a dock where the floppy resides, along with a TDK CD-RW drive. On the dock's SCSI, there is a Syquest SyJet with no disk in the drive. In the PCMCIA slot, I have a 3Com/Megahertz 3CCFEM556 Ethernet/Modem card. Running kernel 2.4.18-27.0, and several others before that, things have been working pretty well. However, when I tried moving to the first 2.4.20 kernel update that RH released, I had trouble. After seeing some discussion here, it looked like it might have been caused by PCMCIA difficulties. It looked more like that once I successfully upgraded my A21p and three desktop systems (none of which use any PCMCIA cards). Now, I've tried upgrading to the latest 2.4.20-18.8, and my system is still locking up at boot time. I am assuming that this is a "panic", but since I can't remember seeing one on Linux before and since it goes on for enough lines to scroll off the top of the screen, who knows? In /var/log/messages, the last few lines from my 2.4.20-18.8 boot attempt are (without the date stamps): pcmcia: cardmgr. cardmgr[598]: starting, version is 3.1.31 rc: Starting pcmcia: succeeded cardmgr[598]: config error, file 'config' line 2129: no function bindings cardmgr[598]: watching 4 sockets cardmgr[598]: Card services release does not match Note that the complaint about line 2129 also happens in the 2.4.18 boot. Note that the "Card services release does not match" complaint also happens in the 2.4.18 boot. In the 2.4.18 boot, what follows are lines from kernel/cs relating to IO port probing, followed by the initialization of the Ethernet/Modem card. So, I guess I'm still suspicious of PCMCIA issues. Further evidence pointing at PCMCIA is that when I pulled the card and re-booted, the system came up fine (althouth, of course, not on the network). Where do I go from here??? Thanks! Ron. -----BEGIN PGP SIGNATURE----- Version: PGP Personal Privacy 6.5.8 Comment: Until recently, the last PGP with full source disclosure. iQA/AwUBPua2dm8pw+2/9pUJEQI7AACeKKQFURSVE3eXq8nBGYVWMC8sVZsAoOuD 7KQ3bTP4K8m9dW0qR1u7CILS =f2iu -----END PGP SIGNATURE----- ===== Message that began my kernel building sub-quest: ===== -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello psyche-list, Because of the problems I have been having with the 2.4.20 kernel updates for RH 8.0, to which no one seems to even be able to provide a hint on tracking down, I decided to have a go at grabbing the latest stable kernel from www.kernel.org. I downloaded 2.4.21 dated 2003-06-13, in hopes that whatever broke between .18 and .20 got fixed in .21. So, I extracted the kernel source onto a filesystem with a bunch of free space, as /data/kernel/linux-2.4.21, and followed the kernel build instructions. Since this is a notebook computer that is not likely to have any changes in sound card, processor, etc., I tried to set everything up that it had (that I use all the time) as "y" to include. Anything that I use only occasionally, I set up as "m", and anything that doesn't exist or never is used, I set up as "n". I've got six saved .config files at this point, but the first couple don't really count, since I didn't even *try* to boot them. I have attached rh5.config (rh for Ron Heiby, not Red Hat) to this message, as it seems to be the closest to working. The change going from this to rh6.config was to make CONFIG_PCMCIA and CONFIG_PCMCIA_3C574 "m", rather than "y". With those changes, the kernel would not build, with the final "ld" complaining about not being able to find drivers/char/pcmcia/pcmcia_char.o, and with me wondering why it would want to look for such a file. Anyway, back to the attached config. The system boots just fine, but no Ethernet. My Ethernet is a Megahertz PCMCIA combo card that runs with no problem on the 2.4.18-27.8.0 stock build, while causing all of the 2.4.20 builds to die horribly. Looking at /var/log/messages, the best clue seems to be that it is trying to "modprobe memory_cs", a module that does not exist. But, it does not exist in any of the kernels I installed from .rpm files, either. So, I'm obviously confused. I ran across <http://www.oreillynet.com/pub/faqs/linux_faq_AEN2325>, which seems to be telling me to turn off all the PCMCIA stuff in the kernel build, and to re-compile the "kernel-pcmcia-cs-3.1.31.9" package, pointing it at my new kernel source tree. While that seems to make some sense, I have to wonder why the kernel-pcmcia-cs package didn't have to be re-done for the 2.4.18 -> 2.4.20 transition. If I don't hear/find something that seems to make more sense by the time I get home from work, that's what I'll be trying next. Thanks! Ron. -----BEGIN PGP SIGNATURE----- Version: PGP Personal Privacy 6.5.8 Comment: Until recently, the last PGP with full source disclosure. iQA/AwUBPzOkfm8pw+2/9pUJEQKGKQCg2x6I8Gz5TsJBTfu4AXwSV20+lrAAoI8G 5zix3VaeTq3aXTtZuIRR10Co =Jcw4 -----END PGP SIGNATURE----- ===== Message that continued my kernel building sub-quest: ===== Monday, August 11, 2003, 9:37:38 PM, Randy wrote: > try renaming dir > /lib/modules/<new_kernel)/pcmcia Finally got a chance to try this. No joy. Here are the messages from /var/log/messages that seem related. Aug 15 23:23:03 rwh pcmcia: Starting PCMCIA services: Aug 15 23:23:03 rwh pcmcia: cardmgr. Aug 15 23:23:03 rwh cardmgr[522]: starting, version is 3.1.31 Aug 15 23:23:03 rwh rc: Starting pcmcia: succeeded Aug 15 23:23:03 rwh cardmgr[522]: config error, file 'config' line 2129: no function bindings Aug 15 23:23:03 rwh cardmgr[522]: watching 4 sockets Aug 15 23:23:03 rwh cardmgr[522]: Card Services release does not match Aug 15 23:23:03 rwh cardmgr[522]: initializing socket 1 Aug 15 23:23:03 rwh kernel: cs: memory probe 0x0c0000-0x0fffff: excluding 0xc0000-0xcbfff 0xf0000-0xfffff Aug 15 23:23:03 rwh cardmgr[522]: socket 1: Anonymous Memory Aug 15 23:23:03 rwh cardmgr[522]: executing: 'modprobe memory_cs' Aug 15 23:23:03 rwh cardmgr[522]: + modprobe: Can't locate module memory_cs Aug 15 23:23:03 rwh cardmgr[522]: modprobe exited with status 255 Aug 15 23:23:03 rwh cardmgr[522]: module /lib/modules/2.4.21.rh1/pcmcia/memory_cs.o not available Aug 15 23:23:05 rwh cardmgr[522]: get dev info on socket 1 failed: Resource temporarily unavailable Ron. ===== (end of song) ===== -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/redhat-list
