From Bugzilla Helper: User-Agent: Mozilla/4.77 [en] (X11; U; Linux 2.2.16-3smp i686) The APA1480 PCMCIA SCSI adapter is recognized but no module is apparently available to get it working. *BIG STEP BACK* wrt previous RedHat distributions, the new kernel based pcmcia stuff really sucks... Reproducible: Always Steps to Reproduce: 1.Boot the machine with an APA1480 (APA1480A in my case) PCMCIA card in 2.The card is recognized, the kernel looks for the driver apa1480_... and does not find it 3.all external SCSI hardware does not work of course Actual Results: The card is not configured Expected Results: The card was used to work flawlessly at least since RH6.1... The new kernel based PCMCIA drivers recognize the APA1480 but there is no module for it (the kernel looks for apa1480_cb). This is the remaining part of bug 30167 for Wolverine. This is really *BAD* since it prevents the use of all SCSI hardware on Laptops with these cards (which are among the few ones available SCSI Pcmcia cards) and it is a huge step back wrt previous RedHat releases. I know all polemics about the new PCMCIA stuff and I am puzzled RH is delivering it despite it still has vastly less hardware supported that the "old" external ones by D. Hinds. This is exactly the way to make people go away from Linux. They had a working system and now suddendly they cannot longer use the two external SCSI HD the SCSI dc-writer and the SCSI MO drive (this is my case). If they are brave enough they make some patch work on the RH kernel with the help of D. Hinds and they get the "old" PCMCIA stuff working again (I did...) What is puzzling me is that during installation, anaconda was perfectly able to configure my APA1480 loading among others aic7xxx and that with Ctrl+Alt+f1 going to console mode I was even able to mount my SCSI partitions on /mnt/sysimage/dev/sda... and look at them. So: a) is APA1480(A) support really missing or it is just a matter to correct the modules loaded by cardmgr? b) if the answer to a) is "yes it is still missing", when does RH plan to bring the new kernel PCMCIA drivers *AT LEAST* at the level of the previous ones?? Again, I stress it is a shame that the hardware supported is shrinking rather than the opposite, particularly in a field where not supporting a card means not supporting all drivers attached to it as well.
DHind's PCMCIA package doesn't work as well as the stock 2.4 code. In your case, it sounds like a config file bug; the apa1480_cb driver is not there as the functionality is provided by aic7xxx_mod.o and the intervention of cardmgr is NOT needed at all. (in fact, it breaks if you use it) Could you please try disabling cardmgr (with a killall cardmgr for all I care) and then "modprobe aic7xxx_mod" ?
I should clarify that. DHind's package only works well if you use it fully. As more and more cardbus drivers are written for native code, and native code has the same drivers (yes it has) using the native code is better for the long-term while getting the same support right now.
Thanks for the explaination! I was hoping in something similar. I'll make the test you suggest and let you know (but tomorrow european time. It's almost midnight and I have to (re)install the original RH7.1 kernel to do it, I am running one modified with the external PCMCIA drivers, alltogether as you correctly pointed out). Perhaps some hints for the users (and maybe a fix in kernel-pcmcia-cs) would help a lot in getting people acquainted with the new pcmcia drivers with this and maybe other cards. Looking on the supported hardware pages for apa1480 on the RH sites gives it as "community knowledge" and suggest to use qlogic_cs which I think does not make sense. A fix for the cardmgr config can help also because it is not easy to guess all ancillary drivers required for getting the card work. When anaconda got it working, lsmod gave: vfat 8784 0 msdos 5136 0 (unused) fat 31840 0 [vfat msdos] raid5 16816 0 (unused) xor 6096 0 [raid5] raid1 12448 0 (unused) raid0 3632 0 (unused) aic7xxx 107536 2 cb_enabler 2560 0 (unused) ds 6896 0 [cb_enabler] yenta_socket 9200 2 pcmcia_core 39680 0 [cb_enabler ds yenta_socket] keybdev 1968 0 (unused) hid 11168 0 (unused) input 3264 0 [keybdev hid] usb-uhci 19856 0 (unused) usbcore 48032 1 [hid usb-uhci] sr_mod 14144 0 (unused) sd_mod 10912 2 scsi_mod 55680 3 [aic7xxx sr_mod sd_mod] ide-cd 26368 0 cdrom 28288 0 [sr_mod ide-cd] I do not know which ones (cb_enabler?) are really needed or not.
cb_enabler isn't need for "new style" cardbus cards aic7xxx_mod is new-style. Once you verify/confirm the aic7xxx_mod works, I'll get the kernel-pcmcia-cs package fixed. ------ UK time ------ :)
Hi, I booted under kernel-2.4.2-2 with the RH7.1 kernel-pcmcia-cs. After the failure to find apa1480_cb I manually modprobed aic7xxx_mod ===> I can see my external SCSI hard drives, MO optical drive and cd-writer with no problem. I made a couple of fsck, measured the speed with hdparm (same as for the "old" drivers as expected) etc. It seems everything is working fine. Of course since it was done manually, the system does not react properly to a card removal or pcmcia shutdown, I was able to oops the machine twice doing these things but I assume they will get fixed as soon as the insertion will go thru the cardmgr or whatever else. So thanks a lot, and I am eagerly waiting for a patched kernel-pcmcia-cs package which deals correctly also with card swapping etc. If you need further tests let me know.
You should be able to get this working OK in the current setup just by editing /etc/pcmcia/config and change all occurences of apa1480_cb to aic7xxx_mod.
This change is in kernel-pcmcia-cs-3.1.24-3, which should be in the next rawhide release.