Red Hat Bugzilla – Bug 11174
Adaptec APA1480 Cardbus adapter nolonger loads under 6.2
Last modified: 2007-04-18 12:26:53 EDT
I have a FOSA 8800C 500Mhz Celeron Laptop. I have A 3Com 3c574 PCMCIA Network Card and a Adaptec APA1480 Cardbus SCSI card. Both cards worked fine under RedHat 6.1, but now under 6.2, I can't get the SCSI card to load. Keep Getting AIC7XXX IO ports already in use no matter what I try.
This is also occuring on my Micron Transport Trek 2 AGP laptop. EXACT SAME PROBLEM.
Same problem on Dell Latitude CPxH laptop. FIxed it by "demoting" the aic7xxx driver to version 5.1.23. After applying the current security patches, I was using kernel 2.2.16 which comes with driver version 5.1.30. Starting with the 5.1.30 patch, I applied successively older patches (in reverse) until the machine recognized the card. version 5.1.23 is the first one (or rather, the LAST one) which worked. Examining the changelog indicates that version 5.1.24 contains a change regarding PCI io port space and how it is reserved (which is exactly what is causing the card to fail). Seems happy as a clam now.
I need the contents of /proc/ioports on some machine that has this problem before I can go further with a diagnosis.
I sent this in email a week or so ago prior to discovering Bugzilla: Hi. I'm having a problem and I hear that you're the person to talk to. I have a brand new Dell Latitude CPx-H laptop which I have just installed RH6.2 on. I have performed the relevant recommended security updates as per the website, but other than that, it's a brand-spanking-new installation. I'm experiencing a problem with the Adaptec SlimSCSI 1480 PCMCIA controller. Essentially, when I insert the card it's detected and the aic7xxx driver is started (as a module) to configure it. The aic7xxx driver complains that the I/O ports are already in use and aborts the configuration. This results in /proc/scsi/scsi saying "Attached devices: none". Unfortunately, I do not see the I/O port conflict. The SCSI adapter asks for 0xa00-0xaff and should be able to get it. But the aic7xxx driver (after a call to "check_region()") thinks it's already allocated to something else. This problem also occurs if the card is plugged into the computer at boot time. I'm running kernel 2.2.16, with aic7xxx driver 5.1.30. Straight from the rpms. I would greatly appreciate it if you could help me out. I've done just about all I can think of to do without debugging this code with which I am unfamiliar. :) Thanks, Bryce Here's the relevant portion of /var/log/messages: ================================================= Aug 10 13:59:29 maggie cardmgr[508]: initializing socket 1 Aug 10 13:59:29 maggie cardmgr[508]: socket 1: Adaptec APA-1480 SCSI Host Adapter Aug 10 13:59:30 maggie cardmgr[508]: executing: 'insmod /lib/modules/2.2.16-3/pcmcia/cb_enabler.o' Aug 10 13:59:30 maggie cardmgr[508]: executing: 'insmod /lib/modules/2.2.16-3/pcmcia/apa1480_cb.o' Aug 10 13:59:30 maggie kernel: cs: cb_config(bus 35) Aug 10 13:59:30 maggie kernel: fn 0 bar 1: io 0xa00-0xaff Aug 10 13:59:30 maggie kernel: fn 0 bar 2: mem 0x60030000-0x60030fff Aug 10 13:59:30 maggie kernel: fn 0 rom: mem 0x60020000-0x6002ffff Aug 10 13:59:30 maggie kernel: apa1480_attach(bus 35, function 0) Aug 10 13:59:30 maggie kernel: aic7xxx: <Adaptec PCMCIA SCSI controller> at PCI 35/0/0 Aug 10 13:59:30 magAug 10 13:59:30 maggie kernel: apa1480_cs: no SCSI devices found Aug 10 13:59:30 maggie kernel: scsi : 0 hosts. Aug 10 13:59:30 maggie cardmgr[508]: get dev info on socket 1 failed: No such device Here's the reserved I/O ports (from /proc/ioports): =================================================== 0000-001f : dma1 0020-003f : pic1 0040-005f : timer 0060-006f : keyboard 0070-007f : rtc 0080-008f : dma page reg 00a0-00bf : pic2 00c0-00df : dma2 00f0-00ff : fpu 0170-0177 : ide1 01f0-01f7 : ide0 0376-0376 : ide1 03c0-03df : vga+ 03f6-03f6 : ide0 03f8-03ff : serial(auto) 0860-0867 : ide0 0868-086f : ide1 d800-d8ff : ESS Maestro 2E dc00-dc7f : eth0
I've heard reports that this is fixed already in the later pcmcia-cs package we are using in our next release. Can someone try the kernel from rawhide or from the public beta and let me know if the problem still exists?
Point me to a URL. I'm pretty ignorant and don't know what a "rawhide" kernel is. I went to the beta release site, but I'm not interested in doing a complete re-install with the next version of redhat. But if it's just a kernel recompile I'd be up for it. :)
Believed fixed, and 6.2 is out of support now