Question: Partial card functionality, memory port probs? 

Forum: PCMCIA Installation and Configuration Issues
Date: 2000, Feb 26
From: Ben hoyt <ben@spectrumwireless.net>

First, vital statistics: Linux version 2.2.13 Linux PCMCIA Card Services 3.1.10 ... Intel i82365sl A step ISA-to-PCMCIA at port 0x3e0 ofs 0x00 host opts [0]: none host opts [1]: none ISA irqs (default) = 3,5,7,9,10,12,14,15 polling interval = 1000 ms cs: IO port probe 0x1000-0x17ff: excluding 0x11f0-0x1207 0x12f8-0x1307 0x1378-0x1387 0x13f0-0x13ff cs: IO port probe 0x0100-0x04ff: excluding 0x1f0-0x207 0x378-0x387 0x3f0-0x3ff cs: IO port probe 0x0a00-0x0aff: excluding 0xaf8-0xaff cs: memory probe 0x0d0000-0x0dffff: clean. And: swlr# cat /proc/bus/pccard/memory 000d0000-000d0fff : card services With that out of the way, I have an interesting problem with multiple cards. The hardware is an embedded system board with an onboard PCMCIA controller. The processor is an AMD Elan SC400 (http://www.amd.com/products/lpd/elan400/21026b.html) that has PCMCIA functionality 'built-in'. I've had 3c589_cs working on older kernel versions and older pcmcia-cs versions, but with my existing software, I have a problem running 3c589_cs, and more specifically Elmer Joandi's Aironet 4800 driver. I have never been able to make the Aironet 4800 card work on this hardware, no matter what software versions I've used. With both cards/drivers, the card is found, but the driver has a problem doing anything with it. I can configure it with an address, and try pinging, but I get errors, such as these with 3c589: eth1: 3Com 3c589, io 0x310, irq 3, hw_addr FF:FF:FF:F9:FF:FA 32K FIFO split 3:5 Rx:Tx, auto xcvr eth1: interrupt(s) dropped! eth1: interrupt(s) dropped! eth1: interrupt(s) dropped! eth1: interrupt(s) dropped! eth1: command 0x5800 did not complete! eth1: command 0x5800 did not complete! eth1: command 0x5800 did not complete! eth1: interrupt(s) dropped! .... and so on. Note the hw_addr - probably bad! This is the same thing that happens with the 4800 card, I get a bogus MAC address, and various error messages when I try to read/write to it. Pulling the card I'm using and trying to load the drivers produces the logical response - card not found. I tried loading X driver for Y card to see if that was a problem/possibility, and that produced the logical response as well - wrong driver for the card. It seems that the cards are being seen and identified, and I can "kind of" communicate with them, but not all the way. Memory addresses 0xd0000-0xdffff are shown free by the system, and the probe, and I have verified vigorously with the hardware manufacturer that this is the case. Anyone have any ideas?

[ Next-in-Thread ]  [ Next Message ] 

Messages Inline: [ 0 ]  [ 1 ] 

Maybe an IO port issue? Re: Question: Partial card functionality, memory port probs? (Ben hoyt) Date: 2000, Feb 26 From: David Hinds <dhinds@pcmcia.sourceforge.org>

I think I'd reclassify this as a development issue, just because it is an oddball platform... (and one that I didn't know worked at all until now) The first thought I have is that it would be very useful to determine which PCMCIA releases worked, and when problems started to appear. And compare the startup messages from the working configuration with the current one, if that is possible. I take it that you didn't need to use any out-of-the-ordinary settings with the older drivers? I don't know anything about Elmer Joandi's Aironet driver. The current PCMCIA release includes Ben Reed's driver for these cards. > eth1: 3Com 3c589, io 0x310, irq 3, hw_addr FF:FF:FF:F9:FF:FA My first guess here would be that maybe you're seeing an IO port conflict. The hw_addr is read using IO ports. -- Dave

More: IO Mapping Re: Maybe an IO port issue? (David Hinds) Date: 2000, Feb 28 From: Ben hoyt <ben@spectrumwireless.net>

OK, here is the info that I used to (in my mind) eliminate the possibility of an IO overlap, other than bugging the HW manufacturer. I may have missed something, so here it is: swlr# cat interrupts ioports CPU0 0: 25356934 XT-PIC timer 1: 0 XT-PIC keyboard 2: 0 XT-PIC cascade 4: 2107 XT-PIC serial 8: 2 XT-PIC rtc 11: 30250 XT-PIC SMC9194 13: 0 XT-PIC fpu NMI: 0 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 02f8-02ff : serial(auto) 0300-030f : SMC9194 03d4-03d5 : cga 03f8-03ff : serial(auto) swlr# insmod /lib/modules/pcmcia_core.o;insmod /lib/modules/i82365.o ignore=1 swlr# insmod /lib/modules/ds.o swlr# cat interrupts ioports CPU0 0: 25364560 XT-PIC timer 1: 0 XT-PIC keyboard 2: 0 XT-PIC cascade 4: 2107 XT-PIC serial 8: 2 XT-PIC rtc 11: 30697 XT-PIC SMC9194 13: 0 XT-PIC fpu NMI: 0 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 02f8-02ff : serial(auto) 0300-030f : SMC9194 03d4-03d5 : cga 03e0-03e1 : i82365 03f8-03ff : serial(auto) swlr# # Inserting Aironet modules. You can find them in the 2.3.newest kernel, I'm using the 2.2.13 version. swlr# cat interrupts CPU0 0: 25374716 XT-PIC timer 1: 0 XT-PIC keyboard 2: 0 XT-PIC cascade 3: 0 XT-PIC aironet4500_cs 4: 2107 XT-PIC serial 8: 2 XT-PIC rtc 11: 31301 XT-PIC SMC9194 13: 0 XT-PIC fpu NMI: 0 swlr# cat 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 0100-013f : aironet4500_cs 02f8-02ff : serial(auto) 0300-030f : SMC9194 03d4-03d5 : cga 03e0-03e1 : i82365 03f8-03ff : serial(auto) As you can see, I'm running at IRQ 3, IO 0x100. This is on my 'special' hardware. Now, the stats from a different HW platform running 2.0.37, and a slightly older pcmcia-cs: dts-dtn-rtr# cat interrupts ioports 0: 1026540671 timer 1: 8 keyboard 2: 0 cascade 3: 2552978658 aironet4500_cs 4: 10 + serial 8: 0 + rtc 11: 2409110548 eth0 13: 0 math error 15: 2 i82365 0000-001f : dma1 0020-003f : pic1 0040-005f : timer 0060-006f : keyboard 0070-007f : rtc 0080-009f : dma page reg 00a0-00bf : pic2 00c0-00df : dma2 00f0-00ff : npu 0100-013f : aironet4500_cs 03b0-03bb : mda 03bf-03bf : mda 03e0-03e1 : i82365 03f8-03ff : serial(set) ff80-ff9f : eth0 As you can see CS assigns the HW the same IO/IRQ for both platforms. Both platforms have been verified by the HW manufacturer to have no IO conflicts at these addresses. I'll let you guys puzzle over that. In the meantime, I'm going to upgrade to kernel 2.3.latest and see what happens re: these cards.

Not a kernel issue; maybe timing? Re: More: IO Mapping (Ben hoyt) Date: 2000, Feb 28 From: David Hinds <dhinds@pcmcia.sourceforge.org>

Ok, I'm a little confused about exactly what worked and what does not work. It sounds like the 3c589 card worked on this box with [some earlier driver version]. I'd like to see the kernel messages from that case. And both cards work on another box, but you don't really say how that box is different from the one that doesn't work. As for the potential for an IO overlap, you cannot tell much by looking at /proc. Those files show what resources Linux knows are in use. But you can have a conflict with a device Linux doesn't know about. Excluding resources in /etc/pcmcia/config.opts is an easy thing to try. I suspect it won't help, but you should try. I don't believe that changing kernels is likely to make any difference at all. Another thing to try is to load the pcmcia_core module with an option like "io_speed=300" to change the timing of IO accesses. -- Dave

Clarification Re: Not a kernel issue; maybe timing? (David Hinds) Date: 2000, Feb 28 From: Ben hoyt <ben@spectrumwireless.net>

To clarify what does and doesn't work: With an older kernel and pcmcia-cs, 3c589 was functioning on my 'new' hardware. With 2.2.12 and pcmcia-cs 3.1.10, 3c589 does not work. The Aironet driver that worked on all kernel/pcmcia versions on my 'old' hardware has not worked yet on the 'new' hardware, regardless of software versions. I've tried excluding, no luck. I just tested pcmcia_core with different io_speed parameters, including 300, no help. I'm going to revert back to something that does work on my new hardware - an older kernel and pcmcia-cs. Then, I'll move up from there, and hopefully identify if it is the different kernel version, or pcmcia-cs version that is making the difference.

[ Add Message ]  to: "Partial card functionality, memory port probs?"

[ Members ]  [ Subscribe ]  [ Admin Mode ] 
[ Show Frames ]  [ Help for HyperNews at pcmcia-cs.sourceforge.net 1.10 ]