On Tuesday 23 November 2004 04:34, patheve2 wrote:
Hello
Yes, I confirm that the infos I gave are the right ones for kernel 2.6.8 . For no confusion (right english word ?) I join the 3 files : serial.opts, config and config.opts. I used these for he following tests. Curious behaviour of the modem card, in particular the Com One which should work according this website : http://tuxmobil.org/pcmcia_ci10478.html So, with these below details, I hope that it will help you to find the reason why the modem cards are so "hard" to work properly :-)
Regards
Pascal
1/ case of the Com Oen card MC227 (modem) So, before pluggin :
- lsmod :
ds 21128 4 yenta_socket 21248 0 pcmcia_core 66188 2 ds,yenta_socket
And after pluggin :
- dmesg : cs: memory probe 0xf8a00000-0xfd1ff000: excluding 0xf8a00000-0xfa4f3fff 0xfadf0000-0xfb26dfff 0xfb6ec000-0xfd65dfff
this is *not* good. On my system serial_cs says which serial port it did assign to the card (I'm not at the system, so I cannot quote the exact message). Once it worked, it did not mention the memory ranges, which it did scan, anymore. This looks as if you insert the card, then cs gets started, scans the memory, but does not find anything. Thus the card does not get a serial port assigned. You need to work on the memory / port ranges, as long as there is no serial port assigned to your card by serial_cs. (If the card does not need a special driver, then of course serial_cs is irrelevant, but that driver would need to get loaded).
- more /var/log/messages under root login :
Nov 22 21:37:13 snidre kernel: cs: memory probe 0xf8a00000-0xfd1ff000: excluding 0xf8a00000-0xfa4f3fff 0xfadf0000-0xfb26dfff 0xfb6ec000-0xfd65dfff Nov 22 21:37:13 snidre cardmgr[2816]: socket 0: Serial or Modem
- lsmod : serial_cs 11176 1 ds 21128 5 serial_cs yenta_socket 21248 1 pcmcia_core 66188 3 serial_cs,ds,yenta_socket
- ls -al /dev/mod* :
lrwxrwxrwx 1 root root 9 Nov 22 21:37 /dev/modem -> /dev/cua4
2/ case of the Hayes Optima288 card (modem) :
- more /var/log/messages under root login (not enough data with only dmesg) :
Nov 22 21:44:28 snidre cardmgr[2816]: socket 0: Hayes Optima 288 Nov 22 21:44:29 snidre kernel: ttyS5 at I/O 0xd000 (irq = 9) is a 16450
Here ttyS5 seems to get assigned. I wonder why dmesg does not show it. Is there really no mention of ttyS5 in dmesg?
- lsmod :
serial_cs 11176 1 ds 21128 5 serial_cs yenta_socket 21248 1 pcmcia_core 66188 3 serial_cs,ds,yenta_socket
- ls -al /dev/mod* :
lrwxrwxrwx 1 root root 9 Nov 22 21:44 /dev/modem -> /dev/cua5
I do not think that /dev/cua5 is the good device. Try to make a link /dev/modem -> /dev/ttyS5 instead, and try to talk to the modem (minicom should be OK, or dial with kppp) If this does not work, then your back at the problem that serial_cs does not read your card correctly. Sorry, in a rush, the car is leaving. HTH, Matt
3/ case of the ethernet NetGear FA411 card :
[snip]
=============================
Hello all
Are you ready to read this very detailed and technical mail :-) ?
Hi Pascal,
first of all I can only help you from my experience with a 2.6 kernels, and not with 2.4 kernels. I did not bother to go back to a 2.4 kernel for my AMD64 notebook.
I will go through your description more detailled later today, I have to leave now for a few hours.
But in the meantime please confirm that the info you give below (configs etc.) are the ones used with the 2.6 kernels.
Also I do see a dmesg output from *before* the card is plugged in ( during boot without PCMCIA card)", but I do not see it *after* the card is plugged in. This is however very useful to see how serial_cs or pcnet_cs accepted and configured your card, if recognized at all. From my experience this is the critical step.
So please add the dmesg output *after* you inserted the card(s), if possible with a 2.6 kernel.
Once that step was fine on my system, it simply worked, no further fiddling was necessary.
HTH, Matt
=============================
Hello all
Are you ready to read this very detailed and technical mail :-) ?
Pascal, I see only now that you have posted replies in the older thread about this same subject. In the following I will comment on the post from you about the 2.6.8 kernel.
I do also understand that the ethernet card works now, and it is only the modem card which does not work yet. Is that correct?
On Sunday 21 November 2004 16:35, patheve2 wrote:
Hereafter the results about PCMCIA cards with the kernel 2.6.8-24.3 (see the previous mail for kernel 2.4.21)
---- 3rd step : case of the FA411 card with the kernel 2.6.8-24.3
[snip details, as this part seems to work]
---- 4th step : case of the Hayes modem card with the kernel 2.6.8-24.3
- cardctl ident : Socket 0: product info: "HAYES", "OPT288", "YG01 GO", "V7.171;A" manfid: 0x010a, 0x0000 function: 2 (serial)
good.
Socket 1: no product info available
- cardctl status : Socket 0: 5V 16-bit PC Card function 0: [ready] Socket 1: no card
- cardctl config : Socket 0: Vcc 5.0V Vpp1 0.0V Vpp2 0.0V interface type is "memory and I/O" irq 9 [exclusive] [level] speaker output is enabled function 0: config base 0x0100 option 0x62 status 0x08 io 0xd000-0xd007 [8bit]
- to force the loading of the right module, I added in the config file :
# insertion le 04-07-2004 card "Hayes Optima 288" version "HAYES", "OPT288", "YG01 GO" bind "serial_cs" #
This step is surely not wrong, however as the card had been recognized (see cardctl ident ) as "function: 2 (serial)" I assume that serial_cs would get loaded anyway.
As you have done the configuration of ports and memory range already, now the next step is to see if serial_cs does indeed find your card and is able to configure it.
This is easy, just do a dmesg right after you plugged tee card in. Then you see - if serial_cs did get loaded, and - how serial_cs did configure the card.
Please do that dmesg after inserting the card, and post the output.
- lsmod gives :
serial_cs 11176 1 ds 21128 5 serial_cs yenta_socket 21248 1 pcmcia_core 66188 3 serial_cs,ds,yenta_socket
- in /dev/, the symbolic link /dev/modem -> /dev/cua5 is created, instead of /dev/ttyS2
The dmesg output ( see before) probably shows here that serial_cs did configure the card as /dev/cua5
As far as I remember this is bad, it did happen when I did work on that problem for me as long as I had a kernel < 2.6.7, and all efforts to communicate with that device failed.
But lets see the dmesg output first, it might give us a hint.
Same goes for the Com One modem card below.
Regards, Matt