WIFI initialization causes system stall
 

Hi,

I have big problem with my a600 with speedtouch 110 wi-fi card. EVERY attempt to initialize network stack causes system hang/stall (only Ctrl+A+A or hard reset helps).

This very same card with this very same network stack and this very same system (well, started from this very same partition) initializes and runs fine on my a1200.

Funny thing is, about month ago I was able to surf the net from this a600 with this wi-fi card setup; only system software was different, and then I've accidentally damaged system partition. After fresh system install every wi-fi initialization attempt stalls a600 (but works, with same system software, on a1200).


This broken system partition had OS3.9, but now a600+wi-fi stalls both in 3.1 and in 3.9.

And, in this a600 i have aca630 installed; I was thinking at first that maybe I have aca630-PCMCIA problem, but wi-fi stalls even with aca630 totally disconnected (jumpered off).

Network stack is AmiTCP 4.0 demo at the moment, but same effect with Genesis on OS3.9. And even networkbootdisk.adf stalls (on a600, on a1200 all works fine).


Any ideas? What freaking lib/patch could I have on this unlucky losted system partition, that would help start wi-fi in a600? Maybe some sort of lib/patch for 68060 (my a1200 has 060 card, and I still am using same hard disk in both machines...). Or is my a600 physically damaged? Should I check with another wi-fi card, another PCMCIA card (e.g. CF adapter)?


Thanks in advance,
Regards,
PPascal

Cardpatch is in use?

What rkauer said.
I had same problem in 3 different A600 revision mobos.

CardReset is not needed but CardPatch IS! Just download it from Aminet and put before SetPatch declaration the following lines:

FailAt 11
C:CardPatch

and you're set :)

I wish... :-). Stil no luck, even with cardpatch in s-s (it was there before, but now I moved it before SetPatch).

And card itself is working in 1200 and PC, and card_info program (from cnetdevice.lha) even reads some info from card. But amitcp:bin/online, bin/ifconfig or any other network initialization program hangs with whole machine (even fdd stops clicking).

Month ago it worked flawlessly; soon I will have some other pcmcia cards test (wifi, cable and CF), for now I am considering some weird software incompatibility or damaged a600 pcmcia port.

So, I am still open to any diagnostic ideas :-).

BTW, snoopdos goes till 'card.resource' (or similar) and then all hangs.

Today i returned to this weird a600 behaviour.

Made some more tests, buyed some more a600 boards... And all (3) have the same problem; I don't believe that I either fried all boards immediately or got all three bad boards -- pcmcia port is like least used port in a600, why should it broke often?

But any access to pcmcia card still freezes my little amigas; today I tried with numerous system configurations, witch cardreset/cardpatch and without, with 3.1 and 3.9; I have now 3 a600 boards, 2 supported WIFI cards (both working in a1200) and 1 cnet card (also working in a1200)... and all them fail to initialize :(

Ah, all boards have ROM 3.1 installed, if that matters. I have no CD-ROM, so system is installed either from 3.1 floppies (on real gear), or from 3.9 CD in WinUAE -- if that matters.

When i try run WiFiSignal tool, in snoopdos it goes through many failed libraries opening attempt (prometheus.library, powerpci.library itd.) and then, when it comes to pccard.library, I have this:

121 ramlib *Open LIBS:pccard.library Read OK
122 ramlib #FINDINPUT 1009099D,1000361A,1009179 BOOT OK
123 ramlib #END 40245e80 BOOT OK

And these are last lines in snoopdos -- then wifi led starts to blink and system freezes.
Please, any ideas what it means? Or any ideas how can I debug this issue deeper?

Regards,
PPascal

That card needs a firmware file, which is installed in DEVS:Firmware.

OK, I have this firmware file in this folder. And then? Should I do something with it, or prism2 driver will load it automatically?


Regards,

The driver will load it automatically every time the driver's opened. I presume it wasn't there before.

To sum it up:

Wow... After 6 months of fiddling with this setup, I just discovered the very (an very unexpected) root of the problem. Turns out, that with ACA630 whole system is much pickier on the CF card used as harddrive.

I have dozen of these, and most of them are high-class UDMA SLC CF cards (very fast,UDMA mode, I used them in PC as home-made SSDs). In my desperate tries, I started installing OS3.9 on every card I have, and EUREKA: on some of them, initialization of PCMCIA device doesn't stall whole system!! And, on cheapest CF card I bought two days ago, my a600+ACA630 was in Internet again, after all this time! :-)

Turns out, ROM version is in fact irrelevant (I've tested all available >37.299 versions). PCMCIA Network card brand is almost irrelevant (I've tested 4 different cards). A600 mainboard revision is irrelevant (and I have tested 4 different boards!). CF-IDE adapter brand is irrelevant (well, I've tested only two). ACATUNE setting is irrelevant. Prism firmware was there, but it also was irrelevant. ONLY "system disk CF" type matters! :-)

P.S. On some of my CF cards I've tested lately, PCMCIA initialization did not stall system at all, but every transfer from PCMCIA was broken -- e.g. my amiga gave my wireless router MAC address that was different from actual card MAC by one or two random digits (obviously network couldn't work in this case), and CF card in PCMCIA adapter,when prepared as HD, was recognized with errors (e.g. Sajdisk instead of Sandisk).

Well, I won this fight, but it was truly exhausting... Thank for your help and encouragement.

For the record, partial list of tested CF cards/real HDDs:
Delkin CompactFlash PRO UDMA -- PCMCIA stalls whole system
Lexar Professional UDMA -- PCMCIA stalls whole system
No name UDMA-marked CF card -- PCMCIA stalls whole system
Kingston Ultimate 266x -- PCMCIA unreadable
Sandisk Extreme III -- PCMCIA unreadable

Real IDE HDD -- PCMCIAFULLY WORKING :-)
cheap Kingston CF (green, with some bright flower) -- PCMCIA FULLY WORKING :-)