PCMCIA breakage

[Francois-Rene Rideau]

On Sun, May 02, 2004 at 06:43:28PM +0200, Oleg Gusev wrote:
> 1. with the old setup, the 3.3V CF cards simply don't work (ask Dave)
With your inverting setup, my network card doesn't work correctly on Socket 0.
I've experienced electrical problems on the ethernet in the past,
and the massive packet drop looks just like it.
Without your inverting setup, my CF doesn't work on Socket 1.
My conclusion is that there is possibly inversion
on Socket 1 but not on Socket 0.
At least, that's what's required to get things to work for me.

> 2. You can't insert the 3.3V-ONLY cards into the PCMCIA socket.
>     Mechanically impossible. Verified. Your card should be capable
>     of operating at 5V anyway.
I don't know. Maybe it only accepts 5V,
and the kernel just *thinks* it is 3.3V. How do I know?

> 3. I can prove that wince does the same setup as i do.
>     (modified version of hpcboot)
I can prove that my patch is necessary with my network card.

> PS. The old PCMCIA voltage config forces the 5V card to operate
>         at 3.3V
Or maybe lines are inverted and it forces 3.3V card to operate at 5V.
Or maybe what I'm doing is completely different.
In any case, please consider what my patch is doing, i.e.
de-inverting the bits you invert on Socket 0.
  state->state[0].vs_3v  = (status & PCSR_S0_VS1)          ? 1 : 0;
  state->state[0].vs_Xv  = (status & PCSR_S0_VS2)          ? 1 : 0;
  state->state[1].vs_3v  = (status & PCSR_S1_VS1)          ? 0 : 1;
  state->state[1].vs_Xv  = (status & PCSR_S1_VS2)          ? 0 : 1;
I don't understand what that does exactly, but without it,
either of my cards doesn't work.

> PPS. I don't have the dropped packets even under very heavy load.
Yes. It's definitely an electrical problem on the ethernet line.
I don't drop packets with my patch.

>          Please tell me if you have the same problem without the 16M RAM card
>          and with old cardmgr.
It doesn't depend on cardmgr.
I haven't tested on 16M RAM card yet, but I fail to see how it could affect
the issue, especially considering the nature of the "fix"
(which is mostly an undoing of part of a patch of yours,
between the moment when it works and the moment when it doesn't).

Cheers!

[ François-René ÐVB Rideau | Reflection&Cybernethics | http://fare.tunes.org ]
[  TUNES project for a Free Reflective Computing System  | http://tunes.org  ]
Always design a thing by considering it in its next larger context -- a chair
in a room, a room in a house, a house in an environment, an environment in a
city plan. -- Eliel Saarinen