tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: no ndis* at cardbus?

On Thursday 12 February 2009, John Nemeth wrote:
>      There is also PR/32938 with a suggested fix based on information
> found in "The PCMCIA Developer's Guide".  Charles shot it down claiming
> that the "PCMCIA spec" disagrees.  Both documents are published by
> PCMCIA, but the first is only a guide.  The last entry on that PR was
> on Oct. 29th, 2006.  It didn't go very far.

FWIW, I'm still using the fix with current.

The PR reallty contains two parts.

The first is a fix for the problem I was having with fixed IO addresses.  It 
enables the address remap from the rbus addresses to the card addresses for 
the cbb controller.  I do not think that Charles had an objection to this 
part, at least not in principle.

The second part was an attempt on my part to understand the PCMCIA specs and 
to rework some of the code based on those.  This lead me to tweak some other 
parts of the code nott directly related to the address remap issue.  The 
second update excluded cards with multiple address spaces per function.  This, 
quite correctly, Charles objected to.  He also comments that my set of test 
devices is too limited.  Again true.  Note, however, that his "There are three 
potential problems with the existing code" comments refer to the existing code 
(without the patch).  I submitted a third set of patches but I never heard 
bach from Charles.

By submitting the patch I was thinking the "maintainer" of the CardBus/PCMCIA 
code could take it as a suggestion for how the existing code could be improved 
or enhanced.  Christos asked me (privatly) if I thought the patch ought be 
applied as is and I told him not unless it was reviewed and verified by a 
developer familiar with the NetBSD PCMCIA code.

If someone wants to work on this I would be happy to discuss the issue in more 


Home | Main Index | Thread Index | Old Index