I'd like to gather rough opinions about an idea to "simplify" pcmcia.
Our pcmcia stack in Solaris is rather complex, and I think a lot of this
owes to a desire to make it compliant with the quasi-portable Card
Services API/DDK (specified by the PCMCIA org, and largely meant for use
with other operating systems; intended to facilitate porting pccard
drivers, I think.)
I'd like to officially deprecate the Card Services API & DDK. In fact,
what I'd really like to do is "desupport" them entirely, so that no new
card services based drivers can be supported. This is probably not any
significant concern, since nobody (at least nobody sane!) is developing
new pc-card hardware or device drivers.
I firmly believe that the *entire* set of "interesting" pcmcia device
drivers are those that are already integrated into ON.
We can, through a variety of possible approaches, make it so that the
interfaces revert to Consolidation Private within ON, and actually
enforce this technologically (so drivers that are not participating
within ON will simply fail to load at modload time.)
Several folks have asked me about just nuking pcmcia altogether. I
personally don't think this is viable for several reasons:
1) CompactFlash adapters -- CF is still used in some cases, though
there are cheap USB readers
2) pcser is still used with modems. And not just 56K modems, but
also (and possibly more importantly) 3G wireless modems. While USB
alternatives exist, I think it is premature to force people to change
these out. (Cardbus and express card alternatives exist as well, but
are less common. Most of the products released during the early days of
3G were pcmcia.)
3) WiFi. There are (or were) quite a few PRISM 802.11b WLAN devices
on the market, which work with "pcwl". Also, the Cisco Aironet products
are supported by "pcan". Although, the reality is that these days there
are good 802.11g cardbus variants that are a better choice for most people.
The clincher IMO is the 3G and 56K modem issue. I think these products
are (or were until recently) still manufactured, as there is no
compelling reason to use Cardbus if you only need a few Mbps or less and
the logic for Cardbus is more complex than for 16-bit PCMCIA.
Anyway, I'm interested to hear what people think? Is there any need to
retain a public version of the Card Services API/DDK?
- Garrett
