>> Are the specifications for a winmodem a secret? If not, it
>> seems to me that a Linux winmodem driver would be a Simple
>> Matter Of Programming (tm).
>
> They're not widely available, but I don't think much effort has gone
> into making them available. Speaking purely from conjecture, I suppose
> that if the idea is to keep the cost down, they could well be using
> off-the-shelf DSP chips, and so we might not have too much trouble as
> long as they haven't rubbed off all the markings or potted the things.
They do not use off-the-shelf DSP chips. (maybe the MWave does)
Apple just used a D->A converter and everybody else seems to have
implemented an old-style modem that is _only_ a modem.
>> Maybe the issue is that you have to implement some huge modem
> protocol stack?
>
> I don't think that'll be necessary. They're fairly simple
> beasts - after all, it's just a tty, essentially.
It's not just a tty. Let's see, you have to implement:
LAPM, MNP4, MNP7, MNP10EC, v.42bis...
At the very least you'd need to properly refuse compression.
Go to http://www.nb.rockwell.com and do a search for RPI.
Also check http://www.rss.rockwell.com, and in particular:
http://www.rss.rockwell.com/techinfo/pc/resource_files/rpi_faq.html
Interesting stuff: the Windows driver uses 60kB to 100kB and can
sometimes beat modems with the standard hardware compression engine.
You can get uncompressed data going faster than 115.2kb/s because
serial port limits only apply to the compressed data.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu