Hiroaki Nagoya writes:
> at1700.c in standard Linux kernel (version 2.0.x) has a lot of
> problems(send many packet, hang the driver), and I hear that author of
> at1700.c (D. Becker) no longer maintain the driver.
Yes. Surely, it's broken.
As long as I'm concerned, it has race condition between interrupt
handler and send routine. I've sent patches to Don Becker once or
twice but it has not been included (yet). Upon requests, I've posted
the patch to fj.os.linux around 1993 or 1994 several times in Japan.
I believe that fmv18x.c is based on at1700.c with my patch (maybe with
out my patch, and indepentently fixed the race condition).
That's just a history.
> Therefore, I wrote another driver for AT1700 series which based on
> fmv18x.c (I hear that it is very stable). This driver source on the
> http://www.is.titech.ac.jp/~nagoya/at1700/ .
>
> This code tested by 7 people in Japan, no trouble at present (but I
> think some bugs survive in the code). Please comments or suggestions
> to me.
Good job, nice hack. :-) If you would like to get any information
about the card and the chip, please ask me. As I have a contact to
Allied Telesys, I could introduce you to them. They're quite open
acutually.
Thanks,
-- NIIBE Yutaka Mitsubishi Research Institute, Inc.