Ralf Baechle wrote:
> Could somebody of the DECstation people please check if the fix for
> sgiserial.c which I checked into CVS last night also is required for it?
While I'm not a DECstation guy, I've checked the changes it order to
ensure DECstation zs driver suffers from the problem. Why I'm care of
any changes in the serial drivers is the fact I observe the situation
where my Baget with the Zilog chip running 2.3.9 hangs somewhere in
getty and other Baget with the VAC chip not. 2.2.1 works fine.
Well, the zs driver seems to handle this case. Perhaps, we might remove
that unnecessary check in the zs driver too (hint!), but thinking the zs
driver full of other unnecessary things it's not so urgent.
Regarding the patch why to not call transmit_chars() instead of manually
feed the chip. transmit_chars makes that in the right way (and wakes up
others levels, if needed, btw).
Also, do you really need setup new value in the junk local variable just
before exit from rs_put_char :-).