GTA01 shared UART vs. flow control (bug #788)

Werner Almesberger werner at
Tue Jan 22 19:41:24 CET 2008

Mike Westerhof wrote:
> I'm right now imagining the response from RMK to a patch submitted
> upstream that drops characters on the floor whenever hardware flow
> control is asserted on the console port,

I guess he would be rightly annoyed about such a hack ;-)

However, that's not what I'm proposing. First of all, the dropping
should start only after a reasonably amount of time has been spent
waiting, and second, the definition of "reasonable" should be up to
the user, i.e., it should be configurable between "very short" and
infinite. For compatibility, the default should be infinite.

> the justification being to solve a hardware problem in a single,
> limited-run, pre-production special-purpose device.

That would be sad indeed. Fortunately, the possible real-life use
cases go well beyond this. The same patch would also prevent the
kernel from getting stuck due a wayward terminal, cable, etc., on
all other 24xx platforms as well.

So, short of introducing a mechanism to prevent getting stuck in
printk in general, this already provides the most general solution
to this type of problem.

> Actually this issue is now being discussed, so I'm not offended :P

Excellent :-) I'll get to your next post in a bit ...

- Werner

More information about the openmoko-kernel mailing list