Am Mittwoch, den 07.04.2010, 13:31 +0200 schrieb Philippe De Muyter: > On Tue, Apr 06, 2010 at 10:45:51AM +1200, Gavin Lambert wrote: > > Quoth Andrew Kohlsmith: > > > > I would hate to try and do it from software with an interrupt. > > > > Yuck. > > > > > > It's not that bad. :-) IRQ on transmit holding or shift register > > > empty (ideally the latter). If you have only a THRE type interrupt > > > > > then you have to set a timer for 1 word time (usually 9 bits) and > > > after that timer, disable the transmitter. Of course when you go > > > to send data again you have to make sure you turn on the > > > transmitter. > > > > We've gone that route, for a master device (ie. other hardware will > > start transmitting back on the half-duplex channel almost > > immediately after the end of transmission) and had no end of trouble > > with getting the interrupt processed in time to avoid corrupting the > > reply. And this was on a board that wasn't even running an OS, so > > demands were relatively light. Sure, it worked fine *most* of the > > time, but that's not good enough. :) (Of course, it also had a > > pretty low-spec CPU, which didn't help.) > > > > We eventually switched to a UART that had hardware control of that > > I have just discovered that there are now 'standard' linux ioctls to > work with rs485 lines : TIOCGRS485 & TIOCSRS485. > > (see > http://git.kernel.org/?p=linux/kernel/git/gerg/m68knommu.git;a=commitdiff;h=c26c56c0f40e200e61d1390629c806f6adaffbcc) > > Of course, not all hardware drivers implement them. > > And also, I do not know how to use them. > > Philippe
I found this example that seems to use those new ioctls: http://foxlx.acmesystems.it/?id=28 If you look into this example, it doesn't look promising because RTS is released by a call from userspace after sending. I can't see much advantage over using tcdrain() and releasing RTS afterwards, if tx_empty is implemented correctly by the serial driver. -Erwin _______________________________________________ uClinux-dev mailing list uClinux-dev@uclinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev@uclinux.org To unsubscribe see: http://mailman.uclinux.org/mailman/options/uclinux-dev
