On Mon, 2008-12-22 at 16:59 +0100, Wolfram Sang wrote: > Hello Chris, > > On Sun, Dec 21, 2008 at 08:50:04PM +1100, Chris Gibson wrote: > > Two and a half weeks ago I submitted a patch to implement RS485 auto > > direction control on either (or both) the RTS and the DTR lines with > > varying pre and post delay settings that would work for _all_ uarts that > > support mctrl. > > Just observing the case, I assume that this concern about software RS485 > is still valid: > > http://article.gmane.org/gmane.linux.serial/2568 Thank you for the reply. In answer to the above thread - the implementation in the patch I provided addresses some of the concerned mentioned. The polling of the shift empty register is done with a timer, so the loss of CPU time is minimal. The ioctl's provide a mechanism for which the driver turnaround delay, and jitter can be established, allowing a user application the ability to determine the suitability of a particular direction control implementation to fulfil its timing requirements. The short fall of the implementation is that it's timing is only as good as the Linux system timer. This is less of a limitation with the latter kernels with higher speed system timers. My main concern though is not whether or not this software implementation of auto direction control is adopted into the kernel, (though I guess I would feel proud if it was :) but rather that the ioctl that has gone into V2.6.28-rc1 is only part way toward what it should be, and this is starting to be used already. best regards -- Christopher Gibson <chris@xxxxxxxxxxxxxxxx> -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
