Re: interrupt programming

[Jan-Benedict Glaw <jbglaw@lug-owl.de>]

[-- Attachment #1: Type: text/plain, Size: 1838 bytes --]

On Mon, 2004-07-26 09:24:42 +0200, Sayang Oin <sayangoin@hotmail.com>
wrote in message <BAY15-F21ZXn97cvZvK00022be7@hotmail.com>:
> 
> I have to use timer because I have to communicate with other programm via 
> filestream
> and I use this timer also for scheduling the time that I have to send data 
> to my device.

That's quite DOSsish thinking. You've got several variants you can
choose from:

	1. Use non-blocking I/O with select() and toggle between your
	   two (or several) tasks.
	2. Separate all workplaces into one thread each.
	2b. Mix both variants.

> people said if I use interrupt so I can be sure I will not lose my datas.

You can't be *sure* to not loose data. Even with using interrupts, there
are possible conditions in which you'd loose some bytes of data due to
overflows. Think eg. about long-running IDE interrupts. While they're
running, no other IRQs will be serviced (unless you've got several
CPUs--it may be different in that case). That is, if eg. a 16byte FIFO
of a serial UART fills up, it'll overrun. At 115200 Baud, you need
little less than 1.5ms -- IDE interrupts can go up over 500ms easily, at
least with no DMA:)

> here is my routine
> ....
> 
>  while(1)
>  {
> 		res=select(fd+1, &rfds, NULL, NULL, &tv);

Hopefully you re-initialize tv before each call to select:)

>  return rd;
> }


But that's all not your problem. Your UART (or it's driver) thinks it's
not allowed to send out data. There's the problem.

MfG, JBG

-- 
Jan-Benedict Glaw       jbglaw@lug-owl.de    . +49-172-7608481             _ O _
"Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur | Gegen Krieg  _ _ O
 fuer einen Freien Staat voll Freier Bürger" | im Internet! |   im Irak!   O O O
ret = do_actions((curr | FREE_SPEECH) & ~(NEW_COPYRIGHT_LAW | DRM | TCPA));

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]