[Galadrin]

Hi, sorry for my english

I try to get data from a serial port on /dev/ttyS*.
my serial port is open and all is good but when the data arrived on the input, i have te following message and my application down.

"I/O possible"

I have configured my serial port in asynchronous, VMIN = 100, VTIME=0.
have you any idea about my problem?

(my application get GPS data on this serial port)

thank's a lot for your answer

[wje_lf]

What language are you programming in? C?

[Galadrin]

my programmation language is the C, I have test on a linux Debian distibution and on the target, cross compillation with arm-linux-gcc.

my EDI is Eclipse

[wje_lf]

When you say you have the message "I/O possible", do you mean that your program is outputting that message? If so, it's because you put the code there to do that. So the next question is: how do you get to that point in your code? Can you show us a relevant piece of your code?

(To make it easy to read, when you've pasted it in your edit buffer, please highlight the code and click the octothorpe (the # icon) at the top of your edit window.)

But if your program is not itself outputting "I/O possible", I don't know what to say.

[Galadrin]

When you say you have the message "I/O possible", do you mean that your program is outputting that message? If so, it's because you put the code there to do that.

I don't say that from my program, in my program I have et simple

read (my_file_descriptor_not_NULL, My_buffer, 100);

the read on my file descriptor is not O_NONBLOCK so while i have no data in my serial port input, all is ok, I'm stoped.

but when I have data, I have the line "I/O possible" and the program is exited!!
I WOULD MY DATA lol, not the information about they are here

I put my code tomorow morning (I'm french, it's 10h pm).

[wje_lf]

The message "I/O possible" has to come from somewhere: if not in your C program, then in some library you call.

Please know that VMIN = 100, VTIME=0 means don't return until 100 bytes have been input. Is that your intent?

[Galadrin]

yes I would this fonctionnement, I receive a lot of GPS data from my receiver, and I don't have a continuous iteration of read for keep some times for my other thread (she print GPS statistic on the screen).

my issue is realy in the read return, I don't know why he kill my application after "I/O possible" print.... it's not documented

[wje_lf]

my issue is realy in the read return, I don't know why he kill my application after "I/O possible" print

If you write a small program which does nothing but this:

1. open the serial port
2. read one byte
3. print out that byte in hexadecimal ("%X\n" format)
4. exit your program

it would be interesting to see whether you get "I/O possible". It's almost certain you won't get that message.

Then input 100 bytes instead of one byte. See whether the message appears.

Continue to make the program more and more like your original program. When the "I/O possible" message appears after changing your program, it's probably your most recent change that causes that message to appear.

I know this isn't terribly helpful, but there's more going on here than just the read(), and this will help isolate the problem.

[Galadrin]

I have make a test of the errno value only but kill of my program => no print

I will make a little test following your plan, and if I have no result, I have a big issue

with my first test program, all is good, but I havn't touch the serial port part (open, read)

[rclark]

I am getting the same condition on a TS-7800 ARM board running Debian flavor of Linux. Simply doing a read() on an open port. I can transmit from it just fine ... but get an character in and the program aborts with the "I/O possible" error. Researching this now. Thinking there is a 'signal' being generated and not caught, but I am looking. Just letting you know you are no alone with this problem!