Re: [sdcc-devel] Automatic testing of SDCC

[Karl B. <ka...@tu...>]

Hi Scott,

I started writing some test code for MCS51 with the same idea in mind.
The MCS51 simulator bundled with SDCC can dump serial port output to
a file.  I wrote a few test functions
which would send out the serial port at the conclusion "OK" or "FAIL".

I couldn't find a good way to automate starting and stopping the s51.exe
simulator program.  It has abilities to control over a telnet like
connection but I had hoped to use a simple script file for this
purpose.  I tried some simple redirection, but it didn't work for me.
Perhaps I missed something, but otherwise my idea would be to add
script file support to s51.

Good luck on your PIC port, hope to see it in the code base soon!
Are you familiar with the Scenix processors?  Aren't they somewhat
PIC compatible?

Cheers,


Karl.


At 09:29 AM 11/21/2000 -0600, you wrote:
>
>I've created a very simple script that works with the PIC Port of SDCC, gpasm,
>and gpsim to automatically test the code generarted by SDCC. I'm not sure if it
>is useful for the '51 port or not. But this is how it works:
>
>The test .c programs are written such that they all have a function called
>`done()' and a variable called `success'. There will also be code in program to
>service the WDT. I'll use a Makefile to:
>
>1) compile the test program 
>2) Assemble it with gpasm 
>3) Invoke gpsim (the gnupic simulator) in command line mode and generate a log
>file.
>4) gpsim will also be invoked with a script that will do this:
>
>   load the program
>   set a break point at done
>   set a break point at the WDT timeout routine.
>   run the program
>   display the contents of the variable `success'
>
>5) The output of the log file produced by gpsim is grep'd for success like so:
>
># SDCC test script for the PIC Port
># 
>GPSIM_SCRIPT="s.stc"
>LOGFILE="r"
>echo "Test $GPSIM_SCRIPT" > $LOGFILE
>./gpsim --cli -c $GPSIM_SCRIPT $2> test
>grep "success " test | grep "0x0"
>if [ $? -eq 0 ] ; then
>  echo "PASSED" >> $LOGFILE
>else
>  echo "FAILED" >> $LOGFILE
>fi
>
>If the simulation was successful, then the value of success will be a zero
>(0x0) and PASSED will be written to the logfile.
>
>6) After the whole test is complete, the logfile can be grep'd again for FAILED
>to see which if any of the programs failed the test.
>
>
>Now the .c programs that are being tested have to be constructed in such a way
>to conform to the test. Each will have check it's own results if they fail, set
>the error in the variable `success'. Here's a contrived example:
>
>unsigned char uc0;
>unsigned char success=0;  //assume success
>
>void done(void)
>{
>  uc0 = 1;   // won't get executed...
>}
>
>void WDT(void)
>{
>
>  success |= 0x80;   // Set msb to indicate WDT timeout - in case we get stuck
>                     // in a loop
>
>  done();
>}
>
>void test_plus(unsigned char uc)
>{
>
>  uc0 += uc;
> 
>}
>
>
>main()
>{
>
>  uc0 = 5;
>
>  test_plus(6);
>
>  if(uc0 != 11)
>    success = FAILED_TEST1;
>
>
>}
>
>
>I'd appreciate any suggestions on whether this is  good idea or if it can be
>improved. Also, does anyone have some test code that I could borrow?
>
>
>Scott
>
>_______________________________________________
>sdcc-devel mailing list
>sdc...@li...
>http://lists.sourceforge.net/mailman/listinfo/sdcc-devel
>