If you are making changes to µCsim you should both run tests to ensure nothing breaks and write new tests to ensure your changes work and do not get broken in the future!
In all cases the tests are run using the locally built simulators. i.e. you need to have completed a successful build before you can test it! Tests produce output when run which is captured to a file and compared with baseline output held in the source tree. If there are any differences they are output and the test is considered failed. In that case either the code needs to be changed to correct the output or the baseline needs to be updated to reflect the new expected output. Only when there are no differences to the baseline can tests be said to have passed.
To run all tests run make test
in the top-level
ucsim
directory.
E.g.:
.../ucsim$ make test
To run tests for a single MCU type run make test
in the directory for that simulator.
E.g.:
.../ucsim/s51.src$ make testor run
make
in the test
subdirectory..../ucsim/s51.src/test$ make
Note that tests for functionality generic to all simulators and that
does not depend on a particular MCU type lives in
sim.src/test
and should be run there. These tests require
at least one simulator to have been built but unless there is only one
it is indeterminate which simulator will be used.
To run a single test run make test_name
in the test
subdirectory of the relevant simulator. E.g.:
.../ucsim/s51.src/test$ make configIn this case the test is run verbosely and the commands executed by make are displayed. In all other cases only the final differences to baseline (if any) are displayed.
If changes to the simulator code base lead to changes in test outputs
and these changes are as expected then the baselines should be
updated to match the new output by copying the contents of the out
directories to the corresponding baseline
directories.
To update all baselines run make baseline
in the top-level
ucsim
directory.
To update baselines for a single MCU type run make baseline
in the directory for that simulator.
If you need to update the baseline for a single test you should copy the relevant files manually.
Remember that the updated baselines form part of your changes and need to be committed or submitted as part of your patch!
test
directory:
.cmd
files, .asm
or C
files etc, in your test directory:
baseline
directory in your test directory:
baseline
directory for each
output that your test generates:
Makefile
that lists the output(s) to be generated,
the recipes to create them and which includes test-lib.mk
:
out
directory
and then copy it to the baseline
directory:
.../ucsim/s51.src/test$ mkdir name .../ucsim/s51.src/test$ cd name .../ucsim/s51.src/test/name$ vim test.{cmd,asm} .../ucsim/s51.src/test/name$ mkdir baseline .../ucsim/s51.src/test/name$ touch baseline/stdout .../ucsim/s51.src/test/name$ cat > Makefile <<EOF OUTPUTS = stdout stdout: mytest.ihx mytest.cmd $(call run-sim) include test-lib.mk EOF .../ucsim/s51.src/test/name$ cd .. .../ucsim/s51.src/test$ make name .../ucsim/s51.src/test$ cp -r name/out name/baseline/.
test-lib.mk
takes care of building any .ihx
files
needed from corresponding .asm
or .c
files.
The call to run-sim
will invoke the simulator with the
.ihx
and .cmd
prerequisites, capturing the
output.
If there are no .cmd
prerequisites and there is no
-e
in the arguments to run-sim
it
will add a -g
argument to start the simulator running.
It is assumed there is at least one .ihx
file in the
prerequisites in this case and that it places code at the reset
address.
If you wish to pass arguments of your own to the simulator these
can be given as arguments to run-sim. This can be used to pass commands
with the -e
option rather than using a .cmd
file
for instance, as is done with the config tests.
If you create any other output from within the simulator you should
specify a pathname starting with out/
. Anything written
to the out
directory will be compared to the baseline
when the test is complete.