Hi Roland, look bellow.
> - Several of the tests are buggy. See the patch below at least.
you are absolutely right: the tests should be written without any bug,
we will check the issues you wrote in the patch and check that we don't
have
any more bugs in the tests.
> - It would be much more useful if the COMPARE() macro printed the
> expected and actual value on failure.
this macro will be changed to print the values in case of failure.
> - Similarly, other macros should probably also print more context.
> For example, in something like:
>
> CHECK_PTR("ibv_create_qp", qp[i], goto cleanup);
>
> I would probably want to know the value of i on failure.
will see how we can add debug prints in those cases.
> - I don't believe some of the tests are really valid. For example,
> the max number of QPs doesn't have to be precisely correct -- no
> valid app is going to depend on being able to create exactly that
> number of QPs and no more.
This is true, but in a black box testing, we wanted to check that there
isn't
any array overrun or any other bug when one tries to create more
resources than
the HCA/driver supports.
> - In any case, I'm not convinced that this sort of negative testing
> is the most valuable thing to focus on right now. I think it would
> be better to have regression tests of basic functionality (sends,
> receives, RDMA, CQ polling, etc) and stress tests before testing
> whether a buggy app will get the right error value when passing
> invalid parameters.
in this test (gen2_basic) we check many tests cases (good + bad flows)
to check the driver
(to check that the return value indicates that there is/isn't any error
and to check that
there isn't any seg fault or kernel crash).
We started to write several tests (basic functionality + stress tests)
and soon we will check in
the tests to the svn (and the changes that you suggested).
thank you, feedback is always welcome ...
Dotan
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
