Re: [PATCH v3 3/3] mm/hmm/test: add self tests for HMM

[Qian Cai]

On Tue, 2019-10-29 at 17:58 +0000, Jason Gunthorpe wrote:
> On Wed, Oct 23, 2019 at 12:55:15PM -0700, Ralph Campbell wrote:
> > Add self tests for HMM.
> > 
> > Signed-off-by: Ralph Campbell <rcampbell@xxxxxxxxxx>
> > ---
> >  MAINTAINERS                            |    3 +
> >  drivers/char/Kconfig                   |   11 +
> >  drivers/char/Makefile                  |    1 +
> >  drivers/char/hmm_dmirror.c             | 1566 ++++++++++++++++++++++++
> >  include/Kbuild                         |    1 +
> >  include/uapi/linux/hmm_dmirror.h       |   74 ++
> >  tools/testing/selftests/vm/.gitignore  |    1 +
> >  tools/testing/selftests/vm/Makefile    |    3 +
> >  tools/testing/selftests/vm/config      |    3 +
> >  tools/testing/selftests/vm/hmm-tests.c | 1311 ++++++++++++++++++++
> >  tools/testing/selftests/vm/run_vmtests |   16 +
> >  tools/testing/selftests/vm/test_hmm.sh |   97 ++
> >  12 files changed, 3087 insertions(+)
> >  create mode 100644 drivers/char/hmm_dmirror.c
> >  create mode 100644 include/uapi/linux/hmm_dmirror.h
> >  create mode 100644 tools/testing/selftests/vm/hmm-tests.c
> >  create mode 100755 tools/testing/selftests/vm/test_hmm.sh
> 
> This is really big, it would be nice to get a comment from the various
> kernel testing folks if this approach makes sense with the test
> frameworks. Do we have other drivers that are only intended to be used
> by selftests?
> 
> Frankly, I'm not super excited about the idea of a 'test driver', it
> seems more logical for testing to have some way for a test harness to
> call hmm_range_fault() under various conditions and check the results?

Not a big fan of those selftests either. Could it be saner to use the new KUnit
framework for those instead?