[Linux-kernel-mentees] [PATCH] kunit: fix: kunit_binary_assert_format() only prints signed int

Brendan Higgins brendanhiggins at google.com
Wed Aug 19 22:19:15 UTC 2020


On Wed, Aug 19, 2020 at 3:01 PM Vitor Massaru Iha <vitor at massaru.org> wrote:
>
> Hi Brendan,
>
> Actually https://patchwork.kernel.org/patch/11724995/
> solves my problem with unsigned int, since I can customize
> the Expected message with this patch.
>
> I think you can ignore this patch :)

Got it. Sounds good :-)

> On Wed, Aug 19, 2020 at 6:37 PM Vitor Massaru Iha <vitor at massaru.org> wrote:
> >
> > On Wed, Aug 19, 2020 at 6:30 PM Brendan Higgins
> > <brendanhiggins at google.com> wrote:
> > >
> > > On Wed, Aug 19, 2020 at 12:46 PM Vitor Massaru Iha <vitor at massaru.org> wrote:
> > > >
> > > > Some tests, such as overflow_kunit(), uses unsigned int,
> > > > But kunit_binary_assert_format() only prints signed int,
> > > > this commit also deals with the unsigned int print.
> >
> > Oops, Thanks!
> > I'll fix it.
> >
> > > >
> > > > Signed-off-by: Vitor Massaru Iha <vitor at massaru.org>
> > > > ---
> > > >  lib/kunit/assert.c | 15 +++++++++++++--
> > > >  1 file changed, 13 insertions(+), 2 deletions(-)
> > > >
> > > > diff --git a/lib/kunit/assert.c b/lib/kunit/assert.c
> > > > index 202f9fdeed0e..3ae90c09986a 100644
> > > > --- a/lib/kunit/assert.c
> > > > +++ b/lib/kunit/assert.c
> > > > @@ -104,12 +104,23 @@ void kunit_binary_assert_format(const struct kunit_assert *assert,
> > > >                                   binary_assert->left_text,
> > > >                                   binary_assert->operation,
> > > >                                   binary_assert->right_text);
> > > > -               string_stream_add(stream, KUNIT_SUBSUBTEST_INDENT "%s == %lld\n",
> > > > +
> > > > +               if (binary_assert->left_value - 1 < 0) {
> > > > +                       string_stream_add(stream, KUNIT_SUBSUBTEST_INDENT "%s == %lld\n",
> > > > +                                 binary_assert->left_text,
> > > > +                                 binary_assert->left_value);
> > > > +                       string_stream_add(stream, KUNIT_SUBSUBTEST_INDENT "%s == %lld",
> > > > +                                 binary_assert->right_text,
> > > > +                                 binary_assert->right_value);
> > > > +               }
> > > > +               else {
> > > > +                       string_stream_add(stream, KUNIT_SUBSUBTEST_INDENT "%s == %llu\n",
> > > >                                   binary_assert->left_text,
> > > >                                   binary_assert->left_value);
> > > > -               string_stream_add(stream, KUNIT_SUBSUBTEST_INDENT "%s == %lld",
> > > > +                       string_stream_add(stream, KUNIT_SUBSUBTEST_INDENT "%s == %llu",
> > > >                                   binary_assert->right_text,
> > > >                                   binary_assert->right_value);
> > > > +               }
> > >
> > > I agree that you found a bug here; however, I disagree that this is
> > > the correct fix. Given that the value is stored as a long long; isn't
> > > the value always stored as a signed value? So if the value overflows,
> > > won't it still not pass the check you have here?
> > >
> > > >         }
> > > >         kunit_assert_print_msg(assert, stream);
> > > >  }
> > > >
> > > > base-commit: d43c7fb05765152d4d4a39a8ef957c4ea14d8847
> > > > prerequisite-patch-id: bf4b0962b0b955e4e45f5d25fece889562118158
> > > > --
> > > > 2.26.2
> > > >


More information about the Linux-kernel-mentees mailing list