[igt-dev] [PATCH i-g-t 2/4] tests/kms_frontbuffer_tracking: Improve tiling test coverage
Souza, Jose
jose.souza at intel.com
Wed Mar 11 00:00:06 UTC 2020
On Thu, 2020-02-06 at 14:00 +0200, Ville Syrjälä wrote:
> On Wed, Feb 05, 2020 at 06:09:42PM -0800, José Roberto de Souza
> wrote:
> > Lets add some more tiling tests, if tiling is supported by FBC draw
> > some rectangles and compare the CRC against benchmark if not
> > supported run the test to guarantee that FBC is disabled.
> >
> > This is a preparation for when kernel will allow FBC to be enabled
> > without fences, so we can better test linear and Y tiled
> > frontbuffers.
> >
> > Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > Signed-off-by: José Roberto de Souza <jose.souza at intel.com>
> > ---
> > tests/kms_frontbuffer_tracking.c | 50 +++++++++++++++++++++++++---
> > ----
> > 1 file changed, 39 insertions(+), 11 deletions(-)
> >
> > diff --git a/tests/kms_frontbuffer_tracking.c
> > b/tests/kms_frontbuffer_tracking.c
> > index 63b5d12d..d5a64a4d 100644
> > --- a/tests/kms_frontbuffer_tracking.c
> > +++ b/tests/kms_frontbuffer_tracking.c
> > @@ -2159,6 +2159,23 @@ static void format_draw_subtest(const struct
> > test_mode *t)
> > badformat_subtest(t);
> > }
> >
> > +static bool tiling_is_valid(int feature_flags, enum tile_type
> > tile)
> > +{
> > + if (!(feature_flags & FEATURE_FBC))
> > + return true;
> > +
> > + switch (tile) {
> > + case TILE_LINEAR:
> > + return false;
> > + case TILE_X:
> > + case TILE_Y:
> > + return true;
> > + default:
> > + igt_assert(false);
> > + return false;
> > + }
> > +}
> > +
> > /*
> > * slow_draw - sleep a little bit between drawing operations
> > *
> > @@ -2954,22 +2971,22 @@ static void stridechange_subtest(const
> > struct test_mode *t)
> > }
> >
> > /**
> > - * tilingchange - alternate between tiled and untiled in multiple
> > ways
> > + * tiling_disable_fbc_subtest - Check that tiling causes FBC to be
> > disabled
> > *
> > * METHOD
> > - * This test alternates between tiled and untiled frontbuffers
> > of the same
> > - * size and format through multiple different APIs: the page
> > flip IOCTL,
> > - * normal modesets and the plane APIs.
> > + * This test alternates between a FBC supported and non-
> > supported tiled
> > + * frontbuffers of the same size and format through multiple
> > different
> > + * APIs: the page flip IOCTL, normal modesets and the plane
> > APIs.
> > *
> > * EXPECTED RESULTS
> > - * FBC gets properly disabled for the untiled FB and reenabled
> > for the
> > - * tiled FB.
> > + * FBC gets properly disabled for the non-supported tiling and
> > reenabled for
> > + * the supported tiling.
> > *
> > * FAILURES
> > * Bad Kernels may somehow leave FBC enabled, which can cause
> > FIFO underruns
> > * that lead to CRC assertion failures.
> > */
> > -static void tilingchange_subtest(const struct test_mode *t)
> > +static void tiling_disable_fbc_subtest(const struct test_mode *t)
> > {
> > struct igt_fb new_fb, *old_fb;
> > struct modeset_params *params = pick_params(t);
> > @@ -2980,13 +2997,13 @@ static void tilingchange_subtest(const
> > struct test_mode *t)
> > old_fb = params->primary.fb;
> >
> > create_fb(t->format, params->primary.fb->width, params-
> > >primary.fb->height,
> > - LOCAL_DRM_FORMAT_MOD_NONE, t->plane, &new_fb);
> > + t->tile, t->plane, &new_fb);
> > fill_fb(&new_fb, COLOR_PRIM_BG);
> >
> > for (flip_type = 0; flip_type < FLIP_COUNT; flip_type++) {
> > igt_debug("Flip type: %d\n", flip_type);
> >
> > - /* Set a buffer with no tiling. */
> > + /* Set a buffer with new tiling. */
> > params->primary.fb = &new_fb;
> > page_flip_for_params(params, flip_type);
> > do_assertions(ASSERT_FBC_DISABLED);
> > @@ -3486,8 +3503,19 @@ igt_main_args("", long_options, help_str,
> > opt_handler, NULL)
> > igt_subtest_f("%s-stridechange",
> > feature_str(t.feature))
> > stridechange_subtest(&t);
> >
> > - igt_subtest_f("%s-tilingchange",
> > feature_str(t.feature))
> > - tilingchange_subtest(&t);
> > + for (t.tile = TILE_LINEAR; t.tile < TILE_COUNT;
> > + t.tile++) {
>
> Ah, I guess the enum is for this loop.
>
> s/tile/tiling/ whould be more customary language in general.
> Although, maybe people would then confuse it with the uapi
> definitions? Shrug.
Renaming to tiling.
>
> > + igt_subtest_f("%s-tiling-%s",
> > + feature_str(t.feature),
> > + tile_str(t.tile)) {
> > +
> > + if (tiling_is_valid(t.feature,
> > t.tile))
> > + draw_subtest(&t);
> > + else
> > + tiling_disable_fbc_subt
> > est(&t);
> > + }
> > + }
> > + t.tile = opt.tiling;
> > }
> >
> > if ((t.feature & FEATURE_PSR) || (t.feature &
> > FEATURE_DRRS))
> > --
> > 2.25.0
More information about the igt-dev
mailing list