[Intel-gfx] [PATCH i-g-t 2/7] tests/gem_mmap_gtt: Deal with tile sizes on gen2/3

[Ville Syrjälä]

On Tue, Dec 15, 2015 at 06:51:56PM -0200, Paulo Zanoni wrote:
> 2015-12-14 18:15 GMT-02:00  <ville.syrjala at linux.intel.com>:
> > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> >
> > Gen2/3 platforms have some unusual tile dimensions. Account
> > for them to make the test work correctly.
> >
> > Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > ---
> >  tests/gem_mmap_gtt.c | 20 +++++++++++++++++---
> >  1 file changed, 17 insertions(+), 3 deletions(-)
> >
> > diff --git a/tests/gem_mmap_gtt.c b/tests/gem_mmap_gtt.c
> > index 4d146426535b..b9c413b6160b 100644
> > --- a/tests/gem_mmap_gtt.c
> > +++ b/tests/gem_mmap_gtt.c
> > @@ -253,6 +253,18 @@ test_write_gtt(int fd)
> >         munmap(src, OBJECT_SIZE);
> >  }
> >
> > +static int tile_width(uint32_t devid, int tiling)
> > +{
> > +       if (intel_gen(devid) == 2)
> > +               return 128;
> > +       else if (tiling == I915_TILING_X)
> > +               return 512;
> > +       else if (IS_915(devid))
> > +               return 512;
> > +       else
> > +               return 128;
> > +}
> > +
> 
> Would this be useful: http://patchwork.freedesktop.org/patch/66846/ ?
> I mean, we could export some subfunction out of the original patch.

I suppose we might try to consolidate. One snag is that for display
stuff we want to use the modifiers, but I'm not sure we want to pollute
any fence stuff with those.

BTW your Yf width is not correct for 8/64 bpp (width*height != 4k).

> 
> >  static void
> >  test_huge_bo(int fd, int huge, int tiling)
> >  {
> > @@ -261,7 +273,8 @@ test_huge_bo(int fd, int huge, int tiling)
> >         char *tiled_pattern;
> >         char *linear_pattern;
> >         uint64_t size, last_offset;
> > -       int pitch = tiling == I915_TILING_Y ? 128 : 512;
> > +       uint32_t devid = intel_get_drm_devid(fd);
> > +       int pitch = tile_width(devid, tiling);
> >         int i;
> >
> >         switch (huge) {
> > @@ -327,6 +340,7 @@ test_huge_bo(int fd, int huge, int tiling)
> >  static void
> >  test_huge_copy(int fd, int huge, int tiling_a, int tiling_b)
> >  {
> > +       uint32_t devid = intel_get_drm_devid(fd);
> >         uint64_t huge_object_size, i;
> >         uint32_t bo, *pattern_a, *pattern_b;
> >         char *a, *b;
> > @@ -357,7 +371,7 @@ test_huge_copy(int fd, int huge, int tiling_a, int tiling_b)
> >
> >         bo = gem_create(fd, huge_object_size);
> >         if (tiling_a)
> > -               igt_require(__gem_set_tiling(fd, bo, tiling_a, tiling_a == I915_TILING_Y ? 128 : 512) == 0);
> > +               igt_require(__gem_set_tiling(fd, bo, tiling_a, tile_width(devid, tiling_a)) == 0);
> >         a = __gem_mmap__gtt(fd, bo, huge_object_size, PROT_READ | PROT_WRITE);
> >         igt_require(a);
> >         gem_close(fd, bo);
> > @@ -367,7 +381,7 @@ test_huge_copy(int fd, int huge, int tiling_a, int tiling_b)
> >
> >         bo = gem_create(fd, huge_object_size);
> >         if (tiling_b)
> > -               igt_require(__gem_set_tiling(fd, bo, tiling_b, tiling_b == I915_TILING_Y ? 128 : 512) == 0);
> > +               igt_require(__gem_set_tiling(fd, bo, tiling_b, tile_width(devid, tiling_b)) == 0);
> >         b = __gem_mmap__gtt(fd, bo, huge_object_size, PROT_READ | PROT_WRITE);
> >         igt_require(b);
> >         gem_close(fd, bo);
> > --
> > 2.4.10
> >
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx at lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/intel-gfx
> 
> 
> 
> -- 
> Paulo Zanoni

-- 
Ville Syrjälä
Intel OTC