This small patch-set fix the same endian issue in four different tests.
The issue is that in order to be endian-safe when working with GL_UNSIGNED_BYTE,
we need to compare results of individual bytes and not 32-bit values.
In addition, we need to construct the expected results as an array-of-bytes
and not a single 32-bit value.

This fixes the four tests in llvmpipe, softpipe and r600g on a POWER7
big-endian machine.

v2: I changed the initialization of the expected results to be more clear

Thanks,

   Oded

Oded Gabbay (4):
  pbo-read-argb8888: make it endian-safe
  fbo-pbo-readpixels-small: make it endian-safe
  pbo-readpixels-small: make it endian-safe
  GL_MESA_pack_invert: make it endian-safe

 tests/fbo/fbo-pbo-readpixels-small.c     | 27 +++++++++++++++++----------
 tests/general/pbo-read-argb8888.c        | 18 ++++++++++++------
 tests/general/pbo-readpixels-small.c     | 27 +++++++++++++++++----------
 tests/spec/mesa_pack_invert/readpixels.c | 12 ++++++------
 4 files changed, 52 insertions(+), 32 deletions(-)

-- 
2.5.5

_______________________________________________
Piglit mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/piglit

Reply via email to