Commit Graph

3 Commits

Author SHA1 Message Date
Mike Reed
f19421961c don't rely on canvas->readPixels
Bug: skia:3216
Change-Id: I01a8d0083c79c0fe71fbc4d8cfdde4fa6e48b636
Reviewed-on: https://skia-review.googlesource.com/25741
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
2017-07-21 18:49:05 +00:00
Brian Osman
bcf65edf2c Bail out of test if MSAA surface creation fails
Bug: skia:
Change-Id: I44768314fad49fb0e094e30037a3fe159ebcbd0b
Reviewed-on: https://skia-review.googlesource.com/20044
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-15 18:29:56 +00:00
Brian Osman
eee3c09e96 Unit test to demonstrate Galaxy S6 MSAA readPixels bug
After lots of tinkering, this is the smallest test case I've
found to reproduce the bug. Frequently, bitmap b4 will be
cleared to blue, but not contain the oval drawn right before.
Bitmap b5 will contain the oval, so the diff of the two
bitmaps will detect the error.

There may be something else that can be removed, but I
haven't found it.

Interesting note:

In the original version, all surfaces were cleared to
black. When I started trying to change the clear colors to
figure out if we were getting an old surface, the bug went
away. In particular, the first clear color is irrelevant,
but the last three clears must all be the same color. If
any are different, the bug doesn't occur.

Bug: skia:6653
Change-Id: Iacafcc140b60594fab208e82987b0f37416975f3
Reviewed-on: https://skia-review.googlesource.com/19817
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
2017-06-15 17:58:08 +00:00