https://bugs.freedesktop.org/show_bug.cgi?id=64172
Zack Rusin <za...@vmware.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords|regression |
--- Comment #1 from Zack Rusin <za...@vmware.com> ---
Unfortunately, this is not a regression. We were over overflowing without the
patch, now we're simply correctly computing the overflow.
Mesa's transformfeedback.c:compute_transform_feedback_buffer_sizes
incorrectly computes size for gallium's so targets: in particular it does:
GLsizeiptr available_space
= buffer_size <= offset ? 0 : buffer_size - offset;
computed_size = available_space;
and returns that as the size of the buffer. Then when setting gallium stream
output targets it sets the computed_size as the new buffer size and
then it also sets the same offset it used when computing the computed_size. So
essentially for gallium:
buffer_size = buffer_size - offset
offset = offset
which means that gallium's stream output target's buffer size is always
'offset' less than it should be. I'm not going to have time to fix this right
now..
--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
