On 09/10/2013 02:18 PM, Ian Romanick wrote:
On 09/10/2013 03:46 PM, Ian Romanick wrote:
From: Ian Romanick <[email protected]>
This is an OpenGL ES 1.x extension that Mesa has advertised for quite
some time. However, it appears that there is *no* code in Mesa to back
that up!
This test currently fails in all versions of Mesa.
There seem to be additional issues with -fbo on this test. Does that
work with ES1 tests?
Boo... Using -fbo with an ES1 test *should* work. If a context doesn't supports
fbos, the Piglit framework is supposed to gracefully to fallback to running the
test
without an fbo. the eglTerminate in the fallback path causes a segfault in
i965. That's
right: our driver dies horrifically if the user calls eglTerminate. Last time I
checked,
the the problem was that the intel_context gets deleted *after* the
intelScreen, despite
the context belongs to the screen.
I posted Piglit tests for this bug a long time ago, but they got ignored.
I'll repost them.
I see two ways to fix this: either fix eglTerminate, or rewrite how Piglit runs
fbo tests.
_______________________________________________
Piglit mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/piglit
