On 08/03/2015 02:47 PM, Martin Peres wrote:
On 03/08/15 14:35, Tapani Pälli wrote:
On 08/03/2015 01:57 PM, Martin Peres wrote:
On 03/08/15 13:50, Tapani Pälli wrote:
On 08/03/2015 01:48 PM, Martin Peres wrote:
On 03/08/15 11:10, Tapani Pälli wrote:
_mesa_get_program_resource_name has logic to append '[0]' in name
if variable is an array, this should be skipped for XFB varyings
that have array index already appended.
Fixes:
ES31-CTS.program_interface_query.transform-feedback-types
Signed-off-by: Tapani Pälli <tapani.pa...@intel.com>
---
src/mesa/main/shader_query.cpp | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/src/mesa/main/shader_query.cpp
b/src/mesa/main/shader_query.cpp
index c1a7660..a50c348 100644
--- a/src/mesa/main/shader_query.cpp
+++ b/src/mesa/main/shader_query.cpp
@@ -726,6 +726,12 @@ _mesa_get_program_resource_name(struct
gl_shader_program *shProg,
bool add_index = !(((programInterface == GL_PROGRAM_INPUT) &&
res->StageReferences & (1 <<
MESA_SHADER_GEOMETRY)));
+ /* Transform feedback varyings have possible array index already
appended
+ * in their names.
+ */
Possibly have the array index already appended? So how do you detect
that here?
By 'possible array index' I mean that not all the XFB varyings are
arrays. Maybe better text would be just:
"Transform feedback varyings have array index already appended in
their names"
If it's not array, then of course it does not have it.
Yes, this comment makes a lot more sense. But I guess having this
distinction is the problem. How much work would XFB varyings working in
the same as any other resource when it comes to the name?
XFB resources point to 'gl_transform_feedback_varying_info' which gets
generated with names that come as user input from
glTransformFeedbackVaryings function (and should match what comes out
from glGetTransformFeedbackVarying). I don't see a straightforward way
to change this without quite a bit of changes in the XFB code so I
would like to have this distinction, it is a different sort of
resource type.
I see. Yeah, makes sense and the workaround is ok.
Reviewed-by: Martin Peres <martin.pe...@linux.intel.com>
Thanks. Now that I think about this more, the way how transform feedback
varying name works seems not consistent with the program_interface_query
spec and it could be we need to use the 'basename' after all (and then
apply [0] if array). I will take a bit more look in to this. There's
other places, for example GL_NAME_LENGTH query that would not work ok
either.
// Tapani
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
