compiler: Turn on enough to invoke front end during XGL execution
Remove glassy compile checks, fix more things to run glsl
Consume glsl in the stack, store resulting IR
Fix up things enough to invoke our shader compiler with glsl, via
the real XGL calls. Store the result as a gl_shader_program which
contains the IR.
The test still consumes the ISA generated by hand, but it will
print out the IR generated by xglCreateShader. It will continue
to use hand ISA until we implement xglCreateGraphicsPipeline, which
will lower the IR to ISA using NOS.
diff --git a/icd/intel/compiler/shader/linker.cpp b/icd/intel/compiler/shader/linker.cpp
index 6f3b876..a55c7c6 100644
--- a/icd/intel/compiler/shader/linker.cpp
+++ b/icd/intel/compiler/shader/linker.cpp
@@ -74,9 +74,9 @@
#include "link_varyings.h"
#include "ir_optimization.h"
#include "ir_rvalue_visitor.h"
+#include "standalone_scaffolding.h" // LunarG ADD:
extern "C" {
-#include "main/shaderobj.h"
#include "main/enums.h"
}
@@ -1255,8 +1255,8 @@
"layout qualifiers for gl_FragCoord\n");
}
- /* Update the linked shader state. Note that uses_gl_fragcoord should
- * accumulate the results. The other values should replace. If there
+ /* Update the linked shader state. Note that uses_gl_fragcoord should
+ * accumulate the results. The other values should replace. If there
* are multiple redeclarations, all the fields except uses_gl_fragcoord
* are already known to be the same.
*/