R600: New control flow for SI v2

This patch replaces the control flow handling with a new
pass which structurize the graph before transforming it to
machine instruction. This has a couple of different advantages
and currently fixes 20 piglit tests without a single regression.

It is now a general purpose transformation that could be not
only be used for SI/R6xx, but also for other hardware
implementations that use a form of structurized control flow.

v2: further cleanup, fixes and documentation

Patch by: Christian König

Signed-off-by: Christian König <deathsimple@vodafone.de>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@170591 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Tom Stellard
2012-12-19 22:10:31 +00:00
parent 6eebe47060
commit 6b7d99d473
14 changed files with 1500 additions and 385 deletions

View File

@@ -25,13 +25,14 @@ FunctionPass* createR600KernelParametersPass(const DataLayout *TD);
FunctionPass *createR600ExpandSpecialInstrsPass(TargetMachine &tm);
// SI Passes
FunctionPass *createSIAnnotateControlFlowPass();
FunctionPass *createSIAssignInterpRegsPass(TargetMachine &tm);
FunctionPass *createSILowerControlFlowPass(TargetMachine &tm);
FunctionPass *createSICodeEmitterPass(formatted_raw_ostream &OS);
FunctionPass *createSILowerLiteralConstantsPass(TargetMachine &tm);
FunctionPass *createSIFixSGPRLivenessPass(TargetMachine &tm);
// Passes common to R600 and SI
Pass *createAMDGPUStructurizeCFGPass();
FunctionPass *createAMDGPUConvertToISAPass(TargetMachine &tm);
} // End namespace llvm