Currently, codegen may spent some time in SDISel passes even if an entire

function is successfully handled by fast-isel. That's because function
arguments are *always* handled by SDISel. Introduce FastLowerArguments to
allow each target to provide hook to handle formal argument lowering.

As a proof-of-concept, add ARMFastIsel::FastLowerArguments to handle
functions with 4 or fewer scalar integer (i8, i16, or i32) arguments. It
completely eliminates the need for SDISel for trivial functions.

rdar://13163905


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174855 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 10 insertions, 0 deletions

diff --git a/include/llvm/CodeGen/FastISel.h b/include/llvm/CodeGen/FastISel.h
index 05619050a3..705db7e643 100644
--- a/include/llvm/CodeGen/FastISel.h
+++ b/include/llvm/CodeGen/FastISel.h
@@ -90,6 +90,11 @@ public:
 
   /// getCurDebugLoc() - Return current debug location information.
   DebugLoc getCurDebugLoc() const { return DL; }
+  
+  /// LowerArguments - Do "fast" instruction selection for function arguments
+  /// and append machine instructions to the current block. Return true if
+  /// it is successful.
+  bool LowerArguments();
 
   /// SelectInstruction - Do "fast" instruction selection for the given
   /// LLVM IR instruction, and append generated machine instructions to
@@ -160,6 +165,11 @@ protected:
   ///
   virtual bool
   TargetSelectInstruction(const Instruction *I) = 0;
+  
+  /// FastLowerArguments - This method is called by target-independent code to
+  /// do target specific argument lowering. It returns true if it was
+  /// successful.
+  virtual bool FastLowerArguments();
 
   /// FastEmit_r - This method is called by target-independent code
   /// to request that an instruction with the given type and opcode