[llvm] 5f7a4d8 - [AArch64][GlobalISel] Perform load/store extended reg folding with optsize
Jessica Paquette via llvm-commits
llvm-commits at lists.llvm.org
Wed Feb 10 14:42:38 PST 2021
Author: Jessica Paquette
Date: 2021-02-10T14:42:25-08:00
New Revision: 5f7a4d8d05896c2d54a3d67cc3fd1fd81a84b2df
URL: https://github.com/llvm/llvm-project/commit/5f7a4d8d05896c2d54a3d67cc3fd1fd81a84b2df
DIFF: https://github.com/llvm/llvm-project/commit/5f7a4d8d05896c2d54a3d67cc3fd1fd81a84b2df.diff
LOG: [AArch64][GlobalISel] Perform load/store extended reg folding with optsize
GlobalISel was only doing this with minsize. SDAG does this with optsize.
(See: `SelectionDAG::shouldOptForSize()`)
This is a 0.3% code size improvement for CTMark at -Os.
(Best: 1.1% improvements on lencod + pairlocalalign)
Differential Revision: https://reviews.llvm.org/D96451
Added:
Modified:
llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp
llvm/test/CodeGen/AArch64/GlobalISel/load-addressing-modes.mir
Removed:
################################################################################
diff --git a/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp b/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp
index decf71fa8af0..bf44fa73e53d 100644
--- a/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp
+++ b/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp
@@ -5219,7 +5219,7 @@ bool AArch64InstructionSelector::isWorthFoldingIntoExtendedReg(
// Always fold if there is one use, or if we're optimizing for size.
Register DefReg = MI.getOperand(0).getReg();
if (MRI.hasOneNonDBGUse(DefReg) ||
- MI.getParent()->getParent()->getFunction().hasMinSize())
+ MI.getParent()->getParent()->getFunction().hasOptSize())
return true;
// It's better to avoid folding and recomputing shifts when we don't have a
diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/load-addressing-modes.mir b/llvm/test/CodeGen/AArch64/GlobalISel/load-addressing-modes.mir
index 8b8d66541469..91b681b9b9bb 100644
--- a/llvm/test/CodeGen/AArch64/GlobalISel/load-addressing-modes.mir
+++ b/llvm/test/CodeGen/AArch64/GlobalISel/load-addressing-modes.mir
@@ -23,7 +23,7 @@
define void @ldrhrox(i64* %addr) { ret void }
define void @ldbbrox(i64* %addr) { ret void }
define void @ldrqrox(i64* %addr) { ret void }
- attributes #0 = { optsize minsize }
+ attributes #0 = { optsize }
attributes #1 = { "target-features"="+lsl-fast" }
...
More information about the llvm-commits
mailing list