[AArch64] Disable by default MachineSink sink-and-fold

[momchil-velikov]

There is a report about a large compile time regression in V8 when generating debug info.

[llvmbot]

@llvm/pr-subscribers-backend-aarch64

Author: Momchil Velikov (momchil-velikov)

Changes

There is a report about a large compile time regression in V8 when generating debug info.

---

Full diff: https://github.com/llvm/llvm-project/pull/70101.diff

1 Files Affected:

• (modified) llvm/lib/Target/AArch64/AArch64TargetMachine.cpp (+1-1)

diff --git a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp
index fcc30a7cfceaf47..3d818c76bd4b7d7 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp
@@ -200,7 +200,7 @@ static cl::opt<bool> EnableGISelLoadStoreOptPostLegal(
 static cl::opt<bool>
     EnableSinkFold("aarch64-enable-sink-fold",
                    cl::desc("Enable sinking and folding of instruction copies"),
-                   cl::init(true), cl::Hidden);
+                   cl::init(false), cl::Hidden);
 
 extern "C" LLVM_EXTERNAL_VISIBILITY void LLVMInitializeAArch64Target() {
   // Register the target.

[bgra8]

Thanks @momchil-velikov ! My reducer is still running. I needed to trace back a few times as it went on a bad path.

I hope to have the reduced test case tomorrow.

[aeubanks]

@bgra8 any reduced test case?

We've bisected a size regression to this commit.

[aeubanks]

@bgra8 any reduced test case?

We've bisected a size regression to this commit.

Ah I see the discussion on #67432

[momchil-velikov]

@bgra8 any reduced test case?
We've bisected a size regression to this commit.

Ah I see the discussion on #67432

I'm working to re-enable it by default, after #72685