[llvm] [AMDGPU] Do not run GCNNSAReassign pass for GFX12 (PR #78185)

[Jay Foad via llvm-commits]

https://github.com/jayfoad updated https://github.com/llvm/llvm-project/pull/78185

>From 1eac82e1f04ed3036fcc62609bff86556d6cc93a Mon Sep 17 00:00:00 2001
From: Mirko Brkusanin <Mirko.Brkusanin at amd.com>
Date: Fri, 17 Mar 2023 17:50:39 +0100
Subject: [PATCH 1/3] [AMDGPU] Do not run GCNNSAReassign pass for GFX12

GFX12 does not have separate NSA and non-NSA encodings.
---
 llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp b/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp
index 4c9ad9b5bcf757..f87ff2d407e145 100644
--- a/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp
+++ b/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp
@@ -237,7 +237,7 @@ GCNNSAReassign::CheckNSA(const MachineInstr &MI, bool Fast) const {
 
 bool GCNNSAReassign::runOnMachineFunction(MachineFunction &MF) {
   ST = &MF.getSubtarget<GCNSubtarget>();
-  if (!ST->hasNSAEncoding())
+  if (!ST->hasNSAEncoding() || ST->getGeneration() > GCNSubtarget::GFX11)
     return false;
 
   MRI = &MF.getRegInfo();

>From e877830f8507ee71a571cc93914d1e2dde84c212 Mon Sep 17 00:00:00 2001
From: Jay Foad <jay.foad at amd.com>
Date: Wed, 17 Jan 2024 13:27:49 +0000
Subject: [PATCH 2/3] Add hasNonNSAEncoding

---
 llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp | 2 +-
 llvm/lib/Target/AMDGPU/GCNSubtarget.h     | 4 ++++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp b/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp
index f87ff2d407e145..272cc7fa6bc667 100644
--- a/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp
+++ b/llvm/lib/Target/AMDGPU/GCNNSAReassign.cpp
@@ -237,7 +237,7 @@ GCNNSAReassign::CheckNSA(const MachineInstr &MI, bool Fast) const {
 
 bool GCNNSAReassign::runOnMachineFunction(MachineFunction &MF) {
   ST = &MF.getSubtarget<GCNSubtarget>();
-  if (!ST->hasNSAEncoding() || ST->getGeneration() > GCNSubtarget::GFX11)
+  if (!ST->hasNSAEncoding() || !ST->hasNonNSAEncoding())
     return false;
 
   MRI = &MF.getRegInfo();
diff --git a/llvm/lib/Target/AMDGPU/GCNSubtarget.h b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
index 070d165cdaadb8..45d0d3919a42a5 100644
--- a/llvm/lib/Target/AMDGPU/GCNSubtarget.h
+++ b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
@@ -992,6 +992,10 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
 
   bool hasNSAEncoding() const { return HasNSAEncoding; }
 
+  bool hasNonNSAEncoding() const {
+    return getGeneration() < GFX12;
+  }
+
   bool hasPartialNSAEncoding() const { return HasPartialNSAEncoding; }
 
   unsigned getNSAMaxSize(bool HasSampler = false) const {

>From 675f04e7846d6a08f7f71df2803ecf758e0ef2d1 Mon Sep 17 00:00:00 2001
From: Jay Foad <jay.foad at amd.com>
Date: Wed, 17 Jan 2024 13:35:32 +0000
Subject: [PATCH 3/3] clang-format

---
 llvm/lib/Target/AMDGPU/GCNSubtarget.h | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/llvm/lib/Target/AMDGPU/GCNSubtarget.h b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
index 45d0d3919a42a5..290c77e5820d67 100644
--- a/llvm/lib/Target/AMDGPU/GCNSubtarget.h
+++ b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
@@ -992,9 +992,7 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
 
   bool hasNSAEncoding() const { return HasNSAEncoding; }
 
-  bool hasNonNSAEncoding() const {
-    return getGeneration() < GFX12;
-  }
+  bool hasNonNSAEncoding() const { return getGeneration() < GFX12; }
 
   bool hasPartialNSAEncoding() const { return HasPartialNSAEncoding; }