[AMDGPU] Add .entry_point back into PAL metadata

llvm/lib/Target/AMDGPU/Utils/AMDGPUPALMetadata.cpp

@@ -27,6 +27,28 @@
 using namespace llvm;
 using namespace llvm::AMDGPU;
 
+// Return the PAL metadata hardware shader stage name.
+static const char *getStageName(CallingConv::ID CC) {
+  switch (CC) {
+  case CallingConv::AMDGPU_PS:
+    return ".ps";
+  case CallingConv::AMDGPU_VS:
+    return ".vs";
+  case CallingConv::AMDGPU_GS:
+    return ".gs";
+  case CallingConv::AMDGPU_ES:
+    return ".es";
+  case CallingConv::AMDGPU_HS:
+    return ".hs";
+  case CallingConv::AMDGPU_LS:
+    return ".ls";
+  case CallingConv::AMDGPU_Gfx:
+    llvm_unreachable("Callable shader has no hardware stage");
+  default:
+    return ".cs";
+  }
+}
+
 // Read the PAL metadata from IR metadata, where it was put by the frontend.
 void AMDGPUPALMetadata::readFromIR(Module &M) {
   auto *NamedMD = M.getNamedMetadata("amdgpu.pal.metadata.msgpack");
@@ -232,8 +254,18 @@ void AMDGPUPALMetadata::setEntryPoint(unsigned CC, StringRef Name) {
   if (isLegacy())
     return;
   // Msgpack format.
+  // Entry point is updated to .entry_point_symbol and is set to the function
+  // name
   getHwStage(CC)[".entry_point_symbol"] =
       MsgPackDoc.getNode(Name, /*Copy=*/true);
+
+  // Set .entry_point which is defined
+  // to be _amdgpu_<stage> and _amdgpu_cs for non-shader functions
+  SmallString<16> EPName("_amdgpu_");
+  raw_svector_ostream EPNameOS(EPName);
+  EPNameOS << getStageName(CC) + 1;
+  getHwStage(CC)[".entry_point"] =
+      MsgPackDoc.getNode(EPNameOS.str(), /*Copy=*/true);
 }
 
 // Set the number of used vgprs in the metadata. This is an optional
@@ -943,28 +975,6 @@ msgpack::MapDocNode AMDGPUPALMetadata::getGraphicsRegisters() {
   return GraphicsRegisters.getMap();
 }
 
-// Return the PAL metadata hardware shader stage name.
-static const char *getStageName(CallingConv::ID CC) {
-  switch (CC) {
-  case CallingConv::AMDGPU_PS:
-    return ".ps";
-  case CallingConv::AMDGPU_VS:
-    return ".vs";
-  case CallingConv::AMDGPU_GS:
-    return ".gs";
-  case CallingConv::AMDGPU_ES:
-    return ".es";
-  case CallingConv::AMDGPU_HS:
-    return ".hs";
-  case CallingConv::AMDGPU_LS:
-    return ".ls";
-  case CallingConv::AMDGPU_Gfx:
-    llvm_unreachable("Callable shader has no hardware stage");
-  default:
-    return ".cs";
-  }
-}
-
 msgpack::DocNode &AMDGPUPALMetadata::refHwStage() {
   auto &N =
       MsgPackDoc.getRoot()

llvm/test/CodeGen/AMDGPU/amdpal-cs.ll

@@ -8,6 +8,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .cs:
+; GCN-NEXT:         .entry_point:    _amdgpu_cs
 ; GCN-NEXT:         .entry_point_symbol:    cs_amdpal
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; GCN:     .registers:

llvm/test/CodeGen/AMDGPU/amdpal-es.ll

@@ -7,6 +7,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .es:
+; GCN-NEXT:         .entry_point:    _amdgpu_es
 ; GCN-NEXT:         .entry_point_symbol:    es_amdpal
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; GCN:     .registers:

llvm/test/CodeGen/AMDGPU/amdpal-gs.ll

@@ -8,6 +8,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .gs:
+; GCN-NEXT:         .entry_point:    _amdgpu_gs
 ; GCN-NEXT:         .entry_point_symbol:    gs_amdpal
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; GCN:     .registers:

llvm/test/CodeGen/AMDGPU/amdpal-hs.ll

@@ -8,6 +8,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .hs:
+; GCN-NEXT:         .entry_point:    _amdgpu_hs
 ; GCN-NEXT:         .entry_point_symbol:    hs_amdpal
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; GCN:     .registers:

llvm/test/CodeGen/AMDGPU/amdpal-ls.ll

@@ -7,6 +7,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .ls:
+; GCN-NEXT:         .entry_point:    _amdgpu_ls
 ; GCN-NEXT:         .entry_point_symbol:    ls_amdpal
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; GCN:     .registers:

llvm/test/CodeGen/AMDGPU/amdpal-psenable.ll

@@ -11,6 +11,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .ps:
+; GCN-NEXT:         .entry_point:    _amdgpu_ps
 ; GCN-NEXT:         .entry_point_symbol:    amdpal_psenable
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; GCN:     .registers:

llvm/test/CodeGen/AMDGPU/amdpal-vs.ll

@@ -8,6 +8,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .vs:
+; GCN-NEXT:         .entry_point:    _amdgpu_vs
 ; GCN-NEXT:         .entry_point_symbol:    vs_amdpal
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; GCN:     .registers:

llvm/test/CodeGen/AMDGPU/amdpal.ll

@@ -86,6 +86,7 @@ declare void @llvm.amdgcn.raw.ptr.buffer.store.f32(float, ptr addrspace(8), i32,
 ; PAL-NEXT: amdpal.pipelines:
 ; PAL-NEXT:   - .hardware_stages:
 ; PAL-NEXT:       .cs:
+; PAL-NEXT:         .entry_point:    _amdgpu_cs
 ; PAL-NEXT:         .entry_point_symbol:    scratch2_cs
 ; PAL-NEXT:         .scratch_memory_size: 0x10
 ; PAL-NEXT:         .sgpr_count:     0x

llvm/test/CodeGen/AMDGPU/elf-notes.ll

@@ -66,6 +66,7 @@
 ; OSABI-PAL-ELF: amdpal.pipelines:
 ; OSABI-PAL-ELF:   - .hardware_stages:
 ; OSABI-PAL-ELF:       .cs:
+; OSABI-PAL-ELF:         .entry_point:    _amdgpu_cs
 ; OSABI-PAL-ELF:         .entry_point_symbol:    elf_notes
 ; OSABI-PAL-ELF:         .scratch_memory_size: 0
 ; OSABI-PAL-ELF:         .sgpr_count:     96

llvm/test/CodeGen/AMDGPU/pal-metadata-3.0.ll

@@ -53,6 +53,7 @@
 ; CHECK-NEXT:      .cs:
 ; CHECK-NEXT:        .checksum_value: 0x9444d7d0
 ; CHECK-NEXT:        .debug_mode:     false
+; CHECK-NEXT:        .entry_point:    _amdgpu_cs
 ; CHECK-NEXT:        .entry_point_symbol:    _amdgpu_cs_main
 ; CHECK-NEXT:        .excp_en:        0
 ; CHECK-NEXT:        .float_mode:     0xc0
@@ -109,6 +110,7 @@
 ; CHECK-NEXT:        .wgp_mode:       false
 ; CHECK-NEXT:      .gs:
 ; CHECK-NEXT:        .debug_mode:     false
+; CHECK-NEXT:        .entry_point:    _amdgpu_gs
 ; CHECK-NEXT:        .entry_point_symbol:    gs_shader
 ; CHECK-NEXT:        .ieee_mode:      false
 ; CHECK-NEXT:        .lds_size:       0x200
@@ -120,6 +122,7 @@
 ; CHECK-NEXT:        .wgp_mode:       true
 ; CHECK-NEXT:      .hs:
 ; CHECK-NEXT:        .debug_mode:     false
+; CHECK-NEXT:        .entry_point:    _amdgpu_hs
 ; CHECK-NEXT:        .entry_point_symbol:    hs_shader
 ; CHECK-NEXT:        .ieee_mode:      false
 ; CHECK-NEXT:        .lds_size:       0x1000
@@ -131,6 +134,7 @@
 ; CHECK-NEXT:        .wgp_mode:       true
 ; CHECK-NEXT:      .ps:
 ; CHECK-NEXT:        .debug_mode:     false
+; CHECK-NEXT:        .entry_point:    _amdgpu_ps
 ; CHECK-NEXT:        .entry_point_symbol:    ps_shader
 ; CHECK-NEXT:        .ieee_mode:      false
 ; CHECK-NEXT:        .lds_size:       0

llvm/test/CodeGen/AMDGPU/wave_dispatch_regs.ll

@@ -12,6 +12,7 @@
 ; GCN-NEXT: amdpal.pipelines:
 ; GCN-NEXT:   - .hardware_stages:
 ; GCN-NEXT:       .cs:
+; GCN-NEXT:         .entry_point:    _amdgpu_cs
 ; GCN-NEXT:         .entry_point_symbol:    _amdgpu_cs_main
 ; GCN-NEXT:         .scratch_memory_size: 0
 ; SI-NEXT:          .sgpr_count:     0x11