Make thread statics faster (#6108)

* Fix the TODO to inline `.cctor` check into the R2R helper. This avoids some method calls and stack frame setup.
* Remove array cast from the hot path. The runtime doesn't care what type the storage is, as long as it's a managed reference - in fact, it seems to be treated as an `Object` in `pCurrentThread->GetThreadStaticStorageForModule` anyway. I considered replacing the cast with an `Unsafe.As`, but just redeclaring the signatures as returning `object[]` seemed better (because now we actually guarantee the classlib won't pass anything but `object[]` to `RhSetThreadStaticStorageForModule`).

2 files changed, 19 insertions, 1 deletions

diff --git a/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64Emitter.cs b/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64Emitter.cs
index 09ebadeaa..662867f99 100644
--- a/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64Emitter.cs
+++ b/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64Emitter.cs
@@ -88,6 +88,20 @@ namespace ILCompiler.DependencyAnalysis.X64
             }
         }
 
+        public void EmitJE(ISymbolNode symbol)
+        {
+            if (symbol.RepresentsIndirectionCell)
+            {
+                throw new NotImplementedException();
+            }
+            else
+            {
+                Builder.EmitByte(0x0f);
+                Builder.EmitByte(0x84);
+                Builder.EmitReloc(symbol, RelocType.IMAGE_REL_BASED_REL32);
+            }
+        }
+
         public void EmitINT3()
         {
             Builder.EmitByte(0xCC);
diff --git a/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64ReadyToRunHelperNode.cs b/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64ReadyToRunHelperNode.cs
index 026367d85..dbe8319c8 100644
--- a/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64ReadyToRunHelperNode.cs
+++ b/src/ILCompiler.Compiler/src/Compiler/DependencyAnalysis/Target_X64/X64ReadyToRunHelperNode.cs
@@ -127,7 +127,11 @@ namespace ILCompiler.DependencyAnalysis
                         else
                         {
                             encoder.EmitLEAQ(encoder.TargetRegister.Arg2, factory.TypeNonGCStaticsSymbol(target), - NonGCStaticsNode.GetClassConstructorContextStorageSize(factory.Target, target));
-                            // TODO: performance optimization - inline the check verifying whether we need to trigger the cctor
+
+                            AddrMode initialized = new AddrMode(encoder.TargetRegister.Arg2, null, factory.Target.PointerSize, 0, AddrModeSize.Int32);
+                            encoder.EmitCMP(ref initialized, 1);
+                            encoder.EmitJE(factory.HelperEntrypoint(HelperEntrypoint.GetThreadStaticBaseForType));
+
                             encoder.EmitJMP(factory.HelperEntrypoint(HelperEntrypoint.EnsureClassConstructorRunAndReturnThreadStaticBase));
                         }
                     }