Fix enregistered lclFld bug (dotnet/coreclr#18418)

* Fix enregistered lclFld bug

In `impFixupStructReturnType()`, don't transform to `GT_LCL_FLD` if we have a scalar lclVar.
Also, to avoid future bad codegen, add verification and recovery code to Lowering.

Fix dotnet/coreclr#18408

* Extract the full conditions for whether a lclVar is a reg candidate, so it can be called from the assert in Lowering.

* Review feedback

Commit migrated from https://github.com/dotnet/coreclr/commit/82134a002fed96739694b6f085baaeea6c7c41f5

diff --git a/src/coreclr/src/jit/importer.cpp b/src/coreclr/src/jit/importer.cpp
index dc3a383..806a25c 100644 (file)
--- a/src/coreclr/src/jit/importer.cpp
+++ b/src/coreclr/src/jit/importer.cpp
@@ -8654,7 +8654,12 @@ REDO_RETURN_NODE:
     // and no normalizing
     if (op->gtOper == GT_LCL_VAR)
     {
-        op->ChangeOper(GT_LCL_FLD);
+        // It is possible that we now have a lclVar of scalar type.
+        // If so, don't transform it to GT_LCL_FLD.
+        if (varTypeIsStruct(lvaTable[op->AsLclVar()->gtLclNum].lvType))
+        {
+            op->ChangeOper(GT_LCL_FLD);
+        }
     }
     else if (op->gtOper == GT_OBJ)
     {

diff --git a/src/coreclr/src/jit/lower.cpp b/src/coreclr/src/jit/lower.cpp
index f45dadc..defc34c 100644 (file)
--- a/src/coreclr/src/jit/lower.cpp
+++ b/src/coreclr/src/jit/lower.cpp
@@ -269,6 +269,13 @@ GenTree* Lowering::LowerNode(GenTree* node)
             break;
 #endif // FEATURE_HW_INTRINSICS
 
+        case GT_LCL_FLD:
+        {
+            // We should only encounter this for lclVars that are lvDoNotEnregister.
+            verifyLclFldDoNotEnregister(node->AsLclVarCommon()->gtLclNum);
+            break;
+        }
+
         case GT_LCL_VAR:
             WidenSIMD12IfNecessary(node->AsLclVarCommon());
             break;

diff --git a/src/coreclr/src/jit/lower.h b/src/coreclr/src/jit/lower.h
index e98dc08..d9ecb23 100644 (file)
--- a/src/coreclr/src/jit/lower.h
+++ b/src/coreclr/src/jit/lower.h
@@ -355,6 +355,23 @@ private:
         return LIR::AsRange(m_block);
     }
 
+    // Any tracked lclVar accessed by a LCL_FLD or STORE_LCL_FLD should be marked doNotEnregister.
+    // This method checks, and asserts in the DEBUG case if it is not so marked,
+    // but in the non-DEBUG case (asserts disabled) set the flag so that we don't generate bad code.
+    // This ensures that the local's value is valid on-stack as expected for a *LCL_FLD.
+    void verifyLclFldDoNotEnregister(unsigned lclNum)
+    {
+        LclVarDsc* varDsc = &(comp->lvaTable[lclNum]);
+        // Do a couple of simple checks before setting lvDoNotEnregister.
+        // This may not cover all cases in 'isRegCandidate()' but we don't want to
+        // do an expensive check here. For non-candidates it is not harmful to set lvDoNotEnregister.
+        if (varDsc->lvTracked && !varDsc->lvDoNotEnregister)
+        {
+            assert(!m_lsra->isRegCandidate(varDsc));
+            comp->lvaSetVarDoNotEnregister(lclNum DEBUG_ARG(Compiler::DNER_LocalField));
+        }
+    }
+
     LinearScan*   m_lsra;
     unsigned      vtableCallTemp;       // local variable we use as a temp for vtable calls
     SideEffectSet m_scratchSideEffects; // SideEffectSet used for IsSafeToContainMem and isRMWIndirCandidate

diff --git a/src/coreclr/src/jit/lowerarmarch.cpp b/src/coreclr/src/jit/lowerarmarch.cpp
index 47998fe..85d8caa 100644 (file)
--- a/src/coreclr/src/jit/lowerarmarch.cpp
+++ b/src/coreclr/src/jit/lowerarmarch.cpp
@@ -203,6 +203,11 @@ void Lowering::LowerStoreLoc(GenTreeLclVarCommon* storeLoc)
             }
         }
     }
+    if (storeLoc->OperIs(GT_STORE_LCL_FLD))
+    {
+        // We should only encounter this for lclVars that are lvDoNotEnregister.
+        verifyLclFldDoNotEnregister(storeLoc->gtLclNum);
+    }
     ContainCheckStoreLoc(storeLoc);
 }
 

diff --git a/src/coreclr/src/jit/lowerxarch.cpp b/src/coreclr/src/jit/lowerxarch.cpp
index a0a4dfe..c912a2d 100644 (file)
--- a/src/coreclr/src/jit/lowerxarch.cpp
+++ b/src/coreclr/src/jit/lowerxarch.cpp
@@ -94,6 +94,11 @@ void Lowering::LowerStoreLoc(GenTreeLclVarCommon* storeLoc)
             }
         }
     }
+    if (storeLoc->OperIs(GT_STORE_LCL_FLD))
+    {
+        // We should only encounter this for lclVars that are lvDoNotEnregister.
+        verifyLclFldDoNotEnregister(storeLoc->gtLclNum);
+    }
     ContainCheckStoreLoc(storeLoc);
 }
 

diff --git a/src/coreclr/src/jit/lsra.cpp b/src/coreclr/src/jit/lsra.cpp
index 5d5baea..a5e676f 100644 (file)
--- a/src/coreclr/src/jit/lsra.cpp
+++ b/src/coreclr/src/jit/lsra.cpp
@@ -1357,7 +1357,10 @@ void LinearScan::identifyCandidatesExceptionDataflow()
 
 bool LinearScan::isRegCandidate(LclVarDsc* varDsc)
 {
-    // We shouldn't be called if opt settings do not permit register variables.
+    if (!enregisterLocalVars)
+    {
+        return false;
+    }
     assert((compiler->opts.compFlags & CLFLG_REGVAR) != 0);
 
     if (!varDsc->lvTracked)
@@ -1386,6 +1389,95 @@ bool LinearScan::isRegCandidate(LclVarDsc* varDsc)
     {
         return false;
     }
+
+    // Don't enregister if the ref count is zero.
+    if (varDsc->lvRefCnt == 0)
+    {
+        varDsc->lvRefCntWtd = 0;
+        return false;
+    }
+
+    // Variables that are address-exposed are never enregistered, or tracked.
+    // A struct may be promoted, and a struct that fits in a register may be fully enregistered.
+    // Pinned variables may not be tracked (a condition of the GCInfo representation)
+    // or enregistered, on x86 -- it is believed that we can enregister pinned (more properly, "pinning")
+    // references when using the general GC encoding.
+    unsigned lclNum = (unsigned)(varDsc - compiler->lvaTable);
+    if (varDsc->lvAddrExposed || !varTypeIsEnregisterableStruct(varDsc))
+    {
+#ifdef DEBUG
+        Compiler::DoNotEnregisterReason dner = Compiler::DNER_AddrExposed;
+        if (!varDsc->lvAddrExposed)
+        {
+            dner = Compiler::DNER_IsStruct;
+        }
+#endif // DEBUG
+        compiler->lvaSetVarDoNotEnregister(lclNum DEBUGARG(dner));
+        return false;
+    }
+    else if (varDsc->lvPinned)
+    {
+        varDsc->lvTracked = 0;
+#ifdef JIT32_GCENCODER
+        compiler->lvaSetVarDoNotEnregister(lclNum DEBUGARG(Compiler::DNER_PinningRef));
+#endif // JIT32_GCENCODER
+        return false;
+    }
+
+    //  Are we not optimizing and we have exception handlers?
+    //   if so mark all args and locals as volatile, so that they
+    //   won't ever get enregistered.
+    //
+    if (compiler->opts.MinOpts() && compiler->compHndBBtabCount > 0)
+    {
+        compiler->lvaSetVarDoNotEnregister(lclNum DEBUGARG(Compiler::DNER_LiveInOutOfHandler));
+    }
+
+    if (varDsc->lvDoNotEnregister)
+    {
+        return false;
+    }
+
+    switch (genActualType(varDsc->TypeGet()))
+    {
+#if CPU_HAS_FP_SUPPORT
+        case TYP_FLOAT:
+        case TYP_DOUBLE:
+            return !compiler->opts.compDbgCode;
+
+#endif // CPU_HAS_FP_SUPPORT
+
+        case TYP_INT:
+        case TYP_LONG:
+        case TYP_REF:
+        case TYP_BYREF:
+            break;
+
+#ifdef FEATURE_SIMD
+        case TYP_SIMD12:
+        case TYP_SIMD16:
+        case TYP_SIMD32:
+            return !varDsc->lvPromoted;
+
+        // TODO-1stClassStructs: Move TYP_SIMD8 up with the other SIMD types, after handling the param issue
+        // (passing & returning as TYP_LONG).
+        case TYP_SIMD8:
+            return false;
+#endif // FEATURE_SIMD
+
+        case TYP_STRUCT:
+            return false;
+
+        case TYP_UNDEF:
+        case TYP_UNKNOWN:
+            noway_assert(!"lvType not set correctly");
+            varDsc->lvType = TYP_INT;
+            return false;
+
+        default:
+            return false;
+    }
+
     return true;
 }
 
@@ -1544,19 +1636,7 @@ void LinearScan::identifyCandidates()
         }
 #endif // DOUBLE_ALIGN
 
-        /* Track all locals that can be enregistered */
-
-        if (!isRegCandidate(varDsc))
-        {
-            varDsc->lvLRACandidate = 0;
-            if (varDsc->lvTracked)
-            {
-                localVarIntervals[varDsc->lvVarIndex] = nullptr;
-            }
-            continue;
-        }
-
-        assert(varDsc->lvTracked);
+        // Start with the assumption that it's a candidate.
 
         varDsc->lvLRACandidate = 1;
 
@@ -1564,116 +1644,19 @@ void LinearScan::identifyCandidates()
         // the same register assignment throughout
         varDsc->lvRegister = false;
 
-        /* If the ref count is zero */
-        if (varDsc->lvRefCnt == 0)
-        {
-            /* Zero ref count, make this untracked */
-            varDsc->lvRefCntWtd    = 0;
-            varDsc->lvLRACandidate = 0;
-        }
-
-        // Variables that are address-exposed are never enregistered, or tracked.
-        // A struct may be promoted, and a struct that fits in a register may be fully enregistered.
-        // Pinned variables may not be tracked (a condition of the GCInfo representation)
-        // or enregistered, on x86 -- it is believed that we can enregister pinned (more properly, "pinning")
-        // references when using the general GC encoding.
-
-        if (varDsc->lvAddrExposed || !varTypeIsEnregisterableStruct(varDsc))
+        if (!isRegCandidate(varDsc))
         {
             varDsc->lvLRACandidate = 0;
-#ifdef DEBUG
-            Compiler::DoNotEnregisterReason dner = Compiler::DNER_AddrExposed;
-            if (!varDsc->lvAddrExposed)
+            if (varDsc->lvTracked)
             {
-                dner = Compiler::DNER_IsStruct;
+                localVarIntervals[varDsc->lvVarIndex] = nullptr;
             }
-#endif // DEBUG
-            compiler->lvaSetVarDoNotEnregister(lclNum DEBUGARG(dner));
-        }
-        else if (varDsc->lvPinned)
-        {
-            varDsc->lvTracked = 0;
-#ifdef JIT32_GCENCODER
-            compiler->lvaSetVarDoNotEnregister(lclNum DEBUGARG(Compiler::DNER_PinningRef));
-#endif // JIT32_GCENCODER
-        }
-
-        //  Are we not optimizing and we have exception handlers?
-        //   if so mark all args and locals as volatile, so that they
-        //   won't ever get enregistered.
-        //
-        if (compiler->opts.MinOpts() && compiler->compHndBBtabCount > 0)
-        {
-            compiler->lvaSetVarDoNotEnregister(lclNum DEBUGARG(Compiler::DNER_LiveInOutOfHandler));
-        }
-
-        if (varDsc->lvDoNotEnregister)
-        {
-            varDsc->lvLRACandidate                = 0;
-            localVarIntervals[varDsc->lvVarIndex] = nullptr;
             continue;
         }
 
-        var_types type = genActualType(varDsc->TypeGet());
-
-        switch (type)
-        {
-#if CPU_HAS_FP_SUPPORT
-            case TYP_FLOAT:
-            case TYP_DOUBLE:
-                if (compiler->opts.compDbgCode)
-                {
-                    varDsc->lvLRACandidate = 0;
-                }
-#ifdef ARM_SOFTFP
-                if (varDsc->lvIsParam && varDsc->lvIsRegArg)
-                {
-                    type = (type == TYP_DOUBLE) ? TYP_LONG : TYP_INT;
-                }
-#endif // ARM_SOFTFP
-                break;
-#endif // CPU_HAS_FP_SUPPORT
-
-            case TYP_INT:
-            case TYP_LONG:
-            case TYP_REF:
-            case TYP_BYREF:
-                break;
-
-#ifdef FEATURE_SIMD
-            case TYP_SIMD12:
-            case TYP_SIMD16:
-            case TYP_SIMD32:
-                if (varDsc->lvPromoted)
-                {
-                    varDsc->lvLRACandidate = 0;
-                }
-                break;
-
-            // TODO-1stClassStructs: Move TYP_SIMD8 up with the other SIMD types, after handling the param issue
-            // (passing & returning as TYP_LONG).
-            case TYP_SIMD8:
-#endif // FEATURE_SIMD
-
-            case TYP_STRUCT:
-            {
-                varDsc->lvLRACandidate = 0;
-            }
-            break;
-
-            case TYP_UNDEF:
-            case TYP_UNKNOWN:
-                noway_assert(!"lvType not set correctly");
-                varDsc->lvType = TYP_INT;
-
-                __fallthrough;
-
-            default:
-                varDsc->lvLRACandidate = 0;
-        }
-
         if (varDsc->lvLRACandidate)
         {
+            var_types type   = genActualType(varDsc->TypeGet());
             Interval* newInt = newInterval(type);
             newInt->setLocalNumber(compiler, lclNum, this);
             VarSetOps::AddElemD(compiler, registerCandidateVars, varDsc->lvVarIndex);

diff --git a/src/coreclr/tests/src/JIT/Regression/JitBlue/GitHub_18408/GitHub_18408.cs b/src/coreclr/tests/src/JIT/Regression/JitBlue/GitHub_18408/GitHub_18408.cs
new file mode 100644 (file)
index 0000000..6b0399d
--- /dev/null
+++ b/src/coreclr/tests/src/JIT/Regression/JitBlue/GitHub_18408/GitHub_18408.cs
@@ -0,0 +1,122 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+
+using System;
+using System.Reflection;
+using System.Threading;
+using System.Runtime.CompilerServices;
+
+class MetadataReader
+{
+    [MethodImpl(MethodImplOptions.NoInlining)]
+    public Method GetMethod(MethodHandle handle)
+    {
+        return new Method(this, handle, MethodAttributes.Abstract);
+    }
+
+}
+
+struct Handle
+{
+    int _value;
+
+    public MethodHandle ToMethodHandle(MetadataReader reader)
+    {
+        return new MethodHandle(this);
+    }
+
+    public int GetValue()
+    {
+        return _value;
+    }
+}
+
+static class MetadataReaderExtensions
+{
+    public static unsafe Handle AsHandle(this int token)
+    {
+        return *(Handle*)&token;
+    }
+}
+
+struct MethodHandle
+{
+    internal int _value;
+
+    [MethodImpl(MethodImplOptions.NoInlining)]
+    internal MethodHandle(Handle value)
+    {
+        _value = value.GetValue();
+    }
+
+    public Method GetMethod(MetadataReader reader)
+    {
+        return reader.GetMethod(this);
+    }
+}
+
+struct Method
+{
+    internal MetadataReader _reader;
+    internal MethodHandle _handle;
+    internal MethodAttributes _flags;
+
+    public Method(MetadataReader r, MethodHandle h, MethodAttributes f)
+    {
+        _reader = r;
+        _handle = h;
+        _flags = f;
+    }
+
+    public MethodAttributes Flags => _flags;
+}
+
+struct QMethodDefinition
+{
+    private QMethodDefinition(MetadataReader reader, int token)
+    {
+        _reader = reader;
+        _handle = token;
+    }
+
+    public static QMethodDefinition FromObjectAndInt(MetadataReader reader, int token)
+    {
+        return new QMethodDefinition(reader, token);
+    }
+
+    public MetadataReader Reader { get { return _reader; } }
+    public int Token { get { return _handle; } }
+
+    public bool IsValid { get { return _reader == null; } }
+
+    private readonly MetadataReader _reader;
+    private readonly int _handle;
+
+    public MetadataReader NativeFormatReader { get { return _reader; } }
+    public MethodHandle NativeFormatHandle { get { return _handle.AsHandle().ToMethodHandle(NativeFormatReader); } }
+}
+
+class GitHub_18408
+{
+    [MethodImpl(MethodImplOptions.NoInlining)]
+    static object foo(QMethodDefinition methodHandle)
+    {
+        Method method = methodHandle.NativeFormatHandle.GetMethod(methodHandle.NativeFormatReader);
+        return (method.Flags != (MethodAttributes)0) ? new object() : null;
+    }
+
+    public static int Main(string[] args)
+    {
+        MetadataReader r = new MetadataReader();
+
+        if (foo(QMethodDefinition.FromObjectAndInt(r, 1)) == null)
+        {
+            Console.WriteLine("FAIL");
+            return -1;
+        }
+
+        Console.WriteLine("PASS");
+        return 100;
+    }
+}

diff --git a/src/coreclr/tests/src/JIT/Regression/JitBlue/GitHub_18408/GitHub_18408.csproj b/src/coreclr/tests/src/JIT/Regression/JitBlue/GitHub_18408/GitHub_18408.csproj
new file mode 100644 (file)
index 0000000..0b9fa76
--- /dev/null
+++ b/src/coreclr/tests/src/JIT/Regression/JitBlue/GitHub_18408/GitHub_18408.csproj
@@ -0,0 +1,35 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="12.0" DefaultTargets="Build" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <Import Project="$([MSBuild]::GetDirectoryNameOfFileAbove($(MSBuildThisFileDirectory), dir.props))\dir.props" />
+  <PropertyGroup>
+    <Configuration Condition=" '$(Configuration)' == '' ">Debug</Configuration>
+    <Platform Condition=" '$(Platform)' == '' ">AnyCPU</Platform>
+    <AssemblyName>$(MSBuildProjectName)</AssemblyName>
+    <SchemaVersion>2.0</SchemaVersion>
+    <ProjectGuid>{95DFC527-4DC1-495E-97D7-E94EE1F7140D}</ProjectGuid>
+    <OutputType>Exe</OutputType>
+    <ProjectTypeGuids>{786C830F-07A1-408B-BD7F-6EE04809D6DB};{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}</ProjectTypeGuids>
+    <SolutionDir Condition="$(SolutionDir) == '' Or $(SolutionDir) == '*Undefined*'">..\..\</SolutionDir>
+  </PropertyGroup>
+  <!-- Default configurations to help VS understand the configurations -->
+  <PropertyGroup Condition=" '$(Configuration)|$(Platform)' == 'Debug|AnyCPU' "></PropertyGroup>
+  <PropertyGroup Condition=" '$(Configuration)|$(Platform)' == 'Release|AnyCPU' "></PropertyGroup>
+  <ItemGroup>
+    <CodeAnalysisDependentAssemblyPaths Condition=" '$(VS100COMNTOOLS)' != '' " Include="$(VS100COMNTOOLS)..\IDE\PrivateAssemblies">
+      <Visible>False</Visible>
+    </CodeAnalysisDependentAssemblyPaths>
+  </ItemGroup>
+  <PropertyGroup>
+    <DebugType></DebugType>
+    <Optimize>True</Optimize>
+    <AllowUnsafeBlocks>True</AllowUnsafeBlocks>
+  </PropertyGroup>
+  <ItemGroup>
+    <Compile Include="$(MSBuildProjectName).cs" />
+  </ItemGroup>
+  <ItemGroup>
+    <Service Include="{82A7F48D-3B50-4B1E-B82E-3ADA8210C358}" />
+  </ItemGroup>
+  <Import Project="$([MSBuild]::GetDirectoryNameOfFileAbove($(MSBuildThisFileDirectory), dir.targets))\dir.targets" />
+  <PropertyGroup Condition=" '$(MsBuildProjectDirOverride)' != '' "></PropertyGroup>
+</Project>