JIT: simple forward substitution pass (#63720)

Extend ref counting done by local morph so that we can determine
single-def single-use locals.

Add a phase that runs just after local morph that will attempt to
forward single-def single-use local defs to uses when they are in
adjacent statements.

Fix or work around issues uncovered elsewhere:
* `gtFoldExprCompare` might fold "identical" volatile subtrees
* `fgGetStubAddrArg` cannot handle complex trees
* some simd/hw operations can lose struct handles
* some calls cannot handle struct local args
* morph expects args not to interfere
* fix arm; don't forward sub no return calls
* update debuginfo test (we may want to revisit this)
* handle subbing past normalize on store assignment
* clean up nullcheck of new helper

Addresses #6973 and related issues. Still sorting through exactly
which ones are fixed, so list below may need revising.

Fixes #48605.
Fixes #51599.
Fixes #55472.

Improves some but not all cases in #12280 and #62064.

Does not fix #33002, #47082, or #63116; these require handling multiple
uses or bypassing statements.

Author: AndyAyersMS

src/coreclr/jit/CMakeLists.txt

@@ -104,6 +104,7 @@ set( JIT_SOURCES
   fgprofile.cpp
   fgstmt.cpp
   flowgraph.cpp
+  forwardsub.cpp
   gcdecode.cpp
   gcencode.cpp
   gcinfo.cpp

src/coreclr/jit/compiler.cpp

@@ -4657,6 +4657,9 @@ void Compiler::compCompile(void** methodCodePtr, uint32_t* methodCodeSize, JitFl
     // Figure out what locals are address-taken.
     //
     DoPhase(this, PHASE_STR_ADRLCL, &Compiler::fgMarkAddressExposedLocals);
+    // Run a simple forward substitution pass.
+    //
+    DoPhase(this, PHASE_FWD_SUB, &Compiler::fgForwardSub);
 
     // Apply the type update to implicit byref parameters; also choose (based on address-exposed
     // analysis) which implicit byref promotions to keep (requires copy to initialize) or discard.

src/coreclr/jit/compiler.h

@@ -6423,7 +6423,7 @@ class Compiler
 
     bool fgMorphCanUseLclFldForCopy(unsigned lclNum1, unsigned lclNum2);
 
-    GenTreeLclVar* fgMorphTryFoldObjAsLclVar(GenTreeObj* obj);
+    GenTreeLclVar* fgMorphTryFoldObjAsLclVar(GenTreeObj* obj, bool destroyNodes = true);
     GenTreeOp* fgMorphCommutative(GenTreeOp* tree);
     GenTree* fgMorphCastedBitwiseOp(GenTreeOp* tree);
 
@@ -6562,6 +6562,10 @@ class Compiler
     void fgMarkAddressExposedLocals();
     void fgMarkAddressExposedLocals(Statement* stmt);
 
+    PhaseStatus fgForwardSub();
+    bool fgForwardSubBlock(BasicBlock* block);
+    bool fgForwardSubStatement(Statement* statement);
+
     static fgWalkPreFn  fgUpdateSideEffectsPre;
     static fgWalkPostFn fgUpdateSideEffectsPost;
 

src/coreclr/jit/compphases.h

@@ -43,6 +43,7 @@ CompPhaseNameMacro(PHASE_UPDATE_FINALLY_FLAGS,   "Update finally target flags",
 CompPhaseNameMacro(PHASE_COMPUTE_PREDS,          "Compute preds",                  "PREDS",    false, -1, false)
 CompPhaseNameMacro(PHASE_EARLY_UPDATE_FLOW_GRAPH,"Update flow graph early pass",   "UPD-FG-E", false, -1, false)
 CompPhaseNameMacro(PHASE_STR_ADRLCL,             "Morph - Structs/AddrExp",        "MOR-STRAL",false, -1, false)
+CompPhaseNameMacro(PHASE_FWD_SUB,                "Forward Substitution",           "FWD-SUB",  false, -1, false)
 CompPhaseNameMacro(PHASE_MORPH_IMPBYREF,         "Morph - ByRefs",                 "MOR-BYREF",false, -1, false)
 CompPhaseNameMacro(PHASE_PROMOTE_STRUCTS,        "Morph - Promote Structs",        "PROMOTER" ,false, -1, false)
 CompPhaseNameMacro(PHASE_MORPH_GLOBAL,           "Morph - Global",                 "MOR-GLOB", false, -1, false)