[llvm] Recognize GEP (#17844)

* initial impl

* improve jit

* fix copy-paste

* disable for non-netcore

* clean up

4 files changed, 41 insertions, 4 deletions

diff --git a/mono/mini/llvm-jit.cpp b/mono/mini/llvm-jit.cpp
index ecd63cbe5c2..85c9ee92107 100644
--- a/mono/mini/llvm-jit.cpp
+++ b/mono/mini/llvm-jit.cpp
@@ -297,12 +297,13 @@ public:
 		initializeScalarOpts(registry);
 		initializeInstCombine(registry);
 		initializeTarget(registry);
+		initializeLoopIdiomRecognizeLegacyPassPass(registry);
 		linkCoreCLRGC(); // Mono uses built-in "coreclr" GCStrategy
 
 		// FIXME: find optimal mono specific order of passes
 		// see https://llvm.org/docs/Frontend/PerformanceTips.html#pass-ordering
 		// the following order is based on a stripped version of "OPT -O2"
-		const char *default_opts = " -simplifycfg -sroa -lower-expect -instcombine -licm -simplifycfg -lcssa -indvars -loop-deletion -gvn -memcpyopt -sccp -bdce -instcombine -dse -simplifycfg -enable-implicit-null-checks";
+		const char *default_opts = " -simplifycfg -sroa -lower-expect -instcombine -loop-rotate -licm -simplifycfg -lcssa -loop-idiom -indvars -loop-deletion -gvn -memcpyopt -sccp -bdce -instcombine -dse -simplifycfg -enable-implicit-null-checks";
 		const char *opts = g_getenv ("MONO_LLVM_OPT");
 		if (opts == NULL)
 			opts = default_opts;
diff --git a/mono/mini/mini-llvm-cpp.cpp b/mono/mini/mini-llvm-cpp.cpp
index c4523ee0b7a..3c95c3e5c79 100644
--- a/mono/mini/mini-llvm-cpp.cpp
+++ b/mono/mini/mini-llvm-cpp.cpp
@@ -298,6 +298,28 @@ mono_llvm_set_func_nonnull_arg (LLVMValueRef func, int argNo)
 }
 
 gboolean
+mono_llvm_can_be_gep (LLVMValueRef base, LLVMValueRef* gep_base, LLVMValueRef* gep_offset)
+{
+#ifdef ENABLE_NETCORE
+	// Look for a pattern like this:
+	//   %1 = ptrtoint i8* %gep_base to i64
+	//   %2 = add i64 %1, %gep_offset
+	if (Instruction *base_inst = dyn_cast<Instruction> (unwrap (base))) {
+		if (base_inst->getOpcode () == Instruction::Add) {
+			if (Instruction *base_ptr_ins = dyn_cast<Instruction> (base_inst->getOperand (0))) {
+				if (base_ptr_ins->getOpcode () == Instruction::PtrToInt) {
+					*gep_base = wrap (base_ptr_ins->getOperand (0));
+					*gep_offset = wrap (base_inst->getOperand (1));
+					return TRUE;
+				}
+			}
+		}
+	}
+#endif
+	return FALSE;
+}
+
+gboolean
 mono_llvm_is_nonnull (LLVMValueRef wrapped)
 {
 	// Argument to function
diff --git a/mono/mini/mini-llvm-cpp.h b/mono/mini/mini-llvm-cpp.h
index b2aa9392e4a..ef56a6f3790 100644
--- a/mono/mini/mini-llvm-cpp.h
+++ b/mono/mini/mini-llvm-cpp.h
@@ -151,6 +151,9 @@ G_EXTERN_C _Unwind_Reason_Code mono_debug_personality (int a, _Unwind_Action b,
 void*
 mono_llvm_create_di_builder (LLVMModuleRef module);
 
+gboolean
+mono_llvm_can_be_gep (LLVMValueRef base, LLVMValueRef* actual_base, LLVMValueRef* actual_offset);
+
 void*
 mono_llvm_di_create_function (void *di_builder, void *cu, LLVMValueRef func, const char *name, const char *mangled_name, const char *dir, const char *file, int line);
 
diff --git a/mono/mini/mini-llvm.c b/mono/mini/mini-llvm.c
index 88e1d2ab2be..7edd0cd2197 100644
--- a/mono/mini/mini-llvm.c
+++ b/mono/mini/mini-llvm.c
@@ -5762,7 +5762,12 @@ process_bb (EmitContext *ctx, MonoBasicBlock *bb)
 				base = lhs;
 
 				if (ins->inst_offset == 0) {
-					addr = base;
+					LLVMValueRef gep_base, gep_offset;
+					if (mono_llvm_can_be_gep (base, &gep_base, &gep_offset)) {
+						addr = LLVMBuildGEP (builder, convert (ctx, gep_base, LLVMPointerType (LLVMInt8Type (), 0)), &gep_offset, 1, "");
+					} else {
+						addr = base;
+					}
 				} else if (ins->inst_offset % size != 0) {
 					/* Unaligned load */
 					index = LLVMConstInt (LLVMInt32Type (), ins->inst_offset, FALSE);
@@ -5816,7 +5821,10 @@ process_bb (EmitContext *ctx, MonoBasicBlock *bb)
 			t = load_store_to_llvm_type (ins->opcode, &size, &sext, &zext);
 
 			base = values [ins->inst_destbasereg];
-			if (ins->inst_offset % size != 0) {
+			LLVMValueRef gep_base, gep_offset;
+			if (ins->inst_offset == 0 && mono_llvm_can_be_gep (base, &gep_base, &gep_offset)) {
+				addr = LLVMBuildGEP (builder, convert (ctx, gep_base, LLVMPointerType (LLVMInt8Type (), 0)), &gep_offset, 1, "");
+			} else if (ins->inst_offset % size != 0) {
 				/* Unaligned store */
 				index = LLVMConstInt (LLVMInt32Type (), ins->inst_offset, FALSE);
 				addr = LLVMBuildGEP (builder, convert (ctx, base, LLVMPointerType (LLVMInt8Type (), 0)), &index, 1, "");
@@ -5846,7 +5854,10 @@ process_bb (EmitContext *ctx, MonoBasicBlock *bb)
 			t = load_store_to_llvm_type (ins->opcode, &size, &sext, &zext);
 
 			base = values [ins->inst_destbasereg];
-			if (ins->inst_offset % size != 0) {
+			LLVMValueRef gep_base, gep_offset;
+			if (ins->inst_offset == 0 && mono_llvm_can_be_gep (base, &gep_base, &gep_offset)) {
+				addr = LLVMBuildGEP (builder, convert (ctx, gep_base, LLVMPointerType (LLVMInt8Type (), 0)), &gep_offset, 1, "");
+			} else if (ins->inst_offset % size != 0) {
 				/* Unaligned store */
 				index = LLVMConstInt (LLVMInt32Type (), ins->inst_offset, FALSE);
 				addr = LLVMBuildGEP (builder, convert (ctx, base, LLVMPointerType (LLVMInt8Type (), 0)), &index, 1, "");