Create global variable from DWARF (#3985)

* Create global variable from DWARF
* Update test
* Add test "Static variables inside function"
* Clean typedb when apply_dwarf

librz/analysis/dwarf_process.c

@@ -9,7 +9,7 @@
 #include "analysis_private.h"
 
 typedef struct dwarf_parse_context_t {
-	const RzAnalysis *analysis;
+	RzAnalysis *analysis;
 	RzBinDwarfCompUnit *unit;
 	RzBinDWARF *dw;
 } Context;
@@ -1390,7 +1390,12 @@ static bool function_var_parse(
 		v->kind = RZ_ANALYSIS_VAR_KIND_VARIABLE;
 		break;
 	case DW_TAG_unspecified_parameters:
-		*has_unspecified_parameters = f->has_unspecified_parameters = true;
+		if (f) {
+			f->has_unspecified_parameters = true;
+		}
+		if (has_unspecified_parameters) {
+			*has_unspecified_parameters = true;
+		}
 		return true;
 	default:
 		return false;
@@ -1444,7 +1449,8 @@ static bool function_var_parse(
 	return true;
 }
 
-static bool function_children_parse(Context *ctx, const RzBinDwarfDie *die, RzCallable *callable, RzAnalysisDwarfFunction *fn) {
+static bool function_children_parse(
+	Context *ctx, const RzBinDwarfDie *die, RzCallable *callable, RzAnalysisDwarfFunction *fn) {
 	if (!die->has_children) {
 		return false;
 	}
@@ -1462,16 +1468,16 @@ static bool function_children_parse(Context *ctx, const RzBinDwarfDie *die, RzCa
 		RzAnalysisDwarfVariable v = { 0 };
 		bool has_unspecified_parameters = false;
 		if (!function_var_parse(ctx, fn, die, &v, child_die, &has_unspecified_parameters)) {
-			goto err;
+			goto loop_end;
 		}
 		if (has_unspecified_parameters) {
 			callable->has_unspecified_parameters = true;
-			goto err;
+			goto loop_end;
 		}
 		if (!v.type) {
 			RZ_LOG_ERROR("DWARF function %s variable %s failed\n",
 				fn->prefer_name, v.prefer_name);
-			goto err;
+			goto loop_end;
 		}
 		if (v.kind == RZ_ANALYSIS_VAR_KIND_FORMAL_PARAMETER) {
 			RzCallableArg *arg = rz_type_callable_arg_new(
@@ -1481,7 +1487,7 @@ static bool function_children_parse(Context *ctx, const RzBinDwarfDie *die, RzCa
 		rz_vector_push(&fn->variables, &v);
 		ht_up_insert(ctx->analysis->debug_info->variable_by_offset, v.offset, &v);
 		continue;
-	err:
+	loop_end:
 		variable_fini(&v);
 	}
 	rz_pvector_free(children);
@@ -1622,7 +1628,45 @@ static bool function_from_die(
 	return false;
 }
 
+static bool variable_exist_global(RzAnalysis *a, RzAnalysisDwarfVariable *v) {
+	RzAnalysisVarGlobal *existing_glob = NULL;
+	if ((existing_glob = rz_analysis_var_global_get_byaddr_in(a, v->location->address))) {
+		return true;
+	}
+	if ((existing_glob = rz_analysis_var_global_get_byname(a, v->prefer_name))) {
+		return true;
+	}
+	return false;
+}
+
+static bool variable_from_die(
+	RZ_BORROW RZ_IN RZ_NONNULL Context *ctx,
+	RZ_BORROW RZ_IN RZ_NONNULL const RzBinDwarfDie *die) {
+	RzAnalysisDwarfVariable v = { 0 };
+	if (!function_var_parse(ctx, NULL, NULL, &v, die, NULL)) {
+		return false;
+	}
+	if (!(v.type && v.location->kind == RzBinDwarfLocationKind_ADDRESS)) {
+		return false;
+	}
+
+	if (variable_exist_global(ctx->analysis, &v)) {
+		return false;
+	}
+
+	bool result = rz_analysis_var_global_create(
+		ctx->analysis, v.prefer_name, v.type, v.location->address);
+
+	v.type = NULL;
+	variable_fini(&v);
+	return result;
+}
+
 static void parse_die(Context *ctx, RzBinDwarfDie *die) {
+	if (set_u_contains(ctx->analysis->debug_info->visited, die->offset)) {
+		return;
+	}
+	set_u_add(ctx->analysis->debug_info->visited, die->offset);
 	switch (die->tag) {
 	case DW_TAG_structure_type:
 	case DW_TAG_union_type:
@@ -1638,6 +1682,9 @@ static void parse_die(Context *ctx, RzBinDwarfDie *die) {
 	case DW_TAG_subprogram:
 		function_from_die(ctx, die);
 		break;
+	case DW_TAG_variable:
+		variable_from_die(ctx, die);
+		break;
 	default:
 		break;
 	}
@@ -1650,7 +1697,7 @@ static void parse_die(Context *ctx, RzBinDwarfDie *die) {
  * \param dw RzBinDwarf pointer
  */
 RZ_API void rz_analysis_dwarf_preprocess_info(
-	RZ_NONNULL RZ_BORROW const RzAnalysis *analysis,
+	RZ_NONNULL RZ_BORROW RzAnalysis *analysis,
 	RZ_NONNULL RZ_BORROW RzBinDWARF *dw) {
 	rz_return_if_fail(analysis && dw);
 	if (!dw->info) {
@@ -1767,7 +1814,7 @@ static bool store_callable(void *u, ut64 k, const void *v) {
  * \param analysis RzAnalysis pointer
  * \param dw RzBinDwarf pointer
  */
-RZ_API void rz_analysis_dwarf_process_info(const RzAnalysis *analysis, RzBinDWARF *dw) {
+RZ_API void rz_analysis_dwarf_process_info(RzAnalysis *analysis, RzBinDWARF *dw) {
 	rz_return_if_fail(analysis && dw);
 	rz_analysis_dwarf_preprocess_info(analysis, dw);
 	ht_pp_foreach(analysis->debug_info->base_type_by_name, store_base_type, (void *)analysis);
@@ -1866,6 +1913,9 @@ static bool RzBinDwarfLocation_as_RzAnalysisVarStorage(
 		break;
 	}
 	case RzBinDwarfLocationKind_ADDRESS: {
+		if (variable_exist_global(a, dw_var)) {
+			return false;
+		}
 		rz_analysis_var_global_create(a, dw_var->prefer_name,
 			rz_type_clone(dw_var->type), loc->address);
 		rz_analysis_var_fini(var);
@@ -2007,6 +2057,7 @@ RZ_API RzAnalysisDebugInfo *rz_analysis_debug_info_new() {
 	debug_info->callable_by_offset = ht_up_new(NULL, HtUP_RzCallable_free, NULL);
 	debug_info->base_type_by_offset = ht_up_new(NULL, HtUP_RzBaseType_free, NULL);
 	debug_info->base_type_by_name = ht_pp_new(NULL, HtPP_RzPVector_free, NULL);
+	debug_info->visited = set_u_new();
 	return debug_info;
 }
 
@@ -2026,5 +2077,6 @@ RZ_API void rz_analysis_debug_info_free(RzAnalysisDebugInfo *debuginfo) {
 	ht_up_free(debuginfo->base_type_by_offset);
 	ht_pp_free(debuginfo->base_type_by_name);
 	rz_bin_dwarf_free(debuginfo->dw);
+	set_u_free(debuginfo->visited);
 	free(debuginfo);
 }

librz/bin/dwarf/op.c

@@ -692,7 +692,7 @@ RZ_API RZ_OWN RzBinDwarfEvaluation *rz_bin_dwarf_evaluation_new(
 	RZ_OWN RZ_NONNULL RzBinEndianReader *byte_code,
 	RZ_BORROW RZ_NONNULL const RzBinDWARF *dw,
 	RZ_BORROW RZ_NULLABLE const RzBinDwarfCompUnit *unit,
-	RZ_BORROW RZ_NULLABLE const RzBinDwarfDie *die) {
+	RZ_BORROW RZ_NULLABLE const RzBinDwarfDie *fn_die) {
 	rz_return_val_if_fail(byte_code && dw && unit, NULL);
 	RzBinDwarfEvaluation *self = RZ_NEW0(RzBinDwarfEvaluation);
 	RET_NULL_IF_FAIL(self);
@@ -704,7 +704,7 @@ RZ_API RZ_OWN RzBinDwarfEvaluation *rz_bin_dwarf_evaluation_new(
 	self->pc = RzBinEndianReader_clone(byte_code);
 	self->dw = dw;
 	self->unit = unit;
-	self->die = die;
+	self->fn_die = fn_die;
 	rz_vector_init(&self->stack, sizeof(RzBinDwarfValue), vec_Value_fini, NULL);
 	rz_vector_init(&self->expression_stack, sizeof(RzBinDwarfExprStackItem), vec_RzBinDwarfExprStackItem_fini, NULL);
 	rz_vector_init(&self->result, sizeof(RzBinDwarfPiece), (RzVectorFree)RzBinDwarfPiece_fini, NULL);
@@ -802,7 +802,8 @@ RZ_API void RzBinDwarfEvaluationResult_free(RZ_OWN RzBinDwarfEvaluationResult *s
 		break; \
 	}
 
-static bool Evaluation_evaluate_one_operation(RzBinDwarfEvaluation *self, OperationEvaluationResult *out) {
+static bool Evaluation_evaluate_one_operation(
+	RzBinDwarfEvaluation *self, OperationEvaluationResult *out) {
 	RzBinEndianReader *reader = self->pc;
 	Operation operation = { 0 };
 	bool ret = false;
@@ -968,7 +969,13 @@ static bool Evaluation_evaluate_one_operation(RzBinDwarfEvaluation *self, Operat
 		break;
 	}
 	case OPERATION_KIND_FRAME_OFFSET: {
-		RzBinDwarfAttr *fb_attr = rz_bin_dwarf_die_get_attr(self->die, DW_AT_frame_base);
+		if (!self->fn_die) {
+			out->kind = OperationEvaluationResult_WAITING;
+			out->waiting._1 = EvaluationStateWaiting_FbReg;
+			goto ok;
+		}
+
+		RzBinDwarfAttr *fb_attr = rz_bin_dwarf_die_get_attr(self->fn_die, DW_AT_frame_base);
 		ERR_IF_FAIL(fb_attr);
 		if (fb_attr->value.kind == RzBinDwarfAttr_UConstant) {
 			RzBinDwarfValue v = {
@@ -982,7 +989,7 @@ static bool Evaluation_evaluate_one_operation(RzBinDwarfEvaluation *self, Operat
 				rz_bin_dwarf_location_free(v.location));
 			break;
 		} else if (fb_attr->value.kind == RzBinDwarfAttr_Block) {
-			RzBinDwarfLocation *loc = rz_bin_dwarf_location_from_block(rz_bin_dwarf_attr_block(fb_attr), self->dw, self->unit, self->die);
+			RzBinDwarfLocation *loc = rz_bin_dwarf_location_from_block(rz_bin_dwarf_attr_block(fb_attr), self->dw, self->unit, self->fn_die);
 			if (!loc) {
 				RzBinDWARFDumpOption opt = {
 					.loclist_indent = "",
@@ -1107,16 +1114,23 @@ static bool Evaluation_evaluate_one_operation(RzBinDwarfEvaluation *self, Operat
 		goto ok;
 	}
 	case OPERATION_KIND_ADDRESS: {
-		out->kind = OperationEvaluationResult_WAITING;
-		out->waiting._1 = EvaluationStateWaiting_RelocatedAddress;
-		out->waiting._2.requires_relocated_address = operation.address.address;
+		out->kind = OperationEvaluationResult_COMPLETE;
+		out->complete.kind = RzBinDwarfLocationKind_ADDRESS;
+		out->complete.address = operation.address.address;
 		goto ok;
 	}
 	case OPERATION_KIND_ADDRESS_INDEX: {
-		out->kind = OperationEvaluationResult_WAITING;
-		out->waiting._1 = EvaluationStateWaiting_IndexedAddress;
-		out->waiting._2.requires_indexed_address.index = operation.address_index.index;
-		out->waiting._2.requires_indexed_address.relocate = true;
+		ut64 addr = 0;
+		if (self->dw && self->unit && rz_bin_dwarf_addr_get(self->dw->addr, &addr, self->unit->hdr.encoding.address_size, self->unit->addr_base, operation.address_index.index)) {
+			out->kind = OperationEvaluationResult_COMPLETE;
+			out->complete.kind = RzBinDwarfLocationKind_ADDRESS;
+			out->complete.address = addr;
+		} else {
+			out->kind = OperationEvaluationResult_WAITING;
+			out->waiting._1 = EvaluationStateWaiting_IndexedAddress;
+			out->waiting._2.requires_indexed_address.index = operation.address_index.index;
+			out->waiting._2.requires_indexed_address.relocate = true;
+		}
 		goto ok;
 	}
 	case OPERATION_KIND_CONSTANT_INDEX: {

librz/core/cbin.c

@@ -684,6 +684,11 @@ RZ_API bool rz_core_bin_apply_dwarf(RzCore *core, RzBinFile *binfile) {
 		return false;
 	}
 
+	rz_type_db_purge(core->analysis->typedb);
+	char *types_dir = rz_path_system(RZ_SDB_TYPES);
+	rz_type_db_reload(core->analysis->typedb, types_dir);
+	free(types_dir);
+
 	rz_analysis_debug_info_free(core->analysis->debug_info);
 	core->analysis->debug_info = rz_analysis_debug_info_new();
 	core->analysis->debug_info->dw = dw;

librz/core/cdwarf.c

@@ -249,6 +249,7 @@ static bool htup_loclists_cb(void *u, ut64 k, const void *v) {
 			RzBinDWARFDumpOption dump_opt = {
 				.loclist_sep = ",\t",
 				.loclist_indent = "",
+				.expr_sep = ", "
 			};
 			rz_bin_dwarf_expression_dump(
 				&ctx->cu->hdr.encoding, entry->expression, ctx->sb, &dump_opt);

librz/include/rz_analysis.h

@@ -456,6 +456,7 @@ typedef struct {
 	HtPP /*<const char*, RzPVector<const RzBaseType *>>*/ *base_type_by_name; ///< Store all RzBaseType parsed from DWARF by DIE offset
 	DWARF_RegisterMapping dwarf_register_mapping; ///< Store the mapping function between DWARF registers number and register name in current architecture
 	RzBinDWARF *dw; ///< Holds ownership of RzBinDwarf, avoid releasing it prematurely
+	SetU *visited;
 } RzAnalysisDebugInfo;
 
 typedef struct rz_analysis_t {
@@ -2284,9 +2285,9 @@ RZ_API void rz_parse_pdb_types(const RzTypeDB *typedb, const RzPdb *pdb);
 
 /* DWARF */
 RZ_API void rz_analysis_dwarf_preprocess_info(
-	RZ_NONNULL RZ_BORROW const RzAnalysis *analysis,
+	RZ_NONNULL RZ_BORROW RzAnalysis *analysis,
 	RZ_NONNULL RZ_BORROW RzBinDWARF *dw);
-RZ_API void rz_analysis_dwarf_process_info(const RzAnalysis *analysis, RzBinDWARF *dw);
+RZ_API void rz_analysis_dwarf_process_info(RzAnalysis *analysis, RzBinDWARF *dw);
 RZ_API void rz_analysis_dwarf_integrate_functions(RzAnalysis *analysis, RzFlag *flags);
 RZ_API RzAnalysisDebugInfo *rz_analysis_debug_info_new();
 RZ_API void rz_analysis_debug_info_free(RzAnalysisDebugInfo *debuginfo);

librz/include/rz_bin_dwarf.h

@@ -1608,7 +1608,7 @@ typedef struct {
 typedef struct {
 	const RzBinDWARF *dw;
 	const RzBinDwarfCompUnit *unit;
-	const RzBinDwarfDie *die;
+	const RzBinDwarfDie *fn_die;
 	RzBinEndianReader *bytecode;
 	const RzBinDwarfEncoding *encoding;
 	ut64 *object_address;
@@ -1709,7 +1709,7 @@ RZ_API RZ_OWN RzBinDwarfEvaluation *rz_bin_dwarf_evaluation_new(
 	RZ_OWN RZ_NONNULL RzBinEndianReader *byte_code,
 	RZ_BORROW RZ_NONNULL const RzBinDWARF *dw,
 	RZ_BORROW RZ_NULLABLE const RzBinDwarfCompUnit *unit,
-	RZ_BORROW RZ_NULLABLE const RzBinDwarfDie *die);
+	RZ_BORROW RZ_NULLABLE const RzBinDwarfDie *fn_die);
 RZ_API RZ_OWN RzBinDwarfEvaluation *rz_bin_dwarf_evaluation_new_from_block(
 	RZ_BORROW RZ_NONNULL const RzBinDwarfBlock *block,
 	RZ_BORROW RZ_NONNULL const RzBinDWARF *dw,

test/db/abi/compilers/clang

@@ -26,7 +26,7 @@ CMDS=<<EOF
 id~?
 EOF
 EXPECT=<<EOF
-22327
+22330
 EOF
 RUN
 

test/db/abi/compilers/clang_64

@@ -27,7 +27,7 @@ CMDS=<<EOF
 id~?
 EOF
 EXPECT=<<EOF
-22442
+22445
 EOF
 RUN
 

test/db/abi/compilers/gcc

@@ -26,7 +26,7 @@ CMDS=<<EOF
 id~?
 EOF
 EXPECT=<<EOF
-25788
+25791
 EOF
 RUN
 

test/db/abi/compilers/gcc_64

@@ -27,7 +27,7 @@ CMDS=<<EOF
 id~?
 EOF
 EXPECT=<<EOF
-25945
+25948
 EOF
 RUN
 

test/db/cmd/cmd_flags

@@ -69,7 +69,7 @@ fs *
 fsl~?*
 EOF
 EXPECT=<<EOF
-8
+9
 EOF
 RUN