Mark the bytes that is referenced as data and not belonging to any fu…

…nction

* add flag space xdata to label these data
* show hex instead of invalid for labeled position

librz/core/canalysis.c

@@ -4685,6 +4685,60 @@ static void core_analysis_using_plugins(RzCore *core) {
 	}
 }
 
+/**
+ * \brief Mark the bytes that are referenced and don't belong
+ *      to any functions as data.
+ */
+static void core_analysis_referenced_data(RzCore *core) {
+	RzList *xrefs = rz_analysis_xrefs_list(core->analysis);
+	RzListIter *it, *it1;
+	RzAnalysisXRef *x, *x1;
+	rz_list_foreach (xrefs, it, x) {
+		if (x->type != RZ_ANALYSIS_XREF_TYPE_DATA) {
+			continue;
+		}
+		ut64 to = x->to;
+
+		// the location <at> doesn't belong to any function
+		RzList *funcs = rz_analysis_get_functions_in(core->analysis, to);
+		if (!rz_list_empty(funcs)) {
+			rz_list_free(funcs);
+			continue;
+		}
+		rz_list_free(funcs);
+
+		// the location <at> is only referenced as DATA instead of CODE
+		RzList *to_xrefs = rz_analysis_xrefs_get_to(core->analysis, to);
+		bool has_nondata_ref = false;
+		rz_list_foreach (to_xrefs, it1, x1) {
+			if (x1->type != RZ_ANALYSIS_XREF_TYPE_DATA) {
+				has_nondata_ref = true;
+			}
+		}
+		rz_list_free(to_xrefs);
+		if (has_nondata_ref) {
+			continue;
+		}
+
+		// skip the offset that already with flag indicating already analyzed
+		const RzList *flags = rz_flag_get_list(core->flags, to);
+		if (!rz_list_empty(flags)) {
+			continue;
+		}
+
+		char *flagname = rz_str_newf("xdata.%08" PFMT64x, to);
+		if (!flagname) {
+			RZ_LOG_ERROR("Failed allocate flag name buffer for xdata\n");
+			return;
+		}
+
+		rz_flag_space_push(core->flags, RZ_FLAGS_FS_XDATA);
+		rz_flag_set(core->flags, flagname, to, core->analysis->bits / 8);
+		rz_flag_space_pop(core->flags);
+		free(flagname);
+	}
+}
+
 /**
  * Runs all the steps of the deep analysis.
  *
@@ -4894,6 +4948,12 @@ RZ_API bool rz_core_analysis_everything(RzCore *core, bool experimental, char *d
 		rz_core_notify_done(core, "Use -AA or aaaa to perform additional experimental analysis.");
 	}
 
+	notify = "Analyze the referenced data (xdata)";
+	rz_core_notify_begin(core, "%s", notify);
+	core_analysis_referenced_data(core);
+	rz_core_notify_done(core, "%s", notify);
+	rz_core_task_yield(&core->tasks);
+
 	rz_core_seek_undo(core);
 	if (dh_orig) {
 		rz_config_set(core->config, "dbg.backend", dh_orig);

librz/core/disasm.c

@@ -1013,6 +1013,16 @@ static void ds_build_op_str(RzDisasmState *ds, bool print_color) {
 		}
 		return;
 	}
+	// if the <invalid> is referred as data, show bytes
+	if (!strcmp(ds->opstr, "invalid")) {
+		RzFlagItem *fi = rz_flag_get_i(ds->core->flags, ds->at);
+		if (fi && rz_str_startswith(fi->name, "xdata.")) {
+			int op_size = rz_asm_op_get_size(&ds->asmop);
+			ut64 data = rz_read_ble(rz_asm_op_get_buf(&ds->asmop), ds->core->analysis->big_endian, op_size * 8);
+			free(ds->opstr);
+			ds->opstr = rz_str_newf("0x%" PFMT64x, data);
+		}
+	}
 	if (ds->hint && ds->hint->opcode) {
 		free(ds->opstr);
 		ds->opstr = strdup(ds->hint->opcode);

librz/include/rz_core.h

@@ -65,6 +65,7 @@ RZ_LIB_VERSION_HEADER(rz_core);
 #define RZ_FLAGS_FS_GLOBALS                 "globals"
 #define RZ_FLAGS_FS_DEBUG_MAPS              "maps"
 #define RZ_FLAGS_FS_POINTERS                "pointers"
+#define RZ_FLAGS_FS_XDATA                   "xdata"
 
 ///
 #define RZ_CONS_COLOR_DEF(x, def) ((core->cons && core->cons->context->pal.x) ? core->cons->context->pal.x : def)