WIP: Early catch some missbehaves in the class parsing and speedup ma…

libr/bin/bobj.c

@@ -267,10 +267,12 @@ static void filter_classes(RBinFile *bf, RList *list) {
 	r_list_foreach (list, iter, cls) {
 		const char *kname = r_bin_name_tostring (cls->name);
 		char *fname = r_bin_filter_name (bf, db, cls->index, kname);
-		if (fname) {
-			r_bin_name_update (cls->name, fname);
-			free (fname);
+		if (R_STR_ISEMPTY (fname)) {
+			R_LOG_INFO ("Corrupted class storage with nameless classes");
+			break;
 		}
+		r_bin_name_update (cls->name, fname);
+		free (fname);
 		r_list_foreach (cls->methods, iter2, sym) {
 			r_bin_filter_sym (bf, ht, sym->vaddr, sym);
 		}
@@ -325,7 +327,6 @@ R_API int r_bin_object_set_items(RBinFile *bf, RBinObject *bo) {
 	int minlen = (bf->rbin->minstrlen > 0) ? bf->rbin->minstrlen : p->minstrlen;
 	bf->bo = bo;
 
-#if 0
 	bo->info = p->info? p->info (bf): NULL;
 	if (bo->info && bo->info->type) {
 		if (strstr (bo->info->type, "CORE")) {
@@ -337,7 +338,6 @@ R_API int r_bin_object_set_items(RBinFile *bf, RBinObject *bo) {
 			}
 		}
 	}
-#endif
 	// XXX: no way to get info from xtr pluginz?
 	// Note, object size can not be set from here due to potential
 	// inconsistencies
@@ -450,6 +450,9 @@ R_API int r_bin_object_set_items(RBinFile *bf, RBinObject *bo) {
 			filter_classes (bf, bo->classes);
 		}
 		// cache addr=class+method
+#if 0
+		// moved into libr/core/canal.c
+		// XXX SLOW on large binaries. only used when needed by getFunctionName from canal.c
 		if (bo->classes) {
 			RList *klasses = bo->classes;
 			RListIter *iter, *iter2;
@@ -465,6 +468,7 @@ R_API int r_bin_object_set_items(RBinFile *bf, RBinObject *bo) {
 				}
 			}
 		}
+#endif
 	}
 	if (p->lines) {
 		bo->lines = p->lines (bf);

libr/bin/format/objc/mach0_classes.c

@@ -835,12 +835,16 @@ static void get_method_list(RBinFile *bf, RBinClass *klass, const char *class_na
 				name = malloc (name_len + 1);
 				len = r_buf_read_at (bf->buf, r, (ut8 *)name, name_len);
 				name[name_len] = 0;
+				// eprintf ("%d %d\n", name_len, strlen (name));
 				if (len < 1) {
 					goto error;
 				}
 			}
 			if (class_name) { // XXX to save memory we can just ref the RBinName instance from the class
 				method->classname = strdup (class_name);
+			} else {
+				R_LOG_ERROR ("Invalid class name for method. Avoid parsing invalid data");
+				goto error;
 			}
 			method->name = r_bin_name_new (name);
 			R_FREE (name);
@@ -1293,6 +1297,8 @@ static void get_class_ro_t(RBinFile *bf, bool *is_meta_class, RBinClass *klass,
 			return;
 		}
 		if (bin->has_crypto) {
+			R_LOG_ERROR ("Not parsing encrypted data");
+			return;
 			const char kn[] = "some_encrypted_data";
 			klass->name = r_bin_name_new (kn);
 			// klass->name = strdup ("some_encrypted_data");
@@ -1853,7 +1859,13 @@ RList *MACH0_(parse_classes)(RBinFile *bf, objc_cache_opt_info *oi) {
 		}
 		p = r_read_ble (&pp[0], bigendian, 8 * sizeof (mach0_ut));
 		MACH0_(get_class_t) (bf, klass, p, false, relocs, oi);
-		if (!klass->name) {
+		if (klass->name) {
+			const char *klass_name = r_bin_name_tostring (klass->name);
+			if (strlen (klass_name) > 512) {
+				R_LOG_INFO ("Invalid class name, probably corrupted binary");
+				break;
+			}
+		} else {
 			char *klass_name = r_str_newf ("UnnamedClass%" PFMT64d, num_of_unnamed_class);
 			klass->name = r_bin_name_new (klass_name);
 			free (klass_name);
@@ -2049,6 +2061,10 @@ void MACH0_(get_category_t)(RBinFile *bf, RBinClass *klass, mach0_ut p, const RS
 	R_FREE (category_name);
 
 	const char *klass_name = r_bin_name_tostring (klass->name);
+	if (R_STR_ISEMPTY (klass_name)) {
+		R_LOG_ERROR ("Invalid class name");
+		return;
+	}
 	if (c.instanceMethods > 0) {
 		get_method_list (bf, klass, klass_name, false, oi, c.instanceMethods);
 	}

libr/core/canal.c

@@ -89,9 +89,27 @@ static int cmpaddr(const void *_a, const void *_b) {
 	return (a->addr > b->addr)? 1: (a->addr < b->addr)? -1: 0;
 }
 
+static void init_addr2klass(RCore *core, RBinObject *bo) {
+	if (bo->addr2klassmethod) {
+		return;
+	}
+	RList *klasses = bo->classes;
+	RListIter *iter, *iter2;
+	RBinClass *klass;
+	RBinSymbol *method;
+	// this is slow. must be optimized, but at least its cached
+	bo->addr2klassmethod = ht_up_new0 ();
+	r_list_foreach (klasses, iter, klass) {
+		r_list_foreach (klass->methods, iter2, method) {
+			ht_up_insert (bo->addr2klassmethod, method->vaddr, method);
+		}
+	}
+}
+
 static char *get_function_name(RCore *core, ut64 addr) {
 	RBinFile *bf = r_bin_cur (core->bin);
 	if (bf && bf->bo) {
+		init_addr2klass (core, bf->bo);
 		RBinSymbol *sym = ht_up_find (bf->bo->addr2klassmethod, addr, NULL);
 		if (sym && sym->classname && sym->name) {
 			const char *sym_name = r_bin_name_tostring (sym->name);

test/db/formats/mach0/swift

@@ -2044,6 +2044,7 @@ EXPECT=<<EOF
   },
   {
     "classname": "WheresMyBrowser.LoadContentViewController",
+    "rawclassname": "_TtC15WheresMyBrowser25LoadContentViewController",
     "addr": 4295178328,
     "lang": "objc",
     "index": 0,
@@ -2161,6 +2162,7 @@ EXPECT=<<EOF
   },
   {
     "classname": "WheresMyBrowser.UIWebViewController",
+    "rawclassname": "_TtC15WheresMyBrowser19UIWebViewController",
     "addr": 4295178472,
     "lang": "objc",
     "index": 0,