Use RVec instead of RList as class storage

libr/bin/bfile.c

@@ -8,7 +8,7 @@
 #define R_STRING_SCAN_BUFFER_SIZE 4096
 #define R_STRING_MAX_UNI_BLOCKS 4
 
-static RBinClass *__getClass(RBinFile *bf, const char *name) {
+static RBinClass *get_class(RBinFile *bf, const char *name) {
 	R_RETURN_VAL_IF_FAIL (bf && bf->bo && bf->bo->classes_ht && name, NULL);
 	return ht_pp_find (bf->bo->classes_ht, name, NULL);
 }
@@ -1094,20 +1094,49 @@ R_API RBinClass *r_bin_class_new(const char *name, const char *super, ut64 attr)
 	return c;
 }
 
-R_API void r_bin_class_free(RBinClass *k) {
+#if R2_USE_NEW_ABI
+R_API void r_bin_class_init(RBinClass *c, const char *name, const char *super, ut64 attr) {
+	R_RETURN_IF_FAIL (c && name);
+	c->name = r_bin_name_new (name);
+	if (R_STR_ISNOTEMPTY (super)) {
+		c->super = r_list_newf (free);
+		r_list_append (c->super, r_bin_name_new (super));
+	}
+	// TODO: use vectors!
+	c->methods = r_list_newf (r_bin_symbol_free);
+	c->fields = r_list_newf (r_bin_field_free);
+	c->attr = attr;
+}
+
+R_API void r_bin_class_fini(RBinClass *k) {
 	if (k) {
 		free (k->name);
 		r_list_free (k->super);
 		free (k->visibility_str);
 		r_list_free (k->methods);
 		r_list_free (k->fields);
+	}
+}
+#else
+static inline void r_bin_class_fini(RBinClass *k) {
+	free (k->name);
+	r_list_free (k->super);
+	free (k->visibility_str);
+	r_list_free (k->methods);
+	r_list_free (k->fields);
+}
+#endif
+
+R_API void r_bin_class_free(RBinClass *k) {
+	if (k) {
+		r_bin_class_fini (k),
 		free (k);
 	}
 }
 
 R_API RBinClass *r_bin_file_add_class(RBinFile *bf, const char *name, const char *super, ut64 attr) {
 	R_RETURN_VAL_IF_FAIL (name && bf && bf->bo, NULL);
-	RBinClass *c = __getClass (bf, name);
+	RBinClass *c = get_class (bf, name);
 	if (c) {
 		if (R_STR_ISNOTEMPTY (super)) {
 			r_list_free (c->super);
@@ -1116,13 +1145,30 @@ R_API RBinClass *r_bin_file_add_class(RBinFile *bf, const char *name, const char
 		}
 		return c;
 	}
+#if R2_USE_NEW_ABI
+	c = R_NEW0 (RBinClass);
+	if (c) {
+		r_bin_class_init (c, name, super, attr);
+		c->index = RVecRBinClass_length (&bf->bo->classes);
+		RVecRBinClass_push_back (&bf->bo->classes, c);
+		ht_pp_insert (bf->bo->classes_ht, name, c);
+	}
+	// free (c);
+	// const int htidx = bc.index + 1;
+ 	// c = RVecRBinClass_last (&bf->bo->classes);
+	// eprintf ("1-> %s (%s)\n", r_bin_name_tostring (c->name), name);
+	// free (c);
+	// c = RVecRBinClass_at (&bf->bo->classes, 0);
+	// return c;
+#else
 	c = r_bin_class_new (name, super, attr);
 	if (c) {
 		// XXX. no need for a list, the ht is iterable too
 		c->index = r_list_length (bf->bo->classes);
 		r_list_append (bf->bo->classes, c);
 		ht_pp_insert (bf->bo->classes_ht, name, c);
 	}
+#endif
 	return c;
 }
 
@@ -1144,7 +1190,6 @@ R_API RBinSymbol *r_bin_file_add_method(RBinFile *bf, const char *klass, const c
 			sym->lang = lang;
 			char *name = r_str_newf ("%s::%s", klass, method);
 			ht_pp_insert (bf->bo->methods_ht, name, sym);
-			// RBinSymbol *dsym = r_bin_symbol_clone (sym);
 			r_list_append (c->methods, sym);
 			free (name);
 		}
@@ -1222,6 +1267,7 @@ R_API RBinFile *r_bin_file_open(RBin *bin, const char *file, RBinFileOptions *op
 
 // TODO Improve this API
 R_API void r_bin_file_merge(RBinFile *dst, RBinFile *src) {
+	R_RETURN_IF_FAIL (dst && src);
 	// merge imports
 	// merge dbginfo
 	sdb_merge (dst->bo->kv, src->bo->kv);

libr/bin/bin.c

@@ -237,6 +237,7 @@ R_API void r_bin_string_free(void *_str) {
 	}
 }
 
+// R2_600 - return RBinFile instead of bool. avoid toctou
 R_API bool r_bin_open(RBin *bin, const char *file, RBinFileOptions *opt) {
 	R_RETURN_VAL_IF_FAIL (bin && bin->iob.io && opt, false);
 
@@ -273,6 +274,7 @@ R_API bool r_bin_reload(RBin *bin, ut32 bf_id, ut64 baseaddr) {
 	return res;
 }
 
+// R2_600 - return RBinFile instead of bool. avoid toctou
 R_API bool r_bin_open_buf(RBin *bin, RBuffer *buf, RBinFileOptions *opt) {
 	R_RETURN_VAL_IF_FAIL (bin && opt, false);
 
@@ -321,10 +323,12 @@ R_API bool r_bin_open_buf(RBin *bin, RBuffer *buf, RBinFileOptions *opt) {
 		return false;
 	}
 	// r_ref (bf);
+	// return bf instead of bool!
 	bin->cur = bf;
 	return res;
 }
 
+// R2_600 - return RBinFile instead of bool. avoid toctou
 R_API bool r_bin_open_io(RBin *bin, RBinFileOptions *opt) {
 	R_RETURN_VAL_IF_FAIL (bin && opt && bin->iob.io, false);
 	R_RETURN_VAL_IF_FAIL (opt->fd >= 0 && (st64)opt->sz >= 0, false);
@@ -1303,11 +1307,15 @@ R_API RBuffer *r_bin_package(RBin *bin, const char *type, const char *file, RLis
 	return NULL;
 }
 
+#if R2_USE_NEW_ABI
+// this api is deprecated. just access the binobject instead
+#else
 R_API RList* /*<RBinClass>*/ r_bin_get_classes(RBin *bin) {
 	R_RETURN_VAL_IF_FAIL (bin, NULL);
 	RBinObject *bo = r_bin_cur_object (bin);
 	return bo ? bo->classes : NULL;
 }
+#endif
 
 /* returns vaddr, rebased with the baseaddr of bin, if va is enabled for bin, * paddr otherwise */
 R_API ut64 r_bin_get_vaddr(RBin *bin, ut64 paddr, ut64 vaddr) {

libr/bin/bobj.c

@@ -51,7 +51,11 @@ static void object_delete_items(RBinObject *o) {
 	}
 	r_list_free (o->symbols);
 
+#if R2_USE_NEW_ABI
+	RVecRBinClass_fini (&o->classes);
+#else
 	r_list_free (o->classes);
+#endif
 	ht_pp_free (o->classes_ht);
 	ht_pp_free (o->methods_ht);
 	r_list_free (o->lines);
@@ -172,7 +176,7 @@ static void classes_from_symbols2(RBinFile *bf, RBinSymbol *sym) {
 	}
 }
 
-static RList *classes_from_symbols(RBinFile *bf) {
+static void classes_from_symbols(RBinFile *bf) {
 	RBinSymbol *sym;
 	RListIter *iter;
 	// TODO: Use rvec here
@@ -185,7 +189,6 @@ static RList *classes_from_symbols(RBinFile *bf) {
 			classes_from_symbols2 (bf, sym);
 		}
 	}
-	return bf->bo->classes;
 }
 
 // TODO: kill offset and sz, because those should be inferred from binfile->buf
@@ -202,7 +205,11 @@ R_IPI RBinObject *r_bin_object_new(RBinFile *bf, RBinPlugin *plugin, ut64 basead
 	bo->regstate = NULL;
 	bo->kv = sdb_new0 (); // XXX bf->sdb bf->bo->sdb wtf
 	bo->baddr = baseaddr;
+#if R2_USE_NEW_ABI
+	RVecRBinClass_init (&bo->classes);
+#else
 	bo->classes = r_list_newf ((RListFree)r_bin_class_free);
+#endif
 	bo->classes_ht = ht_pp_new0 ();
 	bo->methods_ht = ht_pp_new0 ();
 	bo->baddr_shift = 0;
@@ -262,19 +269,23 @@ R_IPI RBinObject *r_bin_object_new(RBinFile *bf, RBinPlugin *plugin, ut64 basead
 	return bo;
 }
 
-static bool filter_classes(RBinFile *bf, RList *list) {
-	bool rc = true;
+static void filter_classes(RBinFile *bf) {
 	HtSU *db = ht_su_new0 ();
 	HtPP *ht = ht_pp_new0 ();
-	RListIter *iter, *iter2;
 	RBinClass *cls;
 	RBinSymbol *sym;
-	r_list_foreach (list, iter, cls) {
+#if R2_USE_NEW_ABI
+	RListIter *iter2;
+	R_VEC_FOREACH (&bf->bo->classes, cls)
+#else
+	RListIter *iter, *iter2;
+	r_list_foreach (bf->bo->classes, iter, cls)
+#endif
+	{
 		const char *kname = r_bin_name_tostring (cls->name);
 		char *fname = r_bin_filter_name (bf, db, cls->index, kname);
 		if (R_STR_ISEMPTY (fname)) {
 			R_LOG_INFO ("Corrupted class storage with nameless classes");
-			rc = false;
 			free (fname);
 			break;
 		}
@@ -291,7 +302,6 @@ static bool filter_classes(RBinFile *bf, RList *list) {
 	}
 	ht_su_free (db);
 	ht_pp_free (ht);
-	return rc;
 }
 
 static RRBTree *list2rbtree(RList *relocs) {
@@ -313,14 +323,22 @@ static void r_bin_object_rebuild_classes_ht(RBinObject *bo) {
 	ht_pp_free (bo->methods_ht);
 	bo->methods_ht = ht_pp_new0 ();
 
+#if R2_USE_NEW_ABI
+	RListIter *it2;
+#else
 	RListIter *it, *it2;
+#endif
 	RBinClass *klass;
 	RBinSymbol *method;
-	r_list_foreach (bo->classes, it, klass) {
+#if R2_USE_NEW_ABI
+	R_VEC_FOREACH (&bo->classes, klass)
+#else
+	r_list_foreach (bo->classes, it, klass)
+#endif
+	{
 		if (klass->name) {
-			ht_pp_insert (bo->classes_ht, klass->name, klass);
+			const char *klass_name = r_bin_name_tostring (klass->name);
 			r_list_foreach (klass->methods, it2, method) {
-				const char *klass_name = r_bin_name_tostring (klass->name);
 				const char *method_name = r_bin_name_tostring (method->name);
 				char *name = r_str_newf ("%s::%s", klass_name, method_name);
 				ht_pp_insert (bo->methods_ht, name, method);
@@ -445,24 +463,47 @@ R_API int r_bin_object_set_items(RBinFile *bf, RBinObject *bo) {
 			RList *classes = p->classes (bf);
 			if (classes) {
 				// XXX we should probably merge them instead
+#if R2_USE_NEW_ABI
+				RListIter *iter;
+				RBinClass *klass;
+				RVecRBinClass_fini (&bo->classes);
+				r_list_foreach (classes, iter, klass) {
+					RVecRBinClass_push_back (&bo->classes, klass);
+				}
+#else
 				r_list_free (bo->classes);
 				bo->classes = classes;
-				r_bin_object_rebuild_classes_ht (bo);
+#endif
 			}
+			r_bin_object_rebuild_classes_ht (bo);
 			isSwift = r_bin_lang_swift (bf);
 			if (isSwift) {
-				bo->classes = classes_from_symbols (bf);
+				classes_from_symbols (bf);
+				bo->classes = bf->bo->classes; // XXX unnecessary i think
 			}
 		} else {
-			RList *classes = classes_from_symbols (bf);
-			if (classes) {
-				bo->classes = classes;
-			}
+			classes_from_symbols (bf);
+			bo->classes = bf->bo->classes;
 		}
 		if (bin->filter) {
-			filter_classes (bf, bo->classes);
+			filter_classes (bf);
 		}
+#if R2_USE_NEW_ABI
+		if (RVecRBinClass_length (&bo->classes) > 0 && !bo->addr2klassmethod) {
+			RListIter *iter;
+			RBinClass *klass;
+			RBinSymbol *method;
+			bo->addr2klassmethod = ht_up_new0 ();
+			R_VEC_FOREACH (&bo->classes, klass) {
+				// this is slow. must be optimized, but at least its cached
+				r_list_foreach (klass->methods, iter, method) {
+					ht_up_insert (bo->addr2klassmethod, method->vaddr, method);
+				}
+			}
+		}
+#else
 		// cache addr=class+method
+#endif
 #if 0
 		// moved into libr/core/canal.c
 		// XXX SLOW on large binaries. only used when needed by getFunctionName from canal.c

libr/bin/format/dex/dex.h

@@ -142,7 +142,11 @@ typedef struct r_bin_dex_obj_t {
 	RList *methods_list;
 	RList *trycatch_list;
 	RList *imports_list;
+#if R2_USE_NEW_ABI
+	// it's stored in RBinObject
+#else
 	RList *classes_list;
+#endif
 	RList *lines_list;
 	ut64 code_from;
 	ut64 code_to;