Add one similarity against all functions of another binary (#4584)

rizinorg · Aug 7, 2024 · 819812c · 819812c
1 parent c02c0b7
commit 819812c
Show file tree

Hide file tree

Showing 3 changed files with 76 additions and 7 deletions.
diff --git a/librz/arch/similarity.c b/librz/arch/similarity.c
@@ -513,6 +513,40 @@ static void *analysis_match_functions(SharedContext *shared) {
 	return NULL;
 }
 
+static void *analysis_match_one_function(SharedContext *shared) {
+	RzAnalysisFunction *fcn_a = NULL, *fcn_b = NULL;
+	RzAnalysisMatchPair *pair = NULL;
+	ut32 size_a = 0, size_b = 0;
+	ut8 *buf_a = NULL, *buf_b = NULL;
+
+	fcn_b = rz_list_first(shared->list_b);
+	if (!shared_context_alloc_a(shared, fcn_b, &buf_b, &size_b)) {
+		RZ_LOG_ERROR("analysis_match: cannot allocate buffer for function %s (B)\n", fcn_b->name);
+		return NULL;
+	}
+
+	while (rz_atomic_bool_get(shared->loop) && (fcn_a = rz_th_queue_pop(shared->queue, false))) {
+		if (!shared_context_alloc_b(shared, fcn_a, &buf_a, &size_a)) {
+			RZ_LOG_ERROR("analysis_match: cannot allocate buffer for function %s (A)\n", fcn_a->name);
+			free(buf_b);
+			return NULL;
+		}
+
+		double similarity = calculate_similarity(buf_a, size_a, buf_b, size_b);
+		free(buf_a);
+
+		if (!(pair = match_pair_new(fcn_a, fcn_b, similarity))) {
+			RZ_LOG_ERROR("analysis_match: cannot allocate match pair\n");
+			free(buf_b);
+			return NULL;
+		}
+		rz_th_queue_push(shared->matches, pair, true);
+	}
+
+	free(buf_b);
+	return NULL;
+}
+
 /**
  * \brief      Finds matching functions of 2 given lists of functions using the same RzAnalysis core
  *
@@ -524,5 +558,8 @@ static void *analysis_match_functions(SharedContext *shared) {
  */
 RZ_API RZ_OWN RzAnalysisMatchResult *rz_analysis_match_functions(RzList /*<RzAnalysisFunction *>*/ *list_a, RzList /*<RzAnalysisFunction *>*/ *list_b, RZ_NONNULL RzAnalysisMatchOpt *opt) {
 	rz_return_val_if_fail(opt && opt->analysis_a && opt->analysis_b && list_a && list_b, NULL);
+	if (rz_list_length(list_a) == 1) {
+		return analysis_match_result_new(opt, list_b, list_a, (RzThreadFunction)analysis_match_one_function, (AllocateBuffer)function_data_new);
+	}
 	return analysis_match_result_new(opt, list_a, list_b, (RzThreadFunction)analysis_match_functions, (AllocateBuffer)function_data_new);
 }
diff --git a/librz/main/rz-diff.c b/librz/main/rz-diff.c
@@ -229,6 +229,7 @@ static void rz_diff_show_help(bool usage_only) {
 		"",         "",             "   bytes      | compare raw bytes in the files (only for small files)",
 		"",         "",             "   lines      | compare text files",
 		"",         "",             "   functions  | compare functions found in the files",
+		"",         "",             "              | optional -0 <fcn name|offset> to compare only one function",
 		"",         "",             "   classes    | compare classes found in the files",
 		"",         "",             "   command    | compare command output returned when executed in both files",
 		"",         "",             "              | require -0 <cmd> and -1 <cmd> is optional",
@@ -365,9 +366,7 @@ static void rz_diff_parse_arguments(int argc, const char **argv, DiffContext *ct
 		} else if (!strcmp(type, "lines")) {
 			rz_diff_ctx_set_type(ctx, DIFF_TYPE_LINES);
 		} else if (!strcmp(type, "functions")) {
-			if (ctx->input_a) {
-				rz_diff_error_opt(ctx, DIFF_OPT_ERROR, "option -t '%s' does not support -0.\n", type);
-			} else if (ctx->input_b) {
+			if (ctx->input_b) {
 				rz_diff_error_opt(ctx, DIFF_OPT_ERROR, "option -t '%s' does not support -1.\n", type);
 			}
 			ctx->option = DIFF_OPT_GRAPH;
@@ -2072,7 +2071,7 @@ static int comparePairFunctions(const RzAnalysisMatchPair *ma, const RzAnalysisM
  * Then the scores are shown in a table (when in quiet mode, the table
  * is headerless)
  * */
-static void core_diff_show(RzCore *core_a, RzCore *core_b, DiffMode mode, bool verbose) {
+static void core_diff_show(RzCore *core_a, RzCore *core_b, const char *addr_a, DiffMode mode, bool verbose) {
 	rz_return_if_fail(core_a && core_b);
 
 	char *output = NULL;
@@ -2085,8 +2084,22 @@ static void core_diff_show(RzCore *core_a, RzCore *core_b, DiffMode mode, bool v
 	RzAnalysisFunction *fcn_a = NULL, *fcn_b = NULL;
 	RzListIter *iter = NULL;
 	bool color = false, no_name = false;
-
-	fcns_a = rz_list_clone(rz_analysis_function_list(core_a->analysis));
+	if (RZ_STR_ISEMPTY(addr_a)) {
+		fcns_a = rz_list_clone(rz_analysis_function_list(core_a->analysis));
+	} else {
+		fcns_a = rz_list_new();
+		if (!fcns_a) {
+			RZ_LOG_ERROR("rz-diff: failed to create new list.\n");
+			goto fail;
+		}
+		ut64 addr = rz_num_get(core_a->num, addr_a);
+		RzAnalysisFunction *fcn = rz_analysis_get_function_at(core_a->analysis, addr);
+		if (!fcn) {
+			RZ_LOG_ERROR("rz-diff: failed to find %s in file0.\n", addr_a);
+			goto fail;
+		}
+		rz_list_append(fcns_a, fcn);
+	}
 	if (rz_list_empty(fcns_a)) {
 		RZ_LOG_ERROR("rz-diff: No functions found in file0.\n");
 		goto fail;
@@ -2296,7 +2309,7 @@ static bool rz_diff_graphs_files(DiffContext *ctx) {
 		if (ctx->verbose) {
 			fprintf(stderr, "rz-diff: start diffing.\n");
 		}
-		core_diff_show(a->core, b->core, ctx->mode, ctx->verbose);
+		core_diff_show(a->core, b->core, ctx->input_a, ctx->mode, ctx->verbose);
 	}
 
 	success = true;

diff --git a/test/db/tools/rz_diff b/test/db/tools/rz_diff
@@ -980,3 +980,22 @@ EXPECT=<<EOF
 `-------------------------------------------------------------------------------------------------'
 EOF
 RUN
+
+NAME=rz-diff functions comparison against only one function
+FILE==
+CMDS=!rz-diff -e 'analysis.fcnprefix=test' -e 'analysis.limits=true' -e 'analysis.from=0x401460' -e 'analysis.to=0x4033d0' -C -t functions -0 test.00401ae0 bins/other/rz-diff/true bins/other/rz-diff/false
+EXPECT=<<EOF
+.-------------------------------------------------------------------------------------------------.
+| name0         | size0 | addr0      | type     | similarity | addr1      | size1 | name1         |
+)-------------------------------------------------------------------------------------------------(
+| test.00401470 | 41    | 0x00401470 | UNLIKE   | 0.0746     | 0x00401ae0 | 228   | test.00401ae0 |
+| test.004019a0 | 137   | 0x004019a0 | UNLIKE   | 0.1228     | 0x00401ae0 | 228   | test.00401ae0 |
+| test.00401a30 | 162   | 0x00401a30 | UNLIKE   | 0.0746     | 0x00401ae0 | 228   | test.00401ae0 |
+| test.00401af0 | 228   | 0x00401af0 | PARTIAL  | 0.964912   | 0x00401ae0 | 228   | test.00401ae0 |
+| test.00401bf0 | 2728  | 0x00401bf0 | UNLIKE   | 0.0433     | 0x00401ae0 | 228   | test.00401ae0 |
+| test.00402740 | 425   | 0x00402740 | UNLIKE   | 0.1106     | 0x00401ae0 | 228   | test.00401ae0 |
+| test.004029b0 | 50    | 0x004029b0 | UNLIKE   | 0.0702     | 0x00401ae0 | 228   | test.00401ae0 |
+| test.00402e30 | 204   | 0x00402e30 | UNLIKE   | 0.1140     | 0x00401ae0 | 228   | test.00401ae0 |
+`-------------------------------------------------------------------------------------------------'
+EOF
+RUN