Add cache for source line

librz/bin/dbginfo.c

@@ -239,13 +239,117 @@ RZ_DEPRECATE RZ_API bool rz_bin_addr2line(RzBin *bin, ut64 addr, char *file, int
 	return false;
 }
 
-RZ_DEPRECATE RZ_API char *rz_bin_addr2text(RzBin *bin, ut64 addr, int origin) {
+static inline bool read_next_line(FILE *fd, ut64 *psize, RzStrBuf *sb) {
+	rz_return_val_if_fail(fd && sb, NULL);
+
+	ut64 size = 0;
+	char buf[1024] = { 0 };
+	while (true) {
+		ut64 r = fread(buf, 1, sizeof(buf) - 1, fd);
+		if (ferror(fd) || r <= 0) {
+			return false;
+		}
+		char *p = buf;
+		char *q = NULL;
+		do {
+			q = memchr(p, '\n', r);
+			p = q + 1;
+			if (q) {
+				ut64 sz = q - buf;
+				size += sz;
+				fseek(fd, -(long)(r - 1 - sz), SEEK_CUR);
+				if (ferror(fd)) {
+					return false;
+				}
+				rz_strbuf_append_n(sb, buf, sz);
+				if (psize) {
+					*psize = size;
+				}
+				return true;
+			}
+			size += r;
+			rz_strbuf_append_n(sb, buf, r);
+		} while (p - buf <= r && q);
+	}
+	return false;
+}
+
+static char *read_line(const char *file, int line, RzBinSourceLineCache *cache) {
+	rz_return_val_if_fail(file, NULL);
+	if (!(cache && cache->items)) {
+		rz_file_slurp_line(file, line, 0);
+	}
+	bool found = false;
+	bool found_off = false;
+	RzStrBuf sb = { 0 };
+	rz_strbuf_init(&sb);
+	RzBinSourceLineCacheItem *item = ht_pp_find(cache->items, file, &found);
+	if (found) {
+		if (!(item && item->fd)) {
+			return NULL;
+		} else {
+			ut64 offset = ht_uu_find(item->line_by_ln, line, &found_off);
+			if (!found_off) {
+				goto build;
+			}
+			fseek(item->fd, (long)offset, SEEK_SET);
+			read_next_line(item->fd, NULL, &sb);
+			return rz_strbuf_drain_nofree(&sb);
+		}
+	} else {
+	build:
+		if (!rz_file_exists(file)) {
+			ht_pp_insert(cache->items, file, NULL);
+			return NULL;
+		}
+		item = RZ_NEW0(RzBinSourceLineCacheItem);
+		if (!item) {
+			goto err;
+		}
+		item->fd = rz_sys_fopen(file, "r");
+		if (!item->fd) {
+			goto err;
+		}
+		item->line_by_ln = ht_uu_new0();
+		if (!item->line_by_ln) {
+			goto err;
+		}
+
+		ut64 i = 1;
+		ut64 offset = 0;
+		while (i < line) {
+			ut64 tmp_sz = 0;
+			sb.len = sb.ptrlen = 0;
+			if (!read_next_line(item->fd, &tmp_sz, &sb)) {
+				goto err;
+			}
+			++i;
+			offset += tmp_sz + 1;
+		}
+		ht_uu_update(item->line_by_ln, line, offset);
+		ht_pp_update(cache->items, file, item);
+
+		sb.len = sb.ptrlen = 0;
+		read_next_line(item->fd, NULL, &sb);
+		return rz_strbuf_drain_nofree(&sb);
+	}
+err:
+	if (item) {
+		ht_uu_free(item->line_by_ln);
+		fclose(item->fd);
+	}
+	free(item);
+	rz_strbuf_fini(&sb);
+	return NULL;
+}
+
+RZ_DEPRECATE RZ_API char *rz_bin_addr2text(RzBin *bin, ut64 addr, RzDebugInfoOption opt) {
 	rz_return_val_if_fail(bin, NULL);
-	if (!bin->cur || !bin->cur->o || !bin->cur->o->lines) {
+	if(!(bin->cur && bin->cur->o && bin->cur->o->lines)){
 		return NULL;
 	}
 	const RzBinSourceLineSample *s = rz_bin_source_line_info_get_first_at(bin->cur->o->lines, addr);
-	if (s && s->address != addr) {
+	if (!(s && s->address == addr)) {
 		// consider only exact matches, not inside of samples
 		return NULL;
 	}
@@ -255,31 +359,24 @@ RZ_DEPRECATE RZ_API char *rz_bin_addr2text(RzBin *bin, ut64 addr, int origin) {
 	if (!s) {
 		return NULL;
 	}
-	const char *file_nopath;
-	if (origin > 1) {
-		file_nopath = s->file;
-	} else {
-		file_nopath = strrchr(s->file, '/');
-		if (file_nopath) {
-			file_nopath++;
-		} else {
-			file_nopath = s->file;
-		}
-	}
+	const char *filepath = opt.abspath ? s->file : rz_file_basename(s->file);
 	if (!s->line) {
-		return strdup(file_nopath);
-	}
-	char *out = rz_file_slurp_line(s->file, s->line, 0);
-	if (out) {
-		rz_str_trim(out);
-		if (origin) {
-			char *res = rz_str_newf("%s:%d %s",
-				file_nopath, s->line,
-				out ? out : "");
-			free(out);
-			out = res;
-		}
-		return out;
+		return strdup(filepath);
 	}
-	return rz_str_newf("%s:%" PFMT32u, file_nopath, s->line);
+
+	RzStrBuf sb = { 0 };
+	rz_strbuf_initf(&sb, "%s:%" PFMT32u, filepath, s->line);
+	if (!opt.file) {
+		return rz_strbuf_drain_nofree(&sb);
+	}
+
+	char *out = read_line(s->file, s->line, &opt.cache);
+	if (!out) {
+		return rz_strbuf_drain_nofree(&sb);
+	}
+
+	rz_str_trim(out);
+	rz_strbuf_appendf(&sb, " %s", out);
+	free(out);
+	return rz_strbuf_drain_nofree(&sb);
 }

librz/core/disasm.c

@@ -118,7 +118,8 @@ typedef struct {
 	int tracespace;
 	int cyclespace;
 	int show_indent;
-	ut32 debuginfo;
+	RzDebugInfoOption debuginfo;
+
 	bool show_size;
 	bool show_trace;
 	bool show_family;
@@ -281,13 +282,6 @@ static inline void set_flag(uint32_t *flags, uint32_t flag, bool value) {
 	}
 }
 
-enum debuginfo_flag_enum_t {
-	DEBUGINFO = 1 << 0,
-	DEBUGINFO_ABSPATH = 1 << 1,
-	DEBUGINFO_FILE = 1 << 2,
-	DEBUGINFO_LINES = 1 << 3,
-};
-
 static void ds_setup_print_pre(RzDisasmState *ds, bool tail, bool middle);
 static void ds_setup_pre(RzDisasmState *ds, bool tail, bool middle);
 static void ds_print_pre(RzDisasmState *ds, bool fcnline);
@@ -566,6 +560,22 @@ static void ds_print_esil_analysis_fini(RzDisasmState *ds) {
 	}
 }
 
+static void RzBinSourceLineCacheItem_free(RzBinSourceLineCacheItem *x) {
+	if (!x) {
+		return;
+	}
+	fclose(x->fd);
+	ht_uu_free(x->line_by_ln);
+	free(x);
+}
+
+static void RzBinSourceLineCacheItem_HtPPKv_free(HtPPKv *x) {
+	if (!x) {
+		return;
+	}
+	RzBinSourceLineCacheItem_free(x->value);
+}
+
 static RzDisasmState *ds_init(RzCore *core) {
 	RzDisasmState *ds = RZ_NEW0(RzDisasmState);
 	if (!ds) {
@@ -631,10 +641,11 @@ static RzDisasmState *ds_init(RzCore *core) {
 	ds->tracespace = rz_config_get_i(core->config, "asm.tracespace");
 	ds->cyclespace = rz_config_get_i(core->config, "asm.cyclespace");
 
-	set_flag(&ds->debuginfo, DEBUGINFO, rz_config_get_b(core->config, "asm.debuginfo"));
-	set_flag(&ds->debuginfo, DEBUGINFO_FILE, rz_config_get_b(core->config, "asm.debuginfo.file"));
-	set_flag(&ds->debuginfo, DEBUGINFO_ABSPATH, rz_config_get_b(core->config, "asm.debuginfo.abspath"));
-	set_flag(&ds->debuginfo, DEBUGINFO_LINES, rz_config_get_b(core->config, "asm.debuginfo.lines"));
+	ds->debuginfo.enable = rz_config_get_b(core->config, "asm.debuginfo");
+	ds->debuginfo.file = rz_config_get_b(core->config, "asm.debuginfo.file");
+	ds->debuginfo.abspath = rz_config_get_b(core->config, "asm.debuginfo.abspath");
+	ds->debuginfo.lines = rz_config_get_b(core->config, "asm.debuginfo.lines");
+	ds->debuginfo.cache.items = ht_pp_new(NULL, RzBinSourceLineCacheItem_HtPPKv_free, NULL);
 
 	ds->show_lines_call = ds->show_lines ? rz_config_get_b(core->config, "asm.lines.call") : false;
 	ds->show_lines_ret = ds->show_lines ? rz_config_get_b(core->config, "asm.lines.ret") : false;
@@ -838,6 +849,7 @@ static void ds_free(RzDisasmState *ds) {
 	ds_reflines_fini(ds);
 	ds_print_esil_analysis_fini(ds);
 	sdb_free(ds->ssa);
+	ht_pp_free(ds->debuginfo.cache.items);
 	free(ds->comment);
 	free(ds->line);
 	free(ds->line_col);
@@ -3699,16 +3711,12 @@ static void ds_align_comment(RzDisasmState *ds) {
 }
 
 static void ds_print_debuginfo(RzDisasmState *ds) {
-	if (!(ds->debuginfo & DEBUGINFO)) {
+	if (!ds->debuginfo.enable)
 		return;
-	}
 
-	if (ds->debuginfo & DEBUGINFO_LINES) {
-		// TODO: cache value in ds
-		int file_and_abspath = (ds->debuginfo & DEBUGINFO_FILE) ? 1 : 0;
-		file_and_abspath += (ds->debuginfo & DEBUGINFO_ABSPATH) ? 1 : 0;
+	if (ds->debuginfo.lines) {
 		free(ds->sl);
-		ds->sl = rz_bin_addr2text(ds->core->bin, ds->at, file_and_abspath);
+		ds->sl = rz_bin_addr2text(ds->core->bin, ds->at, ds->debuginfo);
 		if (RZ_STR_ISEMPTY(ds->sl))
 			return;
 		if (ds->osl && !(ds->osl && strcmp(ds->sl, ds->osl)))

librz/include/rz_bin.h

@@ -163,7 +163,7 @@ typedef enum {
 } RzBinLanguage;
 
 #define RZ_BIN_LANGUAGE_MASK(x)       ((x) & ~RZ_BIN_LANGUAGE_BLOCKS)
-#define RZ_BIN_LANGUAGE_HAS_BLOCKS(x) ((x)&RZ_BIN_LANGUAGE_BLOCKS)
+#define RZ_BIN_LANGUAGE_HAS_BLOCKS(x) ((x) & RZ_BIN_LANGUAGE_BLOCKS)
 
 enum {
 	RZ_BIN_CLASS_PRIVATE,
@@ -503,6 +503,23 @@ typedef struct rz_bin_source_line_info_builder_t {
 	RzStrConstPool filename_pool;
 } RzBinSourceLineInfoBuilder;
 
+typedef struct {
+	FILE *fd;
+	HtUU /*<ut64, ut64>*/ *line_by_ln;
+} RzBinSourceLineCacheItem;
+
+typedef struct {
+	HtPP /*<const char*, RzBinSourceLineCacheItem *>*/ *items;
+} RzBinSourceLineCache;
+
+typedef struct {
+	RzBinSourceLineCache cache;
+	bool enable : 1;
+	bool file : 1;
+	bool abspath : 1;
+	bool lines : 1;
+} RzDebugInfoOption;
+
 RZ_API void rz_bin_source_line_info_builder_init(RzBinSourceLineInfoBuilder *builder);
 RZ_API void rz_bin_source_line_info_builder_fini(RzBinSourceLineInfoBuilder *builder);
 RZ_API void rz_bin_source_line_info_builder_push_sample(RzBinSourceLineInfoBuilder *builder, ut64 address, ut32 line, ut32 column, const char *file);
@@ -1018,7 +1035,7 @@ RZ_API RZ_BORROW RzBinSection *rz_bin_get_section_at(RzBinObject *o, ut64 off, i
 
 /* dbginfo.c */
 RZ_DEPRECATE RZ_API bool rz_bin_addr2line(RzBin *bin, ut64 addr, char *file, int len, int *line);
-RZ_DEPRECATE RZ_API char *rz_bin_addr2text(RzBin *bin, ut64 addr, int origin);
+RZ_DEPRECATE RZ_API char *rz_bin_addr2text(RzBin *bin, ut64 addr, RzDebugInfoOption opt);
 
 /* filter.c */
 RZ_API void rz_bin_load_filter(RzBin *bin, ut64 rules);

librz/main/rz-bin.c

@@ -586,7 +586,10 @@ static int rabin_do_operation(RzBin *bin, const char *op, int rad, const char *o
 
 static int rabin_show_srcline(RzBin *bin, ut64 at) {
 	char *srcline;
-	if (at != UT64_MAX && (srcline = rz_bin_addr2text(bin, at, true))) {
+	RzDebugInfoOption option = { 0 };
+	option.file = true;
+	option.abspath = true;
+	if (at != UT64_MAX && (srcline = rz_bin_addr2text(bin, at, option))) {
 		printf("%s\n", srcline);
 		free(srcline);
 		return true;

test/db/tools/rz_bin

@@ -1037,7 +1037,7 @@ RUN
 
 NAME=rz-bin -d
 FILE=bins/elf/analysis/dwarftest
-CMDS=!rz-bin -d ${RZ_FILE} | grep -c "0x00400535	/tmp/r2-regressions/.//dwarftest.c"
+CMDS=!rz-bin -d ${RZ_FILE} | grep -c "0x00400535	/tmp/r2-regressions/./dwarftest.c"
 EXPECT=<<EOF
 1
 EOF