Optimization

librz/bin/dbginfo.c

@@ -4,7 +4,8 @@
 // SPDX-License-Identifier: LGPL-3.0-only
 
 #include <rz_types.h>
-#include <rz_bin.h>
+#include <rz_bin_source_line.h>
+#include <ctype.h>
 
 RZ_API void rz_bin_source_line_info_builder_init(RzBinSourceLineInfoBuilder *builder) {
 	rz_vector_init(&builder->samples, sizeof(RzBinSourceLineSample), NULL, NULL);
@@ -216,12 +217,14 @@ RZ_API const RzBinSourceLineSample *rz_bin_source_line_info_get_next(const RzBin
 	return next;
 }
 
-RZ_DEPRECATE RZ_API bool rz_bin_addr2line(RzBin *bin, ut64 addr, char *file, int len, int *line) {
-	rz_return_val_if_fail(bin, false);
-	if (!bin->cur || !bin->cur->o || !bin->cur->o->lines) {
-		return false;
-	}
-	const RzBinSourceLineSample *s = rz_bin_source_line_info_get_first_at(bin->cur->o->lines, addr);
+RZ_API bool rz_bin_source_line_addr2line(
+	RZ_BORROW RZ_IN RZ_NONNULL const RzBinSourceLineInfo *sl,
+	ut64 addr,
+	RZ_BORROW RZ_OUT RZ_NULLABLE char *file,
+	int len,
+	RZ_BORROW RZ_OUT RZ_NULLABLE int *line) {
+	rz_return_val_if_fail(sl, false);
+	const RzBinSourceLineSample *s = rz_bin_source_line_info_get_first_at(sl, addr);
 	if (!s || s->address != addr) {
 		// consider only exact matches, not inside of samples
 		return false;
@@ -236,51 +239,40 @@ RZ_DEPRECATE RZ_API bool rz_bin_addr2line(RzBin *bin, ut64 addr, char *file, int
 			*file = 0;
 		}
 	}
-	return false;
+	return true;
 }
 
-static const char *cache_lines(RzBinSourceLineCacheItem *x, ut64 line) {
-	rz_warn_if_fail(line > x->cached_lines);
+static char *str_trim_left_right(char *l, char *r) {
+	l = (char *)rz_str_trim_head_ro(l);
+	for (; r > l && isspace(*r); --r) {
+		*r = '\0';
+	}
+	return l;
+}
 
+static void cache_lines(RzBinSourceLineCacheItem *x) {
 	if (!x->file_content) {
-		return NULL;
-	}
-	bool found = false;
-	char *str = ht_up_find(x->line_by_ln, x->cached_lines, &found);
-	if (!(found && str)) {
-		rz_warn_if_reached();
-		return NULL;
+		return;
 	}
 
-	ut64 i = x->cached_lines;
-	char *p = str;
+	char *p = x->file_content;
 	char *q = NULL;
 	do {
 		q = strchr(p, '\n');
-		++i;
-		if (q) {
-			p = q + 1;
-			*q = '\0';
-			ht_up_update(x->line_by_ln, i, q + 1);
-			x->cached_lines = i;
+		if (!q) {
+			break;
 		}
+		*q = '\0';
+		p = str_trim_left_right(p, q);
+		rz_pvector_push(x->line_by_ln, p);
+		p = q + 1;
 	} while ((p && p - x->file_content < x->file_size));
-
-	str = ht_up_find(x->line_by_ln, line, &found);
-	if (!(found && str)) {
-		return NULL;
-	}
-	q = rz_str_trim_nc(str);
-	if (q != str) {
-		ht_up_update(x->line_by_ln, line, q);
-	}
-	return q;
 }
 
 static const char *read_line(const char *file, int line, RzBinSourceLineCache *cache) {
-	rz_return_val_if_fail(file, NULL);
+	rz_return_val_if_fail(file && line >= 1, NULL);
 	if (!(cache && cache->items)) {
-		rz_file_slurp_line(file, line, 0);
+		return rz_file_slurp_line(file, line, 0);
 	}
 	bool found = false;
 	char *content = NULL;
@@ -290,15 +282,7 @@ static const char *read_line(const char *file, int line, RzBinSourceLineCache *c
 		if (!(item && item->file_content)) {
 			return NULL;
 		} else {
-			char *str = ht_up_find(item->line_by_ln, line, NULL);
-			if (!str) {
-				return NULL;
-			}
-			char *q = rz_str_trim_nc(str);
-			if (q != str) {
-				ht_up_update(item->line_by_ln, line, q);
-			}
-			return q;
+			return rz_pvector_at(item->line_by_ln, line - 1);
 		}
 	} else {
 		content = rz_file_slurp(file, &sz);
@@ -312,36 +296,35 @@ static const char *read_line(const char *file, int line, RzBinSourceLineCache *c
 		}
 		item->file_content = content;
 		item->file_size = sz;
-		item->line_by_ln = ht_up_new0();
+		item->line_by_ln = rz_pvector_new(NULL);
 		if (!item->line_by_ln) {
 			goto err;
 		}
-		item->cached_lines = 1;
-		ht_up_update(item->line_by_ln, 1, item->file_content);
 		ht_pp_update(cache->items, file, item);
-		return cache_lines(item, line);
+
+		rz_pvector_reserve(item->line_by_ln, line);
+		cache_lines(item);
+		return rz_pvector_at(item->line_by_ln, line - 1);
 	}
 err:
 	if (item) {
-		ht_up_free(item->line_by_ln);
+		rz_pvector_free(item->line_by_ln);
 	}
 	free(content);
 	free(item);
 	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)) {
-		return NULL;
-	}
-	const RzBinSourceLineSample *s = rz_bin_source_line_info_get_first_at(bin->cur->o->lines, addr);
+RZ_API RZ_OWN char *rz_bin_source_line_addr2text(
+	RZ_BORROW RZ_IN RZ_NONNULL const RzBinSourceLineInfo *sl, ut64 addr, RzDebugInfoOption opt) {
+	rz_return_val_if_fail(sl, NULL);
+	const RzBinSourceLineSample *s = rz_bin_source_line_info_get_first_at(sl, addr);
 	if (!(s && s->address == addr)) {
 		// consider only exact matches, not inside of samples
 		return NULL;
 	}
 	while (s && !s->file) {
-		s = rz_bin_source_line_info_get_next(bin->cur->o->lines, s);
+		s = rz_bin_source_line_info_get_next(sl, s);
 	}
 	if (!s) {
 		return NULL;

librz/core/cgraph.c

@@ -416,8 +416,12 @@ static char *block_line(RzCore *core, ut64 addr, RzAnalysisBlock *bb) {
 	int line = 0, oline = 0, idx = 0;
 	int is_html = rz_cons_singleton()->is_html;
 	ut64 end = bb->addr + bb->size - 2;
+	RzBinObject *o = rz_bin_cur_object(core->bin);
+	RzBinSourceLineInfo *sl = o ? o->lines : NULL;
 	for (ut64 at = bb->addr; at < end; at += 2) {
-		rz_bin_addr2line(core->bin, at, file, sizeof(file) - 1, &line);
+		if (sl) {
+			rz_bin_source_line_addr2line(sl, at, file, sizeof(file) - 1, &line);
+		}
 		if (line != 0 && line != oline && strcmp(file, "??") != 0) {
 			file_str = rz_file_slurp_line(file, line, 0);
 			if (file_str) {

librz/core/cmd/cmd_debug.c

@@ -454,7 +454,9 @@ static bool step_line(RzCore *core, int times) {
 	}
 	file[0] = 0;
 	file2[0] = 0;
-	if (rz_bin_addr2line(core->bin, off, file, sizeof(file), &line)) {
+	RzBinObject *o = rz_bin_cur_object(core->bin);
+	RzBinSourceLineInfo *sl = o ? o->lines : NULL;
+	if (sl && rz_bin_source_line_addr2line(sl, off, file, sizeof(file), &line)) {
 		char *ptr = rz_file_slurp_line(file, line, 0);
 		RZ_LOG_INFO("--> 0x%08" PFMT64x " %s : %d\n", off, file, line);
 		RZ_LOG_INFO("--> %s\n", ptr);
@@ -467,7 +469,7 @@ static bool step_line(RzCore *core, int times) {
 	do {
 		rz_debug_step(core->dbg, 1);
 		off = rz_debug_reg_get(core->dbg, "PC");
-		if (!rz_bin_addr2line(core->bin, off, file2, sizeof(file2), &line2)) {
+		if (!(sl && rz_bin_source_line_addr2line(sl, off, file2, sizeof(file2), &line2))) {
 			if (find_meta) {
 				continue;
 			}

librz/core/disasm.c

@@ -271,17 +271,6 @@ typedef struct {
 	RzPVector /*<RzAnalysisDisasmText *>*/ *vec;
 } RzDisasmState;
 
-// Function to set or clear a flag based on a boolean value
-static inline void set_flag(uint32_t *flags, uint32_t flag, bool value) {
-	if (value) {
-		// Set the flag
-		*flags |= flag;
-	} else {
-		// Clear the flag
-		*flags &= ~flag;
-	}
-}
-
 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);
@@ -565,7 +554,7 @@ static void RzBinSourceLineCacheItem_free(RzBinSourceLineCacheItem *x) {
 		return;
 	}
 	free(x->file_content);
-	ht_up_free(x->line_by_ln);
+	rz_pvector_free(x->line_by_ln);
 	free(x);
 }
 
@@ -3714,29 +3703,20 @@ static void ds_print_debuginfo(RzDisasmState *ds) {
 	if (!ds->debuginfo.enable)
 		return;
 
-	if (ds->debuginfo.lines) {
+	RzBinObject *o = rz_bin_cur_object(ds->core->bin);
+	RzBinSourceLineInfo *sl = o ? o->lines : NULL;
+	if (ds->debuginfo.lines && sl) {
 		free(ds->sl);
-		ds->sl = rz_bin_addr2text(ds->core->bin, ds->at, ds->debuginfo);
+		ds->sl = rz_bin_source_line_addr2text(sl, ds->at, ds->debuginfo);
 		if (RZ_STR_ISEMPTY(ds->sl))
 			return;
 		if (ds->osl && !(ds->osl && strcmp(ds->sl, ds->osl)))
 			return;
-		char *line = strdup(ds->sl);
-		rz_str_replace_char(line, '\t', ' ');
-		rz_str_replace_char(line, '\x1b', ' ');
-		rz_str_replace_char(line, '\r', ' ');
-		rz_str_replace_char(line, '\n', '\x00');
-		rz_str_trim(line);
-		if (RZ_STR_ISEMPTY(line)) {
-			free(line);
-			return;
-		}
 		ds_align_comment(ds);
-		theme_printf(comment, "; %s", line);
+		theme_printf(comment, "; %s", ds->sl);
 		free(ds->osl);
 		ds->osl = ds->sl;
 		ds->sl = NULL;
-		free(line);
 	}
 }
 

librz/include/meson.build

@@ -5,6 +5,7 @@ include_files = [
   'rz_basefind.h',
   'rz_bin.h',
   'rz_bin_dwarf.h',
+  'rz_bin_source_line.h',
   'rz_bind.h',
   'rz_bp.h',
   'rz_cmd.h',

librz/include/rz_bin.h

@@ -10,10 +10,10 @@
 #include <rz_util/ht_pu.h>
 #include <rz_demangler.h>
 #include <rz_hash.h>
+#include <rz_bin_source_line.h>
 
 typedef struct rz_bin_t RzBin;
 typedef struct rz_bin_file_t RzBinFile;
-typedef struct rz_bin_source_line_info_t RzBinSourceLineInfo;
 typedef struct rz_bin_reloc_storage_t RzBinRelocStorage;
 
 #include <rz_bin_dwarf.h>
@@ -434,104 +434,6 @@ typedef struct rz_bin_trycatch_t {
 RZ_API RzBinTrycatch *rz_bin_trycatch_new(ut64 source, ut64 from, ut64 to, ut64 handler, ut64 filter);
 RZ_API void rz_bin_trycatch_free(RzBinTrycatch *tc);
 
-/**
- * \brief A single sample of source line info for a specific address
- *
- * If at least one of the line, column and file members is not 0/NULL, such a sample specifies the line info
- * for all addresses greater or equal to address until the next address that has another sample.
- *
- * If all the members line, column and file are 0/NULL, then this is a closing sample, indicating that the
- * previous entry stops here. The address is the first address **not contained** by the previous record.
- * Such a case corresponds for example to what DW_LNE_end_sequence emits in Dwarf.
- * Use rz_bin_source_line_sample_is_closing() for checking if a sample is closing.
- */
-typedef struct rz_bin_source_line_sample_t {
-	/**
-	 * The first address that is covered by the given line and column,
-	 * or, if all other members are 0/NULL, this is the first.
-	 */
-	ut64 address;
-
-	/**
-	 * If > 0, then indicates the line for the given address and the following.
-	 * If == 0, then indicates that no line information is known.
-	 *
-	 * 32bit for this value is an intentional decision to lower memory consumption.
-	 */
-	ut32 line;
-
-	/**
-	 * If > 0, then indicates the column.
-	 * If == 0, then no column information is known.
-	 *
-	 * 32bit for this value is an intentional decision to lower memory consumption.
-	 */
-	ut32 column;
-
-	/**
-	 * Filename, which must come out of the const pool of the owning
-	 * RzBinSourceLineInfo or RzBinSourceLineInfoBuilder.
-	 */
-	const char *file;
-} RzBinSourceLineSample;
-
-/*
- * see documentation of RzBinSourceLineSample about what closing exactly means.
- */
-static inline bool rz_bin_source_line_sample_is_closing(const RzBinSourceLineSample *s) {
-	return !s->line && !s->column && !s->file;
-}
-
-struct rz_bin_source_line_info_t {
-	/**
-	 * \brief All source line references for given adresses
-	 *
-	 * These elements must be sorted by address and addresses must be unique, so binary search can be applied.
-	 * Source file information is not contained within this array because source file changes
-	 * are generally much sparser than line changes.
-	 */
-	RzBinSourceLineSample *samples;
-	size_t samples_count;
-	RzStrConstPool filename_pool;
-}; // RzBinSourceLineInfo
-
-/**
- * Temporary data structure for building an RzBinSourceLineInfo.
- */
-typedef struct rz_bin_source_line_info_builder_t {
-	RzVector /*<RzBinSourceLineSample>*/ samples; //< may be unsorted and will be sorted in the finalization step
-	RzStrConstPool filename_pool;
-} RzBinSourceLineInfoBuilder;
-
-typedef struct {
-	char *file_content;
-	ut64 file_size;
-	ut64 cached_lines;
-	HtUP /*<ut64, const char *>*/ *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);
-RZ_API RzBinSourceLineInfo *rz_bin_source_line_info_builder_build_and_fini(RzBinSourceLineInfoBuilder *builder);
-
-RZ_API bool rz_bin_source_line_info_merge(RZ_BORROW RZ_NONNULL RzBinSourceLineInfo *dst, RZ_BORROW RZ_NONNULL RzBinSourceLineInfo *src);
-RZ_API void rz_bin_source_line_info_free(RzBinSourceLineInfo *sli);
-RZ_API const RzBinSourceLineSample *rz_bin_source_line_info_get_first_at(const RzBinSourceLineInfo *sli, ut64 addr);
-RZ_API const RzBinSourceLineSample *rz_bin_source_line_info_get_next(const RzBinSourceLineInfo *sli, RZ_NONNULL const RzBinSourceLineSample *cur);
-
 typedef struct rz_bin_plugin_t {
 	char *name;
 	char *desc;
@@ -1035,10 +937,6 @@ RZ_API const char *rz_bin_get_meth_flag_string(ut64 flag, bool compact);
 
 RZ_API RZ_BORROW RzBinSection *rz_bin_get_section_at(RzBinObject *o, ut64 off, int va);
 
-/* 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, RzDebugInfoOption opt);
-
 /* filter.c */
 RZ_API void rz_bin_load_filter(RzBin *bin, ut64 rules);
 RZ_API bool rz_bin_strpurge(RzBin *bin, const char *str, ut64 addr);