Merge branch 'jk/grep-binary-attribute'

* jk/grep-binary-attribute:
  grep: pre-load userdiff drivers when threaded
  grep: load file data after checking binary-ness
  grep: respect diff attributes for binary-ness
  grep: cache userdiff_driver in grep_source
  grep: drop grep_buffer's "name" parameter
  convert git-grep to use grep_source interface
  grep: refactor the concept of "grep source" into an object
  grep: move sha1-reading mutex into low-level code
  grep: make locking flag global

5 files changed, 308 insertions, 192 deletions

diff --git a/builtin/grep.c b/builtin/grep.c
index dc6de83ab7..e4ea900783 100644
--- a/builtin/grep.c
+++ b/builtin/grep.c
@@ -29,25 +29,12 @@ static int use_threads = 1;
 #define THREADS 8
 static pthread_t threads[THREADS];
 
-static void *load_sha1(const unsigned char *sha1, unsigned long *size,
-		       const char *name);
-static void *load_file(const char *filename, size_t *sz);
-
-enum work_type {WORK_SHA1, WORK_FILE};
-
 /* We use one producer thread and THREADS consumer
  * threads. The producer adds struct work_items to 'todo' and the
  * consumers pick work items from the same array.
  */
 struct work_item {
-	enum work_type type;
-	char *name;
-
-	/* if type == WORK_SHA1, then 'identifier' is a SHA1,
-	 * otherwise type == WORK_FILE, and 'identifier' is a NUL
-	 * terminated filename.
-	 */
-	void *identifier;
+	struct grep_source source;
 	char done;
 	struct strbuf out;
 };
@@ -85,21 +72,6 @@ static inline void grep_unlock(void)
 		pthread_mutex_unlock(&grep_mutex);
 }
 
-/* Used to serialize calls to read_sha1_file. */
-static pthread_mutex_t read_sha1_mutex;
-
-static inline void read_sha1_lock(void)
-{
-	if (use_threads)
-		pthread_mutex_lock(&read_sha1_mutex);
-}
-
-static inline void read_sha1_unlock(void)
-{
-	if (use_threads)
-		pthread_mutex_unlock(&read_sha1_mutex);
-}
-
 /* Signalled when a new work_item is added to todo. */
 static pthread_cond_t cond_add;
 
@@ -113,7 +85,8 @@ static pthread_cond_t cond_result;
 
 static int skip_first_line;
 
-static void add_work(enum work_type type, char *name, void *id)
+static void add_work(struct grep_opt *opt, enum grep_source_type type,
+		     const char *name, const void *id)
 {
 	grep_lock();
 
@@ -121,9 +94,9 @@ static void add_work(enum work_type type, char *name, void *id)
 		pthread_cond_wait(&cond_write, &grep_mutex);
 	}
 
-	todo[todo_end].type = type;
-	todo[todo_end].name = name;
-	todo[todo_end].identifier = id;
+	grep_source_init(&todo[todo_end].source, type, name, id);
+	if (opt->binary != GREP_BINARY_TEXT)
+		grep_source_load_driver(&todo[todo_end].source);
 	todo[todo_end].done = 0;
 	strbuf_reset(&todo[todo_end].out);
 	todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
@@ -151,21 +124,6 @@ static struct work_item *get_work(void)
 	return ret;
 }
 
-static void grep_sha1_async(struct grep_opt *opt, char *name,
-			    const unsigned char *sha1)
-{
-	unsigned char *s;
-	s = xmalloc(20);
-	memcpy(s, sha1, 20);
-	add_work(WORK_SHA1, name, s);
-}
-
-static void grep_file_async(struct grep_opt *opt, char *name,
-			    const char *filename)
-{
-	add_work(WORK_FILE, name, xstrdup(filename));
-}
-
 static void work_done(struct work_item *w)
 {
 	int old_done;
@@ -192,8 +150,7 @@ static void work_done(struct work_item *w)
 
 			write_or_die(1, p, len);
 		}
-		free(w->name);
-		free(w->identifier);
+		grep_source_clear(&w->source);
 	}
 
 	if (old_done != todo_done)
@@ -216,25 +173,8 @@ static void *run(void *arg)
 			break;
 
 		opt->output_priv = w;
-		if (w->type == WORK_SHA1) {
-			unsigned long sz;
-			void* data = load_sha1(w->identifier, &sz, w->name);
-
-			if (data) {
-				hit |= grep_buffer(opt, w->name, data, sz);
-				free(data);
-			}
-		} else if (w->type == WORK_FILE) {
-			size_t sz;
-			void* data = load_file(w->identifier, &sz);
-			if (data) {
-				hit |= grep_buffer(opt, w->name, data, sz);
-				free(data);
-			}
-		} else {
-			assert(0);
-		}
-
+		hit |= grep_source(opt, &w->source);
+		grep_source_clear_data(&w->source);
 		work_done(w);
 	}
 	free_grep_patterns(arg);
@@ -254,11 +194,12 @@ static void start_threads(struct grep_opt *opt)
 	int i;
 
 	pthread_mutex_init(&grep_mutex, NULL);
-	pthread_mutex_init(&read_sha1_mutex, NULL);
+	pthread_mutex_init(&grep_read_mutex, NULL);
 	pthread_mutex_init(&grep_attr_mutex, NULL);
 	pthread_cond_init(&cond_add, NULL);
 	pthread_cond_init(&cond_write, NULL);
 	pthread_cond_init(&cond_result, NULL);
+	grep_use_locks = 1;
 
 	for (i = 0; i < ARRAY_SIZE(todo); i++) {
 		strbuf_init(&todo[i].out, 0);
@@ -302,17 +243,16 @@ static int wait_all(void)
 	}
 
 	pthread_mutex_destroy(&grep_mutex);
-	pthread_mutex_destroy(&read_sha1_mutex);
+	pthread_mutex_destroy(&grep_read_mutex);
 	pthread_mutex_destroy(&grep_attr_mutex);
 	pthread_cond_destroy(&cond_add);
 	pthread_cond_destroy(&cond_write);
 	pthread_cond_destroy(&cond_result);
+	grep_use_locks = 0;
 
 	return hit;
 }
 #else /* !NO_PTHREADS */
-#define read_sha1_lock()
-#define read_sha1_unlock()
 
 static int wait_all(void)
 {
@@ -371,21 +311,9 @@ static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type
 {
 	void *data;
 
-	read_sha1_lock();
+	grep_read_lock();
 	data = read_sha1_file(sha1, type, size);
-	read_sha1_unlock();
-	return data;
-}
-
-static void *load_sha1(const unsigned char *sha1, unsigned long *size,
-		       const char *name)
-{
-	enum object_type type;
-	void *data = lock_and_read_sha1_file(sha1, &type, size);
-
-	if (!data)
-		error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
-
+	grep_read_unlock();
 	return data;
 }
 
@@ -393,7 +321,6 @@ static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 		     const char *filename, int tree_name_len)
 {
 	struct strbuf pathbuf = STRBUF_INIT;
-	char *name;
 
 	if (opt->relative && opt->prefix_length) {
 		quote_path_relative(filename + tree_name_len, -1, &pathbuf,
@@ -403,87 +330,51 @@ static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 		strbuf_addstr(&pathbuf, filename);
 	}
 
-	name = strbuf_detach(&pathbuf, NULL);
-
 #ifndef NO_PTHREADS
 	if (use_threads) {
-		grep_sha1_async(opt, name, sha1);
+		add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, sha1);
+		strbuf_release(&pathbuf);
 		return 0;
 	} else
 #endif
 	{
+		struct grep_source gs;
 		int hit;
-		unsigned long sz;
-		void *data = load_sha1(sha1, &sz, name);
-		if (!data)
-			hit = 0;
-		else
-			hit = grep_buffer(opt, name, data, sz);
-
-		free(data);
-		free(name);
-		return hit;
-	}
-}
 
-static void *load_file(const char *filename, size_t *sz)
-{
-	struct stat st;
-	char *data;
-	int i;
+		grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, sha1);
+		strbuf_release(&pathbuf);
+		hit = grep_source(opt, &gs);
 
-	if (lstat(filename, &st) < 0) {
-	err_ret:
-		if (errno != ENOENT)
-			error(_("'%s': %s"), filename, strerror(errno));
-		return NULL;
-	}
-	if (!S_ISREG(st.st_mode))
-		return NULL;
-	*sz = xsize_t(st.st_size);
-	i = open(filename, O_RDONLY);
-	if (i < 0)
-		goto err_ret;
-	data = xmalloc(*sz + 1);
-	if (st.st_size != read_in_full(i, data, *sz)) {
-		error(_("'%s': short read %s"), filename, strerror(errno));
-		close(i);
-		free(data);
-		return NULL;
+		grep_source_clear(&gs);
+		return hit;
 	}
-	close(i);
-	data[*sz] = 0;
-	return data;
 }
 
 static int grep_file(struct grep_opt *opt, const char *filename)
 {
 	struct strbuf buf = STRBUF_INIT;
-	char *name;
 
 	if (opt->relative && opt->prefix_length)
 		quote_path_relative(filename, -1, &buf, opt->prefix);
 	else
 		strbuf_addstr(&buf, filename);
-	name = strbuf_detach(&buf, NULL);
 
 #ifndef NO_PTHREADS
 	if (use_threads) {
-		grep_file_async(opt, name, filename);
+		add_work(opt, GREP_SOURCE_FILE, buf.buf, filename);
+		strbuf_release(&buf);
 		return 0;
 	} else
 #endif
 	{
+		struct grep_source gs;
 		int hit;
-		size_t sz;
-		void *data = load_file(filename, &sz);
-		if (!data)
-			hit = 0;
-		else
-			hit = grep_buffer(opt, name, data, sz);
 
-		free(data);
-		free(name);
+		grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename);
+		strbuf_release(&buf);
+		hit = grep_source(opt, &gs);
+
+		grep_source_clear(&gs);
 		return hit;
 	}
 }
@@ -612,10 +503,10 @@ static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 		struct strbuf base;
 		int hit, len;
 
-		read_sha1_lock();
+		grep_read_lock();
 		data = read_object_with_reference(obj->sha1, tree_type,
 						  &size, NULL);
-		read_sha1_unlock();
+		grep_read_unlock();
 
 		if (!data)
 			die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
@@ -1027,8 +918,6 @@ int cmd_grep(int argc, const char **argv, const char *prefix)
 	use_threads = 0;
 #endif
 
-	opt.use_threads = use_threads;
-
 #ifndef NO_PTHREADS
 	if (use_threads) {
 		if (!(opt.name_only || opt.unmatch_name_only || opt.count)
diff --git a/grep.c b/grep.c
index 486230b511..3821400966 100644
--- a/grep.c
+++ b/grep.c
@@ -807,38 +807,43 @@ static void show_line(struct grep_opt *opt, char *bol, char *eol,
 }
 
 #ifndef NO_PTHREADS
+int grep_use_locks;
+
 /*
  * This lock protects access to the gitattributes machinery, which is
  * not thread-safe.
  */
 pthread_mutex_t grep_attr_mutex;
 
-static inline void grep_attr_lock(struct grep_opt *opt)
+static inline void grep_attr_lock(void)
 {
-	if (opt->use_threads)
+	if (grep_use_locks)
 		pthread_mutex_lock(&grep_attr_mutex);
 }
 
-static inline void grep_attr_unlock(struct grep_opt *opt)
+static inline void grep_attr_unlock(void)
 {
-	if (opt->use_threads)
+	if (grep_use_locks)
 		pthread_mutex_unlock(&grep_attr_mutex);
 }
+
+/*
+ * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
+ */
+pthread_mutex_t grep_read_mutex;
+
 #else
-#define grep_attr_lock(opt)
-#define grep_attr_unlock(opt)
+#define grep_attr_lock()
+#define grep_attr_unlock()
 #endif
 
-static int match_funcname(struct grep_opt *opt, const char *name, char *bol, char *eol)
+static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
 {
 	xdemitconf_t *xecfg = opt->priv;
 	if (xecfg && !xecfg->find_func) {
-		struct userdiff_driver *drv;
-		grep_attr_lock(opt);
-		drv = userdiff_find_by_path(name);
-		grep_attr_unlock(opt);
-		if (drv && drv->funcname.pattern) {
-			const struct userdiff_funcname *pe = &drv->funcname;
+		grep_source_load_driver(gs);
+		if (gs->driver->funcname.pattern) {
+			const struct userdiff_funcname *pe = &gs->driver->funcname;
 			xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
 		} else {
 			xecfg = opt->priv = NULL;
@@ -858,33 +863,33 @@ static int match_funcname(struct grep_opt *opt, const char *name, char *bol, cha
 	return 0;
 }
 
-static void show_funcname_line(struct grep_opt *opt, const char *name,
-			       char *buf, char *bol, unsigned lno)
+static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
+			       char *bol, unsigned lno)
 {
-	while (bol > buf) {
+	while (bol > gs->buf) {
 		char *eol = --bol;
 
-		while (bol > buf && bol[-1] != '\n')
+		while (bol > gs->buf && bol[-1] != '\n')
 			bol--;
 		lno--;
 
 		if (lno <= opt->last_shown)
 			break;
 
-		if (match_funcname(opt, name, bol, eol)) {
-			show_line(opt, bol, eol, name, lno, '=');
+		if (match_funcname(opt, gs, bol, eol)) {
+			show_line(opt, bol, eol, gs->name, lno, '=');
 			break;
 		}
 	}
 }
 
-static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
+static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
 			     char *bol, char *end, unsigned lno)
 {
 	unsigned cur = lno, from = 1, funcname_lno = 0;
 	int funcname_needed = !!opt->funcname;
 
-	if (opt->funcbody && !match_funcname(opt, name, bol, end))
+	if (opt->funcbody && !match_funcname(opt, gs, bol, end))
 		funcname_needed = 2;
 
 	if (opt->pre_context < lno)
@@ -893,14 +898,14 @@ static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
 		from = opt->last_shown + 1;
 
 	/* Rewind. */
-	while (bol > buf &&
+	while (bol > gs->buf &&
 	       cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
 		char *eol = --bol;
 
-		while (bol > buf && bol[-1] != '\n')
+		while (bol > gs->buf && bol[-1] != '\n')
 			bol--;
 		cur--;
-		if (funcname_needed && match_funcname(opt, name, bol, eol)) {
+		if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
 			funcname_lno = cur;
 			funcname_needed = 0;
 		}
@@ -908,7 +913,7 @@ static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
 
 	/* We need to look even further back to find a function signature. */
 	if (opt->funcname && funcname_needed)
-		show_funcname_line(opt, name, buf, bol, cur);
+		show_funcname_line(opt, gs, bol, cur);
 
 	/* Back forward. */
 	while (cur < lno) {
@@ -916,7 +921,7 @@ static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
 
 		while (*eol != '\n')
 			eol++;
-		show_line(opt, bol, eol, name, cur, sign);
+		show_line(opt, bol, eol, gs->name, cur, sign);
 		bol = eol + 1;
 		cur++;
 	}
@@ -983,11 +988,10 @@ static void std_output(struct grep_opt *opt, const void *buf, size_t size)
 	fwrite(buf, size, 1, stdout);
 }
 
-static int grep_buffer_1(struct grep_opt *opt, const char *name,
-			 char *buf, unsigned long size, int collect_hits)
+static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
 {
-	char *bol = buf;
-	unsigned long left = size;
+	char *bol;
+	unsigned long left;
 	unsigned lno = 1;
 	unsigned last_hit = 0;
 	int binary_match_only = 0;
@@ -1017,11 +1021,11 @@ static int grep_buffer_1(struct grep_opt *opt, const char *name,
 
 	switch (opt->binary) {
 	case GREP_BINARY_DEFAULT:
-		if (buffer_is_binary(buf, size))
+		if (grep_source_is_binary(gs))
 			binary_match_only = 1;
 		break;
 	case GREP_BINARY_NOMATCH:
-		if (buffer_is_binary(buf, size))
+		if (grep_source_is_binary(gs))
 			return 0; /* Assume unmatch */
 		break;
 	case GREP_BINARY_TEXT:
@@ -1035,6 +1039,11 @@ static int grep_buffer_1(struct grep_opt *opt, const char *name,
 
 	try_lookahead = should_lookahead(opt);
 
+	if (grep_source_load(gs) < 0)
+		return 0;
+
+	bol = gs->buf;
+	left = gs->size;
 	while (left) {
 		char *eol, ch;
 		int hit;
@@ -1083,14 +1092,14 @@ static int grep_buffer_1(struct grep_opt *opt, const char *name,
 			if (opt->status_only)
 				return 1;
 			if (opt->name_only) {
-				show_name(opt, name);
+				show_name(opt, gs->name);
 				return 1;
 			}
 			if (opt->count)
 				goto next_line;
 			if (binary_match_only) {
 				opt->output(opt, "Binary file ", 12);
-				output_color(opt, name, strlen(name),
+				output_color(opt, gs->name, strlen(gs->name),
 					     opt->color_filename);
 				opt->output(opt, " matches\n", 9);
 				return 1;
@@ -1099,23 +1108,23 @@ static int grep_buffer_1(struct grep_opt *opt, const char *name,
 			 * pre-context lines, we would need to show them.
 			 */
 			if (opt->pre_context || opt->funcbody)
-				show_pre_context(opt, name, buf, bol, eol, lno);
+				show_pre_context(opt, gs, bol, eol, lno);
 			else if (opt->funcname)
-				show_funcname_line(opt, name, buf, bol, lno);
-			show_line(opt, bol, eol, name, lno, ':');
+				show_funcname_line(opt, gs, bol, lno);
+			show_line(opt, bol, eol, gs->name, lno, ':');
 			last_hit = lno;
 			if (opt->funcbody)
 				show_function = 1;
 			goto next_line;
 		}
-		if (show_function && match_funcname(opt, name, bol, eol))
+		if (show_function && match_funcname(opt, gs, bol, eol))
 			show_function = 0;
 		if (show_function ||
 		    (last_hit && lno <= last_hit + opt->post_context)) {
 			/* If the last hit is within the post context,
 			 * we need to show this line.
 			 */
-			show_line(opt, bol, eol, name, lno, '-');
+			show_line(opt, bol, eol, gs->name, lno, '-');
 		}
 
 	next_line:
@@ -1133,7 +1142,7 @@ static int grep_buffer_1(struct grep_opt *opt, const char *name,
 		return 0;
 	if (opt->unmatch_name_only) {
 		/* We did not see any hit, so we want to show this */
-		show_name(opt, name);
+		show_name(opt, gs->name);
 		return 1;
 	}
 
@@ -1147,7 +1156,7 @@ static int grep_buffer_1(struct grep_opt *opt, const char *name,
 	 */
 	if (opt->count && count) {
 		char buf[32];
-		output_color(opt, name, strlen(name), opt->color_filename);
+		output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
 		output_sep(opt, ':');
 		snprintf(buf, sizeof(buf), "%u\n", count);
 		opt->output(opt, buf, strlen(buf));
@@ -1182,23 +1191,174 @@ static int chk_hit_marker(struct grep_expr *x)
 	}
 }
 
-int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
+int grep_source(struct grep_opt *opt, struct grep_source *gs)
 {
 	/*
 	 * we do not have to do the two-pass grep when we do not check
 	 * buffer-wide "all-match".
 	 */
 	if (!opt->all_match)
-		return grep_buffer_1(opt, name, buf, size, 0);
+		return grep_source_1(opt, gs, 0);
 
 	/* Otherwise the toplevel "or" terms hit a bit differently.
 	 * We first clear hit markers from them.
 	 */
 	clr_hit_marker(opt->pattern_expression);
-	grep_buffer_1(opt, name, buf, size, 1);
+	grep_source_1(opt, gs, 1);
 
 	if (!chk_hit_marker(opt->pattern_expression))
 		return 0;
 
-	return grep_buffer_1(opt, name, buf, size, 0);
+	return grep_source_1(opt, gs, 0);
+}
+
+int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
+{
+	struct grep_source gs;
+	int r;
+
+	grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL);
+	gs.buf = buf;
+	gs.size = size;
+
+	r = grep_source(opt, &gs);
+
+	grep_source_clear(&gs);
+	return r;
+}
+
+void grep_source_init(struct grep_source *gs, enum grep_source_type type,
+		      const char *name, const void *identifier)
+{
+	gs->type = type;
+	gs->name = name ? xstrdup(name) : NULL;
+	gs->buf = NULL;
+	gs->size = 0;
+	gs->driver = NULL;
+
+	switch (type) {
+	case GREP_SOURCE_FILE:
+		gs->identifier = xstrdup(identifier);
+		break;
+	case GREP_SOURCE_SHA1:
+		gs->identifier = xmalloc(20);
+		memcpy(gs->identifier, identifier, 20);
+		break;
+	case GREP_SOURCE_BUF:
+		gs->identifier = NULL;
+	}
+}
+
+void grep_source_clear(struct grep_source *gs)
+{
+	free(gs->name);
+	gs->name = NULL;
+	free(gs->identifier);
+	gs->identifier = NULL;
+	grep_source_clear_data(gs);
+}
+
+void grep_source_clear_data(struct grep_source *gs)
+{
+	switch (gs->type) {
+	case GREP_SOURCE_FILE:
+	case GREP_SOURCE_SHA1:
+		free(gs->buf);
+		gs->buf = NULL;
+		gs->size = 0;
+		break;
+	case GREP_SOURCE_BUF:
+		/* leave user-provided buf intact */
+		break;
+	}
+}
+
+static int grep_source_load_sha1(struct grep_source *gs)
+{
+	enum object_type type;
+
+	grep_read_lock();
+	gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
+	grep_read_unlock();
+
+	if (!gs->buf)
+		return error(_("'%s': unable to read %s"),
+			     gs->name,
+			     sha1_to_hex(gs->identifier));
+	return 0;
+}
+
+static int grep_source_load_file(struct grep_source *gs)
+{
+	const char *filename = gs->identifier;
+	struct stat st;
+	char *data;
+	size_t size;
+	int i;
+
+	if (lstat(filename, &st) < 0) {
+	err_ret:
+		if (errno != ENOENT)
+			error(_("'%s': %s"), filename, strerror(errno));
+		return -1;
+	}
+	if (!S_ISREG(st.st_mode))
+		return -1;
+	size = xsize_t(st.st_size);
+	i = open(filename, O_RDONLY);
+	if (i < 0)
+		goto err_ret;
+	data = xmalloc(size + 1);
+	if (st.st_size != read_in_full(i, data, size)) {
+		error(_("'%s': short read %s"), filename, strerror(errno));
+		close(i);
+		free(data);
+		return -1;
+	}
+	close(i);
+	data[size] = 0;
+
+	gs->buf = data;
+	gs->size = size;
+	return 0;
+}
+
+int grep_source_load(struct grep_source *gs)
+{
+	if (gs->buf)
+		return 0;
+
+	switch (gs->type) {
+	case GREP_SOURCE_FILE:
+		return grep_source_load_file(gs);
+	case GREP_SOURCE_SHA1:
+		return grep_source_load_sha1(gs);
+	case GREP_SOURCE_BUF:
+		return gs->buf ? 0 : -1;
+	}
+	die("BUG: invalid grep_source type");
+}
+
+void grep_source_load_driver(struct grep_source *gs)
+{
+	if (gs->driver)
+		return;
+
+	grep_attr_lock();
+	gs->driver = userdiff_find_by_path(gs->name);
+	if (!gs->driver)
+		gs->driver = userdiff_find_by_name("default");
+	grep_attr_unlock();
+}
+
+int grep_source_is_binary(struct grep_source *gs)
+{
+	grep_source_load_driver(gs);
+	if (gs->driver->binary != -1)
+		return gs->driver->binary;
+
+	if (!grep_source_load(gs))
+		return buffer_is_binary(gs->buf, gs->size);
+
+	return 0;
 }
diff --git a/grep.h b/grep.h
index fb205f3542..36e49d8255 100644
--- a/grep.h
+++ b/grep.h
@@ -9,6 +9,7 @@ typedef int pcre_extra;
 #endif
 #include "kwset.h"
 #include "thread-utils.h"
+#include "userdiff.h"
 
 enum grep_pat_token {
 	GREP_PATTERN,
@@ -116,7 +117,6 @@ struct grep_opt {
 	int show_hunk_mark;
 	int file_break;
 	int heading;
-	int use_threads;
 	void *priv;
 
 	void (*output)(struct grep_opt *opt, const void *data, size_t size);
@@ -128,7 +128,33 @@ extern void append_grep_pattern(struct grep_opt *opt, const char *pat, const cha
 extern void append_header_grep_pattern(struct grep_opt *, enum grep_header_field, const char *);
 extern void compile_grep_patterns(struct grep_opt *opt);
 extern void free_grep_patterns(struct grep_opt *opt);
-extern int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size);
+extern int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size);
+
+struct grep_source {
+	char *name;
+
+	enum grep_source_type {
+		GREP_SOURCE_SHA1,
+		GREP_SOURCE_FILE,
+		GREP_SOURCE_BUF,
+	} type;
+	void *identifier;
+
+	char *buf;
+	unsigned long size;
+
+	struct userdiff_driver *driver;
+};
+
+void grep_source_init(struct grep_source *gs, enum grep_source_type type,
+		      const char *name, const void *identifier);
+int grep_source_load(struct grep_source *gs);
+void grep_source_clear_data(struct grep_source *gs);
+void grep_source_clear(struct grep_source *gs);
+void grep_source_load_driver(struct grep_source *gs);
+int grep_source_is_binary(struct grep_source *gs);
+
+int grep_source(struct grep_opt *opt, struct grep_source *gs);
 
 extern struct grep_opt *grep_opt_dup(const struct grep_opt *opt);
 extern int grep_threads_ok(const struct grep_opt *opt);
@@ -138,7 +164,25 @@ extern int grep_threads_ok(const struct grep_opt *opt);
  * Mutex used around access to the attributes machinery if
  * opt->use_threads.  Must be initialized/destroyed by callers!
  */
+extern int grep_use_locks;
 extern pthread_mutex_t grep_attr_mutex;
+extern pthread_mutex_t grep_read_mutex;
+
+static inline void grep_read_lock(void)
+{
+	if (grep_use_locks)
+		pthread_mutex_lock(&grep_read_mutex);
+}
+
+static inline void grep_read_unlock(void)
+{
+	if (grep_use_locks)
+		pthread_mutex_unlock(&grep_read_mutex);
+}
+
+#else
+#define grep_read_lock()
+#define grep_read_unlock()
 #endif
 
 #endif
diff --git a/revision.c b/revision.c
index c97d834484..819ff012ff 100644
--- a/revision.c
+++ b/revision.c
@@ -2149,7 +2149,6 @@ static int commit_match(struct commit *commit, struct rev_info *opt)
 	if (!opt->grep_filter.pattern_list && !opt->grep_filter.header_list)
 		return 1;
 	return grep_buffer(&opt->grep_filter,
-			   NULL, /* we say nothing, not even filename */
 			   commit->buffer, strlen(commit->buffer));
 }
 
diff --git a/t/t7008-grep-binary.sh b/t/t7008-grep-binary.sh
index 917a264eea..fd6410fc71 100755
--- a/t/t7008-grep-binary.sh
+++ b/t/t7008-grep-binary.sh
@@ -99,4 +99,28 @@ test_expect_success 'git grep y<NUL>x a' "
 	test_must_fail git grep -f f a
 "
 
+test_expect_success 'grep respects binary diff attribute' '
+	echo text >t &&
+	git add t &&
+	echo t:text >expect &&
+	git grep text t >actual &&
+	test_cmp expect actual &&
+	echo "t -diff" >.gitattributes &&
+	echo "Binary file t matches" >expect &&
+	git grep text t >actual &&
+	test_cmp expect actual
+'
+
+test_expect_success 'grep respects not-binary diff attribute' '
+	echo binQary | q_to_nul >b &&
+	git add b &&
+	echo "Binary file b matches" >expect &&
+	git grep bin b >actual &&
+	test_cmp expect actual &&
+	echo "b diff" >.gitattributes &&
+	echo "b:binQary" >expect &&
+	git grep bin b | nul_to_q >actual &&
+	test_cmp expect actual
+'
+
 test_done