X-Git-Url: http://git.tuebingen.mpg.de/?p=adu.git;a=blobdiff_plain;f=adu.c;h=8077bd923cb2836816cb434b2280a19dc330d52d;hp=20f0f61bbc114ffc9b270ab31991970907ff142a;hb=042654d412af1e85a264db553cdcbfa875f61df4;hpb=c3d144c0f9cf04370ef885fdff43817d3f967fa6

diff --git a/adu.c b/adu.c
index 20f0f61..8077bd9 100644
--- a/adu.c
+++ b/adu.c
@@ -2,15 +2,27 @@
 #include <dirent.h> /* readdir() */
 
 #include "gcc-compat.h"
-#include "osl.h"
+#include "cmdline.h"
 #include "fd.h"
-#include "hash.h"
 #include "string.h"
 #include "error.h"
+#include "portable_io.h"
 
 DEFINE_ERRLIST;
 
-#define DATABASE_DIR "/tmp/adu"
+/** Command line and config file options. */
+static struct gengetopt_args_info conf;
+
+struct user_info {
+	uint32_t uid;
+	struct osl_table *table;
+	uint64_t files;
+	uint64_t bytes;
+	uint64_t dirs;
+	struct osl_table_description *desc;
+};
+
+static struct user_info *uid_hash_table;
 
 /** evaluates to 1 if x < y, to -1 if x > y and to 0 if x == y */
 #define NUM_COMPARE(x, y) ((int)((x) < (y)) - (int)((x) > (y)))
@@ -32,7 +44,7 @@ __printf_2_3 void __log(int ll, const char* fmt,...)
 	time_t t1;
 	char str[255] = "";
 
-	if (ll < 4)
+	if (ll < conf.loglevel_arg)
 		return;
 	outfd = stderr;
 	time(&t1);
@@ -76,18 +88,42 @@ static int size_compare(const struct osl_object *obj1, const struct osl_object *
  * are taken into account.
  *
  * \return It returns an integer less than, equal to, or greater than zero if
- * \a obj1 is found, respectively, to be less than, to match, or be greater than
- * obj2.
+ * \a obj1 is found, respectively, to be less than, to match, or be greater
+ * than obj2.
  *
  * \sa strcmp(3), strncmp(3), osl_compare_func.
  */
-int string_compare(const struct osl_object *obj1, const struct osl_object *obj2)
+static int string_compare(const struct osl_object *obj1,
+		const struct osl_object *obj2)
 {
 	const char *str1 = (const char *)obj1->data;
 	const char *str2 = (const char *)obj2->data;
 	return strncmp(str1, str2, MIN(obj1->size, obj2->size));
 }
 
+/**
+ * Compare two osl objects pointing to unsigned integers of 64 bit size.
+ *
+ * \param obj1 Pointer to the first integer.
+ * \param obj2 Pointer to the second integer.
+ *
+ * \return The values required for an osl compare function.
+ *
+ * \sa osl_compare_func, osl_hash_compare().
+ */
+static int uint64_compare(const struct osl_object *obj1,
+		const struct osl_object *obj2)
+{
+	uint64_t d1 = read_u64((const char *)obj1->data);
+	uint64_t d2 = read_u64((const char *)obj2->data);
+
+	if (d1 < d2)
+		return 1;
+	if (d1 > d2)
+		return -1;
+	return 0;
+}
+
 /** The columns of the directory table. */
 enum dir_table_columns {
 	/** The name of the directory. */
@@ -113,8 +149,8 @@ static struct osl_column_description dir_table_cols[] = {
 		.storage_type = OSL_MAPPED_STORAGE,
 		.storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
 		.name = "num",
-		.compare_function = uint32_compare,
-		.data_size = sizeof(uint32_t)
+		.compare_function = uint64_compare,
+		.data_size = sizeof(uint64_t)
 	},
 	[DT_BYTES] = {
 		.storage_type = OSL_MAPPED_STORAGE,
@@ -137,7 +173,6 @@ static struct osl_table_description dir_table_desc = {
 	.num_columns = NUM_DT_COLUMNS,
 	.flags = 0,
 	.column_descriptions = dir_table_cols,
-	.dir = DATABASE_DIR
 };
 
 /** The columns of the id table. */
@@ -157,8 +192,8 @@ static struct osl_column_description user_table_cols[] = {
 		.storage_type = OSL_MAPPED_STORAGE,
 		.storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
 		.name = "dir_num",
-		.compare_function = uint32_compare,
-		.data_size = sizeof(uint32_t)
+		.compare_function = uint64_compare,
+		.data_size = sizeof(uint64_t)
 	},
 	[UT_BYTES] = {
 		.storage_type = OSL_MAPPED_STORAGE,
@@ -178,12 +213,12 @@ static struct osl_column_description user_table_cols[] = {
 
 static struct osl_table *dir_table;
 
-int add_directory(char *dirname, uint32_t dir_num, uint64_t *dir_size,
+static int add_directory(char *dirname, uint64_t dir_num, uint64_t *dir_size,
 		uint64_t *dir_files)
 {
 	struct osl_object dir_objects[NUM_DT_COLUMNS];
 
-	INFO_LOG("adding #%u: %s\n", dir_num, dirname);
+	INFO_LOG("adding #%llu: %s\n", (long long unsigned)dir_num, dirname);
 	dir_objects[DT_NAME].data = dirname;
 	dir_objects[DT_NAME].size = strlen(dirname) + 1;
 	dir_objects[DT_NUM].data = &dir_num;
@@ -196,85 +231,131 @@ int add_directory(char *dirname, uint32_t dir_num, uint64_t *dir_size,
 	return osl_add_row(dir_table, dir_objects);
 }
 
-int create_and_open_user_table(uint32_t uid, struct osl_table **t)
+static uint32_t num_uids;
+
+static int open_user_table(struct user_info *ui, int create)
 {
 	int ret;
-	struct osl_table_description *desc = para_malloc(sizeof(*desc));
-
-	desc->num_columns = NUM_UT_COLUMNS;
-	desc->flags = 0;
-	desc->column_descriptions = user_table_cols;
-	desc->dir = para_strdup(DATABASE_DIR);
-	desc->name = make_message("%u", uid);
-	INFO_LOG("................................. %u\n", uid);
-//	user_table_desc.name = make_message("%u", uid);
-	ret = osl_create_table(desc);
+
+	ui->desc = para_malloc(sizeof(*ui->desc));
+	ui->desc->num_columns = NUM_UT_COLUMNS;
+	ui->desc->flags = 0;
+	ui->desc->column_descriptions = user_table_cols;
+	ui->desc->dir = para_strdup(conf.database_dir_arg);
+	ui->desc->name = make_message("%u", (unsigned)ui->uid);
+	num_uids++;
+	INFO_LOG(".............................uid #%u: %u\n",
+		(unsigned)num_uids, (unsigned)ui->uid);
+	if (create) {
+		ret = osl_create_table(ui->desc);
+		if (ret < 0)
+			goto err;
+	}
+	ret = osl_open_table(ui->desc, &ui->table);
 	if (ret < 0)
-		return ret;
-	return osl_open_table(desc, t);
+		goto err;
+	return 1;
+err:
+	free((char *)ui->desc->name);
+	free((char *)ui->desc->dir);
+	free(ui->desc);
+	ui->desc->name = NULL;
+	ui->desc->dir = NULL;
+	ui->desc = NULL;
+	ui->table = NULL;
+	return ret;
 }
 
-
 #define uid_hash_bits 8
 static uint32_t uid_hash_table_size = 1 << uid_hash_bits;
 #define PRIME1 0x811c9dc5
 #define PRIME2 0x01000193
 
-struct user_info {
-	uint32_t uid;
-	struct osl_table *table;
-	uint64_t files;
-	uint64_t bytes;
-};
-
-static struct user_info *uid_hash_table;
-
 static void create_hash_table(void)
 {
 	uid_hash_table = para_calloc(uid_hash_table_size
 		* sizeof(struct user_info));
 }
 
+static void free_hash_table(void)
+{
+	free(uid_hash_table);
+	uid_hash_table = NULL;
+}
+
 static int create_tables(void)
 {
+	int ret;
+
+	dir_table_desc.dir = para_strdup(conf.database_dir_arg);
+	ret = osl_create_table(&dir_table_desc);
+	if (ret < 0)
+		return ret;
 	create_hash_table();
-	return osl_create_table(&dir_table_desc);
+	return 1;
 }
 
-
+/*
+ * We use a hash table of size s=2^uid_hash_bits to map the uids into the
+ * interval [0..s]. Hash collisions are treated by open addressing, i.e.
+ * unused slots in the table are used to store different uids that hash to the
+ * same slot.
+ *
+ * If a hash collision occurs, different slots are successively probed in order
+ * to find an unused slot for the new uid. Probing is implemented via a second
+ * hash function that maps the uid to h=(uid * PRIME2) | 1, which is always an
+ * odd number.
+ *
+ * An odd number is sufficient to make sure each entry of the hash table gets
+ * probed for probe_num between 0 and s-1 because s is a power of two, hence
+ * the second hash value never hash a common divisor with the hash table size.
+ * IOW: h is invertible in the ring [0..s].
+ */
 static uint32_t double_hash(uint32_t uid, uint32_t probe_num)
 {
-	return (uid * PRIME1 + ((uid * PRIME2) | 1) * probe_num) % uid_hash_table_size;
+	return (uid * PRIME1 + ((uid * PRIME2) | 1) * probe_num)
+		% uid_hash_table_size;
 }
 
-#define FOR_EACH_USER(ui) for (ui = uid_hash_table; ui < uid_hash_table \
+#define FOR_EACH_USER(ui) for (ui = uid_hash_table; ui && ui < uid_hash_table \
 		+ uid_hash_table_size; ui++)
 
-static int search_uid(uint32_t uid, int insert, struct user_info **ui)
+enum search_uid_flags {
+	OPEN_USER_TABLE = 1,
+	CREATE_USER_TABLE = 2,
+};
+
+static int search_uid(uint32_t uid, enum search_uid_flags flags,
+		struct user_info **ui_ptr)
 {
 	uint32_t p;
 
 	for (p = 0; p < uid_hash_table_size; p++) {
-		struct user_info *i = uid_hash_table + double_hash(uid, p);
-		if (!i->table) {
-			if (!insert)
+		struct user_info *ui = uid_hash_table + double_hash(uid, p);
+
+		if (!ui->table) {
+			int ret;
+
+			if (!flags)
 				return -E_BAD_UID;
-			int ret = create_and_open_user_table(uid, &i->table);
+			ui->uid = uid;
+			ret = open_user_table(ui, flags & CREATE_USER_TABLE);
 			if (ret < 0)
 				return ret;
-			i->uid = uid;
-			*ui = i;
+			if (ui_ptr)
+				*ui_ptr = ui;
 			return 1;
 		}
-		if (i->uid != uid)
+		if (ui->uid != uid)
 			continue;
-		*ui = i;
+		if (ui_ptr)
+			*ui_ptr = ui;
 		return 0;
 	}
-	return insert? -E_HASH_TABLE_OVERFLOW : -E_BAD_UID;
+	return flags? -E_HASH_TABLE_OVERFLOW : -E_BAD_UID;
 }
 
-static int update_user_row(struct osl_table *t, uint32_t dir_num,
+static int update_user_row(struct osl_table *t, uint64_t dir_num,
 		uint64_t *add)
 {
 	struct osl_row *row;
@@ -317,8 +398,8 @@ static int update_user_row(struct osl_table *t, uint32_t dir_num,
 	}
 }
 
-static uint32_t num_dirs;
-static uint32_t num_files;
+static uint64_t num_dirs;
+static uint64_t num_files;
 static uint64_t num_bytes;
 
 int scan_dir(char *dirname)
@@ -327,9 +408,9 @@ int scan_dir(char *dirname)
 	struct dirent *entry;
 	int ret, cwd_fd, ret2;
 	uint64_t dir_size = 0, dir_files = 0;
-	uint32_t this_dir_num = num_dirs++;
+	uint64_t this_dir_num = num_dirs++;
 
-	DEBUG_LOG("----------------- %u: %s\n", num_dirs, dirname);
+	DEBUG_LOG("----------------- %llu: %s\n", (long long unsigned)num_dirs, dirname);
 	ret = para_opendir(dirname, &dir, &cwd_fd);
 	if (ret < 0) {
 		if (ret != -ERRNO_TO_ERROR(EACCES))
@@ -372,7 +453,7 @@ int scan_dir(char *dirname)
 		dir_files++;
 		num_files++;
 		uid = s.st_uid;
-		ret = search_uid(uid, 1, &ui);
+		ret = search_uid(uid, CREATE_USER_TABLE | OPEN_USER_TABLE, &ui);
 		if (ret < 0)
 			goto out;
 		ui->bytes += size;
@@ -402,41 +483,105 @@ static int get_dir_name(struct osl_row *row, char **name)
 	return 1;
 }
 
-static int print_dirname_and_size(struct osl_row *row, void *data)
+const uint64_t size_unit_divisors[] = {
+	[size_unit_arg_b] = 1ULL,
+	[size_unit_arg_k] = 1024ULL,
+	[size_unit_arg_m] = 1024ULL * 1024ULL,
+	[size_unit_arg_g] = 1024ULL * 1024ULL * 1024ULL,
+	[size_unit_arg_t] = 1024ULL * 1024ULL * 1024ULL * 1024ULL,
+};
+
+const uint64_t count_unit_divisors[] = {
+
+	[count_unit_arg_n] = 1ULL,
+	[count_unit_arg_k] = 1000ULL,
+	[count_unit_arg_m] = 1000ULL * 1000ULL,
+	[count_unit_arg_g] = 1000ULL * 1000ULL * 1000ULL,
+	[count_unit_arg_t] = 1000ULL * 1000ULL * 1000ULL * 1000ULL,
+};
+
+const char size_unit_abbrevs[] = " BKMGT";
+const char count_unit_abbrevs[] = "  KMGT";
+
+static void format_size_value(enum enum_size_unit unit, uint64_t value, char *result)
 {
-	unsigned *count = data;
-	struct osl_object obj;
-	char *name;
-	int ret;
+	if (unit == size_unit_arg_h) /* human readable */
+		for (unit = size_unit_arg_b; unit < size_unit_arg_t && value > size_unit_divisors[unit + 1]; unit++)
+				; /* nothing */
+	sprintf(result, "%llu%c", (long long unsigned)value / size_unit_divisors[unit], size_unit_abbrevs[unit]);
+}
 
-	if ((*count)++ > 100)
-		return -E_LOOP_COMPLETE;
-	ret = get_dir_name(row, &name);
-	if (ret < 0)
-		return ret;
-	ret = osl_get_object(dir_table, row, DT_BYTES, &obj);
-	if (ret < 0)
-		return ret;
-	printf("%s\t%llu\n", name, *(long long unsigned *)obj.data);
-	return 1;
+static void format_count_value(enum enum_count_unit unit, uint64_t value, char *result)
+{
+	if (unit == count_unit_arg_h) /* human readable */
+		for (unit = count_unit_arg_n; unit < count_unit_arg_t && value > count_unit_divisors[unit + 1]; unit++)
+				; /* nothing */
+	sprintf(result, "%llu%c", (long long unsigned)value / count_unit_divisors[unit], count_unit_abbrevs[unit]);
 }
 
-static int print_dirname_and_file_count(struct osl_row *row, void *data)
+enum global_stats_flags {
+	GSF_PRINT_DIRNAME = 1,
+	GSF_PRINT_BYTES = 2,
+	GSF_PRINT_FILES = 4,
+	GSF_COMPUTE_SUMMARY = 8,
+};
+
+struct global_stats_info {
+	uint32_t count;
+	enum global_stats_flags flags;
+};
+
+static int global_stats_loop_function(struct osl_row *row, void *data)
 {
-	unsigned *count = data;
+	struct global_stats_info *gsi = data;
 	struct osl_object obj;
-	char *name;
-	int ret;
+	char *dirname, formated_value[25];
+	int ret, summary = gsi->flags & GSF_COMPUTE_SUMMARY;
 
-	if ((*count)++ > 100)
+	if (!gsi->count && !summary)
 		return -E_LOOP_COMPLETE;
-	ret = get_dir_name(row, &name);
-	if (ret < 0)
-		return ret;
-	ret = osl_get_object(dir_table, row, DT_FILES, &obj);
-	if (ret < 0)
-		return ret;
-	printf("%s\t%llu\n", name, *(long long unsigned *)obj.data);
+	if (gsi->count && (gsi->flags & GSF_PRINT_DIRNAME)) {
+		ret = get_dir_name(row, &dirname);
+		if (ret < 0)
+			return ret;
+		printf("%s%s", dirname,
+			(gsi->flags & (GSF_PRINT_FILES | GSF_PRINT_BYTES))?
+				"\t" : "\n"
+		);
+	}
+	if (summary || (gsi->count && (gsi->flags & GSF_PRINT_FILES))) {
+		uint64_t files;
+		ret = osl_get_object(dir_table, row, DT_FILES, &obj);
+		if (ret < 0)
+			return ret;
+		files = *(uint64_t *)obj.data;
+		if (gsi->count && (gsi->flags & GSF_PRINT_FILES)) {
+			format_size_value(conf.size_unit_arg, files,
+				formated_value);
+			printf("%s%s", formated_value,
+				(gsi->flags & GSF_PRINT_BYTES)? "\t" : "\n");
+		}
+		if (summary)
+			num_files += files;
+	}
+	if (summary || (gsi->count && (gsi->flags & GSF_PRINT_BYTES))) {
+		uint64_t bytes;
+		ret = osl_get_object(dir_table, row, DT_BYTES, &obj);
+		if (ret < 0)
+			return ret;
+		bytes = *(uint64_t *)obj.data;
+		if (gsi->count && (gsi->flags & GSF_PRINT_BYTES)) {
+			format_size_value(conf.size_unit_arg, bytes,
+				formated_value);
+			printf("%s\n", formated_value);
+		}
+		if (summary) {
+			num_bytes += bytes;
+			num_dirs++;
+		}
+	}
+	if (gsi->count > 0)
+		gsi->count--;
 	return 1;
 }
 
@@ -444,101 +589,323 @@ static void print_id_stats(void)
 {
 	struct user_info *ui;
 
+	printf("--------------------- user summary (uid/dirs/files/bytes):\n");
 	FOR_EACH_USER(ui) {
+		char formated_dir_count[25], formated_file_count[25],
+			formated_bytes[25];
 		if (!ui->table)
 			continue;
-		printf("%u\t%llu\t%llu\n", (unsigned)ui->uid, (long long unsigned)ui->files,
-			(long long unsigned)ui->bytes);
+		format_count_value(conf.count_unit_arg, ui->dirs,
+			formated_dir_count);
+		format_count_value(conf.count_unit_arg, ui->files,
+			formated_file_count);
+		format_size_value(conf.size_unit_arg, ui->bytes,
+			formated_bytes);
+		printf("%u\t%s\t%s\t%s\n", (unsigned)ui->uid,
+			formated_dir_count,
+			formated_file_count,
+			formated_bytes
+		);
 	}
 }
 
-struct big_dir_info {
-	unsigned count;
-	struct osl_table *user_table;
+enum user_stats_flags {
+	USF_PRINT_DIRNAME = 1,
+	USF_PRINT_BYTES = 2,
+	USF_PRINT_FILES = 4,
+	USF_COMPUTE_SUMMARY = 8,
+};
+
+struct user_stats_info {
+	uint32_t count;
+	enum user_stats_flags flags;
+	struct user_info *ui;
 };
 
-static int print_big_dir(struct osl_row *row, void *data)
+static int user_stats_loop_function(struct osl_row *row, void *data)
 {
-	struct big_dir_info *bdi = data;
-	int ret;
+	struct user_stats_info *usi = data;
 	struct osl_row *dir_row;
-	char *dirname;
-	uint64_t bytes;
 	struct osl_object obj;
+	int ret, summary = usi->flags & GSF_COMPUTE_SUMMARY;
+	char formated_value[25];
 
-	if (bdi->count++ > 10)
+	if (!usi->count && !summary)
 		return -E_LOOP_COMPLETE;
-	ret = osl_get_object(bdi->user_table, row, UT_BYTES, &obj);
-	if (ret < 0)
-		return ret;
-	bytes = *(uint64_t *)obj.data;
-	ret = osl_get_object(bdi->user_table, row, UT_DIR_NUM, &obj);
-	if (ret < 0)
-		return ret;
-	ret = osl_get_row(dir_table, DT_NUM, &obj, &dir_row);
-	if (ret < 0)
-		return ret;
-	ret = osl_get_object(dir_table, dir_row, DT_NAME, &obj);
-	if (ret < 0)
-		return ret;
-	dirname = obj.data;
-	printf("%s: %llu\n", dirname, (long long unsigned)bytes);
+	if (usi->count && (usi->flags & USF_PRINT_DIRNAME)) {
+		char *dirname;
+		ret = osl_get_object(usi->ui->table, row, UT_DIR_NUM, &obj);
+		if (ret < 0)
+			return ret;
+		ret = osl_get_row(dir_table, DT_NUM, &obj, &dir_row);
+		if (ret < 0)
+			return ret;
+		ret = osl_get_object(dir_table, dir_row, DT_NAME, &obj);
+		if (ret < 0)
+			return ret;
+		dirname = obj.data;
+		printf("%s%s",
+			dirname,
+			(usi->flags & (USF_PRINT_FILES | USF_PRINT_BYTES))?
+				"\t" : "\n"
+		);
+	}
+	if (summary || (usi->count && (usi->flags & USF_PRINT_FILES))) {
+		uint64_t files;
+		ret = osl_get_object(usi->ui->table, row, UT_FILES, &obj);
+		if (ret < 0)
+			return ret;
+		files = *(uint64_t *)obj.data;
+		if (usi->count && (usi->flags & USF_PRINT_FILES)) {
+			format_size_value(conf.size_unit_arg, files,
+				formated_value);
+			printf("%s%s", formated_value,
+				(usi->flags & USF_PRINT_BYTES)? "\t" : "\n"
+			);
+		}
+		if (summary)
+			usi->ui->files += files;
+	}
+	if (summary || (usi->count && (usi->flags & USF_PRINT_BYTES))) {
+		uint64_t bytes;
+		ret = osl_get_object(usi->ui->table, row, UT_BYTES, &obj);
+		if (ret < 0)
+			return ret;
+		bytes = *(uint64_t *)obj.data;
+		if (usi->count && (usi->flags & USF_PRINT_BYTES)) {
+			format_size_value(conf.size_unit_arg, bytes,
+				formated_value);
+			printf("%s\n", formated_value);
+		}
+		if (summary) {
+			usi->ui->bytes += bytes;
+			usi->ui->dirs++;
+		}
+
+	}
+	if (usi->count > 0)
+		usi->count--;
 	return 1;
 }
 
-static void print_id_dir_stats(void)
+static void print_user_stats(void)
 {
 	struct user_info *ui;
 
 	FOR_EACH_USER(ui) {
-		struct big_dir_info bdi = {.count = 0};
+		struct user_stats_info usi = {
+			.count = conf.limit_arg,
+			.ui = ui
+		};
 		if (!ui->table)
 			continue;
-		bdi.user_table = ui->table;
-		printf("************************* Big dirs owned by uid %u\n", (unsigned) ui->uid);
-		osl_rbtree_loop_reverse(ui->table, UT_BYTES, &bdi, print_big_dir);
+		usi.flags = USF_PRINT_DIRNAME | USF_PRINT_BYTES | USF_COMPUTE_SUMMARY;
+		printf("************************************************ uid %u\n",
+			(unsigned) ui->uid);
+		if (!ui->table)
+			continue;
+		printf("----------------- Largest dirs -------------------\n");
+		osl_rbtree_loop_reverse(ui->table, UT_BYTES, &usi,
+			user_stats_loop_function);
+		printf("---------- dirs containing most files ------------\n");
+		usi.count = conf.limit_arg,
+		usi.flags = USF_PRINT_DIRNAME | USF_PRINT_FILES;
+		osl_rbtree_loop_reverse(ui->table, UT_FILES, &usi,
+			user_stats_loop_function);
 	}
 }
 
 static int print_statistics(void)
 {
-	unsigned count = 0;
 	int ret;
-
-	printf("Summary: %u dirs, %u files, %llu bytes\n", (unsigned)num_dirs,
-		(unsigned)num_files, (long long unsigned)num_bytes);
-	printf("************************* Biggest dirs\n");
-	ret = osl_rbtree_loop_reverse(dir_table, DT_BYTES, &count, print_dirname_and_size);
+	struct global_stats_info gsi = {
+		.count = conf.limit_arg,
+		.flags = GSF_PRINT_DIRNAME | GSF_PRINT_BYTES | GSF_COMPUTE_SUMMARY
+	};
+
+	printf("----------------- Largest dirs -------------------\n");
+	ret = osl_rbtree_loop_reverse(dir_table, DT_BYTES, &gsi,
+		global_stats_loop_function);
 	if (ret < 0 && ret != -E_LOOP_COMPLETE)
 		return ret;
-	count = 0;
-	printf("************************* dirs containing many files\n");
-	ret = osl_rbtree_loop_reverse(dir_table, DT_FILES, &count, print_dirname_and_file_count);
+	gsi.count = conf.limit_arg;
+
+	gsi.flags = GSF_PRINT_DIRNAME | GSF_PRINT_FILES;
+	printf("---------- dirs containing most files ------------\n");
+	ret = osl_rbtree_loop_reverse(dir_table, DT_FILES, &gsi,
+		global_stats_loop_function);
 	if (ret < 0 && ret != -E_LOOP_COMPLETE)
 		return ret;
 
-	printf("************************* dirs stats by owner\n");
+	printf("------------------ Global summary (dirs/files/bytes)\n"
+		"%llu\t%llu\t%llu\n",
+		(long long unsigned)num_dirs, (long long unsigned)num_files,
+		(long long unsigned)num_bytes);
+	print_user_stats();
 	print_id_stats();
-	print_id_dir_stats();
 	return 1;
 }
 
+static char *get_uid_list_name(void)
+{
+	return make_message("%s/uid_list", conf.database_dir_arg);
+}
 
-int main(int argc, char **argv)
+static int write_uid_list(void)
+{
+	char *buf, *filename;
+	uint32_t count = 0;
+	struct user_info *ui;
+	size_t size = num_uids * sizeof(uint32_t);
+	int ret;
+
+	if (!num_uids)
+		return 0;
+	buf = para_malloc(size);
+	FOR_EACH_USER(ui) {
+		if (!ui->table)
+			continue;
+		write_u32(buf + count++ * sizeof(uint32_t), ui->uid);
+	}
+	filename = get_uid_list_name();
+	ret = para_write_file(filename, buf, size);
+	free(filename);
+	free(buf);
+	return ret;
+}
+
+static int open_dir_table(void)
+{
+	if (!dir_table_desc.dir) /* we did not create the table */
+		dir_table_desc.dir = para_strdup(conf.database_dir_arg);
+	return osl_open_table(&dir_table_desc, &dir_table);
+}
+
+static void close_dir_table(void)
+{
+	int ret;
+
+	if (!dir_table)
+		return;
+	ret = osl_close_table(dir_table, OSL_MARK_CLEAN);
+	if (ret < 0)
+		ERROR_LOG("failed to close dir table: %s\n", error_txt(-ret));
+	free((char *)dir_table_desc.dir);
+	dir_table = NULL;
+}
+
+static void close_user_table(struct user_info *ui)
+{
+	int ret;
+
+	if (!ui || !ui->table)
+		return;
+	ret = osl_close_table(ui->table, OSL_MARK_CLEAN);
+	if (ret < 0)
+		ERROR_LOG("failed to close user table %u: %s\n",
+			(unsigned) ui->uid, error_txt(-ret));
+	free((char *)ui->desc->name);
+	ui->desc->name = NULL;
+	free((char *)ui->desc->dir);
+	ui->desc->dir = NULL;
+	free(ui->desc);
+	ui->desc = NULL;
+	ui->table = NULL;
+}
+
+static void close_user_tables(void)
+{
+	struct user_info *ui;
+
+	FOR_EACH_USER(ui)
+		close_user_table(ui);
+}
+
+static void close_all_tables(void)
+{
+	close_dir_table();
+	close_user_tables();
+	free_hash_table();
+}
+
+static int com_create()
 {
 	int ret = create_tables();
 	if (ret < 0)
-		goto out;
-	ret = osl_open_table(&dir_table_desc, &dir_table);
+		return ret;
+	ret = open_dir_table();
+	if (ret < 0)
+		return ret;
+	ret = scan_dir(conf.base_dir_arg);
 	if (ret < 0)
 		goto out;
+	ret = write_uid_list();
+out:
+	close_all_tables();
+	return ret;
+}
+
+static int read_uid_file(void)
+{
+	size_t size;
+	uint32_t n;
+	char *filename = get_uid_list_name(), *map;
+	int ret = mmap_full_file(filename, O_RDONLY, (void **)&map, &size, NULL);
+
+	free(filename);
+	if (ret < 0)
+		return ret;
+	num_uids = size / 4;
+	/* hash table size should be a power of two and larger than the number of uids */
+	uid_hash_table_size = 4;
+	while (uid_hash_table_size < num_uids)
+		uid_hash_table_size *= 2;
+	create_hash_table();
+	for (n = 0; n < num_uids; n++) {
+		uint32_t uid = read_u32(map + n * sizeof(uid));
+		ret = search_uid(uid, OPEN_USER_TABLE, NULL);
+		if (ret < 0)
+			goto out;
+	}
+out:
+	para_munmap(map, size);
+	return ret;
+}
+
+static int com_select(void)
+{
+	int ret;
+
+	ret = open_dir_table();
+	if (ret < 0)
+		return ret;
+	ret = read_uid_file();
+	if (ret < 0)
+		return ret;
+	print_statistics();
+	close_all_tables();
+	return 1;
+}
+
+int main(int argc, char **argv)
+{
+	int ret;
+	struct cmdline_parser_params params = {
+		.override = 0,
+		.initialize = 1,
+		.check_required = 0,
+		.check_ambiguity = 0,
+		.print_errors = 1
+	};
+
+	cmdline_parser_ext(argc, argv, &conf, &params); /* aborts on errors */
 	ret = -E_SYNTAX;
-	if (argc != 2)
-		goto out;
-	ret = scan_dir(argv[1]);
+	if (conf.select_given)
+		ret = com_select();
+	else
+		ret = com_create();
 	if (ret < 0)
 		goto out;
-	print_statistics();
 out:
 	if (ret < 0) {
 		ERROR_LOG("%s\n", error_txt(-ret));
@@ -546,4 +913,3 @@ out:
 	}
 	return EXIT_SUCCESS;
 }
-