X-Git-Url: http://git.tuebingen.mpg.de/?a=blobdiff_plain;f=adu.c;h=8f5ff8d655b439191108d372459fe15a153ec64d;hb=66dc72e7781e232e7fd166a0d420eb586250a4ea;hp=8068d205f2274dded3074158a105243bb4de1b11;hpb=6c043b44a1cec206f3b4f8063fb8383e66ae6d53;p=adu.git

diff --git a/adu.c b/adu.c
index 8068d20..8f5ff8d 100644
--- a/adu.c
+++ b/adu.c
@@ -3,22 +3,26 @@
 
 #include "gcc-compat.h"
 #include "cmdline.h"
-#include "osl.h"
 #include "fd.h"
-#include "hash.h"
 #include "string.h"
 #include "error.h"
+#include "portable_io.h"
 
 DEFINE_ERRLIST;
 
-#define DATABASE_DIR "/tmp/adu"
-#define UID_LIST DATABASE_DIR "/" "uid_list"
-
 /** Command line and config file options. */
 static struct gengetopt_args_info conf;
 
+enum uid_info_flags {
+	/** whether this slot of the hash table is used. */
+	UI_FL_SLOT_USED = 1,
+	/** whether this uid should be taken into account. */
+	UI_FL_ADMISSIBLE = 2,
+};
+
 struct user_info {
 	uint32_t uid;
+	uint32_t flags;
 	struct osl_table *table;
 	uint64_t files;
 	uint64_t bytes;
@@ -26,12 +30,93 @@ struct user_info {
 	struct osl_table_description *desc;
 };
 
+/**
+ * Contains info for each user that owns at least one regular file.
+ *
+ * Even users that are not taken into account because of the --uid
+ * option occupy a slot in this hash table. This allows to find out
+ * quicky whether a uid is admissible. And yes, this has to be fast.
+ */
 static struct user_info *uid_hash_table;
 
+static inline int ui_used(struct user_info *ui)
+{
+	return ui->flags & UI_FL_SLOT_USED;
+}
+
+static inline int ui_admissible(struct user_info *ui)
+{
+	return ui->flags & UI_FL_ADMISSIBLE;
+}
+
+struct uid_range {
+	uint32_t low;
+	uint32_t high;
+};
+
+static struct uid_range *admissible_uids;
+
+static inline int check_uid_arg(const char *arg, uint32_t *uid)
+{
+	const uint32_t max = ~0U;
+	/*
+	 * we need an 64-bit int for string -> uid conversion because strtoll()
+	 * returns a signed value.
+	 */
+	int64_t val;
+	int ret = para_atoi64(arg, &val);
+
+	if (ret < 0)
+		return ret;
+	if (val < 0 || val > max)
+		return -ERRNO_TO_ERROR(EINVAL);
+	*uid = val;
+	return 1;
+}
+
+static int parse_uid_range(const char *orig_arg, struct uid_range *ur)
+{
+	int ret;
+	char *arg = para_strdup(orig_arg), *p = strchr(arg, '-');
+
+	if (!p || p == arg) {
+		if (p == arg) /* -42 */
+			p++;
+		ret = check_uid_arg(p, &ur->high);
+		if (ret < 0)
+			goto out;
+		ur->low = p? 0 : ur->high;
+		ret = 1;
+		goto out;
+	}
+	/* 42- or 42-4711 */
+	*p = '\0';
+	p++;
+	ret = check_uid_arg(arg, &ur->low);
+	if (ret < 0)
+		goto out;
+	ur->high = ~0U;
+	if (*p) { /* 42-4711 */
+		ret = check_uid_arg(p, &ur->high);
+		if (ret < 0)
+			goto out;
+	}
+	if (ur->low > ur->high)
+		ret = -ERRNO_TO_ERROR(EINVAL);
+out:
+	if (ret < 0)
+		ERROR_LOG("bad uid option: %s\n", orig_arg);
+	else
+		INFO_LOG("admissible uid range: %u - %u\n", ur->low,
+			ur->high);
+	free(arg);
+	return ret;
+}
+
+
 /** 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)))
 
-
 /**
  * The log function.
  *
@@ -177,7 +262,6 @@ 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. */
@@ -246,15 +330,15 @@ static int open_user_table(struct user_info *ui, int create)
 	ui->desc->num_columns = NUM_UT_COLUMNS;
 	ui->desc->flags = 0;
 	ui->desc->column_descriptions = user_table_cols;
-	ui->desc->dir = para_strdup(DATABASE_DIR);
+	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;
+		num_uids++;
 	}
 	ret = osl_open_table(ui->desc, &ui->table);
 	if (ret < 0)
@@ -268,6 +352,7 @@ err:
 	ui->desc->dir = NULL;
 	ui->desc = NULL;
 	ui->table = NULL;
+	ui->flags = 0;
 	return ret;
 }
 
@@ -292,6 +377,7 @@ 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;
@@ -312,7 +398,7 @@ static int create_tables(void)
  *
  * 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.
+ * the second hash value has never 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)
@@ -329,6 +415,22 @@ enum search_uid_flags {
 	CREATE_USER_TABLE = 2,
 };
 
+static int uid_is_admissible(uint32_t uid)
+{
+	int i;
+
+	for (i = 0; i < conf.uid_given; i++) {
+		struct uid_range *ur = admissible_uids + i;
+
+		if (ur->low <= uid && ur->high >= uid)
+			break;
+	}
+	i = !conf.uid_given || i < conf.uid_given;
+	DEBUG_LOG("uid %u is %sadmissible\n", (unsigned)uid,
+		i? "" : "not ");
+	return i;
+}
+
 static int search_uid(uint32_t uid, enum search_uid_flags flags,
 		struct user_info **ui_ptr)
 {
@@ -337,15 +439,19 @@ static int search_uid(uint32_t uid, enum search_uid_flags flags,
 	for (p = 0; p < uid_hash_table_size; p++) {
 		struct user_info *ui = uid_hash_table + double_hash(uid, p);
 
-		if (!ui->table) {
+		if (!ui_used(ui)) {
 			int ret;
-
 			if (!flags)
 				return -E_BAD_UID;
 			ui->uid = uid;
+			ui->flags |= UI_FL_SLOT_USED;
+			if (!uid_is_admissible(uid))
+				return 0;
+			ui->flags |= UI_FL_ADMISSIBLE;
 			ret = open_user_table(ui, flags & CREATE_USER_TABLE);
 			if (ret < 0)
 				return ret;
+
 			if (ui_ptr)
 				*ui_ptr = ui;
 			return 1;
@@ -487,6 +593,42 @@ static int get_dir_name(struct osl_row *row, char **name)
 	return 1;
 }
 
+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)
+{
+	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]);
+}
+
+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]);
+}
+
 enum global_stats_flags {
 	GSF_PRINT_DIRNAME = 1,
 	GSF_PRINT_BYTES = 2,
@@ -503,7 +645,7 @@ static int global_stats_loop_function(struct osl_row *row, void *data)
 {
 	struct global_stats_info *gsi = data;
 	struct osl_object obj;
-	char *dirname;
+	char *dirname, formated_value[25];
 	int ret, summary = gsi->flags & GSF_COMPUTE_SUMMARY;
 
 	if (!gsi->count && !summary)
@@ -523,9 +665,12 @@ static int global_stats_loop_function(struct osl_row *row, void *data)
 		if (ret < 0)
 			return ret;
 		files = *(uint64_t *)obj.data;
-		if (gsi->count && (gsi->flags & GSF_PRINT_FILES))
-			printf("%llu%s", (long long unsigned)files,
+		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;
 	}
@@ -535,14 +680,17 @@ static int global_stats_loop_function(struct osl_row *row, void *data)
 		if (ret < 0)
 			return ret;
 		bytes = *(uint64_t *)obj.data;
-		if (gsi->count && (gsi->flags & GSF_PRINT_BYTES))
-			printf("%llu\n", (long long unsigned)bytes);
+		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)
+	if (gsi->count > 0)
 		gsi->count--;
 	return 1;
 }
@@ -553,12 +701,21 @@ static void print_id_stats(void)
 
 	printf("--------------------- user summary (uid/dirs/files/bytes):\n");
 	FOR_EACH_USER(ui) {
-		if (!ui->table)
+		char formated_dir_count[25], formated_file_count[25],
+			formated_bytes[25];
+		if (!ui_used(ui))
 			continue;
-		printf("%u\t%llu\t%llu\t%llu\n", (unsigned)ui->uid,
-			(long long unsigned)ui->dirs,
-			(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
+		);
 	}
 }
 
@@ -581,6 +738,7 @@ static int user_stats_loop_function(struct osl_row *row, void *data)
 	struct osl_row *dir_row;
 	struct osl_object obj;
 	int ret, summary = usi->flags & GSF_COMPUTE_SUMMARY;
+	char formated_value[25];
 
 	if (!usi->count && !summary)
 		return -E_LOOP_COMPLETE;
@@ -608,11 +766,13 @@ static int user_stats_loop_function(struct osl_row *row, void *data)
 		if (ret < 0)
 			return ret;
 		files = *(uint64_t *)obj.data;
-		if (usi->count && (usi->flags & USF_PRINT_FILES))
-			printf("%llu%s",
-				(long long unsigned)files,
+		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;
 	}
@@ -622,15 +782,18 @@ static int user_stats_loop_function(struct osl_row *row, void *data)
 		if (ret < 0)
 			return ret;
 		bytes = *(uint64_t *)obj.data;
-		if (usi->count && (usi->flags & USF_PRINT_BYTES))
-			printf("%llu\n", (long long unsigned)bytes);
+		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)
+	if (usi->count > 0)
 		usi->count--;
 	return 1;
 }
@@ -641,21 +804,19 @@ static void print_user_stats(void)
 
 	FOR_EACH_USER(ui) {
 		struct user_stats_info usi = {
-			.count = 10,
+			.count = conf.limit_arg,
 			.ui = ui
 		};
-		if (!ui->table)
+		if (!ui_used(ui) || !ui_admissible(ui))
 			continue;
 		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 = 10;
+		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);
@@ -666,7 +827,7 @@ static int print_statistics(void)
 {
 	int ret;
 	struct global_stats_info gsi = {
-		.count = 10,
+		.count = conf.limit_arg,
 		.flags = GSF_PRINT_DIRNAME | GSF_PRINT_BYTES | GSF_COMPUTE_SUMMARY
 	};
 
@@ -675,7 +836,7 @@ static int print_statistics(void)
 		global_stats_loop_function);
 	if (ret < 0 && ret != -E_LOOP_COMPLETE)
 		return ret;
-	gsi.count = 10;
+	gsi.count = conf.limit_arg;
 
 	gsi.flags = GSF_PRINT_DIRNAME | GSF_PRINT_FILES;
 	printf("---------- dirs containing most files ------------\n");
@@ -693,9 +854,14 @@ static int print_statistics(void)
 	return 1;
 }
 
+static char *get_uid_list_name(void)
+{
+	return make_message("%s/uid_list", conf.database_dir_arg);
+}
+
 static int write_uid_list(void)
 {
-	char *buf;
+	char *buf, *filename;
 	uint32_t count = 0;
 	struct user_info *ui;
 	size_t size = num_uids * sizeof(uint32_t);
@@ -705,17 +871,22 @@ static int write_uid_list(void)
 		return 0;
 	buf = para_malloc(size);
 	FOR_EACH_USER(ui) {
-		if (!ui->table)
+		if (!ui_used(ui) || !ui_admissible(ui))
 			continue;
+		DEBUG_LOG("saving uid %u\n", (unsigned) ui->uid);
 		write_u32(buf + count++ * sizeof(uint32_t), ui->uid);
 	}
-	ret = para_write_file(UID_LIST, buf, size);
+	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);
 }
 
@@ -728,6 +899,7 @@ static void close_dir_table(void)
 	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;
 }
 
@@ -735,7 +907,7 @@ static void close_user_table(struct user_info *ui)
 {
 	int ret;
 
-	if (!ui || !ui->table)
+	if (!ui || !ui_used(ui) || !ui_admissible(ui))
 		return;
 	ret = osl_close_table(ui->table, OSL_MARK_CLEAN);
 	if (ret < 0)
@@ -748,6 +920,7 @@ static void close_user_table(struct user_info *ui)
 	free(ui->desc);
 	ui->desc = NULL;
 	ui->table = NULL;
+	ui->flags = 0;
 }
 
 static void close_user_tables(void)
@@ -784,14 +957,19 @@ out:
 
 static int read_uid_file(void)
 {
-	char *map;
 	size_t size;
-	int ret = mmap_full_file(UID_LIST, O_RDONLY, (void **)&map, &size, NULL);
 	uint32_t n;
+	char *filename = get_uid_list_name(), *map;
+	int ret = mmap_full_file(filename, O_RDONLY, (void **)&map, &size, NULL);
 
-	if (ret < 0)
+	if (ret < 0) {
+		INFO_LOG("failed to map %s\n", filename);
+		free(filename);
 		return ret;
+	}
 	num_uids = size / 4;
+	INFO_LOG("found %u uids in %s\n", (unsigned)num_uids, filename);
+	free(filename);
 	/* 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)
@@ -823,6 +1001,27 @@ static int com_select(void)
 	return 1;
 }
 
+static int check_args(void)
+{
+	int i, ret;
+
+	if (!conf.uid_given)
+		return 0;
+
+	admissible_uids = para_malloc(conf.uid_given * sizeof(*admissible_uids));
+
+	for (i = 0; i < conf.uid_given; i++) {
+		ret = parse_uid_range(conf.uid_arg[i], admissible_uids + i);
+		if (ret < 0)
+			goto err;
+	}
+	return 1;
+err:
+	free(admissible_uids);
+	admissible_uids = NULL;
+	return ret;
+}
+
 int main(int argc, char **argv)
 {
 	int ret;
@@ -835,6 +1034,9 @@ int main(int argc, char **argv)
 	};
 
 	cmdline_parser_ext(argc, argv, &conf, &params); /* aborts on errors */
+	ret = check_args();
+	if (ret < 0)
+		goto out;
 	ret = -E_SYNTAX;
 	if (conf.select_given)
 		ret = com_select();
@@ -843,6 +1045,7 @@ int main(int argc, char **argv)
 	if (ret < 0)
 		goto out;
 out:
+	free(admissible_uids);
 	if (ret < 0) {
 		ERROR_LOG("%s\n", error_txt(-ret));
 		return -EXIT_FAILURE;