snapraid/cmdline/search.c

// SPDX-License-Identifier: GPL-3.0-or-later
// Copyright (C) 2014 Andrea Mazzoleni

#include "os/portable.h"

#include "support.h"
#include "search.h"

#ifndef _WIN32
static const char* es(int err)
{
	if (is_hw(err))
		return "error_io";
	else
		return "error";
}
#endif

/****************************************************************************/
/* search */

static void search_file(struct snapraid_state* state, const char* path, data_off_t size, int64_t mtime_sec, int mtime_nsec)
{
	struct snapraid_search_file* file;
	tommy_hash_t file_hash;

	file = malloc_nofail(sizeof(struct snapraid_search_file));
	file->path = strdup_nofail(path);
	file->size = size;
	file->mtime_sec = mtime_sec;
	file->mtime_nsec = mtime_nsec;

	file_hash = file_stamp_hash(file->size, file->mtime_sec, file->mtime_nsec);

	tommy_hashdyn_insert(&state->searchset, &file->node, file, file_hash);
}

void search_file_free(void* void_file)
{
	struct snapraid_search_file* file = void_file;

	free(file->path);
	free(file);
}

struct search_file_compare_arg {
	const struct snapraid_state* state;
	const struct snapraid_block* block;
	const struct snapraid_file* file;
	unsigned char* buffer;
	data_off_t offset;
	size_t read_size;
	int prevhash;
};

int search_file_compare(const void* void_arg, const void* void_data)
{
	const struct search_file_compare_arg* arg = void_arg;
	const struct snapraid_search_file* file = void_data;
	const struct snapraid_state* state = arg->state;
	unsigned char buffer_hash[HASH_MAX];
	const char* path = file->path;
	int f;
	ssize_t ret;
	size_t count;

	/* compare file info */
	if (arg->file->size != file->size)
		return -1;

	if (arg->file->mtime_sec != file->mtime_sec)
		return -1;

	if (arg->file->mtime_nsec != file->mtime_nsec)
		return -1;

	/* read the block and compare the hash */
	f = open(path, O_RDONLY | O_BINARY);
	if (f == -1) {
		if (errno == ENOENT)
			log_error(EUSER, "WARNING! Search candidate file '%s' disappeared.\n", path);
		else
			log_error(errno, "WARNING! Error opening search candidate '%s'. %s.\n", path, strerror(errno));
		return -1;
	}

	count = 0;
	do {
		ret = pread(f, (char*)arg->buffer + count, arg->read_size - count, arg->offset + count);
		if (ret < 0) {
			if (errno == EINTR)
				continue;

			log_error(errno, "WARNING! Error reading search candidate '%s'. %s.\n", path, strerror(errno));
			if (close(f) != 0)
				log_error(errno, "WARNING! Error closing search candidate '%s'. %s.\n", path, strerror(errno));
			return -1;
		}
		if (ret == 0) {
			log_error(EUSER, "WARNING! Unexpected end of file in search candidate '%s'.\n", path);
			if (close(f) != 0)
				log_error(errno, "WARNING! Error closing search candidate '%s'. %s.\n", path, strerror(errno));
			return -1;
		}

		count += ret;
	} while (count < arg->read_size);

	ret = close(f);
	if (ret != 0) {
		log_error(errno, "WARNING! Error closing search candidate '%s'. %s.\n", path, strerror(errno));
		return -1;
	}

	if (arg->read_size != state->block_size) {
		/* fill the remaining with 0 */
		memset(arg->buffer + arg->read_size, 0, state->block_size - arg->read_size);
	}

	/* compute the block hash from the canonical buffer */
	if (arg->prevhash)
		memhash_block(state->prevhash, state->prevhashseed, buffer_hash, arg->buffer, arg->read_size, state->block_size);
	else
		memhash_block(state->hash, state->hashseed, buffer_hash, arg->buffer, arg->read_size, state->block_size);

	/* check if the hash is matching */
	if (memcmp(buffer_hash, arg->block->hash, BLOCK_HASH_SIZE) != 0)
		return -1;

	return 0;
}

int state_search_fetch(struct snapraid_state* state, int prevhash, struct snapraid_file* missing_file, block_off_t missing_file_pos, struct snapraid_block* missing_block, unsigned char* buffer)
{
	struct snapraid_search_file* file;
	tommy_hash_t file_hash;
	struct search_file_compare_arg arg;

	arg.state = state;
	arg.block = missing_block;
	arg.file = missing_file;
	arg.buffer = buffer;
	arg.offset = state->block_size * missing_file_pos;
	arg.read_size = file_block_size(missing_file, missing_file_pos, state->block_size);
	arg.prevhash = prevhash;

	file_hash = file_stamp_hash(arg.file->size, arg.file->mtime_sec, arg.file->mtime_nsec);

	/* search in the hashtable, and also check if the data matches the hash */
	file = tommy_hashdyn_search(&state->searchset, search_file_compare, &arg, file_hash);
	if (!file)
		return -1;

	/* if found, buffer is already set with data */
	return 0;
}

static void search_dir(struct snapraid_state* state, struct snapraid_disk* disk, const struct fsidentity* root_identity, const char* dir, const char* sub)
{
	DIR* d;

	size_t mount_point_len = disk != 0 ? strlen(disk->mount_point) : 0;
	size_t sub_len = strlen(sub);

	d = opendir(dir);
	if (!d) {
		/* LCOV_EXCL_START */
		log_fatal(errno, "Error opening directory '%s'. %s.\n", dir, strerror(errno));
		exit(EXIT_FAILURE);
		/* LCOV_EXCL_STOP */
	}

	while (1) {
		char path_next[PATH_MAX];
		char sub_next[PATH_MAX];
		char out[PATH_MAX];
		struct snapraid_filter* reason = 0;
		struct stat st;
		const char* name;
		struct dirent* dd;

		/* clear errno to detect erroneous conditions */
		errno = 0;
		dd = readdir(d);
		if (dd == 0 && errno != 0) {
			/* LCOV_EXCL_START */
			log_fatal(errno, "Error reading directory '%s'. %s.\n", dir, strerror(errno));
			exit(EXIT_FAILURE);
			/* LCOV_EXCL_STOP */
		}
		if (dd == 0) {
			break; /* finished */
		}

		/* skip "." and ".." files */
		name = dd->d_name;
		if (name[0] == '.' && (name[1] == 0 || (name[1] == '.' && name[2] == 0)))
			continue;

		pathprint(path_next, sizeof(path_next), "%s%s", dir, name);
		pathprint(sub_next, sizeof(sub_next), "%s%s", sub, name);

		/* exclude hidden files even before calling lstat() */
		if (disk != 0 && filter_hidden(state->filter_hidden, dd) != 0) {
			if (pathcmp(".snapraidignore", dd->d_name) != 0) {
				msg_verbose("Excluding hidden '%s'\n", path_next);
				continue;
			}
		}

		/* exclude content files even before calling lstat() */
		if (disk != 0 && filter_content(&state->contentlist, disk->mount_point, mount_point_len, sub, sub_len, name) != 0) {
			msg_verbose("Excluding content '%s'\n", path_next);
			continue;
		}

		/* exclude snapshot container even before calling lstat() */
		if (disk != 0 && filter_snapshot(state->snapshot, sub, name) != 0) {
			msg_verbose("Excluding snapshots directory '%s'\n", path_next);
			continue;
		}

#if HAVE_STRUCT_DIRENT_D_STAT
		/* convert dirent to lstat result */
		dirent_lstat(dd, &st);

		/*
		 * If the st_mode field is missing, takes care to fill it using normal lstat()
		 * at now this can happen only in Windows (with HAVE_STRUCT_DIRENT_D_STAT defined),
		 * because we use a directory reading method that doesn't read info about ReparsePoint.
		 * Note that here we cannot call here lstat_sync(), because we don't know what kind
		 * of file is it, and lstat_sync() doesn't always work
		 */
		if (st.st_mode == 0) {
			if (lstat(path_next, &st) != 0) {
				/* LCOV_EXCL_START */
				log_fatal(errno, "Error in stat file/directory '%s'. %s.\n", path_next, strerror(errno));
				exit(EXIT_FAILURE);
				/* LCOV_EXCL_STOP */
			}
		}
#else
		/* get lstat info about the file */
		if (lstat(path_next, &st) != 0) {
			/* LCOV_EXCL_START */
			log_fatal(errno, "Error in stat file/directory '%s'. %s.\n", path_next, strerror(errno));
			exit(EXIT_FAILURE);
			/* LCOV_EXCL_STOP */
		}
#endif

		if (S_ISREG(st.st_mode)) {
			if (disk == 0
				|| pathcmp(".snapraidignore", name) == 0
				|| filter_path(&state->filterlist, &reason, disk->name, sub_next) == 0) {
				search_file(state, path_next, st.st_size, st.st_mtime, STAT_NSEC(&st));
			} else {
				msg_verbose("Excluding file '%s' for rule '%s'\n", path_next, filter_type(reason, out, sizeof(out)));
			}
		} else if (S_ISDIR(st.st_mode)) {
			if (disk == 0 || filter_subdir(&state->filterlist, &reason, disk->name, sub_next) == 0) {
#ifndef _WIN32
				/*
				 * Automatic array search must stay within the same filesystem,
				 * mount, and subvolume boundaries as the normal data-disk scan.
				 * Explicit -i imports pass no root identity and may cross them.
				 */
				if (root_identity) {
					struct fsidentity identity;

					if (fsidentity(path_next, &st, &identity) != 0) {
						/* LCOV_EXCL_START */
						log_tag("%s:%u:%s:%s: Statx error. %s.\n", es(errno), 0, disk->name, esc_tag(path_next), strerror(errno));
						log_fatal(errno, "Error accessing directory '%s'. %s.\n", path_next, strerror(errno));
						exit(EXIT_FAILURE);
						/* LCOV_EXCL_STOP */
					}

					if (identity.device != root_identity->device) {
						log_tag("%s:%u:%s:%s: Ignoring mount point.\n", es(ESOFT), 0, disk->name, esc_tag(path_next));
						log_error(ESOFT, "WARNING! Ignoring mount point '%s' because it appears to be in a different device\n", path_next);
						continue;
					} else if (identity.mnt_id != root_identity->mnt_id) {
						log_tag("%s:%u:%s:%s: Ignoring mount point.\n", es(ESOFT), 0, disk->name, esc_tag(path_next));
						log_error(ESOFT, "WARNING! Ignoring mount point '%s' because it belongs to a different mount\n", path_next);
						continue;
					} else if (identity.subvol != root_identity->subvol) {
						log_tag("%s:%u:%s:%s: Ignoring nested subvolume.\n", es(ESOFT), 0, disk->name, esc_tag(path_next));
						log_error(ESOFT, "WARNING! Ignoring nested subvolume '%s' because it belongs to a different subvolume\n", path_next);
						continue;
					}
				}
#endif

				pathslash(path_next, sizeof(path_next));
				pathslash(sub_next, sizeof(sub_next));
				search_dir(state, disk, root_identity, path_next, sub_next);
			} else {
				msg_verbose("Excluding directory '%s' for rule '%s'\n", path_next, filter_type(reason, out, sizeof(out)));
			}
		}
	}

	if (closedir(d) != 0) {
		/* LCOV_EXCL_START */
		log_fatal(errno, "Error closing directory '%s'. %s.\n", dir, strerror(errno));
		exit(EXIT_FAILURE);
		/* LCOV_EXCL_STOP */
	}
}

void state_search(struct snapraid_state* state, const char* dir)
{
	char path[PATH_MAX];

	msg_progress("Importing...\n");

	/* add the final slash */
	pathimport(path, sizeof(path), dir);
	pathslash(path, sizeof(path));

	search_dir(state, 0, 0, path, "");
}

void state_search_array(struct snapraid_state* state)
{
	tommy_node* i;

	/* import from all the disks */
	for (i = state->disklist; i != 0; i = i->next) {
		struct snapraid_disk* disk = i->data;

		/* skip data disks that are not accessible */
		if (disk->skip_access)
			continue;

		msg_progress("Searching disk %s...\n", disk->name);

#ifndef _WIN32
		struct stat st;
		struct fsidentity identity;

		if (lstat(disk->dir, &st) != 0) {
			/* LCOV_EXCL_START */
			log_tag("%s:%u:%s:%s: Stat error. %s.\n", es(errno), 0, disk->name, esc_tag(disk->dir), strerror(errno));
			log_fatal(errno, "Error accessing directory '%s'. %s.\n", disk->dir, strerror(errno));
			exit(EXIT_FAILURE);
			/* LCOV_EXCL_STOP */
		}

		if (fsidentity(disk->dir, &st, &identity) != 0) {
			/* LCOV_EXCL_START */
			log_tag("%s:%u:%s:%s: Statx error. %s.\n", es(errno), 0, disk->name, esc_tag(disk->dir), strerror(errno));
			log_fatal(errno, "Error accessing directory '%s'. %s.\n", disk->dir, strerror(errno));
			exit(EXIT_FAILURE);
			/* LCOV_EXCL_STOP */
		}

		search_dir(state, disk, &identity, disk->dir, "");
#else
		search_dir(state, disk, 0, disk->dir, "");
#endif
	}
}