genquashfs: detect/filter hardlinks when scanning a directory

The new behavior is enabled by default (except on Windows) and needs
to be turned off explicitly.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

5 files changed, 37 insertions, 5 deletions

diff --git a/bin/gensquashfs/gensquashfs.1 b/bin/gensquashfs/gensquashfs.1
index 88f58c4..5b53bcd 100644
--- a/bin/gensquashfs/gensquashfs.1
+++ b/bin/gensquashfs/gensquashfs.1
@@ -106,6 +106,11 @@ Generate an export table for NFS support.
 Do not perform tail end packing on files that are larger than the specified
 block size.
 .TP
+\fB\-\-no\-hard\-links\fR, \fB\-H\fR
+Do not perform hard link detection when scanning directories. By default,
+gensquashfs records device and inode numbers to find hard links in the input
+directory hierarchy.
+.TP
 \fB\-\-force\fR, \fB\-f\fR
 Overwrite the output file if it exists.
 .TP
@@ -206,6 +211,7 @@ T}
 \-xdev;Do not cross mount points during a recursive glob.
 \-mount;An alias for \fB\-xdev\fR
 \-keeptime;Use the time stamps from the scanned files.
+\-nohardlinks;Do not perform hard link detection.
 \-nonrecursive;T{
 Do not descend into directories.
 
diff --git a/bin/gensquashfs/src/glob.c b/bin/gensquashfs/src/glob.c
index 27b7baa..bc6edee 100644
--- a/bin/gensquashfs/src/glob.c
+++ b/bin/gensquashfs/src/glob.c
@@ -47,6 +47,7 @@ static const struct {
 	{ "-mount", DIR_SCAN_ONE_FILESYSTEM },
 	{ "-keeptime", DIR_SCAN_KEEP_TIME },
 	{ "-nonrecursive", DIR_SCAN_NO_RECURSION },
+	{ "-nohardlinks", DIR_SCAN_NO_HARDLINKS },
 };
 
 static bool set_scan_flag(const char *arg, dir_tree_cfg_t *cfg)
@@ -176,6 +177,10 @@ int glob_files(fstree_t *fs, const char *filename, size_t line_num,
 	cfg.prefix = prefix;
 	cfg.flags = glob_flags;
 
+#if defined(_WIN32) || defined(__WINDOWS__)
+	cfg.flags |= DIR_SCAN_NO_HARDLINKS;
+#endif
+
 	while (sep->count != 0) {
 		if (sep->args[0][0] != '-')
 			break;
diff --git a/bin/gensquashfs/src/options.c b/bin/gensquashfs/src/options.c
index f263bce..c8bf64b 100644
--- a/bin/gensquashfs/src/options.c
+++ b/bin/gensquashfs/src/options.c
@@ -35,6 +35,7 @@ static struct option long_opts[] = {
 #ifdef WITH_SELINUX
 	{ "selinux", required_argument, NULL, 's' },
 #endif
+	{ "no-hard-links", no_argument, NULL, 'H' },
 	{ "xattr-file", required_argument, NULL, 'A' },
 	{ "sort-file", required_argument, NULL, 'S' },
 	{ "version", no_argument, NULL, 'V' },
@@ -42,7 +43,7 @@ static struct option long_opts[] = {
 	{ NULL, 0, NULL, 0 },
 };
 
-static const char *short_opts = "F:D:X:c:b:B:d:u:g:j:Q:S:A:kxoefqThV"
+static const char *short_opts = "HF:D:X:c:b:B:d:u:g:j:Q:S:A:kxoefqThV"
 #ifdef WITH_SELINUX
 "s:"
 #endif
@@ -123,6 +124,10 @@ const char *extra_options =
 "  --exportable, -e            Generate an export table for NFS support.\n"
 "  --no-tail-packing, -T       Do not perform tail end packing on files that\n"
 "                              are larger than block size.\n"
+#if !defined(_WIN32) && !defined(__WINDOWS__)
+"  --no-hard-links, -H         When scanning a directory, do not attempt to\n"
+"                              detect hard links.\n"
+#endif
 "  --force, -f                 Overwrite the output file if it exists.\n"
 "  --quiet, -q                 Do not print out progress reports.\n"
 "  --help, -h                  Print help text and exit.\n"
@@ -223,6 +228,9 @@ void process_command_line(options_t *opt, int argc, char **argv)
 		case 'T':
 			opt->no_tail_packing = true;
 			break;
+		case 'H':
+			opt->dirscan_flags |= DIR_SCAN_NO_HARDLINKS;
+			break;
 		case 'c':
 			have_compressor = true;
 			ret = sqfs_compressor_id_from_name(optarg);
@@ -332,6 +340,10 @@ void process_command_line(options_t *opt, int argc, char **argv)
 		}
 	}
 
+#if defined(_WIN32) || defined(__WINDOWS__)
+	opt->dirscan_flags |= DIR_SCAN_NO_HARDLINKS;
+#endif
+
 	if (opt->cfg.num_jobs < 1)
 		opt->cfg.num_jobs = 1;
 
diff --git a/include/io/dir_iterator.h b/include/io/dir_iterator.h
index 210316f..7154b08 100644
--- a/include/io/dir_iterator.h
+++ b/include/io/dir_iterator.h
@@ -26,7 +26,8 @@ enum {
 
 	DIR_SCAN_ONE_FILESYSTEM = 0x1000,
 	DIR_SCAN_NO_RECURSION = 0x2000,
-	DIR_SCAN_MATCH_FULL_PATH = 0x4000,
+	DIR_SCAN_NO_HARDLINKS = 0x4000,
+	DIR_SCAN_MATCH_FULL_PATH = 0x8000,
 };
 
 typedef struct {
diff --git a/lib/io/src/dir_tree_iterator.c b/lib/io/src/dir_tree_iterator.c
index 3c25529..5688489 100644
--- a/lib/io/src/dir_tree_iterator.c
+++ b/lib/io/src/dir_tree_iterator.c
@@ -229,9 +229,15 @@ sqfs_dir_iterator_t *dir_tree_iterator_create(const char *path,
 
 	ret = sqfs_dir_iterator_create_recursive(&it->rec, dir);
 	sqfs_drop(dir);
-	if (ret) {
-		fprintf(stderr, "%s: out of memory\n", path);
-		goto fail;
+	if (ret)
+		goto fail_oom;
+
+	if (!(cfg->flags & DIR_SCAN_NO_HARDLINKS)) {
+		ret = sqfs_hard_link_filter_create(&dir, it->rec);
+		sqfs_drop(it->rec);
+		it->rec = dir;
+		if (ret)
+			goto fail_oom;
 	}
 
 	sqfs_object_init(it, destroy, NULL);
@@ -243,6 +249,8 @@ sqfs_dir_iterator_t *dir_tree_iterator_create(const char *path,
 	((sqfs_dir_iterator_t *)it)->read_xattr = read_xattr;
 
 	return (sqfs_dir_iterator_t *)it;
+fail_oom:
+	fprintf(stderr, "%s: out of memory\n", path);
 fail:
 	free(it);
 	return NULL;