mtree: Make reading additional information from the fs optional

This feature is not needed if users just want to read in the content of
an mtree file and do validation against the file system themselves.

It is needed for `bsdtar cvf out.tar @input.mtree` which is why the
option is enabled in bsdtar.

Since the mtree tests rely on this feature, this patch also enables it
there.

Signed-off-by: Florian Pritz <bluewind@xinu.at>

diff --git a/libarchive/archive_read_support_format_mtree.c b/libarchive/archive_read_support_format_mtree.c
index 44799dfc95d7b624b603f560afcd48ee9d3c1492..d82d4c157353e9cb6429a0e241986b00d0cc12e6 100644 (file)
--- a/libarchive/archive_read_support_format_mtree.c
+++ b/libarchive/archive_read_support_format_mtree.c
@@ -104,6 +104,7 @@ struct mtree {
        struct archive_entry_linkresolver *resolver;
 
        int64_t                  cur_size;
+       char checkfs;
 };
 
 static int     bid_keycmp(const char *, const char *, ssize_t);
@@ -174,6 +175,29 @@ get_time_t_min(void)
 #endif
 }
 
+static int
+archive_read_format_mtree_options(struct archive_read *a,
+    const char *key, const char *val)
+{
+       struct mtree *mtree;
+
+       mtree = (struct mtree *)(a->format->data);
+       if (strcmp(key, "checkfs")  == 0) {
+               /* Allows to read information missing from the mtree from the file system */
+               if (val == NULL || val[0] == 0) {
+                       mtree->checkfs = 0;
+               } else {
+                       mtree->checkfs = 1;
+               }
+               return (ARCHIVE_OK);
+       }
+
+       /* Note: The "warn" return is just to inform the options
+        * supervisor that we didn't handle it.  It will generate
+        * a suitable error if no one used this option. */
+       return (ARCHIVE_WARN);
+}
+
 static void
 free_options(struct mtree_option *head)
 {
@@ -206,7 +230,7 @@ archive_read_support_format_mtree(struct archive *_a)
        mtree->fd = -1;
 
        r = __archive_read_register_format(a, mtree, "mtree",
-           mtree_bid, NULL, read_header, read_data, skip, NULL, cleanup, NULL, NULL);
+           mtree_bid, archive_read_format_mtree_options, read_header, read_data, skip, NULL, cleanup, NULL, NULL);
 
        if (r != ARCHIVE_OK)
                free(mtree);
@@ -1104,162 +1128,164 @@ parse_file(struct archive_read *a, struct archive_entry *entry,
                        mtree->current_dir.length = n;
        }
 
-       /*
-        * Try to open and stat the file to get the real size
-        * and other file info.  It would be nice to avoid
-        * this here so that getting a listing of an mtree
-        * wouldn't require opening every referenced contents
-        * file.  But then we wouldn't know the actual
-        * contents size, so I don't see a really viable way
-        * around this.  (Also, we may want to someday pull
-        * other unspecified info from the contents file on
-        * disk.)
-        */
-       mtree->fd = -1;
-       if (archive_strlen(&mtree->contents_name) > 0)
-               path = mtree->contents_name.s;
-       else
-               path = archive_entry_pathname(entry);
-
-       if (archive_entry_filetype(entry) == AE_IFREG ||
-           archive_entry_filetype(entry) == AE_IFDIR) {
-               mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC);
-               __archive_ensure_cloexec_flag(mtree->fd);
-               if (mtree->fd == -1 &&
-                   (errno != ENOENT ||
-                    archive_strlen(&mtree->contents_name) > 0)) {
-                       archive_set_error(&a->archive, errno,
-                           "Can't open %s", path);
-                       r = ARCHIVE_WARN;
+       if (mtree->checkfs) {
+               /*
+                * Try to open and stat the file to get the real size
+                * and other file info.  It would be nice to avoid
+                * this here so that getting a listing of an mtree
+                * wouldn't require opening every referenced contents
+                * file.  But then we wouldn't know the actual
+                * contents size, so I don't see a really viable way
+                * around this.  (Also, we may want to someday pull
+                * other unspecified info from the contents file on
+                * disk.)
+                */
+               mtree->fd = -1;
+               if (archive_strlen(&mtree->contents_name) > 0)
+                       path = mtree->contents_name.s;
+               else
+                       path = archive_entry_pathname(entry);
+
+               if (archive_entry_filetype(entry) == AE_IFREG ||
+                               archive_entry_filetype(entry) == AE_IFDIR) {
+                       mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC);
+                       __archive_ensure_cloexec_flag(mtree->fd);
+                       if (mtree->fd == -1 &&
+                                       (errno != ENOENT ||
+                                        archive_strlen(&mtree->contents_name) > 0)) {
+                               archive_set_error(&a->archive, errno,
+                                               "Can't open %s", path);
+                               r = ARCHIVE_WARN;
+                       }
                }
-       }
 
-       st = &st_storage;
-       if (mtree->fd >= 0) {
-               if (fstat(mtree->fd, st) == -1) {
-                       archive_set_error(&a->archive, errno,
-                           "Could not fstat %s", path);
-                       r = ARCHIVE_WARN;
-                       /* If we can't stat it, don't keep it open. */
-                       close(mtree->fd);
-                       mtree->fd = -1;
+               st = &st_storage;
+               if (mtree->fd >= 0) {
+                       if (fstat(mtree->fd, st) == -1) {
+                               archive_set_error(&a->archive, errno,
+                                               "Could not fstat %s", path);
+                               r = ARCHIVE_WARN;
+                               /* If we can't stat it, don't keep it open. */
+                               close(mtree->fd);
+                               mtree->fd = -1;
+                               st = NULL;
+                       }
+               } else if (lstat(path, st) == -1) {
                        st = NULL;
                }
-       } else if (lstat(path, st) == -1) {
-               st = NULL;
-       }
 
-       /*
-        * Check for a mismatch between the type in the specification and
-        * the type of the contents object on disk.
-        */
-       if (st != NULL) {
-               if (
-                   ((st->st_mode & S_IFMT) == S_IFREG &&
-                    archive_entry_filetype(entry) == AE_IFREG)
+               /*
+                * Check for a mismatch between the type in the specification and
+                * the type of the contents object on disk.
+                */
+               if (st != NULL) {
+                       if (
+                                       ((st->st_mode & S_IFMT) == S_IFREG &&
+                                        archive_entry_filetype(entry) == AE_IFREG)
 #ifdef S_IFLNK
-                   || ((st->st_mode & S_IFMT) == S_IFLNK &&
-                       archive_entry_filetype(entry) == AE_IFLNK)
+                                       || ((st->st_mode & S_IFMT) == S_IFLNK &&
+                                               archive_entry_filetype(entry) == AE_IFLNK)
 #endif
 #ifdef S_IFSOCK
-                   || ((st->st_mode & S_IFSOCK) == S_IFSOCK &&
-                       archive_entry_filetype(entry) == AE_IFSOCK)
+                                       || ((st->st_mode & S_IFSOCK) == S_IFSOCK &&
+                                               archive_entry_filetype(entry) == AE_IFSOCK)
 #endif
 #ifdef S_IFCHR
-                   || ((st->st_mode & S_IFMT) == S_IFCHR &&
-                       archive_entry_filetype(entry) == AE_IFCHR)
+                                       || ((st->st_mode & S_IFMT) == S_IFCHR &&
+                                               archive_entry_filetype(entry) == AE_IFCHR)
 #endif
 #ifdef S_IFBLK
-                   || ((st->st_mode & S_IFMT) == S_IFBLK &&
-                       archive_entry_filetype(entry) == AE_IFBLK)
+                                       || ((st->st_mode & S_IFMT) == S_IFBLK &&
+                                               archive_entry_filetype(entry) == AE_IFBLK)
 #endif
-                   || ((st->st_mode & S_IFMT) == S_IFDIR &&
-                       archive_entry_filetype(entry) == AE_IFDIR)
+                                       || ((st->st_mode & S_IFMT) == S_IFDIR &&
+                                               archive_entry_filetype(entry) == AE_IFDIR)
 #ifdef S_IFIFO
-                   || ((st->st_mode & S_IFMT) == S_IFIFO &&
-                       archive_entry_filetype(entry) == AE_IFIFO)
+                                       || ((st->st_mode & S_IFMT) == S_IFIFO &&
+                                                       archive_entry_filetype(entry) == AE_IFIFO)
 #endif
-                   ) {
-                       /* Types match. */
-               } else {
-                       /* Types don't match; bail out gracefully. */
-                       if (mtree->fd >= 0)
-                               close(mtree->fd);
-                       mtree->fd = -1;
-                       if (parsed_kws & MTREE_HAS_OPTIONAL) {
-                               /* It's not an error for an optional entry
-                                  to not match disk. */
-                               *use_next = 1;
-                       } else if (r == ARCHIVE_OK) {
-                               archive_set_error(&a->archive,
-                                   ARCHIVE_ERRNO_MISC,
-                                   "mtree specification has different type for %s",
-                                   archive_entry_pathname(entry));
-                               r = ARCHIVE_WARN;
-                       }
-                       return r;
+                                       ) {
+                                               /* Types match. */
+                                       } else {
+                                               /* Types don't match; bail out gracefully. */
+                                               if (mtree->fd >= 0)
+                                                       close(mtree->fd);
+                                               mtree->fd = -1;
+                                               if (parsed_kws & MTREE_HAS_OPTIONAL) {
+                                                       /* It's not an error for an optional entry
+                                                          to not match disk. */
+                                                       *use_next = 1;
+                                               } else if (r == ARCHIVE_OK) {
+                                                       archive_set_error(&a->archive,
+                                                                       ARCHIVE_ERRNO_MISC,
+                                                                       "mtree specification has different type for %s",
+                                                                       archive_entry_pathname(entry));
+                                                       r = ARCHIVE_WARN;
+                                               }
+                                               return r;
+                                       }
                }
-       }
 
-       /*
-        * If there is a contents file on disk, pick some of the metadata
-        * from that file.  For most of these, we only set it from the contents
-        * if it wasn't already parsed from the specification.
-        */
-       if (st != NULL) {
-               if (((parsed_kws & MTREE_HAS_DEVICE) == 0 ||
-                    (parsed_kws & MTREE_HAS_NOCHANGE) != 0) &&
-                   (archive_entry_filetype(entry) == AE_IFCHR ||
-                    archive_entry_filetype(entry) == AE_IFBLK))
-                       archive_entry_set_rdev(entry, st->st_rdev);
-               if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) == 0 ||
-                   (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
-                       archive_entry_set_gid(entry, st->st_gid);
-               if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) == 0 ||
-                   (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
-                       archive_entry_set_uid(entry, st->st_uid);
-               if ((parsed_kws & MTREE_HAS_MTIME) == 0 ||
-                   (parsed_kws & MTREE_HAS_NOCHANGE) != 0) {
+               /*
+                * If there is a contents file on disk, pick some of the metadata
+                * from that file.  For most of these, we only set it from the contents
+                * if it wasn't already parsed from the specification.
+                */
+               if (st != NULL) {
+                       if (((parsed_kws & MTREE_HAS_DEVICE) == 0 ||
+                                               (parsed_kws & MTREE_HAS_NOCHANGE) != 0) &&
+                                       (archive_entry_filetype(entry) == AE_IFCHR ||
+                                        archive_entry_filetype(entry) == AE_IFBLK))
+                               archive_entry_set_rdev(entry, st->st_rdev);
+                       if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) == 0 ||
+                                       (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
+                               archive_entry_set_gid(entry, st->st_gid);
+                       if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) == 0 ||
+                                       (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
+                               archive_entry_set_uid(entry, st->st_uid);
+                       if ((parsed_kws & MTREE_HAS_MTIME) == 0 ||
+                                       (parsed_kws & MTREE_HAS_NOCHANGE) != 0) {
 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
-                       archive_entry_set_mtime(entry, st->st_mtime,
-                           st->st_mtimespec.tv_nsec);
+                               archive_entry_set_mtime(entry, st->st_mtime,
+                                               st->st_mtimespec.tv_nsec);
 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
-                       archive_entry_set_mtime(entry, st->st_mtime,
-                           st->st_mtim.tv_nsec);
+                               archive_entry_set_mtime(entry, st->st_mtime,
+                                               st->st_mtim.tv_nsec);
 #elif HAVE_STRUCT_STAT_ST_MTIME_N
-                       archive_entry_set_mtime(entry, st->st_mtime,
-                           st->st_mtime_n);
+                               archive_entry_set_mtime(entry, st->st_mtime,
+                                               st->st_mtime_n);
 #elif HAVE_STRUCT_STAT_ST_UMTIME
-                       archive_entry_set_mtime(entry, st->st_mtime,
-                           st->st_umtime*1000);
+                               archive_entry_set_mtime(entry, st->st_mtime,
+                                               st->st_umtime*1000);
 #elif HAVE_STRUCT_STAT_ST_MTIME_USEC
-                       archive_entry_set_mtime(entry, st->st_mtime,
-                           st->st_mtime_usec*1000);
+                               archive_entry_set_mtime(entry, st->st_mtime,
+                                               st->st_mtime_usec*1000);
 #else
-                       archive_entry_set_mtime(entry, st->st_mtime, 0);
+                               archive_entry_set_mtime(entry, st->st_mtime, 0);
 #endif
+                       }
+                       if ((parsed_kws & MTREE_HAS_NLINK) == 0 ||
+                                       (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
+                               archive_entry_set_nlink(entry, st->st_nlink);
+                       if ((parsed_kws & MTREE_HAS_PERM) == 0 ||
+                                       (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
+                               archive_entry_set_perm(entry, st->st_mode);
+                       if ((parsed_kws & MTREE_HAS_SIZE) == 0 ||
+                                       (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
+                               archive_entry_set_size(entry, st->st_size);
+                       archive_entry_set_ino(entry, st->st_ino);
+                       archive_entry_set_dev(entry, st->st_dev);
+
+                       archive_entry_linkify(mtree->resolver, &entry, &sparse_entry);
+               } else if (parsed_kws & MTREE_HAS_OPTIONAL) {
+                       /*
+                        * Couldn't open the entry, stat it or the on-disk type
+                        * didn't match.  If this entry is optional, just ignore it
+                        * and read the next header entry.
+                        */
+                       *use_next = 1;
+                       return ARCHIVE_OK;
                }
-               if ((parsed_kws & MTREE_HAS_NLINK) == 0 ||
-                   (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
-                       archive_entry_set_nlink(entry, st->st_nlink);
-               if ((parsed_kws & MTREE_HAS_PERM) == 0 ||
-                   (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
-                       archive_entry_set_perm(entry, st->st_mode);
-               if ((parsed_kws & MTREE_HAS_SIZE) == 0 ||
-                   (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
-                       archive_entry_set_size(entry, st->st_size);
-               archive_entry_set_ino(entry, st->st_ino);
-               archive_entry_set_dev(entry, st->st_dev);
-
-               archive_entry_linkify(mtree->resolver, &entry, &sparse_entry);
-       } else if (parsed_kws & MTREE_HAS_OPTIONAL) {
-               /*
-                * Couldn't open the entry, stat it or the on-disk type
-                * didn't match.  If this entry is optional, just ignore it
-                * and read the next header entry.
-                */
-               *use_next = 1;
-               return ARCHIVE_OK;
        }
 
        mtree->cur_size = archive_entry_size(entry);

diff --git a/libarchive/test/test_read_format_mtree.c b/libarchive/test/test_read_format_mtree.c
index 830fa0a9cf40dd906378bc3cf5ef164f1baaf2ad..f96529daf7a9c2292e83a5bf6ab051ccfb7bc1fa 100644 (file)
--- a/libarchive/test/test_read_format_mtree.c
+++ b/libarchive/test/test_read_format_mtree.c
@@ -57,6 +57,8 @@ test_read_format_mtree1(void)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_filename(a, reffile, 11));
 
@@ -208,6 +210,8 @@ test_read_format_mtree2(void)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_memory(a, archive, sizeof(archive)));
        assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
@@ -245,6 +249,8 @@ test_read_format_mtree3(void)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_memory(a, archive, sizeof(archive)));
        assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
@@ -298,6 +304,8 @@ DEFINE_TEST(test_read_format_mtree_filenames_only)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_memory(a, archive, sizeof(archive)));
        assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
@@ -364,6 +372,8 @@ DEFINE_TEST(test_read_format_mtree_nochange)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_memory(a, archive, sizeof(archive)));
        assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
@@ -401,6 +411,8 @@ DEFINE_TEST(test_read_format_mtree_nochange)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_memory(a, archive2, sizeof(archive2)));
        assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae));
@@ -448,6 +460,8 @@ DEFINE_TEST(test_read_format_mtree_nomagic_v1_form)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_filename(a, reffile, 11));
 
@@ -551,6 +565,8 @@ DEFINE_TEST(test_read_format_mtree_nomagic_v2_form)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_filename(a, reffile, 11));
 
@@ -616,6 +632,8 @@ DEFINE_TEST(test_read_format_mtree_nomagic_v2_netbsd_form)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_filename(a, reffile, 11));
 
@@ -679,6 +697,8 @@ DEFINE_TEST(test_read_format_mtree_nonexistent_contents_file)
            archive_read_support_filter_all(a));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_support_format_all(a));
+       assertEqualIntA(a, ARCHIVE_OK,
+           archive_read_set_options(a, "mtree:checkfs"));
        assertEqualIntA(a, ARCHIVE_OK,
            archive_read_open_memory(a, archive, sizeof(archive)));
        assertEqualIntA(a, ARCHIVE_WARN, archive_read_next_header(a, &ae));

diff --git a/tar/write.c b/tar/write.c
index 40d2fb0a9f52cf63d9e0363458799611a7b99a40..7e8cb13f8ce1f8cd404621c1634e81645ceaf5ce 100644 (file)
--- a/tar/write.c
+++ b/tar/write.c
@@ -648,6 +648,7 @@ append_archive_filename(struct bsdtar *bsdtar, struct archive *a,
        archive_read_support_format_all(ina);
        archive_read_support_filter_all(ina);
        set_reader_options(bsdtar, a);
+       archive_read_set_options(ina, "mtree:checkfs");
        if (archive_read_open_filename(ina, filename,
                                        bsdtar->bytes_per_block)) {
                lafe_warnc(0, "%s", archive_error_string(ina));