Add support for unpacking sparse files as sparse files

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

7 files changed, 37 insertions, 6 deletions

diff --git a/include/data_reader.h b/include/data_reader.h
index 68788b4..7de8e1a 100644
--- a/include/data_reader.h
+++ b/include/data_reader.h
@@ -26,8 +26,12 @@ void data_reader_destroy(data_reader_t *data);
   file and its fragment, if it has one. The entire data is dumped to the
   given file descriptor.
 
+  If allow_sparse is true, try to truncate and seek forward on outfd if a
+  zero block is found. If false, always write blocks of zeros to outfd.
+
   Returns 0 on success, prints error messages to stderr on failure.
  */
-int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd);
+int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd,
+			  bool allow_sparse);
 
 #endif /* DATA_READER_H */
diff --git a/lib/sqfs/data_reader.c b/lib/sqfs/data_reader.c
index 421d914..f8c80af 100644
--- a/lib/sqfs/data_reader.c
+++ b/lib/sqfs/data_reader.c
@@ -59,7 +59,8 @@ void data_reader_destroy(data_reader_t *data)
 	free(data);
 }
 
-int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd)
+int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd,
+			  bool allow_sparse)
 {
 	size_t i, count, fragsz;
 	bool compressed;
@@ -82,6 +83,14 @@ int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd)
 			if (bs > data->block_size)
 				goto fail_bs;
 
+			if (bs == 0 && allow_sparse) {
+				if (ftruncate(outfd, i * data->block_size))
+					goto fail_sparse;
+				if (lseek(outfd, 0, SEEK_END) == (off_t)-1)
+					goto fail_sparse;
+				continue;
+			}
+
 			if (bs == 0) {
 				bs = data->block_size;
 				memset(data->buffer, 0, bs);
@@ -123,6 +132,12 @@ int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd)
 	if (fragsz > 0) {
 		if (fi->fragment == 0xFFFFFFFF ||
 		    fi->fragment_offset == 0xFFFFFFFF) {
+			if (allow_sparse) {
+				if (ftruncate(outfd, fragsz))
+					goto fail_sparse;
+				return 0;
+			}
+
 			memset(data->buffer, 0, fragsz);
 		} else {
 			if (data->frag == NULL)
@@ -144,6 +159,9 @@ int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd)
 	}
 
 	return 0;
+fail_sparse:
+	perror("creating sparse output file");
+	return -1;
 fail_seek:
 	perror("seek on squashfs");
 	return -1;
diff --git a/tar/sqfs2tar.c b/tar/sqfs2tar.c
index 1be1b7e..e8b8736 100644
--- a/tar/sqfs2tar.c
+++ b/tar/sqfs2tar.c
@@ -131,7 +131,7 @@ static int write_tree_dfs(fstree_t *fs, tree_node_t *n, data_reader_t *data)
 
 		if (S_ISREG(n->mode)) {
 			if (data_reader_dump_file(data, n->data.file,
-						  STDOUT_FILENO)) {
+						  STDOUT_FILENO, false)) {
 				return -1;
 			}
 
diff --git a/unpack/options.c b/unpack/options.c
index 84c17cf..c42f16a 100644
--- a/unpack/options.c
+++ b/unpack/options.c
@@ -11,6 +11,7 @@ static struct option long_opts[] = {
 	{ "no-fifo", no_argument, NULL, 'F' },
 	{ "no-slink", no_argument, NULL, 'L' },
 	{ "no-empty-dir", no_argument, NULL, 'E' },
+	{ "no-sparse", no_argument, NULL, 'Z' },
 	{ "describe", no_argument, NULL, 'd' },
 	{ "chmod", no_argument, NULL, 'C' },
 	{ "chown", no_argument, NULL, 'O' },
@@ -19,7 +20,7 @@ static struct option long_opts[] = {
 	{ "version", no_argument, NULL, 'V' },
 };
 
-static const char *short_opts = "l:c:u:p:DSFLCOEdqhV";
+static const char *short_opts = "l:c:u:p:DSFLCOEZdqhV";
 
 static const char *help_string =
 "Usage: %s [OPTIONS] <squashfs-file>\n"
@@ -47,6 +48,8 @@ static const char *help_string =
 "  --no-slink, -L            Do not unpack symbolic links.\n"
 "  --no-empty-dir, -E        Do not unpack directories that would end up\n"
 "                            empty after applying the above rules.\n"
+"  --no-sparse, -Z           Do not create sparse files, always write zero\n"
+"                            blocks to disk.\n"
 "  --chmod, -C               Change permission flags of unpacked files to\n"
 "                            those store in the squashfs image.\n"
 "  --chown, -O               Change ownership of unpacked files to the\n"
@@ -118,6 +121,9 @@ void process_command_line(options_t *opt, int argc, char **argv)
 		case 'O':
 			opt->flags |= UNPACK_CHOWN;
 			break;
+		case 'Z':
+			opt->flags |= UNPACK_NO_SPARSE;
+			break;
 		case 'c':
 			opt->op = OP_CAT;
 			opt->cmdpath = get_path(opt->cmdpath, optarg);
diff --git a/unpack/rdsquashfs.c b/unpack/rdsquashfs.c
index fcfc9c9..42bddff 100644
--- a/unpack/rdsquashfs.c
+++ b/unpack/rdsquashfs.c
@@ -100,7 +100,8 @@ int main(int argc, char **argv)
 		if (data == NULL)
 			goto out_fs;
 
-		if (data_reader_dump_file(data, n->data.file, STDOUT_FILENO))
+		if (data_reader_dump_file(data, n->data.file,
+					  STDOUT_FILENO, false))
 			goto out_fs;
 		break;
 	case OP_UNPACK:
diff --git a/unpack/rdsquashfs.h b/unpack/rdsquashfs.h
index 9d1ab4f..0ee48fa 100644
--- a/unpack/rdsquashfs.h
+++ b/unpack/rdsquashfs.h
@@ -24,6 +24,7 @@ enum UNPACK_FLAGS {
 	UNPACK_CHMOD = 0x01,
 	UNPACK_CHOWN = 0x02,
 	UNPACK_QUIET = 0x04,
+	UNPACK_NO_SPARSE = 0x08,
 };
 
 enum {
diff --git a/unpack/restore_fstree.c b/unpack/restore_fstree.c
index 976b15c..ab10f5d 100644
--- a/unpack/restore_fstree.c
+++ b/unpack/restore_fstree.c
@@ -63,7 +63,8 @@ static int create_node(tree_node_t *n, data_reader_t *data, int flags)
 			return -1;
 		}
 
-		if (data_reader_dump_file(data, n->data.file, fd)) {
+		if (data_reader_dump_file(data, n->data.file, fd,
+					  (flags & UNPACK_NO_SPARSE) == 0)) {
 			close(fd);
 			return -1;
 		}