Add utility to turn a POSIX/PAX tar archive into a squashfs image

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

6 files changed, 722 insertions, 1 deletions

diff --git a/.gitignore b/.gitignore
index 532b328..a0fc326 100644
--- a/.gitignore
+++ b/.gitignore
@@ -21,3 +21,4 @@ config.h
 gensquashfs
 rdsquashfs
 sqfs2tar
+tar2sqfs
diff --git a/tar/Makemodule.am b/tar/Makemodule.am
index d192380..6c9edc7 100644
--- a/tar/Makemodule.am
+++ b/tar/Makemodule.am
@@ -2,4 +2,8 @@ sqfs2tar_SOURCES = tar/sqfs2tar.c tar/tar.h tar/write_header.c
 sqfs2tar_LDADD = libsquashfs.a libfstree.a libcompress.a libutil.a
 sqfs2tar_LDADD += $(XZ_LIBS) $(ZLIB_LIBS) $(LZO_LIBS) $(LZ4_LIBS) $(ZSTD_LIBS)
 
-bin_PROGRAMS += sqfs2tar
+tar2sqfs_SOURCES = tar/tar2sqfs.c tar/skip.c tar/read_header.c tar/tar.h
+tar2sqfs_LDADD = libsquashfs.a libfstree.a libcompress.a libutil.a
+tar2sqfs_LDADD += $(XZ_LIBS) $(ZLIB_LIBS) $(LZO_LIBS) $(LZ4_LIBS) $(ZSTD_LIBS)
+
+bin_PROGRAMS += sqfs2tar tar2sqfs
diff --git a/tar/read_header.c b/tar/read_header.c
new file mode 100644
index 0000000..099a9ce
--- /dev/null
+++ b/tar/read_header.c
@@ -0,0 +1,379 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "util.h"
+#include "tar.h"
+
+#include <string.h>
+#include <stdlib.h>
+#include <ctype.h>
+#include <stdio.h>
+
+enum {
+	PAX_SIZE = 0x01,
+	PAX_UID = 0x02,
+	PAX_GID = 0x04,
+	PAX_DEV_MAJ = 0x08,
+	PAX_DEV_MIN = 0x10,
+	PAX_NAME = 0x20,
+	PAX_SLINK_TARGET = 0x40,
+};
+
+static int read_octal(const char *str, int digits, uint64_t *out)
+{
+	uint64_t result = 0;
+
+	while (digits > 0 && *str >= '0' && *str <= '7') {
+		if (result > 0x1FFFFFFFFFFFFFFFUL) {
+			fputs("numeric overflow parsing tar header\n", stderr);
+			return -1;
+		}
+
+		result = (result << 3) | (*(str++) - '0');
+		--digits;
+	}
+
+	*out = result;
+	return 0;
+}
+
+static int read_binary(const char *str, int digits, uint64_t *out)
+{
+	uint64_t x, result = 0;
+
+	while (digits > 0) {
+		if (result > 0x00FFFFFFFFFFFFFFUL) {
+			fputs("numeric overflow parsing tar header\n", stderr);
+			return -1;
+		}
+
+		x = *((const unsigned char *)str++);
+		result = (result << 8) | x;
+		--digits;
+	}
+
+	*out = result;
+	return 0;
+}
+
+static int read_number(const char *str, int digits, uint64_t *out)
+{
+	if (*((unsigned char *)str) & 0x80) {
+		if (read_binary(str, digits, out))
+			return -1;
+		*out &= 0x7FFFFFFFFFFFFFFF;
+		return 0;
+	}
+
+	return read_octal(str, digits, out);
+}
+
+static bool is_zero_block(const tar_header_t *hdr)
+{
+	const unsigned char *ptr = (const unsigned char *)hdr;
+
+	return ptr[0] == '\0' && memcmp(ptr, ptr + 1, sizeof(*hdr) - 1) == 0;
+}
+
+static bool is_checksum_valid(const tar_header_t *hdr)
+{
+	unsigned int chksum = 0;
+	tar_header_t copy;
+	uint64_t ref;
+	size_t i;
+
+	if (read_octal(hdr->chksum, sizeof(hdr->chksum), &ref))
+		return false;
+
+	memcpy(&copy, hdr, sizeof(*hdr));
+	memset(copy.chksum, ' ', sizeof(copy.chksum));
+
+	for (i = 0; i < sizeof(copy); ++i)
+		chksum += ((unsigned char *)&copy)[i];
+
+	return chksum == ref;
+}
+
+static bool is_magic_valid(const tar_header_t *hdr)
+{
+	if (memcmp(hdr->magic, TAR_MAGIC, sizeof(hdr->magic)) != 0)
+		return false;
+
+	if (memcmp(hdr->version, TAR_VERSION, sizeof(hdr->version)) != 0)
+		return false;
+
+	return true;
+}
+
+static int pax_read_decimal(const char *str, uint64_t *out)
+{
+	uint64_t result = 0;
+
+	while (*str >= '0' && *str <= '9') {
+		if (result > 0xFFFFFFFFFFFFFFFFUL / 10) {
+			fputs("numeric overflow parsing pax header\n", stderr);
+			return -1;
+		}
+
+		result = (result * 10) + (*(str++) - '0');
+	}
+
+	*out = result;
+	return 0;
+}
+
+static int read_pax_header(int fd, uint64_t entsize, unsigned int *set_by_pax,
+			   tar_header_decoded_t *out)
+{
+	char *buffer, *line;
+	ssize_t ret;
+	uint64_t i;
+
+	buffer = malloc(entsize + 1);
+	if (buffer == NULL)
+		goto fail_errno;
+
+	ret = read_retry(fd, buffer, entsize);
+	if (ret < 0)
+		goto fail_errno;
+	if ((size_t)ret < entsize)
+		goto fail_eof;
+
+	if (skip_padding(fd, entsize))
+		goto fail;
+
+	buffer[entsize] = '\0';
+
+	for (i = 0; i < entsize; ++i) {
+		while (i < entsize && isspace(buffer[i]))
+			++i;
+		while (i < entsize && isdigit(buffer[i]))
+			++i;
+		while (i < entsize && isspace(buffer[i]))
+			++i;
+		if (i >= entsize)
+			break;
+
+		line = buffer + i;
+
+		while (i < entsize && buffer[i] != '\n')
+			++i;
+
+		buffer[i] = '\0';
+
+		if (!strncmp(line, "uid=", 4)) {
+			pax_read_decimal(line + 4, &out->uid);
+			*set_by_pax |= PAX_UID;
+		} else if (!strncmp(line, "gid=", 4)) {
+			pax_read_decimal(line + 4, &out->gid);
+			*set_by_pax |= PAX_GID;
+		} else if (!strncmp(line, "path=", 5)) {
+			free(out->name);
+			out->name = strdup(line + 5);
+			if (out->name == NULL)
+				goto fail_errno;
+			*set_by_pax |= PAX_NAME;
+		} else if (!strncmp(line, "size=", 5)) {
+			pax_read_decimal(line + 5, &out->size);
+			*set_by_pax |= PAX_SIZE;
+		} else if (!strncmp(line, "linkpath=", 9)) {
+			free(out->link_target);
+			out->link_target = strdup(line + 9);
+			if (out->link_target == NULL)
+				goto fail_errno;
+			*set_by_pax |= PAX_SLINK_TARGET;
+		}
+	}
+
+	free(buffer);
+	return 0;
+fail_errno:
+	perror("reading pax header");
+	goto fail;
+fail_eof:
+	fputs("reading pax header: unexpected end of file\n", stderr);
+	goto fail;
+fail:
+	free(buffer);
+	return -1;
+}
+
+static int decode_header(const tar_header_t *hdr, unsigned int set_by_pax,
+			 tar_header_decoded_t *out)
+{
+	size_t count;
+
+	if (!(set_by_pax & PAX_NAME)) {
+		if (hdr->prefix[0] == '\0') {
+			count = strlen(hdr->name) + 1;
+			count += strlen(hdr->prefix) + 1;
+
+			out->name = malloc(count);
+
+			if (out->name != NULL) {
+				sprintf(out->name, "%s/%s",
+					hdr->prefix, hdr->name);
+			}
+		} else {
+			out->name = strdup(hdr->name);
+		}
+
+		if (out->name == NULL) {
+			perror("decoding filename");
+			return -1;
+		}
+	}
+
+	if (!(set_by_pax & PAX_SIZE)) {
+		if (read_number(hdr->size, sizeof(hdr->size), &out->size))
+			return -1;
+	}
+
+	if (!(set_by_pax & PAX_UID)) {
+		if (read_number(hdr->uid, sizeof(hdr->uid), &out->uid))
+			return -1;
+	}
+
+	if (!(set_by_pax & PAX_GID)) {
+		if (read_number(hdr->gid, sizeof(hdr->gid), &out->gid))
+			return -1;
+	}
+
+	if (!(set_by_pax & PAX_DEV_MAJ)) {
+		if (read_number(hdr->devmajor, sizeof(hdr->devmajor),
+				&out->dev_maj)) {
+			return -1;
+		}
+	}
+
+	if (!(set_by_pax & PAX_DEV_MIN)) {
+		if (read_number(hdr->devminor, sizeof(hdr->devminor),
+				&out->dev_min)) {
+			return -1;
+		}
+	}
+
+	if (read_octal(hdr->mode, sizeof(hdr->mode), &out->mode))
+		return -1;
+
+	out->mode &= 07777;
+
+	if (hdr->typeflag == TAR_TYPE_LINK ||
+	    hdr->typeflag == TAR_TYPE_SLINK) {
+		if (!(set_by_pax & PAX_SLINK_TARGET)) {
+			out->link_target = strdup(hdr->linkname);
+			if (out->link_target == NULL) {
+				perror("decoding symlink target");
+				return -1;
+			}
+		}
+	}
+
+	out->unknown_record = false;
+
+	switch (hdr->typeflag) {
+	case '\0':
+	case TAR_TYPE_FILE:
+		out->mode |= S_IFREG;
+		break;
+	case TAR_TYPE_LINK:
+		/* XXX: hard links are not support yet */
+		out->mode = S_IFLNK | 0777;
+		break;
+	case TAR_TYPE_SLINK:
+		out->mode = S_IFLNK | 0777;
+		break;
+	case TAR_TYPE_CHARDEV:
+		out->mode |= S_IFCHR;
+		break;
+	case TAR_TYPE_BLOCKDEV:
+		out->mode |= S_IFBLK;
+		break;
+	case TAR_TYPE_DIR:
+		out->mode |= S_IFDIR;
+		break;
+	case TAR_TYPE_FIFO:
+		out->mode |= S_IFIFO;
+		break;
+	default:
+		out->unknown_record = true;
+		break;
+	}
+
+	return 0;
+}
+
+int read_header(int fd, tar_header_decoded_t *out)
+{
+	unsigned int set_by_pax = 0;
+	bool prev_was_zero = false;
+	uint64_t pax_size;
+	tar_header_t hdr;
+	int ret;
+
+	memset(out, 0, sizeof(*out));
+
+	for (;;) {
+		ret = read_retry(fd, &hdr, sizeof(hdr));
+		if (ret == 0)
+			goto out_eof;
+		if (ret < 0)
+			goto fail_errno;
+		if (ret < (int)sizeof(hdr))
+			goto fail_eof;
+
+		if (is_zero_block(&hdr)) {
+			if (prev_was_zero)
+				goto out_eof;
+			prev_was_zero = true;
+			continue;
+		}
+
+		prev_was_zero = false;
+
+		if (!is_magic_valid(&hdr))
+			goto fail_magic;
+
+		if (!is_checksum_valid(&hdr))
+			goto fail_chksum;
+
+		if (hdr.typeflag == TAR_TYPE_PAX) {
+			clear_header(out);
+			if (read_number(hdr.size, sizeof(hdr.size), &pax_size))
+				return -1;
+			set_by_pax = 0;
+			if (read_pax_header(fd, pax_size, &set_by_pax, out))
+				return -1;
+			continue;
+		}
+		break;
+	}
+
+	if (decode_header(&hdr, set_by_pax, out))
+		goto fail;
+
+	return 0;
+out_eof:
+	clear_header(out);
+	return 1;
+fail_errno:
+	perror("reading tar header");
+	goto fail;
+fail_eof:
+	fputs("reading tar header: unexpected end of file\n", stderr);
+	goto fail;
+fail_magic:
+	fputs("input is not a ustar tar archive!\n", stderr);
+	goto fail;
+fail_chksum:
+	fputs("invalid tar header checksum!\n", stderr);
+	goto fail;
+fail:
+	clear_header(out);
+	return -1;
+}
+
+void clear_header(tar_header_decoded_t *hdr)
+{
+	free(hdr->name);
+	free(hdr->link_target);
+	memset(hdr, 0, sizeof(*hdr));
+}
diff --git a/tar/skip.c b/tar/skip.c
new file mode 100644
index 0000000..69208b2
--- /dev/null
+++ b/tar/skip.c
@@ -0,0 +1,48 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "util.h"
+#include "tar.h"
+
+#include <stdio.h>
+
+static int skip_bytes(int fd, uint64_t size)
+{
+	unsigned char buffer[1024];
+	ssize_t ret;
+	size_t diff;
+
+	while (size != 0) {
+		diff = sizeof(buffer);
+		if (diff > size)
+			diff = size;
+
+		ret = read_retry(fd, buffer, diff);
+
+		if (ret < 0) {
+			perror("reading tar record data");
+			return -1;
+		}
+
+		if ((size_t)ret < diff) {
+			fputs("unexpected end of file\n", stderr);
+			return -1;
+		}
+
+		size -= diff;
+	}
+
+	return 0;
+}
+
+int skip_padding(int fd, uint64_t size)
+{
+	size_t tail = size % 512;
+
+	return tail ? skip_bytes(fd, 512 - tail) : 0;
+}
+
+int skip_entry(int fd, uint64_t size)
+{
+	size_t tail = size % 512;
+
+	return skip_bytes(fd, tail ? (size + 512 - tail) : size);
+}
diff --git a/tar/tar.h b/tar/tar.h
index 62c612a..11af6ee 100644
--- a/tar/tar.h
+++ b/tar/tar.h
@@ -24,6 +24,18 @@ typedef struct {
 	char padding[12];
 } tar_header_t;
 
+typedef struct {
+	uint64_t size;
+	uint64_t mode;
+	uint64_t uid;
+	uint64_t gid;
+	uint64_t dev_maj;
+	uint64_t dev_min;
+	char *name;
+	char *link_target;
+	bool unknown_record;
+} tar_header_decoded_t;
+
 #define TAR_TYPE_FILE '0'
 #define TAR_TYPE_LINK '1'
 #define TAR_TYPE_SLINK '2'
@@ -44,4 +56,14 @@ typedef struct {
 int write_tar_header(int fd, const fstree_t *fs, const tree_node_t *n,
 		     const char *name);
 
+/* calcuate and skip the zero padding */
+int skip_padding(int fd, uint64_t size);
+
+/* round up to block size and skip the entire entry */
+int skip_entry(int fd, uint64_t size);
+
+int read_header(int fd, tar_header_decoded_t *out);
+
+void clear_header(tar_header_decoded_t *hdr);
+
 #endif /* TAR_H */
diff --git a/tar/tar2sqfs.c b/tar/tar2sqfs.c
new file mode 100644
index 0000000..f00f281
--- /dev/null
+++ b/tar/tar2sqfs.c
@@ -0,0 +1,267 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "data_writer.h"
+#include "highlevel.h"
+#include "squashfs.h"
+#include "compress.h"
+#include "id_table.h"
+#include "fstree.h"
+#include "util.h"
+#include "tar.h"
+
+#include <sys/sysmacros.h>
+#include <stdlib.h>
+#include <getopt.h>
+#include <unistd.h>
+#include <string.h>
+#include <ctype.h>
+#include <stdio.h>
+#include <errno.h>
+#include <fcntl.h>
+
+static struct option long_opts[] = {
+	{ "help", no_argument, NULL, 'h' },
+	{ "version", no_argument, NULL, 'V' },
+};
+
+static const char *short_opts = "hV";
+
+static const char *usagestr =
+"Usage: tar2sqfs [OPTIONS...] <sqfsfile>\n"
+"\n"
+"Read an uncompressed tar archive from stdin and turn it into a squashfs\n"
+"filesystem image.\n"
+"\n"
+"Possible options:\n"
+"\n"
+"  --help, -h                Print help text and exit.\n"
+"  --version, -V             Print version information and exit.\n"
+"\n"
+"Examples:\n"
+"\n"
+"\ttar2sqfs rootfs.sqfs < rootfs.tar\n"
+"\tzcat rootfs.tar.gz | tar2sqfs rootfs.sqfs\n"
+"\txzcat rootfs.tar.xz | tar2sqfs rootfs.sqfs\n"
+"\n";
+
+static const char *filename;
+static int block_size = SQFS_DEFAULT_BLOCK_SIZE;
+static uint32_t def_mtime = 0;
+static uint16_t def_mode = 0755;
+static uint32_t def_uid = 0;
+static uint32_t def_gid = 0;
+static size_t devblksize = SQFS_DEVBLK_SIZE;
+
+static void process_args(int argc, char **argv)
+{
+	int i;
+
+	for (;;) {
+		i = getopt_long(argc, argv, short_opts, long_opts, NULL);
+		if (i == -1)
+			break;
+
+		switch (i) {
+		case 'h':
+			fputs(usagestr, stdout);
+			exit(EXIT_SUCCESS);
+		case 'V':
+			print_version();
+			exit(EXIT_SUCCESS);
+		default:
+			goto fail_arg;
+		}
+	}
+
+	if (optind >= argc) {
+		fputs("Missing argument: squashfs image\n", stderr);
+		goto fail_arg;
+	}
+
+	filename = argv[optind++];
+
+	if (optind < argc) {
+		fputs("Unknown extra arguments\n", stderr);
+		goto fail_arg;
+	}
+	return;
+fail_arg:
+	fputs("Try `tar2sqfs --help' for more information.\n", stderr);
+	exit(EXIT_FAILURE);
+}
+
+static int create_node_and_repack_data(tar_header_decoded_t *hdr, fstree_t *fs,
+				       data_writer_t *data)
+{
+	tree_node_t *node;
+	size_t extra = 0;
+
+	if (S_ISLNK(hdr->mode))
+		extra = strlen(hdr->link_target) + 1;
+
+	if (S_ISREG(hdr->mode)) {
+		node = fstree_add_file(fs, hdr->name, hdr->mode,
+				       hdr->uid, hdr->gid, hdr->size, NULL);
+		if (node == NULL)
+			goto fail_errno;
+
+		if (write_data_from_fd(data, node->data.file,
+				       STDIN_FILENO)) {
+			return -1;
+		}
+
+		if (skip_padding(STDIN_FILENO, node->data.file->size))
+			return -1;
+	} else {
+		node = fstree_add(fs, hdr->name, hdr->mode,
+				  hdr->uid, hdr->gid, extra);
+		if (node == NULL)
+			goto fail_errno;
+
+		if (S_ISLNK(hdr->mode)) {
+			strcpy(node->data.slink_target,
+			       hdr->link_target);
+		} else if (S_ISBLK(hdr->mode) || S_ISCHR(hdr->mode)) {
+			node->data.devno = makedev(hdr->dev_maj, hdr->dev_min);
+		}
+	}
+
+	return 0;
+fail_errno:
+	perror(hdr->name);
+	return -1;
+}
+
+static int process_tar_ball(fstree_t *fs, data_writer_t *data)
+{
+	tar_header_decoded_t hdr;
+	int ret;
+
+	for (;;) {
+		ret = read_header(STDIN_FILENO, &hdr);
+		if (ret > 0)
+			break;
+		if (ret < 0)
+			return -1;
+
+		if (hdr.unknown_record) {
+			fprintf(stderr, "skipping '%s' (unknown entry type)\n",
+				hdr.name);
+			if (skip_entry(STDIN_FILENO, hdr.size))
+				goto fail;
+			continue;
+		}
+
+		if (canonicalize_name(hdr.name)) {
+			fprintf(stderr, "skipping '%s' (invalid name)\n",
+				hdr.name);
+			if (skip_entry(STDIN_FILENO, hdr.size))
+				goto fail;
+			continue;
+		}
+
+		if (create_node_and_repack_data(&hdr, fs, data))
+			goto fail;
+
+		clear_header(&hdr);
+	}
+
+	return 0;
+fail:
+	clear_header(&hdr);
+	return -1;
+}
+
+int main(int argc, char **argv)
+{
+	int outfd, status = EXIT_SUCCESS;
+	E_SQFS_COMPRESSOR comp_id;
+	data_writer_t *data;
+	sqfs_super_t super;
+	compressor_t *cmp;
+	id_table_t idtbl;
+	fstree_t fs;
+	int ret;
+
+	process_args(argc, argv);
+
+	outfd = open(filename, O_CREAT | O_EXCL | O_RDWR, 0644);
+	if (outfd < 0) {
+		perror(filename);
+		return EXIT_FAILURE;
+	}
+
+	if (fstree_init(&fs, block_size, def_mtime, def_mode,
+			def_uid, def_gid)) {
+		goto out_fd;
+	}
+
+	comp_id = compressor_get_default();
+
+	cmp = compressor_create(comp_id, true, block_size, NULL);
+	if (cmp == NULL) {
+		fputs("Error creating compressor\n", stderr);
+		goto out_fs;
+	}
+
+	if (sqfs_super_init(&super, block_size, def_mtime, comp_id))
+		goto out_cmp;
+
+	if (sqfs_super_write(&super, outfd))
+		goto out_cmp;
+
+	ret = cmp->write_options(cmp, outfd);
+	if (ret < 0)
+		goto out_cmp;
+
+	if (ret > 0) {
+		super.flags |= SQFS_FLAG_COMPRESSOR_OPTIONS;
+		super.bytes_used += ret;
+	}
+
+	data = data_writer_create(&super, cmp, outfd);
+	if (data == NULL)
+		goto out_cmp;
+
+	if (id_table_init(&idtbl))
+		goto out_data;
+
+	if (process_tar_ball(&fs, data))
+		goto out;
+
+	if (data_writer_flush_fragments(data))
+		goto out;
+
+	fstree_sort(&fs);
+	if (fstree_gen_inode_table(&fs))
+		goto out;
+
+	super.inode_count = fs.inode_tbl_size - 2;
+
+	if (sqfs_serialize_fstree(outfd, &super, &fs, cmp, &idtbl))
+		goto out;
+
+	if (data_writer_write_fragment_table(data))
+		goto out;
+
+	if (id_table_write(&idtbl, outfd, &super, cmp))
+		goto out;
+
+	if (sqfs_super_write(&super, outfd))
+		goto out;
+
+	if (padd_file(outfd, super.bytes_used, devblksize))
+		goto out;
+
+	status = EXIT_SUCCESS;
+out:
+	id_table_cleanup(&idtbl);
+out_data:
+	data_writer_destroy(data);
+out_cmp:
+	cmp->destroy(cmp);
+out_fs:
+	fstree_cleanup(&fs);
+out_fd:
+	close(outfd);
+	return status;
+}