Split generic tar code off to static library

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

4 files changed, 717 insertions, 1 deletions

diff --git a/lib/Makemodule.am b/lib/Makemodule.am
index b2e39b1..a79b9c3 100644
--- a/lib/Makemodule.am
+++ b/lib/Makemodule.am
@@ -7,6 +7,11 @@ libfstree_a_SOURCES += include/fstree.h
 libfstree_a_CFLAGS = $(AM_CFLAGS)
 libfstree_a_CPPFLAGS = $(AM_CPPFLAGS)
 
+libtar_a_SOURCES = lib/tar/read_header.c lib/tar/write_header.c lib/tar/skip.c
+libtar_a_SOURCES += include/tar.h
+libtar_a_CFLAGS = $(AM_CFLAGS)
+libtar_a_CPPFLAGS = $(AM_CPPFLAGS)
+
 libcompress_a_SOURCES = lib/comp/compressor.c lib/comp/internal.h
 libcompress_a_SOURCES += include/compress.h
 libcompress_a_CFLAGS = $(AM_CFLAGS)
@@ -76,4 +81,4 @@ libfstree_a_CFLAGS += $(LIBSELINUX_CFLAGS)
 libfstree_a_CPPFLAGS += -DWITH_SELINUX
 endif
 
-noinst_LIBRARIES += libfstree.a libcompress.a libutil.a libsquashfs.a
+noinst_LIBRARIES += libfstree.a libcompress.a libutil.a libsquashfs.a libtar.a
diff --git a/lib/tar/read_header.c b/lib/tar/read_header.c
new file mode 100644
index 0000000..bdb2d20
--- /dev/null
+++ b/lib/tar/read_header.c
@@ -0,0 +1,415 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "util.h"
+#include "tar.h"
+
+#include <sys/sysmacros.h>
+#include <string.h>
+#include <stdlib.h>
+#include <ctype.h>
+#include <stdio.h>
+
+enum {
+	PAX_SIZE = 0x001,
+	PAX_UID = 0x002,
+	PAX_GID = 0x004,
+	PAX_DEV_MAJ = 0x008,
+	PAX_DEV_MIN = 0x010,
+	PAX_NAME = 0x020,
+	PAX_SLINK_TARGET = 0x040,
+	PAX_ATIME = 0x080,
+	PAX_MTIME = 0x100,
+	PAX_CTIME = 0x200,
+};
+
+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;
+	uint64_t field;
+	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, &field);
+			out->sb.st_uid = field;
+			*set_by_pax |= PAX_UID;
+		} else if (!strncmp(line, "gid=", 4)) {
+			pax_read_decimal(line + 4, &field);
+			out->sb.st_gid = field;
+			*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, &field);
+			out->sb.st_size = field;
+			*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;
+		} else if (!strncmp(line, "atime=", 6)) {
+			pax_read_decimal(line + 6, &field);
+			out->sb.st_atime = field;
+			*set_by_pax |= PAX_ATIME;
+		} else if (!strncmp(line, "mtime=", 6)) {
+			pax_read_decimal(line + 6, &field);
+			out->sb.st_mtime = field;
+			*set_by_pax |= PAX_MTIME;
+		} else if (!strncmp(line, "ctime=", 6)) {
+			pax_read_decimal(line + 6, &field);
+			out->sb.st_ctime = field;
+			*set_by_pax |= PAX_CTIME;
+		}
+	}
+
+	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)
+{
+	uint64_t field;
+	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), &field))
+			return -1;
+		out->sb.st_size = field;
+	}
+
+	if (!(set_by_pax & PAX_UID)) {
+		if (read_number(hdr->uid, sizeof(hdr->uid), &field))
+			return -1;
+		out->sb.st_uid = field;
+	}
+
+	if (!(set_by_pax & PAX_GID)) {
+		if (read_number(hdr->gid, sizeof(hdr->gid), &field))
+			return -1;
+		out->sb.st_gid = field;
+	}
+
+	if (!(set_by_pax & PAX_DEV_MAJ)) {
+		if (read_number(hdr->devmajor, sizeof(hdr->devmajor), &field))
+			return -1;
+
+		out->sb.st_rdev = makedev(field, minor(out->sb.st_rdev));
+	}
+
+	if (!(set_by_pax & PAX_DEV_MIN)) {
+		if (read_number(hdr->devminor, sizeof(hdr->devminor), &field))
+			return -1;
+
+		out->sb.st_rdev = makedev(major(out->sb.st_rdev), field);
+	}
+
+	if (!(set_by_pax & PAX_MTIME)) {
+		if (read_number(hdr->mtime, sizeof(hdr->mtime), &field))
+			return -1;
+		out->sb.st_mtime = field;
+	}
+
+	if (!(set_by_pax & PAX_ATIME))
+		out->sb.st_atime = out->sb.st_mtime;
+
+	if (!(set_by_pax & PAX_CTIME))
+		out->sb.st_ctime = out->sb.st_mtime;
+
+	if (read_octal(hdr->mode, sizeof(hdr->mode), &field))
+		return -1;
+
+	out->sb.st_mode = field & 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->sb.st_mode |= S_IFREG;
+		break;
+	case TAR_TYPE_LINK:
+		/* XXX: hard links are not support yet */
+		out->sb.st_mode = S_IFLNK | 0777;
+		break;
+	case TAR_TYPE_SLINK:
+		out->sb.st_mode = S_IFLNK | 0777;
+		break;
+	case TAR_TYPE_CHARDEV:
+		out->sb.st_mode |= S_IFCHR;
+		break;
+	case TAR_TYPE_BLOCKDEV:
+		out->sb.st_mode |= S_IFBLK;
+		break;
+	case TAR_TYPE_DIR:
+		out->sb.st_mode |= S_IFDIR;
+		break;
+	case TAR_TYPE_FIFO:
+		out->sb.st_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/lib/tar/skip.c b/lib/tar/skip.c
new file mode 100644
index 0000000..69208b2
--- /dev/null
+++ b/lib/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/lib/tar/write_header.c b/lib/tar/write_header.c
new file mode 100644
index 0000000..80db327
--- /dev/null
+++ b/lib/tar/write_header.c
@@ -0,0 +1,248 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "util.h"
+#include "tar.h"
+
+#include <sys/sysmacros.h>
+#include <string.h>
+#include <stdio.h>
+
+static unsigned long pax_hdr_counter = 0;
+static char buffer[4096];
+
+static void write_octal(char *dst, unsigned int value, int digits)
+{
+	char temp[64];
+
+	sprintf(temp, "%0*o ", digits, value);
+	memcpy(dst, temp, strlen(temp));
+}
+
+static int name_to_tar_header(tar_header_t *hdr, const char *path)
+{
+	size_t len = strlen(path);
+	const char *ptr;
+
+	if ((len + 1) <= sizeof(hdr->name)) {
+		memcpy(hdr->name, path, len);
+		return 0;
+	}
+
+	for (ptr = path; ; ++ptr) {
+		ptr = strchr(ptr, '/');
+		if (ptr == NULL)
+			return -1;
+
+		len = ptr - path;
+		if (len >= sizeof(hdr->prefix))
+			continue;
+		if (strlen(ptr + 1) >= sizeof(hdr->name))
+			continue;
+		break;
+	}
+
+	memcpy(hdr->prefix, path, ptr - path);
+	memcpy(hdr->name, ptr + 1, strlen(ptr + 1));
+	return 0;
+}
+
+static void init_header(tar_header_t *hdr, const struct stat *sb,
+			const char *name, const char *slink_target)
+{
+	memset(hdr, 0, sizeof(*hdr));
+
+	name_to_tar_header(hdr, name);
+	memcpy(hdr->magic, TAR_MAGIC, sizeof(hdr->magic));
+	memcpy(hdr->version, TAR_VERSION, sizeof(hdr->version));
+	write_octal(hdr->mode, sb->st_mode & ~S_IFMT, 6);
+	write_octal(hdr->uid, sb->st_uid, 6);
+	write_octal(hdr->gid, sb->st_gid, 6);
+	write_octal(hdr->mtime, sb->st_mtime, 11);
+	write_octal(hdr->size, 0, 11);
+	write_octal(hdr->devmajor, 0, 6);
+	write_octal(hdr->devminor, 0, 6);
+
+	switch (sb->st_mode & S_IFMT) {
+	case S_IFREG:
+		write_octal(hdr->size, sb->st_size & 077777777777L, 11);
+		break;
+	case S_IFLNK:
+		if (sb->st_size < (off_t)sizeof(hdr->linkname))
+			strcpy(hdr->linkname, slink_target);
+		break;
+	case S_IFCHR:
+	case S_IFBLK:
+		write_octal(hdr->devmajor, major(sb->st_rdev), 6);
+		write_octal(hdr->devminor, minor(sb->st_rdev), 6);
+		break;
+	}
+
+	sprintf(hdr->uname, "%u", sb->st_uid);
+	sprintf(hdr->gname, "%u", sb->st_gid);
+}
+
+static void update_checksum(tar_header_t *hdr)
+{
+	unsigned int chksum = 0;
+	size_t i;
+
+	memset(hdr->chksum, ' ', sizeof(hdr->chksum));
+
+	for (i = 0; i < sizeof(*hdr); ++i)
+		chksum += ((unsigned char *)hdr)[i];
+
+	write_octal(hdr->chksum, chksum, 6);
+	hdr->chksum[6] = '\0';
+	hdr->chksum[7] = ' ';
+}
+
+static bool need_pax_header(const struct stat *sb, const char *name)
+{
+	tar_header_t tmp;
+
+	if (sb->st_uid > 0777777 || sb->st_gid > 0777777)
+		return true;
+
+	if (S_ISREG(sb->st_mode) && sb->st_size > 077777777777L)
+		return true;
+
+	if (S_ISLNK(sb->st_mode) && sb->st_size >= (off_t)sizeof(tmp.linkname))
+		return true;
+
+	if (name_to_tar_header(&tmp, name))
+		return true;
+
+	return false;
+}
+
+static char *write_pax_entry(char *dst, const char *key, const char *value)
+{
+	size_t i, len, prefix = 0, oldprefix;
+
+	do {
+		len = prefix + 1 + strlen(key) + 1 + strlen(value) + 1;
+
+		oldprefix = prefix;
+		prefix = 1;
+
+		for (i = len; i >= 10; i /= 10)
+			++prefix;
+	} while (oldprefix != prefix);
+
+	sprintf(dst, "%zu %s=%s\n", len, key, value);
+
+	return dst + len;
+}
+
+static int write_pax_header(int fd, const struct stat *sb, const char *name,
+			    const char *slink_target)
+{
+	char temp[64], *ptr;
+	struct stat fakesb;
+	tar_header_t hdr;
+	ssize_t ret;
+	size_t len;
+
+	memset(buffer, 0, sizeof(buffer));
+	memset(&fakesb, 0, sizeof(fakesb));
+	fakesb.st_mode = S_IFREG | 0644;
+
+	sprintf(temp, "pax%lu", pax_hdr_counter);
+	init_header(&hdr, &fakesb, temp, NULL);
+	hdr.typeflag = TAR_TYPE_PAX;
+
+	sprintf(temp, "%u", sb->st_uid);
+	ptr = buffer;
+	ptr = write_pax_entry(ptr, "uid", temp);
+	ptr = write_pax_entry(ptr, "uname", temp);
+
+	sprintf(temp, "%lu", sb->st_mtime);
+	ptr = write_pax_entry(ptr, "mtime", temp);
+
+	sprintf(temp, "%u", sb->st_gid);
+	ptr = write_pax_entry(ptr, "gid", temp);
+	ptr = write_pax_entry(ptr, "gname", temp);
+
+	ptr = write_pax_entry(ptr, "path", name);
+
+	if (S_ISLNK(sb->st_mode)) {
+		ptr = write_pax_entry(ptr, "linkpath", slink_target);
+	} else if (S_ISREG(sb->st_mode)) {
+		sprintf(temp, "%lu", sb->st_size);
+		ptr = write_pax_entry(ptr, "size", temp);
+	}
+
+	len = strlen(buffer);
+	write_octal(hdr.size, len, 11);
+	update_checksum(&hdr);
+
+	ret = write_retry(fd, &hdr, sizeof(hdr));
+	if (ret < 0)
+		goto fail_wr;
+	if ((size_t)ret < sizeof(hdr))
+		goto fail_trunc;
+
+	ret = write_retry(fd, buffer, len);
+	if (ret < 0)
+		goto fail_wr;
+	if ((size_t)ret < len)
+		goto fail_trunc;
+
+	return padd_file(fd, len, 512);
+fail_wr:
+	perror("writing pax header");
+	return -1;
+fail_trunc:
+	fputs("writing pax header: truncated write\n", stderr);
+	return -1;
+}
+
+int write_tar_header(int fd, const struct stat *sb, const char *name,
+		     const char *slink_target)
+{
+	const char *reason;
+	tar_header_t hdr;
+	ssize_t ret;
+
+	if (need_pax_header(sb, name)) {
+		if (write_pax_header(fd, sb, name, slink_target))
+			return -1;
+
+		sprintf(buffer, "pax%lu_data", pax_hdr_counter++);
+		name = buffer;
+	}
+
+	init_header(&hdr, sb, name, slink_target);
+
+	switch (sb->st_mode & S_IFMT) {
+	case S_IFCHR: hdr.typeflag = TAR_TYPE_CHARDEV; break;
+	case S_IFBLK: hdr.typeflag = TAR_TYPE_BLOCKDEV; break;
+	case S_IFLNK: hdr.typeflag = TAR_TYPE_SLINK; break;
+	case S_IFREG: hdr.typeflag = TAR_TYPE_FILE; break;
+	case S_IFDIR: hdr.typeflag = TAR_TYPE_DIR; break;
+	case S_IFIFO: hdr.typeflag = TAR_TYPE_FIFO; break;
+	case S_IFSOCK:
+		reason = "cannot pack socket";
+		goto out_skip;
+	default:
+		reason = "unknown type";
+		goto out_skip;
+	}
+
+	update_checksum(&hdr);
+
+	ret = write_retry(fd, &hdr, sizeof(hdr));
+
+	if (ret < 0) {
+		perror("writing header record");
+	} else if ((size_t)ret < sizeof(hdr)) {
+		fputs("writing header record: truncated write\n", stderr);
+		ret = -1;
+	} else {
+		ret = 0;
+	}
+
+	return ret;
+out_skip:
+	fprintf(stderr, "WARNING: skipping '%s' (%s)\n", name, reason);
+	return 1;
+}