Add support for reading old style GNU sparse tar file format

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

4 files changed, 266 insertions, 2 deletions

diff --git a/include/tar.h b/include/tar.h
index 4da04f5..6fdb22d 100644
--- a/include/tar.h
+++ b/include/tar.h
@@ -53,9 +53,25 @@ typedef struct {
 } tar_header_t;
 
 typedef struct {
+	struct {
+		char offset[12];
+		char numbytes[12];
+	} sparse[21];
+	char isextended;
+	char padding[7];
+} gnu_sparse_t;
+
+typedef struct tar_sparse_data_t {
+	struct tar_sparse_data_t *next;
+	uint64_t offset;
+	uint64_t count;
+} tar_sparse_data_t;
+
+typedef struct {
 	struct stat sb;
 	char *name;
 	char *link_target;
+	tar_sparse_data_t *sparse;
 	bool unknown_record;
 } tar_header_decoded_t;
 
@@ -69,6 +85,7 @@ typedef struct {
 
 #define TAR_TYPE_GNU_SLINK 'K'
 #define TAR_TYPE_GNU_PATH 'L'
+#define TAR_TYPE_GNU_SPARSE 'S'
 
 #define TAR_TYPE_PAX 'x'
 
diff --git a/lib/tar/read_header.c b/lib/tar/read_header.c
index 76634e1..44a38f9 100644
--- a/lib/tar/read_header.c
+++ b/lib/tar/read_header.c
@@ -390,6 +390,7 @@ static int decode_header(const tar_header_t *hdr, unsigned int set_by_pax,
 	switch (hdr->typeflag) {
 	case '\0':
 	case TAR_TYPE_FILE:
+	case TAR_TYPE_GNU_SPARSE:
 		out->sb.st_mode |= S_IFREG;
 		break;
 	case TAR_TYPE_LINK:
@@ -450,6 +451,104 @@ fail:
 	return NULL;
 }
 
+static void free_sparse_list(tar_sparse_data_t *sparse)
+{
+	tar_sparse_data_t *old;
+
+	while (sparse != NULL) {
+		old = sparse;
+		sparse = sparse->next;
+		free(old);
+	}
+}
+
+static tar_sparse_data_t *read_gnu_old_sparse(int fd, tar_header_t *hdr)
+{
+	tar_sparse_data_t *list = NULL, *end = NULL, *node;
+	gnu_sparse_t sph;
+	uint64_t off, sz;
+	ssize_t ret;
+	int i;
+
+	for (i = 0; i < 4; ++i) {
+		if (!isdigit(hdr->tail.gnu.sparse[i].offset[0]))
+			break;
+		if (!isdigit(hdr->tail.gnu.sparse[i].numbytes[0]))
+			break;
+
+		if (read_octal(hdr->tail.gnu.sparse[i].offset,
+			       sizeof(hdr->tail.gnu.sparse[i].offset), &off))
+			goto fail;
+		if (read_octal(hdr->tail.gnu.sparse[i].numbytes,
+			       sizeof(hdr->tail.gnu.sparse[i].numbytes), &sz))
+			goto fail;
+
+		node = calloc(1, sizeof(*node));
+		if (node == NULL)
+			goto fail_errno;
+
+		node->offset = off;
+		node->count = sz;
+
+		if (list == NULL) {
+			list = end = node;
+		} else {
+			end->next = node;
+			end = node;
+		}
+	}
+
+	if (hdr->tail.gnu.isextended == 0)
+		return list;
+
+	do {
+		ret = read_retry(fd, &sph, sizeof(sph));
+		if (ret < 0)
+			goto fail_errno;
+		if ((size_t)ret < sizeof(sph))
+			goto fail_eof;
+
+		for (i = 0; i < 21; ++i) {
+			if (!isdigit(sph.sparse[i].offset[0]))
+				break;
+			if (!isdigit(sph.sparse[i].numbytes[0]))
+				break;
+
+			if (read_octal(sph.sparse[i].offset,
+				       sizeof(sph.sparse[i].offset), &off))
+				goto fail;
+			if (read_octal(sph.sparse[i].numbytes,
+				       sizeof(sph.sparse[i].numbytes), &sz))
+				goto fail;
+
+			node = calloc(1, sizeof(*node));
+			if (node == NULL)
+				goto fail_errno;
+
+			node->offset = off;
+			node->count = sz;
+
+			if (list == NULL) {
+				list = end = node;
+			} else {
+				end->next = node;
+				end = node;
+			}
+		}
+	} while (sph.isextended != 0);
+
+	return list;
+fail_eof:
+	fputs("parsing GNU sparse header: unexpected end of file", stderr);
+	goto fail;
+fail_errno:
+	perror("parsing GNU sparse header");
+	goto fail;
+fail:
+	free_sparse_list(list);
+	return NULL;
+}
+
 int read_header(int fd, tar_header_decoded_t *out)
 {
 	unsigned int set_by_pax = 0;
@@ -513,6 +612,22 @@ int read_header(int fd, tar_header_decoded_t *out)
 			if (read_pax_header(fd, pax_size, &set_by_pax, out))
 				goto fail;
 			continue;
+		case TAR_TYPE_GNU_SPARSE:
+			if (!(set_by_pax & PAX_SIZE)) {
+				if (read_number(hdr.tail.gnu.realsize,
+						sizeof(hdr.tail.gnu.realsize),
+						&pax_size))
+					goto fail;
+
+				out->sb.st_size = pax_size;
+				set_by_pax |= PAX_SIZE;
+			}
+
+			free_sparse_list(out->sparse);
+			out->sparse = read_gnu_old_sparse(fd, &hdr);
+			if (out->sparse == NULL)
+				goto fail;
+			break;
 		}
 		break;
 	}
@@ -543,6 +658,7 @@ fail:
 
 void clear_header(tar_header_decoded_t *hdr)
 {
+	free_sparse_list(hdr->sparse);
 	free(hdr->name);
 	free(hdr->link_target);
 	memset(hdr, 0, sizeof(*hdr));
diff --git a/tests/Makemodule.am b/tests/Makemodule.am
index e9cb8c4..004f83c 100644
--- a/tests/Makemodule.am
+++ b/tests/Makemodule.am
@@ -46,16 +46,20 @@ test_tar_ustar_SOURCES = tests/tar_ustar.c
 test_tar_ustar_LDADD = libtar.a libutil.a
 test_tar_ustar_CPPFLAGS = $(AM_CPPFLAGS) -DTESTPATH=$(top_srcdir)/tests/tar
 
+test_tar_sparse_gnu_SOURCES = tests/tar_sparse_gnu.c
+test_tar_sparse_gnu_LDADD = libtar.a libutil.a
+test_tar_sparse_gnu_CPPFLAGS = $(AM_CPPFLAGS) -DTESTPATH=$(top_srcdir)/tests/tar
+
 check_PROGRAMS += test_canonicalize_name test_mknode_simple test_mknode_slink
 check_PROGRAMS += test_mknode_reg test_mknode_dir test_gen_inode_table
 check_PROGRAMS += test_add_by_path test_get_path test_fstree_sort
 check_PROGRAMS += test_fstree_from_file test_fstree_init test_fstree_xattr
-check_PROGRAMS += test_tar_ustar test_tar_pax test_tar_gnu
+check_PROGRAMS += test_tar_ustar test_tar_pax test_tar_gnu test_tar_sparse_gnu
 
 TESTS += test_canonicalize_name test_mknode_simple test_mknode_slink
 TESTS += test_mknode_reg test_mknode_dir test_gen_inode_table
 TESTS += test_add_by_path test_get_path test_fstree_sort test_fstree_from_file
 TESTS += test_fstree_init test_fstree_xattr test_tar_ustar test_tar_pax
-TESTS += test_tar_gnu
+TESTS += test_tar_gnu test_tar_sparse_gnu
 
 EXTRA_DIST += $(top_srcdir)/tests/tar
diff --git a/tests/tar_sparse_gnu.c b/tests/tar_sparse_gnu.c
new file mode 100644
index 0000000..2f2e1c5
--- /dev/null
+++ b/tests/tar_sparse_gnu.c
@@ -0,0 +1,127 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "util.h"
+#include "tar.h"
+
+#include <unistd.h>
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include <stdio.h>
+
+#define STR(x) #x
+#define STRVALUE(x) STR(x)
+
+#define TEST_PATH STRVALUE(TESTPATH)
+
+static int open_read(const char *path)
+{
+	int fd = open(path, O_RDONLY);
+
+	if (fd < 0) {
+		perror(path);
+		exit(EXIT_FAILURE);
+	}
+
+	return fd;
+}
+
+int main(void)
+{
+	tar_sparse_data_t *sparse;
+	tar_header_decoded_t hdr;
+	int fd;
+
+	assert(chdir(TEST_PATH) == 0);
+
+	fd = open_read("sparse-files/gnu-small.tar");
+	assert(read_header(fd, &hdr) == 0);
+	assert(hdr.sb.st_mode == (S_IFREG | 0644));
+	assert(hdr.sb.st_uid == 01750);
+	assert(hdr.sb.st_gid == 01750);
+	assert(hdr.sb.st_size == 524288);
+	assert(strcmp(hdr.name, "input.bin") == 0);
+	assert(!hdr.unknown_record);
+
+	sparse = hdr.sparse;
+	assert(sparse != NULL);
+	assert(sparse->offset == 0);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 262144);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 524288);
+	assert(sparse->count == 0);
+
+	assert(sparse->next == NULL);
+
+	clear_header(&hdr);
+	close(fd);
+
+	fd = open_read("sparse-files/gnu.tar");
+	assert(read_header(fd, &hdr) == 0);
+	assert(hdr.sb.st_mode == (S_IFREG | 0644));
+	assert(hdr.sb.st_uid == 01750);
+	assert(hdr.sb.st_gid == 01750);
+	assert(hdr.sb.st_size == 2097152);
+	assert(strcmp(hdr.name, "input.bin") == 0);
+	assert(!hdr.unknown_record);
+
+	sparse = hdr.sparse;
+	assert(sparse != NULL);
+	assert(sparse->offset == 0);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 262144);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 524288);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 786432);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 1048576);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 1310720);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 1572864);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 1835008);
+	assert(sparse->count == 4096);
+
+	sparse = sparse->next;
+	assert(sparse != NULL);
+	assert(sparse->offset == 2097152);
+	assert(sparse->count == 0);
+
+	sparse = sparse->next;
+	assert(sparse == NULL);
+
+	clear_header(&hdr);
+	close(fd);
+
+	return EXIT_SUCCESS;
+}