Add support for gnu pax sparse file format 0.0

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

3 files changed, 140 insertions, 13 deletions

diff --git a/lib/tar/read_header.c b/lib/tar/read_header.c
index 21642d9..23c23c4 100644
--- a/lib/tar/read_header.c
+++ b/lib/tar/read_header.c
@@ -19,8 +19,20 @@ enum {
 	PAX_ATIME = 0x080,
 	PAX_MTIME = 0x100,
 	PAX_CTIME = 0x200,
+	PAX_SPARSE_SIZE = 0x400,
 };
 
+static void free_sparse_list(sparse_map_t *sparse)
+{
+	sparse_map_t *old;
+
+	while (sparse != NULL) {
+		old = sparse;
+		sparse = sparse->next;
+		free(old);
+	}
+}
+
 static int read_octal(const char *str, int digits, uint64_t *out)
 {
 	uint64_t result = 0;
@@ -154,8 +166,9 @@ static int pax_read_decimal(const char *str, uint64_t *out)
 static int read_pax_header(int fd, uint64_t entsize, unsigned int *set_by_pax,
 			   tar_header_decoded_t *out)
 {
+	sparse_map_t *sparse_last = NULL, *sparse;
+	uint64_t field, offset = 0, num_bytes = 0;
 	char *buffer, *line;
-	uint64_t field;
 	ssize_t ret;
 	uint64_t i;
 
@@ -241,6 +254,25 @@ static int read_pax_header(int fd, uint64_t entsize, unsigned int *set_by_pax,
 				out->sb.st_ctime = field;
 			}
 			*set_by_pax |= PAX_CTIME;
+		} else if (!strncmp(line, "GNU.sparse.size=", 16)) {
+			pax_read_decimal(line + 16, &out->actual_size);
+			*set_by_pax |= PAX_SPARSE_SIZE;
+		} else if (!strncmp(line, "GNU.sparse.offset=", 18)) {
+			pax_read_decimal(line + 18, &offset);
+		} else if (!strncmp(line, "GNU.sparse.numbytes=", 20)) {
+			pax_read_decimal(line + 20, &num_bytes);
+			sparse = calloc(1, sizeof(*sparse));
+			if (sparse == NULL)
+				goto fail_errno;
+			sparse->offset = offset;
+			sparse->count = num_bytes;
+			if (sparse_last == NULL) {
+				free_sparse_list(out->sparse);
+				out->sparse = sparse_last = sparse;
+			} else {
+				sparse_last->next = sparse;
+				sparse_last = sparse;
+			}
 		}
 	}
 
@@ -449,17 +481,6 @@ fail:
 	return NULL;
 }
 
-static void free_sparse_list(sparse_map_t *sparse)
-{
-	sparse_map_t *old;
-
-	while (sparse != NULL) {
-		old = sparse;
-		sparse = sparse->next;
-		free(old);
-	}
-}
-
 static sparse_map_t *read_gnu_old_sparse(int fd, tar_header_t *hdr)
 {
 	sparse_map_t *list = NULL, *end = NULL, *node;
diff --git a/tests/Makemodule.am b/tests/Makemodule.am
index 004f83c..f9232dc 100644
--- a/tests/Makemodule.am
+++ b/tests/Makemodule.am
@@ -50,16 +50,22 @@ 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
 
+test_tar_sparse_gnu1_SOURCES = tests/tar_sparse_gnu1.c
+test_tar_sparse_gnu1_LDADD = libtar.a libutil.a
+test_tar_sparse_gnu1_CPPFLAGS = $(AM_CPPFLAGS)
+test_tar_sparse_gnu1_CPPFLAGS += -DTESTPATH=$(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 test_tar_sparse_gnu
+check_PROGRAMS += test_tar_sparse_gnu1
 
 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 test_tar_sparse_gnu
+TESTS += test_tar_gnu test_tar_sparse_gnu test_tar_sparse_gnu1
 
 EXTRA_DIST += $(top_srcdir)/tests/tar
diff --git a/tests/tar_sparse_gnu1.c b/tests/tar_sparse_gnu1.c
new file mode 100644
index 0000000..fac23f1
--- /dev/null
+++ b/tests/tar_sparse_gnu1.c
@@ -0,0 +1,100 @@
+/* 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_header_decoded_t hdr;
+	sparse_map_t *sparse;
+	int fd;
+
+	assert(chdir(TEST_PATH) == 0);
+
+	fd = open_read("sparse-files/pax-gnu0-0.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(hdr.actual_size == 2097152);
+	assert(hdr.record_size == 32768);
+	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;
+}