Extend filesystem diff utility to work on squashfs images

Since it already works on an fstree_t instance (constructed from the
input paths), and we now have a handy sqfs_reader_t, it is quite simple
to extend.

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

5 files changed, 178 insertions, 0 deletions

diff --git a/.gitignore b/.gitignore
index 2fecd17..83e68ac 100644
--- a/.gitignore
+++ b/.gitignore
@@ -22,6 +22,7 @@ gensquashfs
 rdsquashfs
 sqfs2tar
 tar2sqfs
+sqfsdiff
 test_*
 test-*
 fscompare
diff --git a/difftool/Makemodule.am b/difftool/Makemodule.am
index 4a03495..62c9197 100644
--- a/difftool/Makemodule.am
+++ b/difftool/Makemodule.am
@@ -3,4 +3,11 @@ fscompare_SOURCES += difftool/compare_dir.c difftool/node_compare.c
 fscompare_SOURCES += difftool/compare_file.c
 fscompare_LDADD = libfstree.a libutil.a
 
+sqfsdiff_SOURCES = difftool/sqfsdiff.c difftool/difftool.h difftool/util.c
+sqfsdiff_SOURCES += difftool/compare_dir.c difftool/node_compare.c
+sqfsdiff_SOURCES += difftool/compare_files_sqfs.c
+sqfsdiff_LDADD = libsquashfs.a libfstree.a libcompress.a libutil.a
+sqfsdiff_LDADD += $(XZ_LIBS) $(ZLIB_LIBS) $(LZO_LIBS) $(LZ4_LIBS) $(ZSTD_LIBS)
+
 noinst_PROGRAMS += fscompare
+bin_PROGRAMS += sqfsdiff
diff --git a/difftool/compare_files_sqfs.c b/difftool/compare_files_sqfs.c
new file mode 100644
index 0000000..d8bef83
--- /dev/null
+++ b/difftool/compare_files_sqfs.c
@@ -0,0 +1,45 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * compare_file_sfqs.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "difftool.h"
+
+static unsigned char a_buf[MAX_WINDOW_SIZE];
+static unsigned char b_buf[MAX_WINDOW_SIZE];
+
+int compare_files(file_info_t *a, file_info_t *b, const char *path)
+{
+	uint64_t offset, diff;
+	ssize_t ret;
+
+	if (a->size != b->size)
+		return 1;
+
+	for (offset = 0; offset < a->size; offset += diff) {
+		diff = a->size - offset;
+
+		if (diff > MAX_WINDOW_SIZE)
+			diff = MAX_WINDOW_SIZE;
+
+		ret = data_reader_read(sqfs_a.data, a, offset, a_buf, diff);
+		if (ret < 0 || (size_t)ret < diff) {
+			fprintf(stderr, "Failed to read %s from %s\n",
+				path, first_path);
+			return -1;
+		}
+
+		ret = data_reader_read(sqfs_b.data, b, offset, b_buf, diff);
+		if (ret < 0 || (size_t)ret < diff) {
+			fprintf(stderr, "Failed to read %s from %s\n",
+				path, second_path);
+			return -1;
+		}
+
+		if (memcmp(a_buf, b_buf, diff) != 0)
+			return 1;
+	}
+
+	return 0;
+}
diff --git a/difftool/difftool.h b/difftool/difftool.h
index 0dea6d5..f7739c4 100644
--- a/difftool/difftool.h
+++ b/difftool/difftool.h
@@ -8,6 +8,8 @@
 #define DIFFTOOL_H
 
 #include "config.h"
+
+#include "highlevel.h"
 #include "fstree.h"
 #include "util.h"
 
@@ -24,6 +26,8 @@
 extern const char *first_path;
 extern const char *second_path;
 extern int compare_flags;
+extern sqfs_reader_t sqfs_a;
+extern sqfs_reader_t sqfs_b;
 
 enum {
 	COMPARE_NO_PERM = 0x01,
diff --git a/difftool/sqfsdiff.c b/difftool/sqfsdiff.c
new file mode 100644
index 0000000..2fc10a1
--- /dev/null
+++ b/difftool/sqfsdiff.c
@@ -0,0 +1,121 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * sqfsdiff.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "difftool.h"
+
+static struct option long_opts[] = {
+	{ "no-owner", no_argument, NULL, 'O' },
+	{ "no-permissions", no_argument, NULL, 'P' },
+	{ "help", no_argument, NULL, 'h' },
+	{ "version", no_argument, NULL, 'V' },
+};
+
+static const char *short_opts = "OPhV";
+
+static const char *usagestr =
+"Usage: sqfsdiff [OPTIONS...] <first> <second>\n"
+"\n"
+"Compare the contents of two squashfs images. In contrast to doing a direct\n"
+"diff of the images, this actually recovers the file system trees and\n"
+"recursively compares them against each other.\n"
+"\n"
+"Any differences in packed file layout, ordering, compression, inode\n"
+"allocation and so on is ignored, only the contents are compared.\n"
+"\n"
+"The two images are considered equal if each directory contains the same\n"
+"entries, symlink with the same paths have the same targets, device nodes\n"
+"the same device number and files the same size and contents.\n"
+"\n"
+"A report of any difference is printed to stdout. The exit status is similar\n"
+"that of diff(1): 0 means equal, 1 means different, 2 means problem.\n"
+"\n"
+"Possible options:\n"
+"\n"
+"  --no-owner, -O              Do not compare file owners.\n"
+"  --no-permissions, -P        Do not compare permission bits.\n"
+"\n"
+"  --help, -h                  Print help text and exit.\n"
+"  --version, -V               Print version information and exit.\n"
+"\n";
+
+int compare_flags = 0;
+const char *first_path;
+const char *second_path;
+sqfs_reader_t sqfs_a;
+sqfs_reader_t sqfs_b;
+
+static void process_options(int argc, char **argv)
+{
+	int i;
+
+	for (;;) {
+		i = getopt_long(argc, argv, short_opts, long_opts, NULL);
+		if (i == -1)
+			break;
+
+		switch (i) {
+		case 'O':
+			compare_flags |= COMPARE_NO_OWNER;
+			break;
+		case 'P':
+			compare_flags |= COMPARE_NO_PERM;
+			break;
+		case 'h':
+			fputs(usagestr, stdout);
+			exit(EXIT_SUCCESS);
+		case 'V':
+			print_version();
+			exit(EXIT_SUCCESS);
+		default:
+			goto fail_arg;
+		}
+	}
+
+	if (optind >= argc) {
+		fputs("Missing arguments: first filesystem\n", stderr);
+		goto fail_arg;
+	}
+
+	first_path = argv[optind++];
+
+	if (optind >= argc) {
+		fputs("Missing arguments: second filesystem\n", stderr);
+		goto fail_arg;
+	}
+
+	second_path = argv[optind++];
+
+	if (optind < argc) {
+		fputs("Unknown extra arguments\n", stderr);
+		goto fail_arg;
+	}
+	return;
+fail_arg:
+	fprintf(stderr, "Try `sqfsdiff --help' for more information.\n");
+	exit(EXIT_FAILURE);
+}
+
+int main(int argc, char **argv)
+{
+	int ret = EXIT_FAILURE;
+
+	process_options(argc, argv);
+
+	if (sqfs_reader_open(&sqfs_a, first_path, 0))
+		return EXIT_FAILURE;
+
+	if (sqfs_reader_open(&sqfs_b, second_path, 0))
+		goto out_sqfs_a;
+
+	ret = node_compare(sqfs_a.fs.root, sqfs_b.fs.root);
+	if (ret < 0)
+		ret = 2;
+
+	sqfs_reader_close(&sqfs_b);
+out_sqfs_a:
+	sqfs_reader_close(&sqfs_a);
+	return ret;
+}