From 64101a6c24075490c636dd1d2d28057f6514f699 Mon Sep 17 00:00:00 2001
From: Amin Hassani <ahassani@google.com>
Date: Fri, 11 Aug 2017 11:20:02 -0700
Subject: [PATCH] Generate squashfs image file map using unsquashfs

This CL modifies unsquashfs to generate a file map when given the flag
-m or -map. The structure of the file map is a list of per-line of the
following:

path start compressed_block_size_1 ... compressed_block_size_N

where:
- 'path' is the full path to the file.
- 'start' is the start byte address of the file.
- 'compressed_block_size_i' is a the length of the i'th compressed block
in bytes. If the block is uncompressed its 25th LSB is set.

It also first prints the list of all fragments along with their address
and block sizes at the beginning of the aforementioned list. The path
for each fragment is '<fragment_i>' with 'i' be the i'th fragment.

The usage:
unsquashfs -m(ap) <path_to_file_map> <squashfs_image>
---
 squashfs-tools/unsquashfs.c | 66 ++++++++++++++++++++++++++++++++++---
 1 file changed, 61 insertions(+), 5 deletions(-)

diff --git a/squashfs-tools/unsquashfs.c b/squashfs-tools/unsquashfs.c
index 0ac6356..67644b8 100644
--- a/squashfs-tools/unsquashfs.c
+++ b/squashfs-tools/unsquashfs.c
@@ -110,6 +110,10 @@ int no_xattrs = XATTR_DEF;
 regex_t *xattr_exclude_preg = NULL;
 regex_t *xattr_include_preg = NULL;
 
+int lsfilemap = FALSE;
+char* file_map_path = NULL;
+FILE* file_map_file = NULL;
+
 int lookup_type[] = {
 	0,
 	S_IFDIR,
@@ -525,7 +529,6 @@ void dump_cache(struct cache *cache)
 	pthread_mutex_unlock(&cache->mutex);
 }
 
-
 char *modestr(char *str, int mode)
 {
 	int i;
@@ -540,6 +543,15 @@ char *modestr(char *str, int mode)
 	return str;
 }
 
+void print_file_map(char* pathname, long long start, int block_count,
+					unsigned int* block_list)
+{
+	fprintf(file_map_file, "%s %lld", pathname, start);
+	int i;
+	for(i = 0; i < block_count; i++)
+		fprintf(file_map_file, " %u", block_list[i]);
+	fprintf(file_map_file, "\n");
+}
 
 #define TOTALCHARS  25
 void print_filename(char *pathname, struct inode *inode)
@@ -552,7 +564,21 @@ void print_filename(char *pathname, struct inode *inode)
 	struct tm *t;
 
 	if(short_ls) {
-		printf("%s\n", pathname);
+		if(!lsfilemap)
+			printf("%s\n", pathname);
+		else if((SQUASHFS_FILE_TYPE == inode->type ||
+				SQUASHFS_LREG_TYPE == inode->type) &&
+				inode->fragment < 0) {
+			unsigned int* block_list =
+				malloc(inode->blocks * sizeof(unsigned int));
+			if(block_list == NULL)
+				EXIT_UNSQUASH("print_filename: unable to malloc block list\n");
+			s_ops->read_block_list(block_list, inode->block_start,
+						inode->block_offset, inode->blocks);
+			print_file_map(pathname, inode->start, inode->blocks,
+							   block_list);
+			free(block_list);
+		}
 		return;
 	}
 
@@ -2149,7 +2175,7 @@ int dir_scan(char *parent_name, unsigned int start_block, unsigned int offset,
 	if(inumber_lookup(i->inode_number))
 		EXIT_UNSQUASH("File System corrupted: directory loop detected\n");
 
-	if((lsonly || info) && (!concise || dir->dir_count ==0))
+	if(!lsfilemap && (lsonly || info) && (!concise || dir->dir_count ==0))
 		print_filename(parent_name, i);
 
 	if(!lsonly) {
@@ -2210,6 +2236,8 @@ int dir_scan(char *parent_name, unsigned int start_block, unsigned int offset,
 			res = asprintf(&pathname, "%s/%s", parent_name, name);
 			if(res == -1)
 				MEM_ERROR();
+			if(lsfilemap || lsonly || info)
+				print_filename(pathname, i);
 
 			if(type == SQUASHFS_DIR_TYPE) {
 				res = dir_scan(pathname, start_block, offset,
@@ -4024,6 +4052,7 @@ static void print_options(FILE *stream, char *name)
 	fprintf(stream, "\t-exc[f] <exclude file>\tsynonym for -exclude-file\n");
 	fprintf(stream, "\t-L\t\t\tsynonym for -follow-symlinks\n");
 	fprintf(stream, "\t-pseudo-file <file>\talternative name for -pf\n");
+	fprintf(stream, "\t-m[ap] <file>\tcreates file map, but doesn't unsquash\n");
 	fprintf(stream, "\nDecompressors available:\n");
 	display_compressors(stream, "", "");
 
@@ -4425,7 +4454,17 @@ int parse_options(int argc, char *argv[])
 		} else if(strcmp(argv[i], "-full-precision") == 0 ||
 				strcmp(argv[i], "-full") == 0)
 			full_precision = TRUE;
-		else {
+		else if(strcmp(argv[i], "-map") == 0 ||
+				strcmp(argv[i], "-m") == 0) {
+			if(++i == argc) {
+				ERROR("%s: -map missing filename\n",
+					argv[0]);
+				exit(1);
+			}
+			file_map_path = argv[i];
+			lsonly = TRUE;
+			lsfilemap = TRUE;
+		} else {
 			print_options(stderr, argv[0]);
 			exit(1);
 		}
@@ -4622,10 +4661,24 @@ int main(int argc, char *argv[])
 			printf("%u inodes (%lld blocks) to write\n\n",
 				total_inodes, total_blocks);
 		}
-
 		enable_progress_bar();
 	}
 
+	if(lsfilemap) {
+		int i, size;
+		long long start;
+		char tmp[22];
+		file_map_file = fopen(file_map_path, "w");
+		if(file_map_file == NULL)
+			EXIT_UNSQUASH("Failed to open file map \"%s\" because %s\n",
+				file_map_path, strerror(errno));
+		for(i = 0; i < sBlk.s.fragments; i++) {
+			s_ops->read_fragment(i, &start, &size);
+			snprintf(tmp, 22, "<fragment-%d>", i);
+			print_file_map(tmp, start, 1, (unsigned int*)&size);
+		}
+	}
+
 	res = dir_scan(dest, SQUASHFS_INODE_BLK(sBlk.s.root_inode),
 		SQUASHFS_INODE_OFFSET(sBlk.s.root_inode), extracts, excludes, 1);
 	if(res == FALSE && set_exit_code)
@@ -4638,6 +4691,9 @@ int main(int argc, char *argv[])
 			exit_code = 2;
 	}
 
+	if(lsfilemap)
+		fclose(file_map_file);
+
 	disable_progress_bar();
 
 	if(!quiet) {
-- 
2.43.0.472.g3155946c3a-goog