bcachefs undump

Add our own version of 'qemu-img convert', which doesn't have the l1
table size limit or require fixing our missing reflink table.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

8 files changed, 148 insertions, 6 deletions

diff --git a/c_src/bcachefs.c b/c_src/bcachefs.c
index 58fa2549..d91e0475 100644
--- a/c_src/bcachefs.c
+++ b/c_src/bcachefs.c
@@ -96,6 +96,7 @@ void bcachefs_usage(void)
 	     "Debug:\n"
 	     "These commands work on offline, unmounted filesystems\n"
 	     "  dump                     Dump filesystem metadata to a qcow2 image\n"
+	     "  undump                   Convert qcow2 metadata dumps to sparse raw files\n"
 	     "  list                     List filesystem metadata in textual form\n"
 	     "  list_journal             List contents of journal\n"
 	     "\n"
diff --git a/c_src/cmd_dump.c b/c_src/cmd_dump.c
index 5f354272..93e7d743 100644
--- a/c_src/cmd_dump.c
+++ b/c_src/cmd_dump.c
@@ -386,3 +386,85 @@ int cmd_dump(int argc, char *argv[])
 	darray_exit(&dev_names);
 	return ret;
 }
+
+static void undump_usage(void)
+{
+	puts("bcachefs undump - turn qcow2 images from 'bcachefs dump' back into sparse raw images\n"
+	     "Usage: bcachefs undump [OPTION]... <files>\n"
+	     "\n"
+	     "Options:\n"
+	     "  -f, --force     Force; overwrite when needed\n"
+	     "  -h, --help      Display this help and exit\n"
+	     "Report bugs to <linux-bcachefs@vger.kernel.org>");
+}
+
+struct undump {
+	char	*in, *out;
+	int	infd, outfd;
+};
+
+int cmd_undump(int argc, char *argv[])
+{
+	static const struct option longopts[] = {
+		{ "force",		no_argument,		NULL, 'f' },
+		{ "help",		no_argument,		NULL, 'h' },
+		{ NULL }
+	};
+	bool force = false;
+	int opt;
+
+	while ((opt = getopt_long(argc, argv, "fh",
+				  longopts, NULL)) != -1)
+		switch (opt) {
+		case 'f':
+			force = true;
+			break;
+		case 'h':
+			undump_usage();
+			exit(EXIT_SUCCESS);
+		}
+	args_shift(optind);
+
+	if (!argc) {
+		undump_usage();
+		die("Please supply file(s) to convert");
+	}
+
+	DARRAY(struct undump) files = {};
+
+	for (unsigned i = 0; i < argc; i++) {
+		unsigned len = strlen(argv[i]);
+		const char *suffix = ".qcow2";
+		unsigned suffixlen = strlen(suffix);
+
+		if (len <= suffixlen ||
+		    strcmp(suffix, argv[i] + len - suffixlen)) {
+			die("%s not a qcow2 image?", argv[i]);
+		}
+
+		char *out = strdup(argv[i]);
+		out[len - suffixlen] = '\0';
+
+		if (!force && !access(out, F_OK))
+			die("%s already exists", out);
+
+		darray_push(&files, ((struct undump) {
+			.in	= argv[i],
+			.out	= out,
+			.infd	= xopen(argv[i], O_RDONLY),
+		}));
+	}
+
+	darray_for_each(files, i)
+		i->outfd = xopen(i->out, O_WRONLY|O_CREAT|(!force ? O_EXCL : 0), 0600);
+
+	darray_for_each(files, i) {
+		qcow2_to_raw(i->infd, i->outfd);
+		close(i->infd);
+		close(i->outfd);
+		free(i->out);
+	}
+
+	darray_exit(&files);
+	return 0;
+}
diff --git a/c_src/cmds.h b/c_src/cmds.h
index 30b47286..fa4b87a9 100644
--- a/c_src/cmds.h
+++ b/c_src/cmds.h
@@ -34,6 +34,8 @@ int cmd_fsck(int argc, char *argv[]);
 int cmd_recovery_pass(int argc, char *argv[]);
 
 int cmd_dump(int argc, char *argv[]);
+int cmd_undump(int argc, char *argv[]);
+
 int cmd_list_journal(int argc, char *argv[]);
 int cmd_kill_btree_node(int argc, char *argv[]);
 
diff --git a/c_src/qcow2.c b/c_src/qcow2.c
index 53959a00..7afb6812 100644
--- a/c_src/qcow2.c
+++ b/c_src/qcow2.c
@@ -143,8 +143,7 @@ void qcow2_image_finish(struct qcow2_image *img)
 
 	memset(buf, 0, img->block_size);
 	memcpy(buf, &hdr, sizeof(hdr));
-	xpwrite(img->outfd, buf, img->block_size, 0,
-		"qcow2 header");
+	xpwrite(img->outfd, buf, img->block_size, 0, "qcow2 header");
 
 	free(img->l2_table);
 	free(img->l1_table);
@@ -160,3 +159,51 @@ void qcow2_write_image(int infd, int outfd, ranges *data,
 	qcow2_write_ranges(&img, data);
 	qcow2_image_finish(&img);
 }
+
+void qcow2_to_raw(int infd, int outfd)
+{
+	struct qcow2_hdr hdr;
+
+	xpread(infd, &hdr, sizeof(hdr), 0);
+
+	if (hdr.magic != cpu_to_be32(QCOW_MAGIC))
+		die("not a qcow2 image");
+
+	if (hdr.version != cpu_to_be32(QCOW_VERSION))
+		die("incorrect qcow2 version");
+
+	ftruncate(outfd, be64_to_cpu(hdr.size));
+
+	unsigned block_size = 1U << be32_to_cpu(hdr.block_bits);
+
+	unsigned l1_size = be32_to_cpu(hdr.l1_size);
+	unsigned l2_size = block_size / sizeof(u64);
+
+	__be64 *l1_table	= xcalloc(l1_size, sizeof(u64));
+	__be64 *l2_table	= xmalloc(block_size);
+	void *data_buf		= xmalloc(block_size);
+
+	xpread(infd, l1_table, l1_size * sizeof(u64), be64_to_cpu(hdr.l1_table_offset));
+
+	for (u64 i = 0; i < l1_size; i++) {
+		if (!l1_table[i])
+			continue;
+
+		xpread(infd, l2_table, block_size, be64_to_cpu(l1_table[i]) & ~QCOW_OFLAG_COPIED);
+
+		for (unsigned j = 0; j < l2_size; j++) {
+			u64 src_offset = be64_to_cpu(l2_table[j]) & ~QCOW_OFLAG_COPIED;
+			if (!src_offset)
+				continue;
+
+			u64 dst_offset = (i * l2_size + j) * block_size;
+
+			xpread(infd,	data_buf, block_size, src_offset);
+			xpwrite(outfd,	data_buf, block_size, dst_offset, "qcow2 data");
+		}
+	}
+
+	free(data_buf);
+	free(l2_table);
+	free(l1_table);
+}
diff --git a/c_src/qcow2.h b/c_src/qcow2.h
index c7b35627..7ccb2773 100644
--- a/c_src/qcow2.h
+++ b/c_src/qcow2.h
@@ -4,6 +4,8 @@
 #include <linux/types.h>
 #include "tools-util.h"
 
+#define QCOW2_L1_MAX		(4ULL << 20)
+
 struct qcow2_image {
 	int			infd;
 	int			outfd;
@@ -25,4 +27,6 @@ void qcow2_image_finish(struct qcow2_image *);
 
 void qcow2_write_image(int, int, ranges *, unsigned);
 
+void qcow2_to_raw(int, int);
+
 #endif /* _QCOW2_H */
diff --git a/c_src/tools-util.c b/c_src/tools-util.c
index f48d3f21..7733b9e9 100644
--- a/c_src/tools-util.c
+++ b/c_src/tools-util.c
@@ -52,15 +52,15 @@ char *mprintf(const char *fmt, ...)
 	return str;
 }
 
-void xpread(int fd, void *buf, size_t count, off_t offset)
+void __xpread(int fd, void *buf, size_t count, off_t offset, const char *file, unsigned line)
 {
 	while (count) {
 		ssize_t r = pread(fd, buf, count, offset);
 
 		if (r < 0)
-			die("read error: %m");
+			die("read error: %m at %s:%u", file, line);
 		if (!r)
-			die("pread error: unexpected eof");
+			die("pread error: unexpected eof at %s:%u", file, line);
 		count	-= r;
 		offset	+= r;
 	}
diff --git a/c_src/tools-util.h b/c_src/tools-util.h
index b8104002..988e2d62 100644
--- a/c_src/tools-util.h
+++ b/c_src/tools-util.h
@@ -28,8 +28,13 @@ void die(const char *, ...)
 	__attribute__ ((format (printf, 1, 2))) noreturn;
 char *mprintf(const char *, ...)
 	__attribute__ ((format (printf, 1, 2)));
-void xpread(int, void *, size_t, off_t);
+
+void __xpread(int, void *, size_t, off_t, const char *, unsigned);
+#define xpread(_fd, _buf, _count, _offset)				\
+	__xpread(_fd, _buf, _count, _offset, __FILE__, __LINE__)
+
 void xpwrite(int, const void *, size_t, off_t, const char *);
+
 struct stat xfstatat(int, const char *, int);
 struct stat xfstat(int);
 struct stat xstat(const char *);
diff --git a/src/bcachefs.rs b/src/bcachefs.rs
index 4fb0b14c..d0c1d6e8 100644
--- a/src/bcachefs.rs
+++ b/src/bcachefs.rs
@@ -48,6 +48,7 @@ fn handle_c_command(mut argv: Vec<String>, symlink_cmd: Option<&str>) -> i32 {
             "data" => c::data_cmds(argc, argv),
             "device" => c::device_cmds(argc, argv),
             "dump" => c::cmd_dump(argc, argv),
+            "undump" => c::cmd_undump(argc, argv),
             "format" => c::cmd_format(argc, argv),
             "fs" => c::fs_cmds(argc, argv),
             "fsck" => c::cmd_fsck(argc, argv),