Fix a bug where the second saved block was reached by skipping over the size of a gfs2 superblock, which is not always the correct offset. Instead, skip over the length specified by the saved superblock's descriptor.
Signed-off-by: Andrew Price <[email protected]> --- gfs2/edit/savemeta.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/gfs2/edit/savemeta.c b/gfs2/edit/savemeta.c index e20b405d..63cb91af 100644 --- a/gfs2/edit/savemeta.c +++ b/gfs2/edit/savemeta.c @@ -1198,6 +1198,7 @@ static void complain(const char *complaint) static int restore_init(const char *path, struct metafd *mfd, struct savemeta_header *smh, int printonly) { struct gfs2_meta_header *sbmh; + uint16_t sb_siglen; char *end; char *bp; int ret; @@ -1229,6 +1230,7 @@ static int restore_init(const char *path, struct metafd *mfd, struct savemeta_he /* Scan for the position of the superblock. Required to support old formats(?). */ end = &restore_buf[256 + sizeof(struct saved_metablock) + sizeof(*sbmh)]; while (bp <= end) { + sb_siglen = be16_to_cpu(((struct saved_metablock *)bp)->siglen); sbmh = (struct gfs2_meta_header *)(bp + sizeof(struct saved_metablock)); if (sbmh->mh_magic == cpu_to_be32(GFS2_MAGIC) && sbmh->mh_type == cpu_to_be32(GFS2_METATYPE_SB)) @@ -1243,7 +1245,7 @@ static int restore_init(const char *path, struct metafd *mfd, struct savemeta_he if (ret != 0) return ret; - bp += sizeof(struct saved_metablock) + sizeof(sbd.sd_sb); + bp += sizeof(struct saved_metablock) + sb_siglen; restore_off = bp - restore_buf; restore_left -= restore_off; return 0; -- 2.24.1
