On Thu, Aug 16, 2012 at 9:54 AM, Kevin Wolf <kw...@redhat.com> wrote: > The footer takes precedence over the header when it exists. It contains > the real grain directory offset that is missing in the header. Without > this patch, streamOptimized images with a footer cannot be read. > > Signed-off-by: Kevin Wolf <kw...@redhat.com> > --- > block/vmdk.c | 16 ++++++++++++++++ > 1 files changed, 16 insertions(+), 0 deletions(-) > > diff --git a/block/vmdk.c b/block/vmdk.c > index 9648398..c243a96 100644 > --- a/block/vmdk.c > +++ b/block/vmdk.c > @@ -35,6 +35,7 @@ > #define VMDK4_FLAG_RGD (1 << 1) > #define VMDK4_FLAG_COMPRESS (1 << 16) > #define VMDK4_FLAG_MARKER (1 << 17) > +#define VMDK4_GD_AT_END 0xffffffffffffffffULL > > typedef struct { > uint32_t version; > @@ -451,6 +452,21 @@ static int vmdk_open_vmdk4(BlockDriverState *bs, > if (header.capacity == 0 && header.desc_offset) { > return vmdk_open_desc_file(bs, flags, header.desc_offset << 9); > } > + > + if (header.gd_offset == VMDK4_GD_AT_END) { > + /* > + * The footer takes precedence over the header, so read it in. The > + * footer starts at offset -1024 from the end: One sector for the > + * footer, and another one for the end-of-stream marker. > + */ > + ret = bdrv_pread(file, > + bs->file->total_sectors * 512 - 1024 + sizeof(magic), > + &header, sizeof(header)); > + if (ret < 0) { > + return ret; > + } > + } > + > l1_entry_sectors = le32_to_cpu(header.num_gtes_per_gte) > * le64_to_cpu(header.granularity); > if (l1_entry_sectors == 0) {
I think we should check the magic number or marker before trusting the contents of the footer. Stefan