[v2,09/12] block/dmg: use SectorNumber from BLKX header

Previously the sector table parsing relied on the previous offset of
the DMG file. Now it uses the sector number from the BLKX header
(see http://newosxbook.com/DMG.html).

The implementation of dmg2img (from vu1tur) does not base the output
sector on the location of the terminator (0xffffffff) either so it
should be safe to drop this dependency on the previous state.

(It makes somehow makes sense, a terminator should halt further
processing of a block and is perhaps used to preallocate some space.)

Signed-off-by: Peter Wu <peter@lekensteyn.nl>
---
 v2: initial patch after suggestions from John Snow to read and use
     these fields.
---
 block/dmg.c | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

On 01/06/2015 12:48 PM, Peter Wu wrote:
> Previously the sector table parsing relied on the previous offset of
> the DMG file. Now it uses the sector number from the BLKX header
> (see http://newosxbook.com/DMG.html).
>
> The implementation of dmg2img (from vu1tur) does not base the output
> sector on the location of the terminator (0xffffffff) either so it
> should be safe to drop this dependency on the previous state.
>
> (It makes somehow makes sense, a terminator should halt further
> processing of a block and is perhaps used to preallocate some space.)
>
> Signed-off-by: Peter Wu <peter@lekensteyn.nl>
> ---
>   v2: initial patch after suggestions from John Snow to read and use
>       these fields.
> ---
>   block/dmg.c | 12 +++++-------
>   1 file changed, 5 insertions(+), 7 deletions(-)
>
> diff --git a/block/dmg.c b/block/dmg.c
> index 130efac..57922c5 100644
> --- a/block/dmg.c
> +++ b/block/dmg.c
> @@ -189,7 +189,6 @@ typedef struct DmgHeaderState {
>       /* used internally by dmg_read_mish_block to remember offsets of blocks
>        * across calls */
>       uint64_t data_fork_offset;
> -    uint64_t last_out_offset;
>       /* exported for dmg_open */
>       uint32_t max_compressed_size;
>       uint32_t max_sectors_per_chunk;
> @@ -205,6 +204,7 @@ static int dmg_read_mish_block(BDRVDMGState *s, DmgHeaderState *ds,
>       int64_t offset = 0;
>       uint64_t data_offset;
>       uint64_t in_offset = ds->data_fork_offset;
> +    uint64_t out_offset;
>
>       type = buff_read_uint32(buffer, offset);
>       /* skip data that is not a valid MISH block (invalid magic or too small) */
> @@ -213,6 +213,9 @@ static int dmg_read_mish_block(BDRVDMGState *s, DmgHeaderState *ds,
>           return 0;
>       }
>
> +    /* chunk offsets are relative to this sector number */
> +    out_offset = buff_read_uint64(buffer, offset + 8);
> +
>       /* location in data fork for (compressed) blob (in bytes) */
>       data_offset = buff_read_uint64(buffer, offset + 0x18);
>       in_offset += data_offset;
> @@ -233,10 +236,6 @@ static int dmg_read_mish_block(BDRVDMGState *s, DmgHeaderState *ds,
>           offset += 4;
>           if (s->types[i] != 0x80000005 && s->types[i] != 1 &&
>               s->types[i] != 2) {
> -            if (s->types[i] == 0xffffffff && i > 0) {
> -                ds->last_out_offset = s->sectors[i - 1] +
> -                                      s->sectorcounts[i - 1];
> -            }

This is exactly what I was hoping to see could be eliminated by reading 
the mish headers. Thanks for this.

>               chunk_count--;
>               i--;
>               offset += 36;
> @@ -245,7 +244,7 @@ static int dmg_read_mish_block(BDRVDMGState *s, DmgHeaderState *ds,
>           offset += 4;
>
>           s->sectors[i] = buff_read_uint64(buffer, offset);
> -        s->sectors[i] += ds->last_out_offset;
> +        s->sectors[i] += out_offset;
>           offset += 8;
>
>           s->sectorcounts[i] = buff_read_uint64(buffer, offset);
> @@ -419,7 +418,6 @@ static int dmg_open(BlockDriverState *bs, QDict *options, int flags,
>       s->offsets = s->lengths = s->sectors = s->sectorcounts = NULL;
>       /* used by dmg_read_mish_block to keep track of the current I/O position */
>       ds.data_fork_offset = 0;
> -    ds.last_out_offset = 0;
>       ds.max_compressed_size = 1;
>       ds.max_sectors_per_chunk = 1;
>
>

Much nicer.

Reviewed-by: John Snow <jsnow@redhat.com>