[parted-devel] [PATCH 01/14] Preserve first 446 bytes of the pmbr in gpt.

[Jim Meyering]

Joel Granados Moreno wrote:
> * libparted/label/gpt.c (_write_pmbr) : Make sure we read the first 446
>   bytes of the device when we are creating the pmbr.
> ---
>  libparted/labels/gpt.c |   13 ++++++++++++-
>  1 files changed, 12 insertions(+), 1 deletions(-)
>
> diff --git a/libparted/labels/gpt.c b/libparted/labels/gpt.c
> index 765c8a5..89e46fc 100644
> --- a/libparted/labels/gpt.c
> +++ b/libparted/labels/gpt.c
> @@ -958,7 +958,18 @@ _write_pmbr (PedDevice * dev)
>  {
>  	LegacyMBR_t pmbr;
>
> -	memset(&pmbr, 0, sizeof(pmbr));
> +	/* The UEFI spec is not clear about what to do with the following
> +	 * elements of the Protective MBR (pmbr): BootCode (0-440B),
> +	 * UniqueMBRSignature (440B-444B) and Unknown (444B-446B).
> +	 * With this in mind, we try not to modify these elements.
> +	 */
> +	if(ped_device_read(dev, &pmbr, 0, GPT_PMBR_SECTORS) < GPT_PMBR_SECTORS)
> +		memset(&pmbr, 0, sizeof(pmbr));
> +
> +	/* Make sure we zero out all the legacy partitions.
> +	 * There are 4 PartitionRecords.  */
> +	memset(&(pmbr.PartitionRecord), 0, sizeof(PartitionRecord_t) * 4);

In isolation, I'd say this looks fine.
I really appreciate the accompanying test.

However, this change conflicts directly with bug-fix changes on "next"
that allow GPT support to work with >512-byte sectors.  In that case,
a larger sector cannot be read into a fixed-size "pmbr" buffer.

What do you think about applying it only to next?
The alternative is to apply it here, rebase and
adjust "next" accordingly.

If there are too many changes like this,
we may have to start working on "next" in earnest.