Re: [OpenBIOS] [PATCH 2/10] Ignore extraneous partitions

[Mark Cave-Ayland]

On 09/08/11 22:54, William Hahne wrote:
> BootX will only check the first few partitions from mac-parts to find
> the Mac OS partition. So it must expect drivers and void partitions to
> be ignored. This patch makes it so that if you specify the partition
> cd:1 and partition 1 is a driver it will keep checking partitions until
> reaching the first non-driver and non-void partition.
> Index: packages/mac-parts.c
>
> --- packages/mac-parts.c (revision 1041)
> +++ packages/mac-parts.c (working copy)
> @@ -200,15 +203,26 @@
> } else {
> /* Another partition was explicitly requested */
> SEEK( bs * parnum );
> READ( &par, sizeof(par) );
> /* We have to find the first valid partition corresponding to the
>
> given number.
> This is because Apple sometimes includes a bunch of extra partitions
>
> that
> BootX expects to be ignored */
> while (parnum < par.pmMapBlkCnt) {
> SEEK( bs * parnum );
> READ( &par, sizeof(par) );
> if( (__be32_to_cpu(par.pmPartStatus) & kPartitionAUXIsValid) &&
> (__be32_to_cpu(par.pmPartStatus) & kPartitionAUXIsAllocated) &&
> (__be32_to_cpu(par.pmPartStatus) & kPartitionAUXIsReadable) ) {
> /* Check if valid, also device driver partitions are expected to be
>
> ignored */
> if( (__be32_to_cpu(par.pmPartStatus) & kPartitionAUXIsValid) &&
> (__be32_to_cpu(par.pmPartStatus) & kPartitionAUXIsAllocated) &&
> (__be32_to_cpu(par.pmPartStatus) & kPartitionAUXIsReadable) &&
> !(__be32_to_cpu(par.pmPartStatus) & kPartitionIsRealDeviceDriver)) {
> offs = (long long)__be32_to_cpu(par.pmPyPartStart) * bs;
> size = (long long)__be32_to_cpu(par.pmPartBlkCnt) * bs;
> offs = (long long)__be32_to_cpu(par.pmPyPartStart) * bs;
> size = (long long)__be32_to_cpu(par.pmPartBlkCnt) * bs;
> break;
> }
> parnum++;
>
> }
> }

I remember last time I reworked this, I had to spend a lot of time 
reading the Apple HFS documentation with regard to this. Could you add a 
link to the documentation on the Apple site into the commit message, and 
confirm whether this is something I missed which is documented or it's 
new behaviour that you have discoverd through testing.

ATB,

Mark.

[PATCH 2/10] Ignore extraneous partitions

[William Hahne]

On Wed, Aug 10, 2011 at 5:48 AM, Mark Cave-Ayland <
[e-mail]> wrote:
> On 09/08/11 22:54, William Hahne wrote:
>
> BootX will only check the first few partitions from mac-parts to find
> > the Mac OS partition. So it must expect drivers and void partitions to
> > be ignored. This patch makes it so that if you specify the partition
> > cd:1 and partition 1 is a driver it will keep checking partitions until
> > reaching the first non-driver and non-void partition.
> >
> > Index: packages/mac-parts.c
> > ==============================**==============================**=======
> > --- packages/mac-parts.c (revision 1041)
> > +++ packages/mac-parts.c (working copy)
> > @@ -200,15 +203,26 @@
> > } else {
> > /* Another partition was explicitly requested */
> > SEEK( bs * parnum );
> > READ( &par, sizeof(par) );
> > /* We have to find the first valid partition corresponding to the
> >
> > given number.
> > This is because Apple sometimes includes a bunch of extra partitions
> >
> > that
> > BootX expects to be ignored */
> > while (parnum < par.pmMapBlkCnt) {
> > SEEK( bs * parnum );
> > READ( &par, sizeof(par) );
> > if( (__be32_to_cpu(par.**pmPartStatus) & kPartitionAUXIsValid) &&
> > (__be32_to_cpu(par.**pmPartStatus) & kPartitionAUXIsAllocated) &&
> > (__be32_to_cpu(par.**pmPartStatus) & kPartitionAUXIsReadable) ) {
> > /* Check if valid, also device driver partitions are expected to be
> >
> > ignored */
> > if( (__be32_to_cpu(par.**pmPartStatus) & kPartitionAUXIsValid) &&
> > (__be32_to_cpu(par.**pmPartStatus) & kPartitionAUXIsAllocated) &&
> > (__be32_to_cpu(par.**pmPartStatus) & kPartitionAUXIsReadable) &&
> > !(__be32_to_cpu(par.**pmPartStatus) & kPartitionIsRealDeviceDriver))
> >
> > {
> > offs = (long long)__be32_to_cpu(par.**pmPyPartStart) * bs;
> > size = (long long)__be32_to_cpu(par.**pmPartBlkCnt) * bs;
> > offs = (long long)__be32_to_cpu(par.**pmPyPartStart) * bs;
> > size = (long long)__be32_to_cpu(par.**pmPartBlkCnt) * bs;
> > break;
> > }
> > parnum++;
> >
> > }
> > }
>
> I remember last time I reworked this, I had to spend a lot of time reading
> the Apple HFS documentation with regard to this. Could you add a link to the
> documentation on the Apple site into the commit message, and confirm whether
> this is something I missed which is documented or it's new behaviour that
> you have discoverd through testing.

This is something I discovered through testing and carefully looking through
the BootX source code. I don't know if there is documentation on this or
not.

William Hahne
> ATB,
>
> Mark.
>
> Sirius Labs: http://www.siriusit.co.uk/labs