[RFC PATCH 01/10] GPT: Add function to get gpt header and partition entries

[AKASHI Takahiro]

On Wed, Dec 08, 2021 at 01:10:51PM +0530, Sughosh Ganu wrote:
> hi Patrick,
> 
> On Tue, 7 Dec 2021 at 21:05, Patrick DELAUNAY <patrick.delaunay at foss.st.com>
> wrote:
> 
> > Hi Sugosh
> >
> > On 11/25/21 8:01 AM, Sughosh Ganu wrote:
> >
> > Add function to get the gpt header and partition entries filled. These
> >
> > would be used subsequently for multi bank firmware update support on
> >
> > devices where the images reside on GPT partitions.
> >
> >
> >
> > Signed-off-by: Sughosh Ganu <sughosh.ganu at linaro.org> <sughosh.ganu at linaro.org>
> >
> > ---
> >
> >  disk/part_efi.c | 10 ++++++++++
> >
> >  include/part.h  | 14 ++++++++++++++
> >
> >  2 files changed, 24 insertions(+)
> >
> >
> >
> > diff --git a/disk/part_efi.c b/disk/part_efi.c
> >
> > index 0ca7effc32..792b9218a9 100644
> >
> > --- a/disk/part_efi.c
> >
> > +++ b/disk/part_efi.c
> >
> > @@ -216,6 +216,16 @@ int get_disk_guid(struct blk_desc * dev_desc, char *guid)
> >
> >   return 0;
> >
> >  }
> >
> >  +int get_gpt_hdr_parts(struct blk_desc *desc, gpt_header *gpt_head,
> >
> > +              gpt_entry **gpt_pte)
> >
> > +{
> >
> > + /* This function validates and fills in the GPT header and PTE's */
> >
> > + if (find_valid_gpt(desc, gpt_head, gpt_pte) != 1)
> >
> > +        return -EINVAL;
> >
> > +
> >
> > + return 0;
> >
> > +}
> >
> > +
> >
> >  void part_print_efi(struct blk_desc *dev_desc)
> >
> >  {
> >
> >   ALLOC_CACHE_ALIGN_BUFFER_PAD(gpt_header, gpt_head, 1, dev_desc->blksz);
> >
> > diff --git a/include/part.h b/include/part.h
> >
> > index b66b07a1f0..8e86485b97 100644
> >
> > --- a/include/part.h
> >
> > +++ b/include/part.h
> >
> > @@ -345,6 +345,20 @@ struct part_driver {
> >
> >   #if CONFIG_IS_ENABLED(EFI_PARTITION)
> >
> >  /* disk/part_efi.c */
> >
> > +
> >
> > +/**
> >
> > + * get_gpt_hdr_parts() - Get information on the GPT Header and
> >
> > + *                       Partition Table Entries
> >
> > + *
> >
> > + * @param desc - block device descriptor
> >
> > + * @param gpt_h - pointer to GPT header representation
> >
> >
> >
> > gpt_h need to be allocated / DMA alligned by caller
> >
> >
> Will add this comment to the description.
> 
> 
> >
> >
> > + * @param gpt_e - pointer to GPT partition table entries
> >
> >
> >
> > gpt_e is allocated in this fucntion need to freed by caller
> >
> >
>  Will add this comment to the description.
> 
> >
> >
> > + *
> >
> > + * @return - zero on success, otherwise error
> >
> > + */
> >
> > +int get_gpt_hdr_parts(struct blk_desc *desc, gpt_header *gpt_head,
> >
> > +              gpt_entry **gpt_pte);
> >
> > +
> >
> >  /**
> >
> >   * write_gpt_table() - Write the GUID Partition Table to disk
> >
> >   *
> >
> > no need to export the low level function "find_valid_gpt"
> >
> >
> find_valid_gpt is not being exported. It is defined statically in
> part_efi.c. Do i misunderstand your comment?
> 
> 
> >
> >
> > the same services are already provided by part_get_info()  ?
> >
> > ST Restricted
> >
> > 2 information are used in the serie
> >
> >
> >
> > => gpt_head->num_partition_entries
> >
> > => &gpt_pte[i].partition_type_guid
> >
> >
> >
> >
> >
> > type_guid is available in disk_partition,
> >
> > and num_partition_entries is not needed (replaced by MAX_SEARCH_PARTITIONS)
> >
> >
> >
> > something like:
> >
> >
> >
> >         struct disk_partition info;
> >
> >
> >
> >         for (p = 1; p < MAX_SEARCH_PARTITIONS; p++) {
> >
> >                if (part_get_info(desc, p, &info))
> >
> >                        continue;
> >
> >                => info.type_guid
> >
> >         }
> >
> >
> I had actually explored using part_get_info. But the type_guid field is
> actually a char string. While I need the guid value to compare it using
> guidcmp. Which is the reason I introduced get_gpt_hdr_parts.

Think of using uuid_str_to_bin() if this is the only reason.

-Takahiro Akashi


>  -sughosh
> 
> 
> >
> > Regards
> >
> >
> >
> > Patrick
> >
> >