Hi Joshua,
Thank you for the patch.
On mer., août 07, 2024 at 08:13, Joshua Watt <jpewhac...@gmail.com> wrote:
> The backup offset is in bytes, but was incorrectly be interpreted as
> blocks, leading to it being written to the wrong location. Fix the
> calculation and clarify that ANDROID_AB_BACKUP_OFFSET is in bytes and
> must be a multiple of the block size.
Nice finding !
>
> Signed-off-by: Joshua Watt <jpewhac...@gmail.com>
> ---
> boot/android_ab.c | 4 ++--
> common/Kconfig | 3 ++-
> 2 files changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/boot/android_ab.c b/boot/android_ab.c
> index 143f373aae..b253200161 100644
> --- a/boot/android_ab.c
> +++ b/boot/android_ab.c
> @@ -139,8 +139,8 @@ static int ab_control_store(struct blk_desc *dev_desc,
> {
> ulong abc_offset, abc_blocks, ret;
>
Can we add a safeguard in here to make sure that offset
is always blksz aligned ?
We could do something similar to ab_control_create_from_disk():
if (offset % part_info->blksz) {
log_err("ANDROID: offset not block aligned.\n");
return -EINVAL;
}
> - abc_offset = offset +
> - offsetof(struct bootloader_message_ab, slot_suffix) /
> + abc_offset = (offset +
> + offsetof(struct bootloader_message_ab, slot_suffix)) /
> part_info->blksz;
> abc_blocks = DIV_ROUND_UP(sizeof(struct bootloader_control),
> part_info->blksz);
I've also noticed that ab_control_store()'s documentation does not
document the @param[in] offset.
Could we add that as well (either in a seperate patch or in this one) ?
> diff --git a/common/Kconfig b/common/Kconfig
> index 83c81edac2..e1b8557e0c 100644
> --- a/common/Kconfig
> +++ b/common/Kconfig
> @@ -986,7 +986,8 @@ config ANDROID_AB_BACKUP_OFFSET
> help
> If non-zero, a backup bootloader message starting at this offset in
> the partition will tried in the event that the primary one (starting
> - at offset 0) fails its checksum.
> + at offset 0) fails its checksum. The offset is in bytes and must be
> + multiple of the block size.
>
> endmenu
>
> --
> 2.45.2
