Code review comment for lp://qastaging/~berolinux/linaro-image-tools/android-iMX53

Hi, here's the answers to your questions:

>> @@ -77,10 +78,6 @@
>>
>>      @classmethod
>>      def populate_boot_script(cls, boot_partition, boot_disk, consoles):
>> -        cmd_runner.run(['mkdir', '-p', boot_disk]).wait()
>> -        cmd_runner.run(['mount', boot_partition, boot_disk],
>> -            as_root=True).wait()
>> -
>>          boot_env = cls._get_boot_env(consoles)
>>          cmdline_filepath = os.path.join(boot_disk, "cmdline")
>>          cmdline_file = open(cmdline_filepath, 'r')
>> @@ -94,10 +91,6 @@
>>          make_boot_script(boot_env, boot_script_path)
>>
>>          cmd_runner.run(['sync']).wait()
>> -        try:
>> -            cmd_runner.run(['umount', boot_disk], as_root=True).wait()
>> -        except cmd_runner.SubcommandNonZeroReturnValue:
>> -            pass
>
> This change is no longer necessary and will actually break l-a-m-c as of
> r388. You might want to use the new partition_mounted() context manager
> here although that can be done later.

Yes, it was necessary in Saturday's build because the other changes
broke l-a-m-c completely. Probably obsoleted by the rebase, will check
if it's still needed w/ current trunk

>>      @classmethod
>>      def get_sfdisk_cmd(cls, should_align_boot_part=False,
>> @@ -158,6 +151,9 @@
>>      def populate_raw_partition(cls, media, boot_dir):
>>          super(AndroidBoardConfig, cls).populate_raw_partition(boot_dir, media)
>>
>> +    @classmethod
>> +    def install_boot_loader(cls, boot_partition, boot_device_or_file):
>> +        pass
>
> Do we need a new classmethod for this?  ISTM that
> populate_raw_partition(), which is already called in l-a-m-c, should be
> the one used to populate a raw partition with the board-specific
> bootloader instead, no?

Probably populate_raw_partition can be adapted to do the right thing instead.

>> --- linaro_image_tools/media_create/partitions.py     2011-06-29 09:16:39 +0000
>> +++ linaro_image_tools/media_create/partitions.py     2011-07-25 00:21:08 +0000
>> @@ -337,8 +337,14 @@
>>          media.path, 1 + board_config.mmc_part_offset)
>>      system_partition = _get_device_file_for_partition_number(
>>          media.path, 2 + board_config.mmc_part_offset)
>> -    cache_partition = _get_device_file_for_partition_number(
>> -        media.path, 3 + board_config.mmc_part_offset)
>> +    if board_config.mmc_part_offset != 1:
>> +        cache_partition = _get_device_file_for_partition_number(
>> +            media.path, 3 + board_config.mmc_part_offset)
>> +    else:
>> +        # In the current setup, partition 4 is always the
>> +        # extended partition container, so we need to skip 4
>> +        cache_partition = _get_device_file_for_partition_number(
>> +            media.path, 5)
>
> I'm slightly confused by this change as the mmc_part_offset of
> Mx53LoCoConfig is still 0, so it won't use the new code (in the else
> block). Maybe this is not related to the rest of the changes in this
> branch?

Actually it is related to the rest of the changes and is vital to avoid a crash.
The patch removes the mmc_part_offset=0 line, and mmc_part_offset is
set to 1 in Mx5Config (and in turn Mx53Config and Mx53LoCoConfig).