Re: [PATCH 5/5] booty changes for iswmd

[Dan Williams <dan j williams intel com>]

Hans de Goede wrote:
> Hi,
>
> Comments inline.
>
>
> Subject: [PATCH 5/5] booty changes for iswmd
>
> ---
>   booty/util.py |    4 +++-
>   booty/x86.py  |   11 +++++++++--
>   2 files changed, 12 insertions(+), 3 deletions(-)
>
> diff --git a/booty/util.py b/booty/util.py
> index 74ba561..ddc2a51 100644
> --- a/booty/util.py
> +++ b/booty/util.py
> @@ -1,11 +1,13 @@
>   import string
> +from flags import flags
>
>   def getDiskPart(dev, storage):
>       path = storage.devicetree.getDeviceByName(dev).path[5:]
>       cut = len(dev)
>       if (path.startswith('rd/') or path.startswith('ida/') or
>               path.startswith('cciss/') or path.startswith('sx8/') or
> -            path.startswith('mapper/') or path.startswith('mmcblk')):
> +            path.startswith('mapper/') or path.startswith('mmcblk') or
> +            (path.startswith('md') and (flags.cmdline.has_key("iswmd")))):
>           if dev[-2] == 'p':
>               cut = -1
>           elif dev[-3] == 'p':
>
> We may want to support partitionable md in general in the future so I don't
> think the flags.cmdline.has_key("iswmd") check here is needed.

All md arrays are now partitionable by default since 2.6.28 where Tejun 
added block device extended minor support.

>  
> diff --git a/booty/x86.py b/booty/x86.py
> index 84d2625..19ccbc8 100644
> --- a/booty/x86.py
> +++ b/booty/x86.py
> @@ -4,6 +4,7 @@ import string
>   from booty import BootyNoKernelWarning
>   from util import getDiskPart
>   from bootloaderInfo import *
> +from flags import flags
>   import checkbootloader
>   import iutil
>   import rhpl
> @@ -54,6 +55,9 @@ class x86BootloaderInfo(efiBootloaderInfo):
>           if path.startswith("mapper/luks-"):
>               return []
>
> +        if flags.cmdline.has_key("iswmd"):
> +            return [device]
> +
>           if path.startswith('md'):
>               bootable = 0
>               parts = checkbootloader.getRaidDisks(device, self.storage,
>
> Thus hunk seems wrong, as it will make the "return [device]" happen for
> every device if the flag is present, what if I'm doing an install with
> this flag, but not booting from an md isw raid set ?

We also need to consider the case where the user has disabled raid 
support in the bios, but the metadata and the 'iswmd' command line 
parameter still exist.  I would expect that the OS ignores raid arrays 
in that case.  Currently the bios-raid-enabled state can be determined 
by "mdadm --detail-platform -e imsm" returning 'true'.

Regards,
Dan