Re: [PATCH RESEND] Fix and extend raid6check repair

[NeilBrown <neilb@suse.de>]

[-- Attachment #1: Type: text/plain, Size: 1474 bytes --]

On Wed, 15 Aug 2012 14:55:43 +0200 Robert Buchholz
<robert.buchholz@goodpoint.de> wrote:

> Hello,
> 
> this is a resend of my patches dated Jul 20th in the "Find mismatch in 
> data blocks during raid6 repair" thread.
> 
> Currently, the raid6check tool does not compile (due to the xmalloc 
> changes) and my initial set of repair patches are faulty.  Please merge 
> and see my original emails for extended comments
> http://thread.gmane.org/gmane.linux.raid/38922/focus=39454
> http://thread.gmane.org/gmane.linux.raid/38922/focus=39460
> 
> 
> 
> Cheers
> 
> Robert

Hi,
 sorry for the long delay in getting to these.
They mostly look OK.  However:

1/ They way you added 'autorepair' as a new, final positional parameter is
not good.  I understand why you did it (it was easy) but I think we really
need to bite the bullet and introduce proper "getopt" parameter parsing.

2/ your new test scripts (thanks for including them!!) call 
   blockdev --flushbufs $md0
  after
    mdadm -S $md0
  This is pointless. Once $md0 is stopped, there can be no bufs to flush.
  I don't think there is any point in the 'udevadm settle' 'flushbufs' and
  'drop_caches' at the end.

I've applied your patches but if you could send updates to address those
issues, I'd appreciate it.

Also: please send one patch per email, preferably in-line (not an
attachment).  It makes it easier to reply to and also easier to apply.

Thanks,
NeilBrown

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 828 bytes --]