Re: review: log_filter.patch 018,081,082 pv#981362

[Timothy Shimmin <tes@sgi.com>]

David Chinner wrote:
> On Thu, May 08, 2008 at 02:17:55PM +1000, tes@sgi.com wrote:
>> Due to the changes for default mkfs options where we now have
>> version 2 inodes, we no longer update the di_onlink field.
>> To be consistent with previous output, we filter these values out.
>> Also, as part of changes to the inode generation number which uses
>> a random num generator, we need to filter out the gen# too.
>>
>> --Tim
>>
>>  018.op.irix             |  400 +++++++++++------------
>>  018.op.linux            |  400 +++++++++++------------
>>  018.trans_inode         |  800 +++++++++++++++++++++++------------------------
>>  081.ugquota.trans_inode |  816 ++++++++++++++++++++++++------------------------
>>  082.op.irix             |  400 +++++++++++------------
>>  082.op.linux            |  400 +++++++++++------------
>>  082.trans_inode         |  800 +++++++++++++++++++++++------------------------
>>  common.log              |    3 
>>  8 files changed, 2011 insertions(+), 2008 deletions(-)
> 
> Worth noting is that these tests all still fail on a config that
> corrects a problem with the superblock features2 field on mount.
> 
> Probably not worth bothering about - but I thought I'd mention it
> as updating the tests didn't fix the failure on my machine. Updating
> xfsprogs now....
> 
> Cheers,
> 
> Dave.

:-)
Yeah, I had the same failure until I updated mkfs.
I had an extra SB_UNIT sb buffer transaction at the start of
the log for the features2 correction.
I didn't think it was worth bothering about for that interim scenario.

--Tim