Re: exofs_file_fsync

[Boaz Harrosh <bharrosh@panasas.com>]

On 05/31/2010 01:27 PM, Christoph Hellwig wrote:
> On Mon, May 31, 2010 at 01:23:09PM +0300, Boaz Harrosh wrote:
>> On 05/31/2010 01:09 PM, Christoph Hellwig wrote:
>>> Various odd things going on here:
>>>
>>>  - no checks for I_DIRTY and friends, so it will always write out data
>>>  - filemap_write_and_wait is superflous, as it's already done by
>>>    the caller
>>>  - write_inode_now is overkill as it also writes out data, better use
>>>    sync_inode in a similar way to generic_file_fsync
>>
>> I'll look into it, but I think what I did here was to effectively
>> "data sync" because I wanted a data sync on close and that was the only
>> vector I already had that's called on close.
> 
> fsync won't get called at close time.  ->release is called on last close
> and ->flush on every close.
> 

OK, I was just looking at that. thanks you saved me some digging.
should I just open-code the generic_file_fsync minus the blocks
thing then?

I'm busy with the truncate stuff, but I'll do this next.
Do you need this ASAP?

Thanks
Boaz