FILE_SUPPORTS_SPARSE_FILES not set

[ronnie sahlberg]

On Thu, Aug 14, 2014 at 2:23 PM, Jeremy Allison <jra at samba.org> wrote:
> On Thu, Aug 14, 2014 at 05:19:37PM -0400, Ira Cooper wrote:
>>
>>
>>
>> On Thu, Aug 14, 2014 at 4:37 PM, Jeremy Allison <jra at samba.org> wrote:
>>      On Thu, Aug 14, 2014 at 12:51:35PM -0700, Jeremy Allison wrote:
>>      > On Mon, Aug 11, 2014 at 01:17:12PM -0500, Steve French wrote:
>>      > > Why is darwin the only one that supports setting
>>      FILE_SUPPORTS_SPARSE_FILES
>>      > > in source3/smbd/statvfs.c?
>>      > >
>>      > > Yet Samba 4.1 supports setting fsctl set_sparse (and it seems to
>>      > > work), and of course ext4 and multiple local linux fs support
>>      sparse
>>      > > files.
>>      >
>>      > That's an ommision, from before we implemented the fsp->is_sparse
>>      > flag in the file handle struct.
>>      >
>>      > Log a bug and I'll update it for Linux/BSD.
>>
>>      I think this is what you want. Can I get a second
>>      Team reviewer for master ?
>>
>>
>> Yes, they do... but can they report back the sparse ranges?  Which is part of
>> the sparse support?
>
> I think some filesystems have ioctls that
> can do that.

I don't think there are ioctl()s available that can tell you if the
filesystem supports punching holes.
My recollection from when I added sparse file support to TGTD (to map
to the SCSI UNMAP command) was that consensus was basically "try
punching a hole. If it works then the filesystem supports it. If it
didn't work then either the filesystem does not support it or you did
something wrong."


>
>> Or are we stuck giving the bogus answer we give today....
>
> Hmmm. Is it so bad we tell the client we
> can do sparse ? We *can* do sparse.. :-).