Jason Thorpe wrote:
>> Accoring to truncate(2) ftruncate should not return
>> ENOSPC, but if looking at FFS implementation,
> 
> Well, if it can't allocate the space, what else would you expect it  to do?

I think the OP was responding (reasonably) to the wording of our man page,
which says

  truncate() succeeds unless:

  [list of errnos not including ENOSPC]

which is a clear, albeit incorrect, guarantee that nothing else can go
wrong.

SUS, by contrast, is worded:

  The ftruncate() function shall fail if:

  [list of errnos not including ENOSPC]

which does not preclude other things that could go wrong.
The two phrasings do not mean the same thing. We should probably
fix ours.

-Chap