Re: [RFC 0/5] pathconf(3) with _PC_LINK_MAX

[tytso]

On Sun, Dec 06, 2009 at 08:39:58AM +0000, Al Viro wrote:
> 
> Um...  Why do we need that, again?  Note that there is no way whatsoever
> for predicting whether link(2) will fail due to having too many existing
> links before you attempt the call - links can be created or removed between
> stat(2) and link(2).  So any uses of that value are heuristical.
> 
> Can you actually show any use cases of that thing?  Preferably - in existing
> code, but even a theoretical one would be interesting.

I think it's mainly a "if we're going to implement a POSIX interface,
it would be nice if it returned something based on reality instead of
a wild-assed guess".  :-)

The "real life" use case I could think of is that backup programs that
use hard links everywhere would be able to determine ahead of time in
advance when it might need to create a new file instead of using a
hard link, without needing to do the link and getting the EMLINK
error.  I agree that the only way you can know for sure is by actually
trying the link, so it's a pretty feeble use case.

I will note that without a functional, ext3 and ext4 (or ext3
filesystem with dir_nlink file system feature mounted with ext4) file
systems would be indistinguishable.

						- Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/