Re: fcntl() F_GETLK semantics vs. test t_vnops.c

[Tom Spindler <dogcow%babymeat.com@localhost>]

On Fri, Jul 22, 2011 at 11:13:48AM +0100, Alexander Nasonov wrote:
> J. Hannken-Illjes wrote:
> > The test fcntl_getlock_pids() from fs/vfs/t_vnops.c assumes
> > fcntl(fd, F_GETLK, &lock) returns the blocking lock with the
> > lowest start offset.
> > 
> > Our documentation and POSIX.1 document it returning the
> > "first lock that blocks" but doesn't call for any specific order.
> 
> I think it's an ambiguity in POSIX. I wrote the test in assumption
> that "first" means a lock with the lowest start offset but my
> intention was to interate over *all* locks in any order. I think
> it's still possible with the current behaviour but with a less
> straightforward implementation: rather then moving linearly through
> a file, you will have to build a tree.

If you care sufficiently much, you may want to mail austin-group-l@
opengroup.org, as they're the current posix manglers.