[RFC] a corner case of open(2)

From: Al Viro
Date: Tue Apr 26 2016 - 13:55:45 EST


According to POSIX (and behaviour on other Unices) the following
should succeed: open("/tmp", O_CREAT, 0) does not have O_EXCL and the pathname
does refer to existing object, so O_CREAT is ignored and the call is
equivalent to open("/tmp", 0), which succeeds.

We have it rejected with EISDIR. The thing is, the standard behaviour
is actually less messy wrt code, and do_last()/lookup_open()/atomic_open()
badly needs untangling.

Another place where we produce a bogus EISDIR is O_CREAT|O_EXCL on
an existing directory. POSIX (and other Unices) have EEXIST there. In some
cases we produce EEXIST, in some - EISDIR. Uniform EEXIST is actually easier.

It is a change of user-visible behaviour, but I would be very
surprised if anything broke from that change. And it would help to simplify
the awful mess we have in there.

Comments?