[OpenAFS-devel] Aliasing bugs

[Benjamin Kaduk]

On Tue, 12 Jul 2011, Garrett Wollman wrote:

> <<On Tue, 12 Jul 2011 18:31:40 -0400 (EDT), Benjamin Kaduk <kaduk@MIT.EDU> said:
>
>> A few of the errors are more structural in nature, e.g. the "unique
>> identifier" for the vop_advlock routines that always has a 32-bit field
>> but sometimes tries to squash a 64-bit value into it.
>
> I looked at that as well.  That VOP_ADVLOCK argument was apparently a
> pid_t on some system, but now it is multiple different kinds of
> pointer -- it can be a pointer to at least a struct proc or a struct
> file, depending on the kind of lock it is.  (POSIX specifies stupid
> semantics for what happens to locks when a file is closed, which I
> think accounts for some of the weirdness.)
>
> I was going to see about making it an intptr_t, but it looked like a
> bit of a slog so I put off thinking about it any further.

intptr_t lets it compile with -Werror, but is still papering over the root 
issue (which is an ABI-level thing).

-Ben