Re: nevermind Re: allocating a spare from kinfo_proc.

* Garance A Drosihn <drosih_at_rpi.edu> [040711 22:23] wrote:
> At 9:50 PM -0700 7/11/04, Alfred Perlstein wrote:
> >Sorry, got this done now.
> >
> >tested on sparc64, alpha and i386.
> >
> >* Alfred Perlstein <alfred_at_freebsd.org> [040711 21:15] wrote:
> >> I would like to allocate a void * "user pointer" from struct
> >> kinfo_proc.
> >>
> > > This is for convience of programs that want to retrieve a
> > > kinfo_proc but also compute some information and hang it
> > > from the kinfo proc.
> 
> Fwiw, /bin/ps does this by creating it's own struct which has a
> 'struct kinfo_proc' as one variable inside of it.  However, this
> means that `ps' has to copy all the kinfo_proc's that the kernel
> returns into an array of these other structs.  Depending on what
> the program is doing with those kinfo_proc's, this new variable
> would probably make more sense.

Yeah, maybe ps can be fixed now. :)

> > > Are there any guidelines for adjusting this struct that I
> > > should be aware of?
> 
> It would be nice to make sure that alignment issues are not wasting
> space on *any* of the platforms.  When I made my recent changes to
> user.h, I had a program which printed out all the variables and
> their alignments.  If I have time this week I will update that
> program and see what it says on the different hardware platforms.

I think we're ok since sizeof(long) == sizeof(void *) on all the
platforms I could test.  And the compiler is likely to align
a new pointer in that location in the same spot as the array.

At least it would make sense if it did. :)

-- 
- Alfred Perlstein
- Research Engineering Development Inc.
- email: bright_at_mu.org cell: 408-480-4684