Re: [PATCH 03/20] ctf: give shared kernel types a synthetic module owner

[Kris Van Hees <kris.van.hees@oracle.com>]

On Fri, Nov 06, 2026 at 12:40:35PM +0000, Nix wrote:
> On 30 Oct 2026, Kris Van Hees stated:
> 
> > Store the shared dictionary in shared_ctf->dm_ctfp and remove dt_shared_ctf
> > from the DTrace handle.
> 
> This is effectively reversing a change I made ten years ago :P because
> having it as a separate module was too problematic given that it didn't
> correspond to a *real* module any more (in particular, we don't want
> people to be able to look it up by name with `, etc.)
> 
> As far as I could tell, all the advantages of being a module were
> actually disadvantages for the shared dict, and no advantages were left.
> It's not actually a module, and none of the things you can do to modules
> should be doable on the shared CTF, and it didn't make the code simpler
> to conflate the two either.
> 
> So... what benefits does this bring?

It makes it cleaner to be able to e.g. share the module data between multiple
handles within the same process (which we will be doing with the PCP support
Alan contributed).  It also avoids needing to store the shared_ctf data in the
handle itself, because it is truly associated with kernel module data.

	Kris