Re: rights(4) split? CAP_WRITE -> CAP_WRITE_DATA + CAP_WRITE_CTRL
Kyle Evans <[email protected]> Wed, 3 Jun 2026 10:54:29 -0500
| Newsgroups | gmane.os.freebsd.devel.hackers,gmane.os.freebsd.architechture |
|---|---|
| Message-ID | <[email protected]> |
On 6/3/26 10:34, Adrian Chadd wrote: > On Wed, 3 Jun 2026 at 07:32, Kyle Evans <[email protected]> wrote: >> >> Sure, no objection- that was my first inclination, but I renamed it to COMPAT at the last minute to match the COMPAT_FREEBSD* nomenclature. If there aren't any complaints about the notion within the next week or so, I'll go ahead and hack something up. > > That all sounds sane. Do you have some ideas on where you could use it > in freebsd just so there's some > usage of it in base? > The vast majority of CAP_WRITE usage in lib* and *bin today could probably be scoped down to CAP_WRITE_DATA, but I don't know if they'd really see much benefit from it. caph_limit_stdio() could probably also revoke control messages and that may be more broadly useful- stdout is certainly just supposed to be 'writable', but when I think stdio I don't think of something I'd send fds over in the common case.