Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 19 May 2018 15:44:37 -0400
From:      Mark Johnston <markj@FreeBSD.org>
To:        Matthew Macy <mmacy@freebsd.org>
Cc:        Ed Maste <emaste@freebsd.org>, src-committers <src-committers@freebsd.org>, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   Re: svn commit: r333872 - head/cddl/contrib/opensolaris/tools/ctf/cvt
Message-ID:  <20180519194437.GA21485@raichu>
In-Reply-To: <CAPrugNoGudT2djU2F2M%2BjzpBiqHhRzE7fgcga-W6Soz_XBKdVA@mail.gmail.com>
References:  <201805190631.w4J6VHhr094225@repo.freebsd.org> <CAPyFy2CAGvdOKidTha17YzyVUn5ep0bt8fO6Umhsk4zYnwFRZA@mail.gmail.com> <CAPrugNoGudT2djU2F2M%2BjzpBiqHhRzE7fgcga-W6Soz_XBKdVA@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, May 19, 2018 at 11:00:22AM -0700, Matthew Macy wrote:
> On Sat, May 19, 2018 at 4:49 AM, Ed Maste <emaste@freebsd.org> wrote:
> > On 19 May 2018 at 02:31, Matt Macy <mmacy@freebsd.org> wrote:
> >> Author: mmacy
> >> Date: Sat May 19 06:31:17 2018
> >> New Revision: 333872
> >> URL: https://svnweb.freebsd.org/changeset/base/333872
> >>
> >> Log:
> >>   ctfconvert: silence useless enum has too many values warning
> >
> > I agree it's reasonable to silence this warning as it is not
> > actionable for almost everyone who encounters it. It does indicate a
> > real problem in our tool chain though and I added an entry to
> > https://wiki.freebsd.org/DTraceTODO.
> 
> Conrad brought up the need to change the CTF ABI on IRC. This will
> involve an on-disk format change which he and I believe to be
> acceptable, but perhaps interoperability with other operating systems
> is still considered important.

I don't really think it's important. The main consideration is the
toolchain. We use illumos as an upstream, which is pretty inactive at
this point. Joyent's illumos fork has put a lot of work into the CTF
toolchain, and OpenBSD has made some progress towards an ISC-licensed
ctfconvert utility. I'd like to import the latter, since the permissive
license means that we can use it in DDB. It requires more work because
of some missing functionality, though.

At some point I think we'd like to pursue one of these two upstreams,
so it becomes a question of whether they're amenable to modifying the
CTF binary format (and there are some other limitations that ought to be
fixed in the process), and if not, whether it's painful to maintain the
local modifications needed to support large enums.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20180519194437.GA21485>