Re: 2026 TODO list!

Piotr Kubaj <[email protected]> Tue, 20 Jan 2026 21:58:45 +0100
Newsgroups	gmane.os.freebsd.devel.ppc
Message-ID	<aW_shWVzKcBbDDHN@talos-powerpc64le>
On 26-01-20 09:21:13, Adrian Chadd wrote:
> On Mon, 19 Jan 2026 at 00:08, Alexey Dokuchaev <[email protected]> wrote:
> >
> > On Mon, Jan 12, 2026 at 07:33:49AM -0800, Adrian Chadd wrote:
> > > hi!
> >
> > Hey Adrian,
> >
> > Glad to see you stirring the PPC swamp! :-)
> >
> > > I'd like to assemble a 2026 TODO list for PPC64 work. Justin, Ivy,
> > > Warner, I and and others have been poking at the sharp edges and
> > > trying to use/abuse various PPC64 hardware and VMs.
> > >
> > > [...]
> > >
> > > So I'd like some brainstorming/braindumping/ideas here if you've got
> > > them, even on older PPC64 hardware if you have it. I'd like to collate
> > > these into the wiki and figure out what we think is achievable in
> > > 2026.
> >
> > As an avid Mac mini G4 user I'm more interested in PPC32 at the moment,
> > but on a larger scale it would be really nice to have bi-endian support
> > in UFS.  We had briefly talked* about it back in 2022 but without much
> > actionable details except for mention of Stacey's old code which sadly
> > was never upstreamed (perhaps we should ping him about it).
> 
> If someone wishes to step up to offer to help justin maintain the 32
> bit PPC bits and make
> a case to srcmgr then please, by all means do! It's also easily
> emulatable in qemu so
> we can at least use that to sanity check 32 bit PPC big-endian support.
> 
> Right now I have my hands full with 802.11 work and cleaning up sharp
> edges on PPC64.
> I don't have energy to sort out PPC32 too :-)
> 
> 
> -adrian
> 

32-bit powerpc has an issue since 14.3-RELEASE with LLVM. LLVM has some
bug that makes it hang when building many packages with -O2 (which is
the default), including ports-mgmt/pkg, but it seems to work fine when
WITH_LLVM_ASSERTIONS is set, so I've been unable to pinpoint it exactly.
If anyone wants to take over 32-bit base maintainership, this issue is
probably the first to tackle.
signature.asc (application/pgp-signature, 870 B)