Re: git: b13110e9f3ca - main - ufs/ffs: detect endian mismatch between machine and filesystem

On Wed, Dec 21, 2022 at 02:36:03AM +0000, Alexey Dokuchaev wrote:
> On Tue, Dec 20, 2022 at 11:29:04AM -0300, Alfredo Dal'Ava J??nior wrote:
> > On Tue, Dec 20, 2022 at 12:51 AM Warner Losh <imp@bsdimp.com> wrote:
> > > On Mon, Dec 19, 2022, 8:45 PM Alexey Dokuchaev <danfe@freebsd.org> wrote:
> > >> On Tue, Dec 20, 2022 at 03:25:15AM +0000, Alfredo Dal'Ava Junior wrote:
> > >> >   Mount on a LE machine a filesystem formatted for BE is not
> > >> >   supported currently.
> > >> Interesting.  Are there future plans to add full support for this?
> > >> I'd be looking forward to it!
> > >>
> > > NetBSD has this code, but differences in the VM systems made it tricky to
> > > port when I tried it back when I was hacking FreeBSD/mips in the pre BERI
> > > times...
> > 
> > We actually considered it when we added BE support to mpr and mrsas drivers
> > (focused on powerpc64), but there were other priorities. At this moment I'm
> > not aware that anyone is working on it....
> > @danfe, is it something useful for you?
> 
> Not to extent to push someone do the work, but it would be nice to simply
> unplug the HDD from Mac mini G4 and connect it directly to my i386 desktop.
> I'm using a WD 1.0TB SATA drive through the converter which started to fail
> on me recently (i.e. the computer won't boot off of it), so until I manage
> to fix this somehow, I can't access my data.

IIRC sson@ did a bi-endian port of UFS about 10 years ago.

It turns out this isn't want we needed so we never tried to upstream
the work (we needed the ability to make BE filesystems for CHERI-MIPS
systems from amd64 without giving everyone root so makefs was more
useful to us).

I have no idea where the code is today.

-- Brooks