Re: git: d6ec20a30919 - stable/14 - geom_map(4): Garbage collect disconnected driver

From: Marius Strobl <marius_at_freebsd.org>
Date: Sat, 20 Jan 2024 22:57:30 UTC
On Thu, Jan 18, 2024 at 09:23:05PM +0000, Brooks Davis wrote:
> On Thu, Jan 18, 2024 at 08:15:07PM +0000, Marius Strobl wrote:
> > The branch stable/14 has been updated by marius:
> > 
> > URL: https://cgit.FreeBSD.org/src/commit/?id=d6ec20a3091924793576ba3fd71f42d4fcd3371f
> > 
> > commit d6ec20a3091924793576ba3fd71f42d4fcd3371f
> > Author:     Marius Strobl <marius@FreeBSD.org>
> > AuthorDate: 2024-01-14 08:31:26 +0000
> > Commit:     Marius Strobl <marius@FreeBSD.org>
> > CommitDate: 2024-01-18 20:14:46 +0000
> > 
> >     geom_map(4): Garbage collect disconnected driver
> >     
> >     The last MIPS user has been removed in c09981f1 2 years ago, the last
> >     ARM one in 58d5c511 even 5.5 years ago.
> 
> Unless this and geom_redboot were getting in the way of MFCing other
> work, it seems a bit odd to MFC the removal of functionality.
 
Yes, the motivation here is to simplify merges for code that no
longer can be tested in either main or stable. Recently, I wasted
a considerable amount of time trying to figure out how on earth I'm
supposed to even just compile-test the code in sys/arm/ti when only
a fraction of it is hooked up to the build until I found 3416e102
(which removes most but not all glue for sys/arm/ti from the kernel
config files).

Apart from that, unlike as with say le(4) (which also can be used by
emulated and virtualized x86 guests) or sys/mips, I don't see which
net "functionality" is removed in these two cases that wasn't already
gone when sys/mips has been removed prior to branching stable/14.

Marius