git: e013e36939ac - main - linux(4): Get rid of Linuxulator kernel build options.

[Dmitry Chagin]

On Tue, Jun 22, 2021 at 12:49:02PM +0100, Jessica Clarke wrote:
> On 22 Jun 2021, at 12:45, Dmitry Chagin <dchagin at freebsd.org> wrote:
> > On Tue, Jun 22, 2021 at 12:36:26PM +0100, Jessica Clarke wrote:
> >> On 22 Jun 2021, at 12:01, Dmitry Chagin <dchagin at freebsd.org> wrote:
> >>> On Mon, Jun 21, 2021 at 10:56:38PM +0100, Jessica Clarke wrote:
> >>>> On 21 Jun 2021, at 17:56, Dmitry Chagin <dchagin at FreeBSD.org> wrote:
> >>>>> 
> >>>>> The branch main has been updated by dchagin:
> >>>>> 
> >>>>> URL: https://cgit.FreeBSD.org/src/commit/?id=e013e36939ac87b53195370fb5e29f29c1a4b5c6
> >>>>> 
> >>>>> commit e013e36939ac87b53195370fb5e29f29c1a4b5c6
> >>>>> Author:     Dmitry Chagin <dchagin at FreeBSD.org>
> >>>>> AuthorDate: 2021-06-22 05:32:39 +0000
> >>>>> Commit:     Dmitry Chagin <dchagin at FreeBSD.org>
> >>>>> CommitDate: 2021-06-22 05:32:39 +0000
> >>>>> 
> >>>>>  linux(4): Get rid of Linuxulator kernel build options.
> >>>>> 
> >>>>>  Stop confusing people, retire COMPAT_LINUX and COMPAT_LINUX32 kernel
> >>>>>  build options. Since we have 32 and 64 bit Linux emulators, we can't build both
> >>>>>  emulators together into the kernel. I don't think it matters, Linux emulation
> >>>>>  depends on loadable modules (via rc).
> >>>>> 
> >>>>>  Cut LINPROCFS and LINSYSFS for consistency.
> >>>> 
> >>>> I don’t see why these two should be deleted? They currently build fine,
> >>>> and GNU/kFreeBSD kernels enable them. They might work as modules, but I
> >>>> would worry that too many parts of userland would try and read them
> >>>> before /etc/init.d/kldutils (the init script that loads modules) loads
> >>>> them, so then we’d have to mess around with GRUB configs to preload
> >>>> them. If the options work, please leave them in.
> >>>> 
> >>> 
> >>> both FS modules depend on linux.ko on i386 or linux_common.ko on amd64,
> >>> so it doesn't make sense to have options for them
> >> 
> >> But that still worked, and was even in NOTES so being tested by LINT.
> >> 
> >>>> There’s a separate debate of whether this is the “right” fix for
> >>>> COMPAT_LINUX*; arguably that *should* work and it’s a bug that they
> >>>> don’t, not a feature, even if it’s not of much interest to support…
> >>>> 
> >>>> I’d like to see the second half reverted, please, and believe the first
> >>>> should be too, but I feel less strongly about that.
> >>>> 
> >>> 
> >>> I think that descendants should adapt to upstream, esp since there is no
> >>> KBI or ABI breakage. btw, debian wiki says gnu/kFreeBSD unmaintained
> >>> since 2014. That is the reason to worry about dead project?
> >> 
> >> As the current maintainer this is news to me. I see no such comment on
> >> the port’s wiki page[1].
> >> 
> >> Jess
> >> 
> >> [1] https://wiki.debian.org/Debian_GNU/kFreeBSD
> >> 
> > I read: https://www.debian.org/ports/kfreebsd-gnu/index.en.html
> 
> That doesn’t say unmaintained, that says not officially supported. It’s
> the equivalent of Tier 1 vs not Tier 1 in FreeBSD, and falls into the
> same category as powerpc, ppc64 and riscv64, among other architectures.
> It just means that there are no stable releases, only unstable
> (-CURRENT equivalent).
>

ah, ok. I understood. nice to meet you:)

some years ago i tried to rewrite compat/linux/* with Elf(function),
this is so terrible... the cost of this is much more significant than
kernel options removal.

btw, could you please subscribe to Linux emulation on a phab?
You comments are needful, but its better to get them before commit.