Re: reliability of COMPAT_* support

[christos%astron.com@localhost (Christos Zoulas)]

In article <B41C475B-1FB3-435C-809A-B461B4C30514%nmsu.edu@localhost>,
Brook Milligan  <brook%nmsu.edu@localhost> wrote:
>I am curious about the state of COMPAT_* support in NetBSD kernels.  In
>the distant past I know I had some binaries that I had to run using that
>support, but generally don't any longer.  Is it still really possible to
>run 0.8 binaries, for example, on current systems?  How is this support
>validated?  Is that part of the test harness?  Can anyone point to
>examples of known success at  running really old binaries?
>
>Thanks for any information that emerges.

While, there are no test suites... I am currently running a lisp
NetBSD-0.8 (?) binary on amd64-current:

[3:50pm] 2021>ls -l lisp
-rwxrwxr-x  1 christos  christos  1038692 Oct 23  1993 lisp*
[3:50pm] 2022>file lisp
lisp: a.out little-endian 32-bit demand paged pure executable not stripped
[3:50pm] 2023>su -m
christos's password:
[3:50pm] 2014#sysctl -w vm.user_va0_disable=0
vm.user_va0_disable: 1 -> 0
[3:50pm] 2015#./lisp
Franz Lisp, Opus 38.92
-> (+ 1 2)
3
-> ^D
Goodbye
[3:50pm] 2016#uname -a
NetBSD quasar.astron.com 7.99.2 NetBSD 7.99.2 (QUASAR) #128: Fri Dec  5 18:25:31 EST 2014  christos%quasar.astron.com@localhost:/usr/src/sys/arch/amd64/compile/QUASAR amd64

I am currently tracking a bug that makes the lisp compiler reset the machine
due to an uninitialized lock, so not all is rozy ;-)

christos