On Thu, 10 Sep 2026, Ian Jackson wrote:

>Sorry, forgot to CC this to you.  (Hi, btw.  Hope you're doing OK.)

Hi! Fatigued but doing okay, yes, but I’m no longer involved
in d-ports due to… oh my, where do I start… systemd, usrmerge,
nuclear reactors, and “AI”.

But! I can answer some of your questions.

>I suspect that the cause is something to do with erroneous type
>punning between time types, perhaps time_t.  Does x32 have a 64-bit
>time_t ?  What size is its long ?

x32, amd64ilp32, and similar have 32-bit long and 64-bit time_t, yes.
This is not an unusual configuration, MirBSD/i386 has had this since
2004 (sparc joined in 2021), NetBSD did it around 2012, etc.

>I guess we don't know if adns actually functions normally on x32 ?
>I see that it's not in ci.debian.net, for example.  (Since this
>bug was filed, adns has grown autopkgtests.)

I don’t know either, but it’s actually rather easy to test; you need
an amd64 box with “syscall.x32=y” on the kernel command line to enable
the ABI (it’s not fuzz-tested, therefore disabled by default even if
present, to avoid security accidents), then you can debootstrap it
from 'http://deb.debian.org/debian-ports/ unstable main' with
--extra-suites=unreleased into a chroot. It has no kernel packages,
but with dpkg --add-architecture=amd64 you can install one and then
even run it, I did this for years as $dayjob desktop, adding individual
packages from i386/amd64 where the x32 one did not yet work (but for
testing purposes, a chroot should do).

x32 is half on its way out (even the original author lacks interest,
and klibc/x32 is actually amd64), but Gentoo and PLD Linux have working
installations, so it may live longer (I am still fond of the idea).

If you run into problems, #d-ports on OFTC might have people who can
help (though porting work is mostly Adrian), or ask.

Thanks,
//mirabilos
-- 
“It is inappropriate to require that a time represented as
 seconds since the Epoch precisely represent the number of
 seconds between the referenced time and the Epoch.”
        -- IEEE Std 1003.1b-1993 (POSIX) Section B.2.2.2

Reply via email to