[Bug 268479] lib/libc/stdlib/getenv.c may have a problem with putenv()

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=268479

--- Comment #7 from Mark Millard <[email protected]> ---
(In reply to Dennis Clarke from comment #6)

Not simple?

# sysctl kern.supported_archs
kern.supported_archs: aarch64 armv7

That is from a aarch64 Cortex-A72 system. It can have chroots and jails that
are armv7 instead of aarch64:

# uname -m
arm64
# uname -p
aarch64
# chroot /usr/obj/DESTDIRs/main-CA7-chroot
# uname -m
arm
# uname -p
armv7
# file /bin/sh
/bin/sh: ELF 32-bit LSB executable, ARM, EABI5 version 1 (FreeBSD), dynamically
linked, interpreter /libexec/ld-elf.so.1, FreeBSD-style, for FreeBSD 14.0
(1400073), not stripped

Same machine. Same kernel in operation. Different uname -m and uname -r
results and a different execution environment. Are you expecting that?
(This example lacks expect manipulation via UNAME_m and UNAME_p. I'll
use uname -r and UNAME_r for an example of manipulation.)

There is also running older worlds on newer kernels, such as poudriere building
in/for for stable/13 or a releng/13.* on a system that booted main [so: 14]
(both kernel and world).

If I chroot to a tree that contains a stable/13 world (and, possibly an
unused(!) kernel), what should uname -r report? Reporting 14.0-CURRENT
would mess up various things that are trying to figure out what to build
software for. Reporting 13.1-STABLE would make more sense for a port
build targeting stable/13 .

But just using chroot leads to:
(chroot makes for simple to set up examples)

# uname -r
14.0-CURRENT
# bectl mount 13S-CA72 /be-13S-CA72-mnt # It is a 13.1 context here
# chroot /be-13S-CA72-mnt/
# uname -r
14.0-CURRENT

This is not what poudriere would want to provide to a builder process
that is targeting a stable/13 context. Thus poudriere assigns UNAME_r
in the processes for building to what a stable/13 kernel would indicate:
13.1-STABLE instead of 14-CURRENT . Then the builder sees uname -r as
returning 13.1-STABLE . (Similar points could go for buildworld
buildkernel  targeting stable/13 in the chroot.)

That is not to say that all jail/chroot contexts would want to see
13.1-STABLE . It is purpose driven: building things for stable/13
makes 13.1-STABLE appropriate. (The source tree involved has
stable/13 material, including for kernel interface source code.
main's source code is not in use, given the targeting of
stable/13 .)

So which do you expect to see: 14-CURRENT always? 13-STABLE always?
A more context/purpose specific one or the other?

I can not tell from what you have written what you mean by "and
know the results". I can tell you that in various contexts
FreeBSD assigns various UNAME_*'s to produce processes that see
uname -* outputs appropriate to a specific purpose --and they
need not match the boot environment.

-- 
You are receiving this mail because:
You are the assignee for the bug.