As at 2021-04-07:
binutils-is-llvm WORKS for
needrestart -p
(on Debian 11 bullseye).https://bugs.debian.org/986507 is an alternative.
binutils-is-llvm DOES NOT WORK on for
apt install build-essential
, because that has a versioned dependency onbinutils (>= X)
.To fix this, I would need to either make debian/changelog have corresponding version, or change my
Provides: binutils
toProvides: binutils (= X)
.
I want check-support-status
to be happy, but I need needrestart
:
bash5$ check-support-status Limited security support for one or more packages Unfortunately, it has been necessary to limit security support for some packages. The following packages found on this system are affected by this: * Source:binutils Details: Only suitable for trusted content; see https://lists.debian.org/msgid-search/[email protected] Affected binary packages: - binutils (installed version: 2.35.2-2) - binutils-common:amd64 (installed version: 2.35.2-2) - binutils-x86-64-linux-gnu (installed version: 2.35.2-2) - libbinutils:amd64 (installed version: 2.35.2-2) - libctf-nobfd0:amd64 (installed version: 2.35.2-2) - libctf0:amd64 (installed version: 2.35.2-2) bash5$ aptitude why binutils i needrestart Depends binutils bash5$ sudo needrestart -p CRIT - Kernel: 5.10.0-4-amd64!=5.10.0-5-amd64 (!!), Microcode: CURRENT, Services: 6 (!), Containers: none, Sessions: none|Kernel=2;0;;0;2 Microcode=0;0;;0;1 Services=6;;0;0 Containers=0;;0;0 Sessions=0;0;;0 Services: - dbus.service - [email protected] - systemd-logind.service - unattended-upgrades.service - [email protected] - [email protected]
Has anyone tried to solve this already?:
/connect irc.oftc.net /join #debian-security <twb> check-support-status (rightly) complains about binutils, but it's hard to avoid installing it. For example, both build-essential and needrestart need it. Is there a way to replace it with llvm (e.g. /usr/bin/strings -> llvm-strings-10), so binutils can be purged, but without breaking anything? <pabs> add an equivs package perhaps? <twb> I'm thinking along those lines, yes. But I'm not sure how well the different implementation's arguments will line up. So it would need quite a bit of testing, and I was hoping someone else had already gotten annoyed enough to do it :-)