clevis/0002-systemd-account-for-unlocking-failures-in-clevis-luk.patch
2021-10-29 12:10:03 -03:00

42 lines
1.5 KiB
Diff

From 2a16458622be983aba3762079d7a72383bd8be02 Mon Sep 17 00:00:00 2001
From: Sergio Correia <scorreia@redhat.com>
Date: Fri, 29 Oct 2021 10:58:26 -0300
Subject: [PATCH 2/2] systemd: account for unlocking failures in
clevis-luks-askpass
As unlock may fail for some reason, e.g. the network is not up yet,
one way cause problems would be to add extra `rd.luks.uuid' params
to the cmdline, which would then cause such devices to be unlocked
in early boot. If the unlocking fail, those devices might not be
accounted for in the clevis_devices_to_unlock() check, as it is
based on crypttab.
Let's make sure there are no pending ask.* sockets waiting to be
answered, before exiting.
Related: https://bugzilla.redhat.com/show_bug.cgi?id=1878892
---
src/luks/systemd/clevis-luks-askpass.in | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/src/luks/systemd/clevis-luks-askpass.in b/src/luks/systemd/clevis-luks-askpass.in
index 8f54859..a6699c9 100755
--- a/src/luks/systemd/clevis-luks-askpass.in
+++ b/src/luks/systemd/clevis-luks-askpass.in
@@ -67,8 +67,11 @@ while true; do
done
[ "${loop}" != true ] && break
+
# Checking for pending devices to be unlocked.
- if remaining=$(clevis_devices_to_unlock) && [ -z "${remaining}" ]; then
+ remaining_crypttab=$(clevis_devices_to_unlock) ||:
+ remaining_askfiles=$(ls "${path}"/ask.* 2>/dev/null) ||:
+ if [ -z "${remaining_crypttab}" ] && [ -z "${remaining_askfiles}" ]; then
break;
fi
--
2.33.1