diff -Nru flash-kernel-3.106ubuntu13/debian/changelog flash-kernel-3.106ubuntu14/debian/changelog
--- flash-kernel-3.106ubuntu13/debian/changelog	2023-02-27 11:53:49.000000000 +0000
+++ flash-kernel-3.106ubuntu14/debian/changelog	2023-03-23 14:12:19.000000000 +0000
@@ -1,3 +1,10 @@
+flash-kernel (3.106ubuntu14) lunar; urgency=medium
+
+  * Handle "not fully installed" kernels differently by quitting instead of
+    re-triggering to avoid a dpkg trigger cycle (LP: #2011925)
+
+ -- Dave Jones <dave.jones@canonical.com>  Thu, 23 Mar 2023 14:12:19 +0000
+
 flash-kernel (3.106ubuntu13) lunar; urgency=medium
 
   * Ensure that only kernels in fully "installed" state are considered for
diff -Nru flash-kernel-3.106ubuntu13/functions flash-kernel-3.106ubuntu14/functions
--- flash-kernel-3.106ubuntu13/functions	2023-02-27 11:53:49.000000000 +0000
+++ flash-kernel-3.106ubuntu14/functions	2023-03-23 14:09:14.000000000 +0000
@@ -1034,14 +1034,12 @@
 if [ "$kstatus" != "installed" ]; then
 	echo "flash-kernel: requested kernel ${kvers} is in state ${kstatus}." >&2
 	if [ -n "$DPKG_RUNNING_VERSION" ]; then
-		# We're running under dpkg; re-trigger ourselves under the assumption
-		# that the status will resolve
-		if dpkg-trigger --no-await flash-kernel; then
-			echo "flash-kernel: deferring update (trigger activated)" >&2
-			exit 0
-		else
-			echo "flash-kernel: deferral failed; proceeding with install" >&2
-		fi
+		# This is a horrid hack; we *should* re-trigger ourselves, but that
+		# causes dpkg to report a trigger-cycle on some arm64 clouds for
+		# reasons not fully understood. So we exit cleanly in the vague hope
+		# that we'll be re-triggered later in the apt run
+		echo "flash-kernel: quitting cleanly, anticipating a future run" >&2
+		exit 0
 	elif [ "$force" != "yes" ]; then
 		exit 1
 	fi