dbus/backport-bus-Clear-INVOCATION_ID-when-carrying-out-traditiona.patch
licunlong 10e71fd5ca Tell systemd when dbus is ready/shutting down/reloading config
(cherry picked from commit ac18e08d6959d479e839f40f26232a359e5a36fc)
2022-01-30 09:27:31 +08:00

46 lines
1.8 KiB
Diff

From 66246fff2a0bd005817fb9d108544da86973a81e Mon Sep 17 00:00:00 2001
From: Simon McVittie <smcv@collabora.com>
Date: Tue, 20 Mar 2018 12:33:51 +0000
Subject: [PATCH] bus: Clear INVOCATION_ID when carrying out traditional
activation
We weren't sure whether this one should be inherited or not, so I
asked on systemd-devel, and Lennart thought it shouldn't.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=104641
Reviewed-by: Philip Withnall <withnall@endlessm.com>
---
bus/activation.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/bus/activation.c b/bus/activation.c
index 66640ba6..2a427d9c 100644
--- a/bus/activation.c
+++ b/bus/activation.c
@@ -862,7 +862,6 @@ populate_environment (BusActivation *activation)
* - TERM, WATCHDOG_*: Should not be set for dbus-daemon, so not applicable
* - MAINPID, SERVICE_RESULT, EXIT_CODE, EXIT_STATUS: Not set for ExecStart,
* so not applicable
- * - INVOCATION_ID: TODO: Do we want to clear this or not? It isn't clear.
*/
/* We give activated services their own Journal stream to avoid their
@@ -878,6 +877,13 @@ populate_environment (BusActivation *activation)
* (and NotifyAccess wouldn't let it write here anyway) */
_dbus_hash_table_remove_string (activation->environment, "NOTIFY_SOCKET");
+ /* This identifies the dbus-daemon invocation. Whether it should be
+ * inherited by "smaller" services isn't entirely clear-cut, but not
+ * inheriting it makes traditional D-Bus activation under systemd a
+ * little more consistent with systemd activation.
+ * https://lists.freedesktop.org/archives/systemd-devel/2018-March/040467.html */
+ _dbus_hash_table_remove_string (activation->environment, "INVOCATION_ID");
+
return retval;
}
--
2.27.0