From 50a0a550661b3c7ec0d7464661d33c7e2424038e Mon Sep 17 00:00:00 2001
From: Mike Yuan <me@yhndnzj.com>
Date: Wed, 5 Jun 2024 20:06:46 +0200
Subject: [PATCH] core: do not drop CGroupRuntime when unit stops, but only on
 GC

Fixes #33149
Replaces #33145

(cherry picked from commit 4918f14adabba71913cf3abb8b5d50a8ca0fe844)
---
 src/core/cgroup.c | 28 +++++++++++++++-------------
 src/core/cgroup.h |  5 +----
 src/core/unit.c   |  6 +++---
 3 files changed, 19 insertions(+), 20 deletions(-)

diff --git a/src/core/cgroup.c b/src/core/cgroup.c
index 8f74bea7291..70eaaaf283d 100644
--- a/src/core/cgroup.c
+++ b/src/core/cgroup.c
@@ -2685,7 +2685,7 @@ int unit_set_cgroup_path(Unit *u, const char *path) {
         if (crt && streq_ptr(crt->cgroup_path, path))
                 return 0;
 
-        unit_release_cgroup(u);
+        unit_release_cgroup(u, /* drop_cgroup_runtime = */ true);
 
         crt = unit_setup_cgroup_runtime(u);
         if (!crt)
@@ -3483,7 +3483,7 @@ int unit_realize_cgroup(Unit *u) {
         return unit_realize_cgroup_now(u, manager_state(u->manager));
 }
 
-void unit_release_cgroup(Unit *u) {
+void unit_release_cgroup(Unit *u, bool drop_cgroup_runtime) {
         assert(u);
 
         /* Forgets all cgroup details for this cgroup — but does *not* destroy the cgroup. This is hence OK to call
@@ -3514,7 +3514,8 @@ void unit_release_cgroup(Unit *u) {
                 crt->cgroup_memory_inotify_wd = -1;
         }
 
-        *(CGroupRuntime**) ((uint8_t*) u + UNIT_VTABLE(u)->cgroup_runtime_offset) = cgroup_runtime_free(crt);
+        if (drop_cgroup_runtime)
+                *(CGroupRuntime**) ((uint8_t*) u + UNIT_VTABLE(u)->cgroup_runtime_offset) = cgroup_runtime_free(crt);
 }
 
 int unit_cgroup_is_empty(Unit *u) {
@@ -3535,14 +3536,13 @@ int unit_cgroup_is_empty(Unit *u) {
         return r;
 }
 
-bool unit_maybe_release_cgroup(Unit *u) {
+static bool unit_maybe_release_cgroup(Unit *u) {
         int r;
 
-        assert(u);
+        /* Releases the cgroup only if it is recursively empty.
+         * Returns true if the cgroup was released, false otherwise. */
 
-        CGroupRuntime *crt = unit_get_cgroup_runtime(u);
-        if (!crt || !crt->cgroup_path)
-                return true;
+        assert(u);
 
         /* Don't release the cgroup if there are still processes under it. If we get notified later when all
          * the processes exit (e.g. the processes were in D-state and exited after the unit was marked as
@@ -3550,7 +3550,10 @@ bool unit_maybe_release_cgroup(Unit *u) {
          * and cleaned up later. */
         r = unit_cgroup_is_empty(u);
         if (r > 0) {
-                unit_release_cgroup(u);
+                /* Do not free CGroupRuntime when called from unit_prune_cgroup. Various accounting data
+                 * we should keep, especially CPU usage and *_peak ones which would be shown even after
+                 * the unit stops. */
+                unit_release_cgroup(u, /* drop_cgroup_runtime = */ false);
                 return true;
         }
 
@@ -3558,8 +3561,8 @@ bool unit_maybe_release_cgroup(Unit *u) {
 }
 
 void unit_prune_cgroup(Unit *u) {
-        int r;
         bool is_root_slice;
+        int r;
 
         assert(u);
 
@@ -3597,9 +3600,8 @@ void unit_prune_cgroup(Unit *u) {
         if (!unit_maybe_release_cgroup(u)) /* Returns true if the cgroup was released */
                 return;
 
-        crt = unit_get_cgroup_runtime(u); /* The above might have destroyed the runtime object, let's see if it's still there */
-        if (!crt)
-                return;
+        assert(crt == unit_get_cgroup_runtime(u));
+        assert(!crt->cgroup_path);
 
         crt->cgroup_realized = false;
         crt->cgroup_realized_mask = 0;
diff --git a/src/core/cgroup.h b/src/core/cgroup.h
index 72fe275e8ca..1bb642f1f00 100644
--- a/src/core/cgroup.h
+++ b/src/core/cgroup.h
@@ -449,10 +449,7 @@ int unit_watch_cgroup_memory(Unit *u);
 void unit_add_to_cgroup_realize_queue(Unit *u);
 
 int unit_cgroup_is_empty(Unit *u);
-void unit_release_cgroup(Unit *u);
-/* Releases the cgroup only if it is recursively empty.
- * Returns true if the cgroup was released, false otherwise. */
-bool unit_maybe_release_cgroup(Unit *u);
+void unit_release_cgroup(Unit *u, bool drop_cgroup_runtime);
 
 void unit_add_to_cgroup_empty_queue(Unit *u);
 int unit_check_oomd_kill(Unit *u);
diff --git a/src/core/unit.c b/src/core/unit.c
index 852926bcc9b..d645090abde 100644
--- a/src/core/unit.c
+++ b/src/core/unit.c
@@ -484,8 +484,8 @@ bool unit_may_gc(Unit *u) {
         /* If the unit has a cgroup, then check whether there's anything in it. If so, we should stay
          * around. Units with active processes should never be collected. */
         r = unit_cgroup_is_empty(u);
-        if (r <= 0 && r != -ENXIO)
-                return false; /* ENXIO means: currently not realized */
+        if (r <= 0 && !IN_SET(r, -ENXIO, -EOWNERDEAD))
+                return false; /* ENXIO/EOWNERDEAD means: currently not realized */
 
         if (!UNIT_VTABLE(u)->may_gc)
                 return true;
@@ -790,7 +790,7 @@ Unit* unit_free(Unit *u) {
         if (u->on_console)
                 manager_unref_console(u->manager);
 
-        unit_release_cgroup(u);
+        unit_release_cgroup(u, /* drop_cgroup_runtime = */ true);
 
         if (!MANAGER_IS_RELOADING(u->manager))
                 unit_unlink_state_files(u);