Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.37 as permitted sender) client-ip=23.128.96.37;
From:   Joseph Salisbury <joseph.salisbury@canonical.com>
To:     LKML <linux-kernel@vger.kernel.org>,
        linux-rt-users <linux-rt-users@vger.kernel.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Carsten Emde <C.Emde@osadl.org>,
        John Kacur <jkacur@redhat.com>,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        Daniel Wagner <daniel.wagner@suse.com>,
        Tom Zanussi <tom.zanussi@linux.intel.com>,
        Clark Williams <williams@redhat.com>,
        Pavel Machek <pavel@denx.de>,
        Joseph Salisbury <joseph.salisbury@canonical.com>
Cc:     Ido Schimmel <idosch@nvidia.com>
Subject: [PATCH RT 04/12] debugobject: Ensure pool refill (again)
Date:   Wed, 18 Oct 2023 15:48:25 -0400
Message-Id: <20231018194833.651674-5-joseph.salisbury@canonical.com>
In-Reply-To: <20231018194833.651674-1-joseph.salisbury@canonical.com>
References: <20231018194833.651674-1-joseph.salisbury@canonical.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Thomas Gleixner <tglx@linutronix.de>

v5.15.133-rt70-rc1 stable review patch.
If anyone has any objections, please let me know.

-----------


The recent fix to ensure atomicity of lookup and allocation inadvertently
broke the pool refill mechanism.

Prior to that change debug_objects_activate() and debug_objecs_assert_init()
invoked debug_objecs_init() to set up the tracking object for statically
initialized objects. That's not longer the case and debug_objecs_init() is
now the only place which does pool refills.

Depending on the number of statically initialized objects this can be
enough to actually deplete the pool, which was observed by Ido via a
debugobjects OOM warning.

Restore the old behaviour by adding explicit refill opportunities to
debug_objects_activate() and debug_objecs_assert_init().

Fixes: 63a759694eed ("debugobject: Prevent init race with static objects")
Reported-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Ido Schimmel <idosch@nvidia.com>
Link: https://lore.kernel.org/r/871qk05a9d.ffs@tglx

(cherry picked from commit 0af462f19e635ad522f28981238334620881badc)
Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
---
 lib/debugobjects.c | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)

diff --git a/lib/debugobjects.c b/lib/debugobjects.c
index 579406c1e9ed..4c39678c03ee 100644
--- a/lib/debugobjects.c
+++ b/lib/debugobjects.c
@@ -590,6 +590,16 @@ static struct debug_obj *lookup_object_or_alloc(void *addr, struct debug_bucket
 	return NULL;
 }
 
+static void debug_objects_fill_pool(void)
+{
+	/*
+	 * On RT enabled kernels the pool refill must happen in preemptible
+	 * context:
+	 */
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT) || preemptible())
+		fill_pool();
+}
+
 static void
 __debug_object_init(void *addr, const struct debug_obj_descr *descr, int onstack)
 {
@@ -598,12 +608,7 @@ __debug_object_init(void *addr, const struct debug_obj_descr *descr, int onstack
 	struct debug_obj *obj;
 	unsigned long flags;
 
-	/*
-	 * On RT enabled kernels the pool refill must happen in preemptible
-	 * context:
-	 */
-	if (!IS_ENABLED(CONFIG_PREEMPT_RT) || preemptible())
-		fill_pool();
+	debug_objects_fill_pool();
 
 	db = get_bucket((unsigned long) addr);
 
@@ -688,6 +693,8 @@ int debug_object_activate(void *addr, const struct debug_obj_descr *descr)
 	if (!debug_objects_enabled)
 		return 0;
 
+	debug_objects_fill_pool();
+
 	db = get_bucket((unsigned long) addr);
 
 	raw_spin_lock_irqsave(&db->lock, flags);
@@ -897,6 +904,8 @@ void debug_object_assert_init(void *addr, const struct debug_obj_descr *descr)
 	if (!debug_objects_enabled)
 		return;
 
+	debug_objects_fill_pool();
+
 	db = get_bucket((unsigned long) addr);
 
 	raw_spin_lock_irqsave(&db->lock, flags);
-- 
2.34.1