Skip to contentSkip to conversation

debugobjects: Don't call fill_pool() in early boot hardirq context

PATCH In review Waiman Long 2 participants 4 messages Jun 8

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v6 cover debugobjects: Don't call fill_pool() in early boot hardirq context

v6:
- Break the v5 patch into 2 separate patches with the first patch just adding a can_fill_pool() helper with no functional change as suggested by Ingo.

Patch 1 adds a can_fill_pool() helper to break down the conditional check in debug_objects_fill_pool() to determine if fill_pool() should be called into easier to read components.

Patch 2 fix the lockdep warning caused by calling fill_pool() in early boot hardirq context.

These 2 patches together produce the same code change as in the v5 patch [1]. So the Reviewed-by tag from Sebastian is kept in the patch 2.

[1] https://lore.kernel.org/lkml/[email protected]/

Waiman Long (2):
debugobjects: Add a can_fill_pool() helper for
debug_objects_fill_pool()
debugobjects: Don't call fill_pool() in early boot hardirq context

 lib/debugobjects.c | 46 +++++++++++++++++++++++++++++++++++++---------
 1 file changed, 37 insertions(+), 9 deletions(-)
Signature · 1 line
2.54.0
#1 lore raw

PATCH v6 1/2 debugobjects: Add a can_fill_pool() helper for debug_objects_fill_pool()

The conditional check in debug_objects_fill_pool() for determining if fill_pool() can be called is now quite complicated. Add a new can_fill_pool() helper to break down the conditions individually to make them easier to understand. No functional change is expected.

---
 lib/debugobjects.c | 35 ++++++++++++++++++++++++++---------
 1 file changed, 26 insertions(+), 9 deletions(-)
lib/debugobjects.c +26 -9
index b18a682fe3da..1d826689611f 100644
@@ -720,6 +720,30 @@ static inline bool debug_objects_is_pi_blocked_on(void)
 #endif
 }
 
+static inline bool can_fill_pool(void)
+{
+	/*
+	 * On !RT enabled kernels there are no restrictions and spinlock_t and
+	 * raw_spinlock_t are the same types.
+	 */
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+		return true;
+
+	/*
+	 * On RT enabled kernels the pool refill must happen in preemptible
+	 * context and the task must not be blocked on a lock as that could
+	 * corrupt the PI state when blocking on a lock in the allocation path.
+	 */
+	if (preemptible() && !debug_objects_is_pi_blocked_on())
+		return true;
+
+	/*
+	 * Allow refilling to happen early in the boot with disabled interrupts
+	 * as long as the scheduler is not operational.
+	 */
+	return system_state < SYSTEM_SCHEDULING;
+}
+
 static void debug_objects_fill_pool(void)
 {
 	if (!static_branch_likely(&obj_cache_enabled))
@@ -734,18 +758,11 @@ static void debug_objects_fill_pool(void)
 	if (likely(!pool_should_refill(&pool_global)))
 		return;
 
-	/*
-	 * On RT enabled kernels the pool refill must happen in preemptible
-	 * context and not enqueued on an rt_mutex -- for !RT kernels we rely
-	 * on the fact that spinlock_t and raw_spinlock_t are basically the
-	 * same type and this lock-type inversion works just fine.
-	 */
-	if (!IS_ENABLED(CONFIG_PREEMPT_RT) || system_state < SYSTEM_SCHEDULING ||
-	    (preemptible() && !debug_objects_is_pi_blocked_on())) {
+	if (can_fill_pool()) {
 		/*
 		 * Annotate away the spinlock_t inside raw_spinlock_t warning
 		 * by temporarily raising the wait-type to LD_WAIT_CONFIG, matching
-		 * the preemptible() condition above.
+		 * the preemptible() condition in can_fill_pool().
 		 */
 		static DEFINE_WAIT_OVERRIDE_MAP(fill_pool_map, LD_WAIT_CONFIG);
 		lock_map_acquire_try(&fill_pool_map);
Signature · 1 line
2.54.0
#2 lore raw

PATCH v6 2/2 debugobjects: Don't call fill_pool() in early boot hardirq context

In reply to Waiman Long

When booting a debug PREEMPT_RT kernel on an arm64 system with grace processor, a "inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage" lockdep warning message was reported to the console.

During early boot, interrupts are getting enabled before the scheduler is enabled. In this window (before SYSTEM_SCHEDULING is set) interrupts can fire and attempt to fill the pool from within the hardirq. This can lead to a deadlock the interrupt occurred while in the memory allocator.

Reorder the exception rules and forbid this scenario by excluding allocations from hardirq.

Fixes 06e0ae988f6e ("debugobjects: Allow to refill the pool before SYSTEM_SCHEDULING")
---
 lib/debugobjects.c | 25 ++++++++++++++++++-------
 1 file changed, 18 insertions(+), 7 deletions(-)
lib/debugobjects.c +18 -7
index 1d826689611f..6fb00e08a4e2 100644
@@ -730,18 +730,29 @@ static inline bool can_fill_pool(void)
 		return true;
 
 	/*
-	 * On RT enabled kernels the pool refill must happen in preemptible
-	 * context and the task must not be blocked on a lock as that could
-	 * corrupt the PI state when blocking on a lock in the allocation path.
+	 * On RT enabled kernels, the task must not be blocked on a lock as
+	 * that could corrupt the PI state when blocking on a lock in the
+	 * allocation path.
 	 */
-	if (preemptible() && !debug_objects_is_pi_blocked_on())
+	if (debug_objects_is_pi_blocked_on())
+		return false;
+
+	/*
+	 * On RT enabled kernels the pool refill should happen in preemptible
+	 * context.
+	 */
+	if (preemptible())
 		return true;
 
 	/*
-	 * Allow refilling to happen early in the boot with disabled interrupts
-	 * as long as the scheduler is not operational.
+	 * Though during system boot before scheduling is set up, preemption is
+	 * disabled and the pool can get exhausted. Before scheduling is active
+	 * a task cannot be blocked on a sleeping lock, but it might hold a lock
+	 * and if interrupted then hard interrupt context might run into a lock
+	 * inversion. So exclude hard interrupt context from allocations before
+	 * scheduling is active.
 	 */
-	return system_state < SYSTEM_SCHEDULING;
+	return system_state < SYSTEM_SCHEDULING && !in_hardirq();
 }
 
 static void debug_objects_fill_pool(void)
Signature · 1 line
2.54.0