kunit: irq: Ensure timer doesn't fire too frequently

Fix a bug where kunit_run_irq_test() could hang if the system is too slow. This was noticed with the crypto library tests in certain VMs.

Specifically, if kunit_irq_test_timer_func() and the associated hrtimer code took over 5us to run, then the CPU would spend all its time executing that code in hardirq context. As a result, the task executing kunit_run_irq_test() never had a chance to run, exit the loop, and cancel the timer.

To fix it, make kunit_irq_test_timer_func() increase the timer interval when the other contexts aren't having a chance to run.

Fixes 950a81224e8b ("lib/crypto: tests: Add hash-test-template.h and gen-hash-testvecs.py")

Cc [email protected]

Signed-off-by Eric Biggers <[email protected]>

---

This patch applies to v7.0-rc1 and is targeting libcrypto-fixes

  include/kunit/run-in-irq-context.h | 44 +++++++++++++++++++-----------
  1 file changed, 28 insertions(+), 16 deletions(-)

include/kunit/run-in-irq-context.h +28 -16

index c89b1b1b12dd5..bfe60d6cf28d8 100644
@@ -10,36 +10,47 @@
  #include <kunit/test.h>
  #include <linux/timekeeping.h>
  #include <linux/hrtimer.h>
  #include <linux/workqueue.h>
 
-#define KUNIT_IRQ_TEST_HRTIMER_INTERVAL us_to_ktime(5)
-
  struct kunit_irq_test_state {
  	bool (*func)(void *test_specific_state);
  	void *test_specific_state;
  	bool task_func_reported_failure;
  	bool hardirq_func_reported_failure;
  	bool softirq_func_reported_failure;
+	atomic_t task_func_calls;
  	atomic_t hardirq_func_calls;
  	atomic_t softirq_func_calls;
+	ktime_t interval;
  	struct hrtimer timer;
  	struct work_struct bh_work;
  };
 
  static enum hrtimer_restart kunit_irq_test_timer_func(struct hrtimer *timer)
  {
  	struct kunit_irq_test_state *state =
  		container_of(timer, typeof(*state), timer);
+	int task_calls, hardirq_calls, softirq_calls;
 
  	WARN_ON_ONCE(!in_hardirq());
-	atomic_inc(&state->hardirq_func_calls);
+	task_calls = atomic_read(&state->task_func_calls);
+	hardirq_calls = atomic_inc_return(&state->hardirq_func_calls);
+	softirq_calls = atomic_read(&state->softirq_func_calls);
+
+	/*
+	 * If the timer is firing too often for the softirq or task to ever have
+	 * a chance to run, increase the timer interval.  This is needed on very
+	 * slow systems.
+	 */
+	if (hardirq_calls >= 20 && (softirq_calls == 0 || task_calls == 0))
+		state->interval = ktime_add_ns(state->interval, 250);
 
  	if (!state->func(state->test_specific_state))
  		state->hardirq_func_reported_failure = true;
 
-	hrtimer_forward_now(&state->timer, KUNIT_IRQ_TEST_HRTIMER_INTERVAL);
+	hrtimer_forward_now(&state->timer, state->interval);
  	queue_work(system_bh_wq, &state->bh_work);
  	return HRTIMER_RESTART;
  }
 
  static void kunit_irq_test_bh_work_func(struct work_struct *work)

@@ -84,14 +95,18 @@ static inline void kunit_run_irq_test(struct kunit *test, bool (*func)(void *),
  				      void *test_specific_state)
  {
  	struct kunit_irq_test_state state = {
  		.func = func,
  		.test_specific_state = test_specific_state,
+		/*
+		 * Start with a 5us timer interval.  If the system can't keep
+		 * up, kunit_irq_test_timer_func() will increase it.
+		 */
+		.interval = us_to_ktime(5),
  	};
  	unsigned long end_jiffies;
-	int hardirq_calls, softirq_calls;
-	bool allctx = false;
+	int task_calls, hardirq_calls, softirq_calls;
 
  	/*
  	 * Set up a hrtimer (the way we access hardirq context) and a work
  	 * struct for the BH workqueue (the way we access softirq context).
  	 */

@@ -102,25 +117,22 @@ static inline void kunit_run_irq_test(struct kunit *test, bool (*func)(void *),
  	/*
  	 * Run for up to max_iterations (including at least one task, softirq,
  	 * and hardirq), or 1 second, whichever comes first.
  	 */
  	end_jiffies = jiffies + HZ;
-	hrtimer_start(&state.timer, KUNIT_IRQ_TEST_HRTIMER_INTERVAL,
-		      HRTIMER_MODE_REL_HARD);
-	for (int task_calls = 0, calls = 0;
-	     ((calls < max_iterations) || !allctx) &&
-	     !time_after(jiffies, end_jiffies);
-	     task_calls++) {
+	hrtimer_start(&state.timer, state.interval, HRTIMER_MODE_REL_HARD);
+	do {
  		if (!func(test_specific_state))
  			state.task_func_reported_failure = true;
 
+		task_calls = atomic_inc_return(&state.task_func_calls);
  		hardirq_calls = atomic_read(&state.hardirq_func_calls);
  		softirq_calls = atomic_read(&state.softirq_func_calls);
-		calls = task_calls + hardirq_calls + softirq_calls;
-		allctx = (task_calls > 0) && (hardirq_calls > 0) &&
-			 (softirq_calls > 0);
-	}
+	} while ((task_calls + hardirq_calls + softirq_calls < max_iterations ||
+		  (task_calls == 0 || hardirq_calls == 0 ||
+		   softirq_calls == 0)) &&
+		 !time_after(jiffies, end_jiffies));
 
  	/* Cancel the timer and work. */
  	hrtimer_cancel(&state.timer);
  	flush_work(&state.bh_work);
 

base-commit: 6de23f81a5e08be8fbf5e8d7e9febc72a5b5f27f

Fix a bug where kunit_run_irq_test() could hang if the system is too slow. This was noticed with the crypto library tests in certain VMs.

Specifically, if kunit_irq_test_timer_func() and the associated hrtimer code took over 5us to run, then the CPU would spend all its time executing that code in hardirq context. As a result, the task executing kunit_run_irq_test() never had a chance to run, exit the loop, and cancel the timer.

To fix it, make kunit_irq_test_timer_func() increase the timer interval when the other contexts aren't having a chance to run.

Fixes 950a81224e8b ("lib/crypto: tests: Add hash-test-template.h and gen-hash-testvecs.py")

Cc [email protected]

Signed-off-by Eric Biggers <[email protected]>

---

This patch applies to v7.0-rc1 and is targeting libcrypto-fixes

 include/kunit/run-in-irq-context.h | 44 +++++++++++++++++++-----------
 1 file changed, 28 insertions(+), 16 deletions(-)