[10/14] rcuscale: Make rcu_scale_writer() tolerate repeated GFP_KERNEL failure

Commit Message

Neeraj Upadhyay Aug. 16, 2024, 7:02 a.m. UTC

From: "Paul E. McKenney" <[e-mail]>

Under some conditions, kmalloc(GFP_KERNEL) allocations have been
observed to repeatedly fail.  This situation has been observed to
cause one of the rcu_scale_writer() instances to loop indefinitely
retrying memory allocation for an asynchronous grace-period primitive.
The problem is that if memory is short, all the other instances will
allocate all available memory before the looping task is awakened from
its rcu_barrier*() call.  This in turn results in hangs, so that rcuscale
fails to complete.

This commit therefore removes the tight retry loop, so that when this
condition occurs, the affected task is still passing through the full
loop with its full set of termination checks.  This spreads the risk
of indefinite memory-allocation retry failures across all instances of
rcu_scale_writer() tasks, which in turn prevents the hangs.

Signed-off-by: "Paul E. McKenney" <[e-mail]>
Signed-off-by: Neeraj Upadhyay <[e-mail]>
---
 kernel/rcu/rcuscale.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

Patch

diff --git a/kernel/rcu/rcuscale.c b/kernel/rcu/rcuscale.c
index dfe8e0faa4d8..80518662273b 100644
--- a/kernel/rcu/rcuscale.c
+++ b/kernel/rcu/rcuscale.c
@@ -520,6 +520,8 @@  rcu_scale_writer(void *arg)
 
 	jdone = jiffies + minruntime * HZ;
 	do {
+		bool gp_succeeded = false;
+
 		if (writer_holdoff)
 			udelay(writer_holdoff);
 		if (writer_holdoff_jiffies)
@@ -527,23 +529,24 @@  rcu_scale_writer(void *arg)
 		wdp = &wdpp[i];
 		*wdp = ktime_get_mono_fast_ns();
 		if (gp_async && !WARN_ON_ONCE(!cur_ops->async)) {
-retry:
 			if (!rhp)
 				rhp = kmalloc(sizeof(*rhp), GFP_KERNEL);
 			if (rhp && atomic_read(this_cpu_ptr(&n_async_inflight)) < gp_async_max) {
 				atomic_inc(this_cpu_ptr(&n_async_inflight));
 				cur_ops->async(rhp, rcu_scale_async_cb);
 				rhp = NULL;
+				gp_succeeded = true;
 			} else if (!kthread_should_stop()) {
 				cur_ops->gp_barrier();
-				goto retry;
 			} else {
 				kfree(rhp); /* Because we are stopping. */
 			}
 		} else if (gp_exp) {
 			cur_ops->exp_sync();
+			gp_succeeded = true;
 		} else {
 			cur_ops->sync();
+			gp_succeeded = true;
 		}
 		t = ktime_get_mono_fast_ns();
 		*wdp = t - *wdp;
@@ -599,7 +602,7 @@  rcu_scale_writer(void *arg)
 				__func__, me, started, done, writer_done[me], atomic_read(&n_rcu_scale_writer_finished), i, jiffies - jdone);
 			selfreport = true;
 		}
-		if (started && !alldone && i < MAX_MEAS - 1)
+		if (gp_succeeded && started && !alldone && i < MAX_MEAS - 1)
 			i++;
 		rcu_scale_wait_shutdown();
 	} while (!torture_must_stop());