tools/sched_ext: scx_userland: fix data races on shared counters

The stats thread reads nr_vruntime_enqueues, nr_vruntime_dispatches,
nr_vruntime_failed, and nr_curr_enqueued concurrently with the main
thread writing them, with no synchronization.

Use __atomic builtins with relaxed ordering for all accesses to these
counters to eliminate the data races.

Only display accuracy is affected, not scheduling correctness.

Signed-off-by: David Carlier <devnexen@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>

diff --git a/tools/sched_ext/scx_userland.c b/tools/sched_ext/scx_userland.c
index 504a808..3f2aba6 100644 (file)
--- a/tools/sched_ext/scx_userland.c
+++ b/tools/sched_ext/scx_userland.c
@@ -157,9 +157,9 @@ static int dispatch_task(__s32 pid)
 
        err = bpf_map_update_elem(dispatched_fd, NULL, &pid, 0);
        if (err) {
-               nr_vruntime_failed++;
+               __atomic_add_fetch(&nr_vruntime_failed, 1, __ATOMIC_RELAXED);
        } else {
-               nr_vruntime_dispatches++;
+               __atomic_add_fetch(&nr_vruntime_dispatches, 1, __ATOMIC_RELAXED);
        }
 
        return err;
@@ -202,8 +202,8 @@ static int vruntime_enqueue(const struct scx_userland_enqueued_task *bpf_task)
                return ENOENT;
 
        update_enqueued(curr, bpf_task);
-       nr_vruntime_enqueues++;
-       nr_curr_enqueued++;
+       __atomic_add_fetch(&nr_vruntime_enqueues, 1, __ATOMIC_RELAXED);
+       __atomic_add_fetch(&nr_curr_enqueued, 1, __ATOMIC_RELAXED);
 
        /*
         * Enqueue the task in a vruntime-sorted list. A more optimal data
@@ -279,9 +279,9 @@ static void dispatch_batch(void)
                        LIST_INSERT_HEAD(&vruntime_head, task, entries);
                        break;
                }
-               nr_curr_enqueued--;
+               __atomic_sub_fetch(&nr_curr_enqueued, 1, __ATOMIC_RELAXED);
        }
-       skel->bss->nr_scheduled = nr_curr_enqueued;
+       skel->bss->nr_scheduled = __atomic_load_n(&nr_curr_enqueued, __ATOMIC_RELAXED);
 }
 
 static void *run_stats_printer(void *arg)
@@ -306,9 +306,9 @@ static void *run_stats_printer(void *arg)
                printf("|-----------------------|\n");
                printf("| VRUNTIME / USER       |\n");
                printf("|-----------------------|\n");
-               printf("|  enq:      %10llu |\n", nr_vruntime_enqueues);
-               printf("|  disp:     %10llu |\n", nr_vruntime_dispatches);
-               printf("|  failed:   %10llu |\n", nr_vruntime_failed);
+               printf("|  enq:      %10llu |\n", __atomic_load_n(&nr_vruntime_enqueues, __ATOMIC_RELAXED));
+               printf("|  disp:     %10llu |\n", __atomic_load_n(&nr_vruntime_dispatches, __ATOMIC_RELAXED));
+               printf("|  failed:   %10llu |\n", __atomic_load_n(&nr_vruntime_failed, __ATOMIC_RELAXED));
                printf("o-----------------------o\n");
                printf("\n\n");
                fflush(stdout);
@@ -376,10 +376,10 @@ static void bootstrap(char *comm)
 {
        exit_req = 0;
        min_vruntime = 0.0;
-       nr_vruntime_enqueues = 0;
-       nr_vruntime_dispatches = 0;
-       nr_vruntime_failed = 0;
-       nr_curr_enqueued = 0;
+       __atomic_store_n(&nr_vruntime_enqueues, 0, __ATOMIC_RELAXED);
+       __atomic_store_n(&nr_vruntime_dispatches, 0, __ATOMIC_RELAXED);
+       __atomic_store_n(&nr_vruntime_failed, 0, __ATOMIC_RELAXED);
+       __atomic_store_n(&nr_curr_enqueued, 0, __ATOMIC_RELAXED);
        memset(tasks, 0, pid_max * sizeof(*tasks));
        LIST_INIT(&vruntime_head);