tools/sched_ext: scx_userland: fix restart and stats thread lifecycle bugs

Fix three issues in scx_userland's restart path:

- exit_req is not reset on restart, causing sched_main_loop() to exit
  immediately without doing any scheduling work.

- stats_printer thread handle is local to spawn_stats_thread(), making
  it impossible to join from main(). Promote it to file scope.

- The stats thread continues reading skel->bss after the skeleton is
  destroyed on restart, causing a use-after-free. Join the stats thread
  before destroying the skeleton to ensure it has exited.

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 10b3102..63f89b3 100644 (file)
--- a/tools/sched_ext/scx_userland.c
+++ b/tools/sched_ext/scx_userland.c
@@ -54,6 +54,7 @@ static bool verbose;
 static volatile int exit_req;
 static int enqueued_fd, dispatched_fd;
 
+static pthread_t stats_printer;
 static struct scx_userland *skel;
 static struct bpf_link *ops_link;
 
@@ -319,8 +320,6 @@ static void *run_stats_printer(void *arg)
 
 static int spawn_stats_thread(void)
 {
-       pthread_t stats_printer;
-
        return pthread_create(&stats_printer, NULL, run_stats_printer, NULL);
 }
 
@@ -375,6 +374,7 @@ static void pre_bootstrap(int argc, char **argv)
 
 static void bootstrap(char *comm)
 {
+       exit_req = 0;
        skel = SCX_OPS_OPEN(userland_ops, scx_userland);
 
        skel->rodata->num_possible_cpus = libbpf_num_possible_cpus();
@@ -428,6 +428,7 @@ restart:
 
        exit_req = 1;
        bpf_link__destroy(ops_link);
+       pthread_join(stats_printer, NULL);
        ecode = UEI_REPORT(skel, uei);
        scx_userland__destroy(skel);