public inbox for sched-ext@lists.linux.dev
 help / color / mirror / Atom feed
* Re: [PATCH] sched_ext/selftests: Fix bpf_link leak on assertion failure
       [not found] <20260317174402.48082-1-devnexen@gmail.com>
@ 2026-03-17 18:09 ` Tejun Heo
  0 siblings, 0 replies; only message in thread
From: Tejun Heo @ 2026-03-17 18:09 UTC (permalink / raw)
  To: David Carlier
  Cc: David Vernet, Andrea Righi, linux-kernel, Emil Tsalapatis,
	sched-ext

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 829 bytes --]

Hello,

In numa.c and rt_stall.c, the assertions check for SCX_EXIT_NONE — i.e.
the scheduler ran without error. bpf_link__destroy() triggers unregistration
which changes uei.kind to SCX_EXIT_UNREG, so moving it before the assertions
would make them always fail. Did you run the selftests after this patch?

A simpler approach would be using __attribute__((cleanup())) to auto-destroy
the link on any return path:

  static inline void autolink_destroy(struct bpf_link **linkp)
  {
  	if (*linkp)
  		bpf_link__destroy(*linkp);
  }

  #define __autolink __attribute__((cleanup(autolink_destroy)))

Then each test just does:

  struct bpf_link *link __autolink = bpf_map__attach_struct_ops(...);

No reordering needed — the link auto-destroys on any return, including
early returns from assertion failures.

Thanks.

--
tejun

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2026-03-17 18:09 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <20260317174402.48082-1-devnexen@gmail.com>
2026-03-17 18:09 ` [PATCH] sched_ext/selftests: Fix bpf_link leak on assertion failure Tejun Heo

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox