Re: [PATCH v2 2/4] ftrace: introduce workqueue_handler_exittracepoint and rename workqueue_execution to workqueue_handler_entry

[Oleg Nesterov]

On 04/15, Ingo Molnar wrote:
>
> The patches try to map and instrument the life cycle of a worklet,
> and the main actions that occur in the workqueue subsystem in
> general.
>
> The purpose is instrumentation: for debugging purposes, for
> improving kernel code and for just understanding how the system
> functions and what its dynamic actions are.
>
> In that sense the worklet 'key' possibly getting reallocated and
> reused before the 'completed' action was traced is probably not a
> big deal - but tools have to be aware of it possibly happening (and
> most not hard-code any assumption to the contrary).
>
> Plus the exit handler must not dereference the worklet either.
> Safest would be to make this sure in the prototype already: pass in
> a void * key, not a work structure.

Yes, I think you are right.

And thanks for your explanations!

Oleg.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/