Re: [PATCH V2 15/15] perf tools: fix ppid in thread__fork()

[David Ahern]

On 6/28/13 12:47 AM, Adrian Hunter wrote:
> > knowledge of the specific thread within the parent that created the process
> > can be valuable -- that's what we get now. This change drops that
> > information. I'd prefer this stays at thread id -- or save the parent thread
> > id in addition to the pid.
>
> That means renaming ppid to ptid.  Do you want to do that?

Don't lose information, so if you want the process id add ppid and ptid. 
I like knowing the exact thread.

> Isn't it possible that the parent could exit and the pid or tid be re-used
> for another process?  In that case, to reliable identify the parent a
> pointer to its struct thread would be needed.  i.e.
>
> diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h
> --- a/tools/perf/util/thread.h
> +++ b/tools/perf/util/thread.h
> @@ -14,6 +14,7 @@ struct thread {
>          struct map_groups       mg;
>          pid_t                   pid_; /* Not all tools update this */
>          pid_t                   tid;
> -       pid_t                   ppid;
> +       struct thread           *parent;
>          char                    shortname[3];
>          bool                    comm_set;
>
> However that means tracking the lifetime of 'parent' to ensure that it is
> not left dangling.
>
> Do you want to do that?

No, I do not think we want to save pointers to other thread structs. We 
actually need to be able to clean up the dead_threads list for long 
running sessions and dangling pointers would add to the problem.

David

--
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/