Re: [PATCH 1/3] stop_machine: check work->done while handlingenqueued works

[Tejun Heo]

On Wed, Feb 06, 2013 at 08:36:01PM +0800, Hillf Danton wrote:
> To advoid invalid reference, updating work result is moved to
> cpu_stop_signal_done().

Can you please elaborate "why" in addition to "what"?  Do we actually
have a bug here, is it meant to be an independent cleanup, or is it
something which is necessary for later patches in the series?  What
are we achieving here?

>  /* signal completion unless @done is NULL */
> -static void cpu_stop_signal_done(struct cpu_stop_done *done, bool executed)
> +static void cpu_stop_signal_done(struct cpu_stop_done *done, bool
> executed, int ret)

Patch is corrupt.

> @@ -279,8 +281,6 @@ repeat:
>  		preempt_disable();
> 
>  		ret = fn(arg);
> -		if (ret)
> -			done->ret = ret;

If this is meant as a pure cleanup, I'm not sure it's an improvement.
You own your cpu_stop_done until you call cpu_stop_signal_done() on
it, so I don't think there's anything wrong with the current code.
The new code is different, not necessarily better.

Thanks.

-- 
tejun
--
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/