Re: [PATCH 1/2] refscale: simplify the errexit checkpoint

From: Paul E. McKenney
Date: Fri Oct 22 2021 - 19:10:58 EST


On Fri, Oct 22, 2021 at 06:51:10PM +0800, Li Zhijian wrote:
> Signed-off-by: Li Zhijian <lizhijian@xxxxxxxxxxxxxx>

Good catch!

But given that errexit is only set once at the beginning, why not
eliminate this local variable in favor of a goto at the point that it
is currently assigned to? That would permit further simplification.

Thanx, Paul

> ---
> kernel/rcu/refscale.c | 17 ++++++++---------
> 1 file changed, 8 insertions(+), 9 deletions(-)
>
> diff --git a/kernel/rcu/refscale.c b/kernel/rcu/refscale.c
> index 66dc14cf5687..2cbe2a2ba387 100644
> --- a/kernel/rcu/refscale.c
> +++ b/kernel/rcu/refscale.c
> @@ -698,26 +698,25 @@ static int main_func(void *arg)
> // Print the average of all experiments
> SCALEOUT("END OF TEST. Calculating average duration per loop (nanoseconds)...\n");
>
> - if (!errexit) {
> - buf[0] = 0;
> - strcat(buf, "\n");
> - strcat(buf, "Runs\tTime(ns)\n");
> - }
> + if (errexit)
> + goto err;
> +
> + buf[0] = 0;
> + strcat(buf, "\n");
> + strcat(buf, "Runs\tTime(ns)\n");
>
> for (exp = 0; exp < nruns; exp++) {
> u64 avg;
> u32 rem;
>
> - if (errexit)
> - break;
> avg = div_u64_rem(result_avg[exp], 1000, &rem);
> sprintf(buf1, "%d\t%llu.%03u\n", exp + 1, avg, rem);
> strcat(buf, buf1);
> }
>
> - if (!errexit)
> - SCALEOUT("%s", buf);
> + SCALEOUT("%s", buf);
>
> +err:
> // This will shutdown everything including us.
> if (shutdown) {
> shutdown_start = 1;
> --
> 2.33.0
>
>
>