On Fri, Jun 02, 2017 at 05:12:52PM +0200, Andrew Jones wrote:
> on_cpu() and friends are a bit risky when implemented without IPIs
> (no preemption), because we can easily get deadlocks. Luckily, it's
> also easy to detect those deadlocks, and assert, to at least make
> them easier to debug.
>
> Signed-off-by: Andrew Jones <drjones@xxxxxxxxxx>
> ---
> lib/arm/smp.c | 31 ++++++++++++++++++++++++++-----
> 1 file changed, 26 insertions(+), 5 deletions(-)
>
> diff --git a/lib/arm/smp.c b/lib/arm/smp.c
> index b4b43237e32e..34204955a549 100644
> --- a/lib/arm/smp.c
> +++ b/lib/arm/smp.c
> @@ -76,9 +76,25 @@ typedef void (*on_cpu_func)(void *);
> struct on_cpu_info {
> on_cpu_func func;
> void *data;
> + cpumask_t waiters;
> };
> static struct on_cpu_info on_cpu_info[NR_CPUS];
>
> +static void cpu_wait(int cpu)
> +{
> + int me = smp_processor_id();
> +
> + if (cpu == me)
> + return;
> +
> + assert_msg(!cpumask_test_cpu(cpu, &on_cpu_info[me].waiters), "CPU%d <=> CPU%d deadlock detected", me, cpu);
> +
> + cpumask_set_cpu(me, &on_cpu_info[cpu].waiters);
Doh, this bit set should come before the assert, otherwise there's
a small race here. v2 coming up.
drew
