Re: [PATCH v6 11/17] static_call: Simple self-test

From: Peter Zijlstra
Date: Sat Jul 11 2020 - 06:31:30 EST


On Fri, Jul 10, 2020 at 06:42:29PM -0400, Steven Rostedt wrote:
> On Fri, 10 Jul 2020 15:38:42 +0200
> Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:

> > +static int __init test_static_call_init(void)
> > +{
> > + WARN_ON(static_call(sc_selftest)(2) != 3);
> > + static_call_update(sc_selftest, &func_b);
> > + WARN_ON(static_call(sc_selftest)(2) != 4);
> > + static_call_update(sc_selftest, &func_a);
> > + WARN_ON(static_call(sc_selftest)(2) != 3);
> > +
> > + return 0;
> > +}
>
> I wonder if this would be better if we were testing the same static call each time?

Makes sense, I suppose.

> static int __init run_static_call(int val)
> {
> return static_call(sc_selftest)(val);
> }

Don't think we need this, or are you afraid of loop unrolling, in which
case you also want a noinline here I suppose.

>
> static struct {
> int (*func)(int);
> int val;
> int expect;
> } static_call_data [] = {
> { NULL, 2, 3 }
> ( func_b, 2 , 4},
> { func_a, 2, 3}
> } __initdata;
>
> static int __init test_static_call_init(void)
> {
> int i;
>
> for (i = 0; i < ARRAY_SIZE(static_call_data); i++ ) {
> if (static_call_data[i].func)
> static_call_update(sc_selftest, static_call_data[i].func);
> WARN_ON(run_static_call(static_call_data[i].val) != static_call_data[i].expect);
> }
>
> return 0;
> }

Lots of compile errors with that, fixed them all :-)