Re: [PATCH RESEND] lib/test_printf.c: call wait_for_random_bytes() before plain %p tests

[Thierry Escande]

On 08/06/2018 13:22, Petr Mladek wrote:
> On Fri 2018-06-08 12:32:33, Thierry Escande wrote:
> > On 08/06/2018 11:46, Andy Shevchenko wrote:
> > > On Fri, Jun 8, 2018 at 12:07 PM, Thierry Escande
> > > <thierry.escande@xxxxxxxxxx> wrote:
> > >
> > > > But as I type I realize it's not necessary. I will simply enclose the call
> > > > to wait_for_random_bytes() by #if IS_MODULE() #endif so it gets called only
> > > > if built as a module, which is how run_kselftest.sh wants it... If
> > > > test_printf is compiled built-in and the crng is not yet initialized the
> > > > test will fail anyway so there is no need to add an extra check.
> > >
> > > Unfortunately I can't support this as is.
> > > We have environments where crng will be ready minutes after the boot.
> > > It's unacceptable.
> > >
> > > So, we need to have means to not delay test for so long.
> >
> > I agree we can't delay test execution for too long. In my case the crng is
> > ready only a few seconds after the boot. So we may just skip this plain 'p'
> > printf test if crng is not ready then.
>
> Alternative solution would be to accept
> const char *str = sizeof(ptr) == 8 ? "(____ptrval____)" : "(ptrval)";
> as a valid result. It would make sense to print some warning in that case.
>
> In each case, it would look ugly to use add_random_ready_callback()
> wihtout passing a callback. If you really needed to check crng_ready(),
> it would be better to make it public.

Agree but even with crng_ready() public we would have to block the test 
until it's ready which is not good either.

Accepting "(ptrval)" as a valid result seems the least bad alternative...

Regards,
Thierry