Re: [RESEND PATCH 1/2] spi: rockchip: Convert to late and early system PM callbacks

[Brian Norris]

Hi,

On Fri, Oct 13, 2017 at 08:32:12AM -0700, Doug Anderson wrote:
> Hi,
> 
> On Fri, Oct 13, 2017 at 3:41 AM, Jeffy Chen <jeffy.chen@xxxxxxxxxxxxxx> wrote:
> > Currently we are suspending the spi master in it's ->suspend callback,
> > which is racy as some other drivers may still want to transmit messages
> > on the bus(e.g. spi based pwm backlight).
> >
> > Convert to late and early system PM callbacks to avoid the race.
> >
> > Signed-off-by: Jeffy Chen <jeffy.chen@xxxxxxxxxxxxxx>
> > ---
> >
> >  drivers/spi/spi-rockchip.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> It shouldn't hurt to do this, but I'm curious if you did any digging
> about why this happens?  As I understood it suspend order is supposed
> to be opposite of probe order.  Thus anything that was able to get a
> reference to the cros-ec PWM at its probe time should get suspended
> before cros-ec suspends.

Yes, this does seem odd to me too. This looks like an arms race hack
that should be avoided unless we know a legit root cause. Also,
"probe order implies suspend order" doesn't quite work for async suspend
anyway, so we'd probably want to express the dependency properly
anyway.

Any chance this is related? Seems like that might break the parent/child
relationship for master/slave:

commit d7e2ee257038baeb03baef602500368a51ee9eef
Author: Linus Walleij <linus.walleij@xxxxxxxxxx>
Date:   Mon Apr 11 13:51:03 2016 +0200

    spi: let SPI masters ignore their children for PM

Brian