Hi Geert,
On Monday, September 24, 2018, Geert Uytterhoeven wrote:
> Thanks for your patch!
Thanks for your review!
> > +struct cpg_mssr_priv *early_priv;
>
> static
>
> Just call the pointer cpg_mssr_priv, as you're gonna need it in both
> cases
> (see below)?
Seems strange to have a variable with the same name of the struct type
that it is....but I changed as you suggested.
I also made all the other changes you suggested for both patches and
retested.
> > + if (early_priv) {
> > + priv = early_priv;
> > + priv->dev = &pdev->dev;
> > + } else {
> > + error = cpg_mssr_common_init(dev, dev->of_node, info);
>
> Oops, priv is not set => BOOM.
> I guess you have no other hardware using the renesas-cpg-mssr driver?
> You can probably still test this case on RZ/A2 without patch 2 (or with
> the
> OF_CLK_DECLARE_DRIVER() line commented out), and with the OSTM timer
> hack
> you posted before.
Well...I have a RZ/G1 board in a box somewhere...
But instead, like you suggested, I applied the OSTM subsys_initcall hack
back in and changed r7s9210-cpg-mssr.c to be non-early like the other
R-Car SoCs.
And now, no crash. So I think I'm good this time.
Thanks!
Chris
