[Skiboot] [PATCH] opal/fsp: Fix opal boot failure on systems with redundant FSP

[Mahesh J Salgaonkar]

On 2025-04-04 09:27:26 Fri, Reza Arbab wrote:
> On Mon, Mar 17, 2025 at 09:30:03AM +0530, Mahesh Salgaonkar wrote:
> > --- a/hw/fsp/fsp.c
> > +++ b/hw/fsp/fsp.c
> > @@ -76,6 +76,7 @@ struct fsp {
> > 	unsigned int		iopath_count;
> > 	int			active_iopath;	/* -1: no active IO path */
> > 	struct fsp_iopath	iopath[FSP_MAX_IOPATH];
> > +	bool			primary;
> > };
> > 
> > enum ipl_state {
> > @@ -1959,7 +1960,7 @@ static void fsp_update_links_states(struct fsp *fsp)
> > 	}
> > 
> > 	if (fsp->active_iopath >= 0) {
> > -		if (!active_fsp || (active_fsp != fsp))
> > +		if (active_fsp && (active_fsp != fsp))
> > 			active_fsp = fsp;
> 
> What happens if none of the fsp nodes in the device tree has the "primary"
> property? With this change, won't active_fsp stay NULL forever? Maybe the
> above line should just be

Not sure if that counts as HDAT bug. But yes, if that happens active_fsp
will stay NULL and OPAL may not communicate to any of fsp.

> 
> 		if (active_fsp != fsp)
> 
> or even remove the conditional completely and just always do the assignment?

Agree. Will go with unconditional assignment.

Thanks for the review.
-Mahesh.