Re: [PATCH v7 1/5] powerpc/85xx: implement hardware timebase sync

[Zhao Chenhui <chenhui.zhao@freescale.com>]

On Wed, Jul 04, 2012 at 10:19:54AM -0500, Tabi Timur-B04825 wrote:
> Zhao Chenhui wrote:
> > On Tue, Jul 03, 2012 at 10:17:12PM -0500, Tabi Timur-B04825 wrote:
> >> Zhao Chenhui wrote:
> >>> If the guts variable is NULL, it indicates there is error in dts or kernel.
> >>> We should fix the error, rather than ignore it.
> >>
> >> And that's why there's a warning message.  Crashing the kernel is not
> >> going to fix anything.
> >>
> >
> > This error likely crashes the kenel somewhere.
> 
> Can you test this, please?
> 
> The point I'm trying to make is that it's wrong to intentionally halt the 
> kernel unless you're sure that it's the best option.  A missing device 
> tree node is supposed to only disable a given feature, not break everything.
> 

I think there is some misunderstanding here.

	np = of_find_matching_node(NULL, mpc85xx_smp_guts_ids);
	if (np) {
		guts = of_iomap(np, 0);
		of_node_put(np);
		if (!guts) {
			pr_err("%s: Could not map guts node address\n",
								__func__);
			return;
		}
		smp_85xx_ops.give_timebase = mpc85xx_give_timebase;
		smp_85xx_ops.take_timebase = mpc85xx_take_timebase;
	}

If the guts node is missing, this code snippet will be skipped. If the guts node is existed,
the return value of of_iomap(), namely guts, will be tested. If it is NULL, it shows
that there is error in dts, or the ioremap() in of_iomap() failed. I think
these errors are fatal errors, so I print an error info and return.

-Chenhui