[Skiboot] [PATCH v3 0/6] Enable basic VAS support

Sukadev Bhattiprolu sukadev at linux.vnet.ibm.com
Fri Dec 2 14:16:41 AEDT 2016

Balbir Singh [bsingharora at gmail.com] wrote:
> On 01/12/16 11:46, Sukadev Bhattiprolu wrote:
> > Oliver O'Halloran [oohall at gmail.com] wrote:
> >> Half the reason we have the DT is to avoid having to guess about what
> >> hardware is present based on things like the processor version. Is
> >> there a compelling reason not to create a DT node?
> > 
> > Not really. I was keeping the code minimal. Will add a patch on top of this
> > set to create a device node for VAS under the xscom node.
> > 
> That sounds reasonable, but I think it would be nice to discuss the design
> of the DT interface. I assume it won't be too late to revisit these patches then?

With this change to my patchset:

diff --git a/core/vas.c b/core/vas.c
index 48d9505..9107e6b 100644
--- a/core/vas.c
+++ b/core/vas.c
@@ -372,6 +372,8 @@ static int init_one_chip(struct proc_chip *chip)
        if (alloc_irqs(chip))
                goto out;
+       (void)dt_new(chip->devnode, "vas");
        prlog(PR_INFO, "VAS: Initialized chip %d\n", chip->id);
        return 0;

I get:

	# ls /proc/device-tree/xscom at 603fc00000000/vas/
	name     phandle
	# od -xc  /proc/device-tree/xscom at 603fc00000000/vas/name
	0000000   v   a   s  \0
		7661 7300

	# od -xc  /proc/device-tree/xscom at 603fc00000000/vas/phandle
	0000000  \0  \0  \0   $
		0000 0024

'vas' should show up under each chip of course. I can fold that change
into one of my patches, but do we need to do anything else at this point?



More information about the Skiboot mailing list