On Wed, 2014-11-05 at 10:11 -0800, Bjorn Andersson wrote:

On Tue, Nov 4, 2014 at 5:33 AM, Ivan T. Ivanov [off-list ref] 
wrote:
[..]

quoted

@@ -28,11 +144,27 @@ static int pmic_spmi_probe(struct spmi_device 

*sdev)
 {
        struct device_node *root = sdev->dev.of_node;
        struct regmap *regmap;
+       struct property *prop;
+       int major, minor, ret;
+       char *name, compatible[32];

        regmap = devm_regmap_init_spmi_ext(sdev, 
&spmi_regmap_config);
        if (IS_ERR(regmap))
                return PTR_ERR(regmap);

+       ret = pmic_spmi_read_revid(regmap, &name, &major, &minor);
+       if (!ret) {
+               snprintf(compatible, ARRAY_SIZE(compatible), 
"qcom,%s-v%d.%d",
+                        name, major, minor);
+               prop = kzalloc(sizeof(*prop), GFP_KERNEL);
+               if (prop) {
+                       prop->name = kstrdup("compatible", 
GFP_KERNEL);
+                       prop->value = kstrdup(compatible, 
GFP_KERNEL);
+                       prop->length = strlen(prop->value);
+                       of_update_property(root, prop);
+               }
+       }
+

Why would you do this?
What benefit does it give to patch the of_node to have a more 
specific
compatible?

Some of the child device drivers have to know PMIC chip revision.

Regards,
Ivan