On 30/09/2020 05:37, Weiyi Lu wrote:
On Mon, 2020-09-28 at 15:14 +0800, Nicolas Boichat wrote:
On Thu, Aug 6, 2020 at 5:22 PM Weiyi Lu <weiyi.lu@xxxxxxxxxxxx> wrote:
Try to list all the power domains of under power controller
node to show the dependency between each power domain directly
instead of filling the dependency in scp_soc_data.
And could be more clearly to group subsys clocks into power domain
sub node to introduce subsys clocks of bus protection in next patch.
Signed-off-by: Weiyi Lu <weiyi.lu@xxxxxxxxxxxx>
---
[snip]
+static int traverse_scp(struct platform_device *pdev, struct scp *scp,
+ const struct scp_domain_data *scp_domain_data)
+{
+ struct device *dev = &pdev->dev;
+ struct device_node *np = dev->of_node;
+ struct device_node *sub;
+ int ret;
+
+ INIT_LIST_HEAD(&scp->dep_links);
+
+ for_each_available_child_of_node(np, sub) {
+ ret = scpsys_get_domain(pdev, scp, sub, scp_domain_data);
+ if (ret) {
+ dev_err(&pdev->dev, "failed to handle node %pOFn: %d\n", sub, ret);
minor comment: this error should not be printed if ret ==
-EPROBE_DEFER (use the new dev_err_probe?)
You're right! I'll use dev_err_probe() instead if anyone is interested
in this series. Thank you!
I'd propose that we put all our effort of reviewing and testing into the series
Enric send:
https://lore.kernel.org/linux-mediatek/20200910172826.3074357-1-enric.balletbo@xxxxxxxxxxxxx/
Regards,
Matthias
+ goto err;
+ }
+ }
+
+ return 0;
+
+err:
+ of_node_put(sub);
+ return ret;
+}
[snip]