Re: [PATCH net-next v6 04/12] net: pse-pd: Add support for PSE power domains

On Mon, 17 Mar 2025 10:59:02 +0100
Oleksij Rempel [off-list ref] wrote:

On Tue, Mar 04, 2025 at 11:18:53AM +0100, Kory Maincent wrote:

quoted

+static int pse_register_pw_ds(struct pse_controller_dev *pcdev)
+{
+	int i;
+
+	for (i = 0; i < pcdev->nr_lines; i++) {
+		struct regulator_dev *rdev = pcdev->pi[i].rdev;
+		struct pse_power_domain *pw_d;
+		struct regulator *supply;
+		bool present = false;
+		unsigned long index;
+
+		/* No regulator or regulator parent supply registered.
+		 * We need a regulator parent to register a PSE power
domain
+		 */
+		if (!rdev || !rdev->supply)
+			continue;
+  

Should we use xa_lock() before iteration over the map?

I am scratching my head to find out the case where it may be needed.
The only case I think of is two PSE controller with PSE PI using the same power
supply. Which is not a good idea for well distributing power.
In this case we could have two pse_register_pw_ds running concurrently and we
could have issues. Not only the iteration over the map should be protected but
also the content of the pw_d entry. Also when we unbind one of the PSE
controller it will remove all the PSE power domains event if they are used
by another PSE. :/
I need to add refcount and lock on the PSE power domains to fix this case! 

quoted

+		xa_for_each(&pse_pw_d_map, index, pw_d) {
+			/* Power supply already registered as a PSE power
+			 * domain.
+			 */
+			if (regulator_is_equal(pw_d->supply,
rdev->supply)) {
+				present = true;
+				pcdev->pi[i].pw_d = pw_d;
+				break;
+			}
+		}
+		if (present)
+			continue;
+
+		pw_d = devm_pse_alloc_pw_d(pcdev->dev);
+		if (IS_ERR_OR_NULL(pw_d))
+			return PTR_ERR(pw_d);  

It is better to break the loop and roll back previous allocations.

This will be done by pse_flush_pw_ds called by devm_pse_controller_release as
the pse_controller_register failed.

Regards,
-- 
Köry Maincent, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com