Re: [PATCH 2/2] PCI: rockchip: Simplify optional regulator handling

On Mon, Nov 18, 2019 at 08:24:28AM -0600, Bjorn Helgaas wrote:

On Mon, Nov 18, 2019 at 11:59:30AM +0000, Mark Brown wrote:

quoted

On Sat, Nov 16, 2019 at 12:54:20PM +0000, Robin Murphy wrote:

quoted

Null checks are both cheaper and more readable than having !IS_ERR()
splattered everywhere.

quoted

-	if (IS_ERR(rockchip->vpcie3v3))
+	if (!rockchip->vpcie3v3)
 		return;
 
 	/*

@@ -611,6 +611,7 @@ static int rockchip_pcie_parse_host_dt(struct rockchip_pcie *rockchip)
 		if (PTR_ERR(rockchip->vpcie12v) != -ENODEV)
 			return PTR_ERR(rockchip->vpcie12v);
 		dev_info(dev, "no vpcie12v regulator found\n");
+		rockchip->vpcie12v = NULL;

According to the API NULL is a valid regulator.  We don't currently
actually do this but it's storing up surprises if you treat it as
invalid.

I don't know anything about the regulator API, but the fact that NULL
can be a valid regulator is itself a little surprising :)

if (rockchip->vpcie3v3 == NULL) is true the driver would currently
panic the kernel AFAICS.

rockchip_pcie_set_power_limit()
->regulator_get_current_limit(NULL)
 -> _regulator_get_current_limit(NULL)
   -> regulator_lock(NULL)
     -> regulator_lock_nested(NULL, NULL)
       -> ww_mutex_trylock(&NULL->mutex)

Also, by making NULL a valid regulator, it means that regulators
(ie pointers) with default values are valid whether we call
devm_regulator_get* or not. I understand this patch can be dropped
but that per-se does not make this driver code any more robust AFAICS.

Lorenzo

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel