Re: Re: [PATCH] pci: controller: mobiveil: Hold reference returned by of_parse_phandle()

[Bjorn Helgaas <helgaas@kernel.org>]

On Fri, Jul 15, 2022 at 08:06:46AM +0800, Liang He wrote:
> At 2022-07-15 00:37:51, "Bjorn Helgaas" <helgaas@kernel.org> wrote:
> >On Mon, Jul 04, 2022 at 02:26:08PM +0800, Liang He wrote:
> >> In ls_g4_pcie_probe(), we should hold the reference returned by
> >> of_parse_phandle() and use it to call of_node_put() for refcount
> >> balance.
> >> 
> >> Fixes: d29ad70a813b ("PCI: mobiveil: Add PCIe Gen4 RC driver for Layerscape SoCs")
> >> Co-authored-by: Miaoqian Lin <linmq006@gmail.com>
> >> Signed-off-by: Liang He <windhl@126.com>
> >> ---
> >>  drivers/pci/controller/mobiveil/pcie-layerscape-gen4.c | 6 ++++--
> >>  1 file changed, 4 insertions(+), 2 deletions(-)
> >> 
> >> diff --git a/drivers/pci/controller/mobiveil/pcie-layerscape-gen4.c b/drivers/pci/controller/mobiveil/pcie-layerscape-gen4.c
> >> index d7b7350f02dd..075aa487f92e 100644
> >> --- a/drivers/pci/controller/mobiveil/pcie-layerscape-gen4.c
> >> +++ b/drivers/pci/controller/mobiveil/pcie-layerscape-gen4.c
> >> @@ -204,13 +204,15 @@ static int __init ls_g4_pcie_probe(struct platform_device *pdev)
> >>  	struct pci_host_bridge *bridge;
> >>  	struct mobiveil_pcie *mv_pci;
> >>  	struct ls_g4_pcie *pcie;
> >> -	struct device_node *np = dev->of_node;
> >> +	struct device_node *np = dev->of_node, *parse_np;
> >>  	int ret;
> >>  
> >> -	if (!of_parse_phandle(np, "msi-parent", 0)) {
> >> +	parse_np = of_parse_phandle(np, "msi-parent", 0);
> >
> >I don't understand what's going on here.  Where is "msi-parent"
> >actually used?  If we just need to know whether "msi-parent" exists,
> >can we use of_property_read_bool() instead?  Or can we call
> >of_parse_phandle() closer to where it is used?  
> 
> I think this code is used to check the existence of the device_node
> whose property is "msi-parent". of_property_read_bool() can only be
> used to check current device_node 'np'.
>
> If this is right, I think we can use a helper to make thing simple
> like this:
> 
> https://lore.kernel.org/all/20220712061417.363145-1-windhl@126.com/
> 
> If it is ok, I will prepare the helper in new version.

I dunno.  The helper in that patch has nothing ftgmac100-specific in
it, so if the helper is correct and useful, it seems like it should be
somewhere generic so we don't end up with a zillion copies of it.
Maybe Rob will chime in.

I just want drivers/pci/controller/* to deal with "msi-parent"
consistently.

dw_pcie_host_init() uses of_property_read_bool().

ls_g4_pcie_probe(), brcm_pcie_probe(), and iproc_pcie_msi_enable() use
of_parse_phandle().

It's not clear to me that these uses are fundamentally different and
need to use different interfaces.

> >> +	if (!parse_np) {
> >>  		dev_err(dev, "Failed to find msi-parent\n");
> >>  		return -EINVAL;
> >>  	}
> >> +	of_node_put(parse_np);
> >>  
> >>  	bridge = devm_pci_alloc_host_bridge(dev, sizeof(*pcie));
> >>  	if (!bridge)
> >> -- 
> >> 2.25.1
> >>