Re: [PATCH 6/6] net: mvpp2: use device_for_each_child_node() to access device child nodes

[Javier Carrasco]

On 29/07/2024 10:23, Russell King (Oracle) wrote:
> On Sat, Jul 06, 2024 at 05:23:38PM +0200, Javier Carrasco wrote:
>> The iterated nodes are direct children of the device node, and the
>> `device_for_each_child_node()` macro accounts for child node
>> availability.
>>
>> `fwnode_for_each_available_child_node()` is meant to access the child
>> nodes of an fwnode, and therefore not direct child nodes of the device
>> node.
>>
>> The child nodes within mvpp2_probe are not accessed outside the lopps,
> 
> "lopps" ?
> 
>> and the socped version of the macro can be used to automatically
> 
> "socped" ?
> 

I'll fix the typos for v3.

>> decrement the refcount on early exits.
>>
>> Use `device_for_each_child_node()` and its scoped variant to indicate
>> device's direct child nodes.
>>
>> Signed-off-by: Javier Carrasco <javier.carrasco.cruz@xxxxxxxxx>
>> ---
>>  drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c | 13 ++++---------
>>  1 file changed, 4 insertions(+), 9 deletions(-)
>>
>> diff --git a/drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c b/drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c
>> index 9adf4301c9b1..97f1faab6f28 100644
>> --- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c
>> +++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c
>> @@ -7417,8 +7417,6 @@ static int mvpp2_get_sram(struct platform_device *pdev,
>>  
>>  static int mvpp2_probe(struct platform_device *pdev)
>>  {
>> -	struct fwnode_handle *fwnode = pdev->dev.fwnode;
>> -	struct fwnode_handle *port_fwnode;
>>  	struct mvpp2 *priv;
>>  	struct resource *res;
>>  	void __iomem *base;
>> @@ -7591,7 +7589,7 @@ static int mvpp2_probe(struct platform_device *pdev)
>>  	}
>>  
>>  	/* Map DTS-active ports. Should be done before FIFO mvpp2_init */
>> -	fwnode_for_each_available_child_node(fwnode, port_fwnode) {
>> +	device_for_each_child_node_scoped(&pdev->dev, port_fwnode) {
>>  		if (!fwnode_property_read_u32(port_fwnode, "port-id", &i))
>>  			priv->port_map |= BIT(i);
>>  	}
>> @@ -7614,7 +7612,7 @@ static int mvpp2_probe(struct platform_device *pdev)
>>  		goto err_axi_clk;
>>  
>>  	/* Initialize ports */
>> -	fwnode_for_each_available_child_node(fwnode, port_fwnode) {
>> +	device_for_each_child_node_scoped(&pdev->dev, port_fwnode) {
>>  		err = mvpp2_port_probe(pdev, port_fwnode, priv);
>>  		if (err < 0)
>>  			goto err_port_probe;
>> @@ -7653,10 +7651,8 @@ static int mvpp2_probe(struct platform_device *pdev)
>>  	return 0;
>>  
>>  err_port_probe:
>> -	fwnode_handle_put(port_fwnode);
>> -
>>  	i = 0;
>> -	fwnode_for_each_available_child_node(fwnode, port_fwnode) {
>> +	device_for_each_child_node_scoped(&pdev->dev, port_fwnode) {
>>  		if (priv->port_list[i])
>>  			mvpp2_port_remove(priv->port_list[i]);
>>  		i++;
>> @@ -7677,13 +7673,12 @@ static int mvpp2_probe(struct platform_device *pdev)
>>  static void mvpp2_remove(struct platform_device *pdev)
>>  {
>>  	struct mvpp2 *priv = platform_get_drvdata(pdev);
>> -	struct fwnode_handle *fwnode = pdev->dev.fwnode;
>>  	int i = 0, poolnum = MVPP2_BM_POOLS_NUM;
>>  	struct fwnode_handle *port_fwnode;
>>  
>>  	mvpp2_dbgfs_cleanup(priv);
>>  
>> -	fwnode_for_each_available_child_node(fwnode, port_fwnode) {
>> +	device_for_each_child_node(&pdev->dev, port_fwnode) {
>>  		if (priv->port_list[i]) {
>>  			mutex_destroy(&priv->port_list[i]->gather_stats_lock);
>>  			mvpp2_port_remove(priv->port_list[i]);
> 
> This loop is just silly. There is no need to iterate the child nodes.
> port_fwnode is not used, and the loop boils down to:
> 
> 	for (i = 0; i < priv->port_count; i++) {
> 		mutex_destroy(&priv->port_list[i]->gather_stats_lock);
> 		mvpp2_port_remove(priv->port_list[i]);
> 	}
> 
> Not only is walking the child nodes not necessary, but checking whether
> the pointer is NULL is also unnecessary. mvpp2_port_probe() populates
> the array using:
> 
>         priv->port_list[priv->port_count++] = port;
> 
> and "port" can not be NULL here, so we're guaranteed that all port_list
> entries for 0..priv->port_count will be non-NULL, and the driver makes
> this assumption in multiple places.
> 
> In fact, I'd say that using fwnode_for_each_available_child_node() or
> device_for_each_child_node() is buggy here if the availability of the
> children change - it could leave ports not cleaned up.
> 

I will add your suggestions in a separate patch with the corresponding
Suggested-by: tag. In that case, and taking into account that the
pointer check is unnecessary, the loop after a goto err_port_probe will
turn into this:

err_port_probe:
	for (i = 0; i < priv->port_count; i++)
		mvpp2_port_remove(priv->port_list[i]);

and the loop in mvpp2_remove() will be exactly the one you suggested.

Apart from that, there is a suspicious check towards the end of the same
function:

 if (is_acpi_node(port_fwnode))
		return;

At the point it is called in the current implementation, port_fwnode
could have been cleaned. And after removing the loop, it is simply
uninitialized. Was that meant to be pdev->dev->fwnode?

Thanks and best regards,
Javier Carrasco