Re: [PATCH v4 08/23] ata: libahci_platform: Sanity check the DT child nodes number

[Damien Le Moal <damien.lemoal@opensource.wdc.com>]

On 6/18/22 05:18, Serge Semin wrote:
> On Thu, Jun 16, 2022 at 09:25:48AM +0900, Damien Le Moal wrote:
>> On 2022/06/16 5:53, Serge Semin wrote:
>>> On Tue, Jun 14, 2022 at 05:23:33PM +0900, Damien Le Moal wrote:
>>>> On 6/10/22 17:17, Serge Semin wrote:
>>>>> Having greater than AHCI_MAX_PORTS (32) ports detected isn't that critical
>>>>> from the further AHCI-platform initialization point of view since
>>>>> exceeding the ports upper limit will cause allocating more resources than
>>>>> will be used afterwards. But detecting too many child DT-nodes doesn't
>>>>> seem right since it's very unlikely to have it on an ordinary platform. In
>>>>> accordance with the AHCI specification there can't be more than 32 ports
>>>>> implemented at least due to having the CAP.NP field of 5 bits wide and the
>>>>> PI register of dword size. Thus if such situation is found the DTB must
>>>>> have been corrupted and the data read from it shouldn't be reliable. Let's
>>>>> consider that as an erroneous situation and halt further resources
>>>>> allocation.
>>>>>
>>>>> Note it's logically more correct to have the nports set only after the
>>>>> initialization value is checked for being sane. So while at it let's make
>>>>> sure nports is assigned with a correct value.
>>>>>
>>>>> Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
>>>>> Reviewed-by: Hannes Reinecke <hare@suse.de>
>>>>>
>>>>> ---
>>>>>
>>>>> Changelog v2:
>>>>> - Drop the else word from the child_nodes value checking if-else-if
>>>>>   statement (@Damien) and convert the after-else part into the ternary
>>>>>   operator-based statement.
>>>>>
>>>>> Changelog v4:
>>>>> - Fix some logical mistakes in the patch log. (@Sergei Shtylyov)
>>>>> ---
>>>>>  drivers/ata/libahci_platform.c | 13 ++++++++++---
>>>>>  1 file changed, 10 insertions(+), 3 deletions(-)
>>>>>
>>>>> diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
>>>>> index 814804582d1d..8aed7b29c7ab 100644
>>>>> --- a/drivers/ata/libahci_platform.c
>>>>> +++ b/drivers/ata/libahci_platform.c
>>>>> @@ -451,15 +451,22 @@ struct ahci_host_priv *ahci_platform_get_resources(struct platform_device *pdev,
>>>>>  		}
>>>>>  	}
>>>>>  
>>>>> -	hpriv->nports = child_nodes = of_get_child_count(dev->of_node);
>>>>> +	/*
>>>>> +	 * Too many sub-nodes most likely means having something wrong with
>>>>> +	 * the firmware.
>>>>> +	 */
>>>>> +	child_nodes = of_get_child_count(dev->of_node);
>>>>> +	if (child_nodes > AHCI_MAX_PORTS) {
>>>>> +		rc = -EINVAL;
>>>>> +		goto err_out;
>>>>> +	}
>>>>>  
>>>>>  	/*
>>>>>  	 * If no sub-node was found, we still need to set nports to
>>>>>  	 * one in order to be able to use the
>>>>>  	 * ahci_platform_[en|dis]able_[phys|regulators] functions.
>>>>>  	 */
>>>>> -	if (!child_nodes)
>>>>> -		hpriv->nports = 1;
>>>>> +	hpriv->nports = child_nodes ?: 1;
>>>>
>>>
>>>> This change is not necessary and makes the code far less easy to read.
>>>
>>> elaborate please. What change? What part of this change makes the code
>>> less easy to read?
>>
> 
>> You changed:
>>
>> 	if (!child_nodes)
>> 		hpriv->nports = 1;
>>
>> to:
>>
>> 	hpriv->nports = child_nodes ?: 1;
>>
>> That is the same. So the change is not needed in the first place, and worse,
>> makes the code way harder to read for no good reason.
> 
> No, they aren't the same:
> +	if (!child_nodes)
> +		hpriv->nports = 1;
> and
> +	hpriv->nports = child_nodes ?: 1;
> aren't equivalent. The equivalent implementation would be:
> +	if (child_nodes)
> +		hpriv->nports = child_nodes;
> +	else
> +		hpriv->nports = 1;

Then use this code. That cryptic C code is hard to read.

> 
> As I said in the patchlog, hpriv->nports is updated now only if
> of_get_child_count() returns a valid number of the child nodes,
> ports, which semantically is more correct. In the previous
> implementation it was always set to the number of child nodes
> no matter whether that value was correct or not.
> 
> Regarding the ternary operator with omitted operand. Well, it's not
> that rare beast in the kernel:
> $ grep -r "?:" kernel/ drivers/ mm/ fs/ block/ | wc -l
> 699
> But if you insist in it being not that readable, I can replace it with
> more bulky if-else statement. Do you?

Yes please, use the spelled out if/else. I prefer easy to read code rather
than loosing time trying to understand that cryptic C syntax, which  I
actually did not know about.

> 
> -Sergey
> 
>>
>>>
>>> -Sergey
>>>
>>>>
>>>>>  
>>>>>  	hpriv->phys = devm_kcalloc(dev, hpriv->nports, sizeof(*hpriv->phys), GFP_KERNEL);
>>>>>  	if (!hpriv->phys) {
>>>>
>>>>
>>>> -- 
>>>> Damien Le Moal
>>>> Western Digital Research
>>
>>
>> -- 
>> Damien Le Moal
>> Western Digital Research


-- 
Damien Le Moal
Western Digital Research