[6/6] mtd: docg3: Don't do ERR_PTR(0)

Don't return a obfuscated null pointer using ERR_PTR(0).
If the no device is found clearly return -ENODEV.
This makes the code more clear and matches the comment
of doc_probe_device().

Signed-off-by: Richard Weinberger <richard@nod.at>
---
 drivers/mtd/devices/docg3.c | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)

On Mon, Jun 01, 2015 at 11:10:54PM +0200, Richard Weinberger wrote:
> Don't return a obfuscated null pointer using ERR_PTR(0).
> If the no device is found clearly return -ENODEV.
> This makes the code more clear and matches the comment
> of doc_probe_device().
> 
> Signed-off-by: Richard Weinberger <richard@nod.at>
> ---

Have you tested this patch?

>  drivers/mtd/devices/docg3.c | 16 ++++++----------
>  1 file changed, 6 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/mtd/devices/docg3.c b/drivers/mtd/devices/docg3.c
> index 5e67b4a..630e29a 100644
> --- a/drivers/mtd/devices/docg3.c
> +++ b/drivers/mtd/devices/docg3.c
> @@ -1902,7 +1902,7 @@ doc_probe_device(struct docg3_cascade *cascade, int floor, struct device *dev)
>  	chip_id = doc_register_readw(docg3, DOC_CHIPID);
>  	chip_id_inv = doc_register_readw(docg3, DOC_CHIPID_INV);
>  
> -	ret = 0;
> +	ret = -ENODEV;
>  	if (chip_id != (u16)(~chip_id_inv)) {
>  		goto nomem4;
>  	}
> @@ -2068,13 +2068,10 @@ static int __init docg3_probe(struct platform_device *pdev)
>  		mtd = doc_probe_device(cascade, floor, dev);
>  		if (IS_ERR(mtd)) {
>  			ret = PTR_ERR(mtd);
> -			goto err_probe;
> -		}
> -		if (!mtd) {
> -			if (floor == 0)
> -				goto notfound;
> -			else
> +			if (ret == -ENODEV && floor == 0)

I think you might have changed the logic when refactoring here. I think
the right refactoring would be the following, no?

			if (ret == -ENODEV && floor != 0)
				continue;
			else
				goto err_probe;

>  				continue;
> +			else
> +				goto err_probe;
>  		}
>  		cascade->floors[floor] = mtd;
>  		ret = mtd_device_parse_register(mtd, part_probes, NULL, NULL,
> @@ -2091,10 +2088,9 @@ static int __init docg3_probe(struct platform_device *pdev)
>  	doc_dbg_register(cascade->floors[0]->priv);
>  	return 0;
>  
> -notfound:
> -	ret = -ENODEV;
> -	dev_info(dev, "No supported DiskOnChip found\n");
>  err_probe:
> +	if (ret == -ENODEV)
> +		dev_info(dev, "No supported DiskOnChip found\n");
>  	free_bch(cascade->bch);
>  	for (floor = 0; floor < DOC_MAX_NBFLOORS; floor++)
>  		if (cascade->floors[floor])

Brian

Am 17.06.2015 um 20:41 schrieb Brian Norris:
> On Mon, Jun 01, 2015 at 11:10:54PM +0200, Richard Weinberger wrote:
>> Don't return a obfuscated null pointer using ERR_PTR(0).
>> If the no device is found clearly return -ENODEV.
>> This makes the code more clear and matches the comment
>> of doc_probe_device().
>>
>> Signed-off-by: Richard Weinberger <richard@nod.at>
>> ---
> 
> Have you tested this patch?

nah, I don't own such a device.

>>  drivers/mtd/devices/docg3.c | 16 ++++++----------
>>  1 file changed, 6 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/mtd/devices/docg3.c b/drivers/mtd/devices/docg3.c
>> index 5e67b4a..630e29a 100644
>> --- a/drivers/mtd/devices/docg3.c
>> +++ b/drivers/mtd/devices/docg3.c
>> @@ -1902,7 +1902,7 @@ doc_probe_device(struct docg3_cascade *cascade, int floor, struct device *dev)
>>  	chip_id = doc_register_readw(docg3, DOC_CHIPID);
>>  	chip_id_inv = doc_register_readw(docg3, DOC_CHIPID_INV);
>>  
>> -	ret = 0;
>> +	ret = -ENODEV;
>>  	if (chip_id != (u16)(~chip_id_inv)) {
>>  		goto nomem4;
>>  	}
>> @@ -2068,13 +2068,10 @@ static int __init docg3_probe(struct platform_device *pdev)
>>  		mtd = doc_probe_device(cascade, floor, dev);
>>  		if (IS_ERR(mtd)) {
>>  			ret = PTR_ERR(mtd);
>> -			goto err_probe;
>> -		}
>> -		if (!mtd) {
>> -			if (floor == 0)
>> -				goto notfound;
>> -			else
>> +			if (ret == -ENODEV && floor == 0)
> 
> I think you might have changed the logic when refactoring here. I think
> the right refactoring would be the following, no?
> 
> 			if (ret == -ENODEV && floor != 0)
> 				continue;
> 			else
> 				goto err_probe;
> 

You are right, the logic changed. ;-\
Please drop this patch, I'll resend it for 4.3.

Thanks,
//richard

Richard Weinberger <richard@nod.at> writes:

> Am 17.06.2015 um 20:41 schrieb Brian Norris:
>> On Mon, Jun 01, 2015 at 11:10:54PM +0200, Richard Weinberger wrote:
>>> Don't return a obfuscated null pointer using ERR_PTR(0).
>>> If the no device is found clearly return -ENODEV.
>>> This makes the code more clear and matches the comment
>>> of doc_probe_device().
>>>
>>> Signed-off-by: Richard Weinberger <richard@nod.at>
>>> ---
>> 
>> Have you tested this patch?
>
> nah, I don't own such a device.
But I do. If you resend a patch, please Cc me. You can even ask for a test from
time to time if you want a confirmation, I have a 2 floors docg3 device.

Cheers.

--
Robert

Hi Robert,

On Tue, Jun 23, 2015 at 10:41:33PM +0200, Robert Jarzmik wrote:
> Richard Weinberger <richard@nod.at> writes:
> 
> > Am 17.06.2015 um 20:41 schrieb Brian Norris:
> >> Have you tested this patch?
> >
> > nah, I don't own such a device.
> But I do. If you resend a patch, please Cc me. You can even ask for a test from
> time to time if you want a confirmation, I have a 2 floors docg3 device.

Do you want to be on a MAINTAINERS entry for this driver?

Brian