Re: [PATCH v3 12/14] intel_gna: add a 'misc' device

[Maciej Kwapulinski <maciej.kwapulinski@linux.intel.com>]

Greg Kroah-Hartman <gregkh@linuxfoundation.org> writes:

> On Thu, May 13, 2021 at 01:00:38PM +0200, Maciej Kwapulinski wrote:
>> The new 'misc' device is the node for applications in user space to
>> interact with the driver.
>> 
>> Signed-off-by: Maciej Kwapulinski <maciej.kwapulinski@linux.intel.com>
>> Tested-by: Savo Novakovic <savox.novakovic@intel.com>
>> ---
>>  drivers/misc/intel/gna/device.c | 52 +++++++++++++++++++++++++++++++--
>>  drivers/misc/intel/gna/device.h | 11 +++----
>>  2 files changed, 55 insertions(+), 8 deletions(-)
>> 
>> diff --git a/drivers/misc/intel/gna/device.c b/drivers/misc/intel/gna/device.c
>> index 0e31b8c6bb70..1e6345a8325b 100644
>> --- a/drivers/misc/intel/gna/device.c
>> +++ b/drivers/misc/intel/gna/device.c
>> @@ -20,6 +20,18 @@ module_param(recovery_timeout, int, 0644);
>>  MODULE_PARM_DESC(recovery_timeout, "Recovery timeout in seconds");
>>  #endif
>>  
>> +struct file;
>> +
>> +static int gna_open(struct inode *inode, struct file *f)
>> +{
>> +	return -EPERM;
>> +}
>
> That sucks, why have an open that does nothing but fail?

next patch provides complete implementation of gna_open(), here it's
just a protection if someone would incidentally run gna in the middle of patch series

>
>> +
>> +static const struct file_operations gna_file_ops = {
>> +	.owner		=	THIS_MODULE,
>> +	.open		=	gna_open,
>> +};
>> +
>>  static void gna_devm_idr_destroy(void *data)
>>  {
>>  	struct idr *idr = data;
>> @@ -27,6 +39,36 @@ static void gna_devm_idr_destroy(void *data)
>>  	idr_destroy(idr);
>>  }
>>  
>> +static void gna_devm_deregister_misc_dev(void *data)
>
> Why is this a void *?

it goes to devm_add_action() api.

>
> This isn't windows, use real pointer types everywhere in the kernel
> please.
>
>> +{
>> +	struct miscdevice *misc = data;
>> +
>> +	misc_deregister(misc);
>> +}
>> +
>> +static int gna_devm_register_misc_dev(struct device *parent, struct miscdevice *misc)
>> +{
>> +	int ret;
>> +
>> +	ret = misc_register(misc);
>> +	if (ret) {
>> +		dev_err(parent, "misc device %s registering failed. errcode: %d\n",
>> +			misc->name, ret);
>> +		gna_devm_deregister_misc_dev(misc);
>> +	} else {
>> +		dev_dbg(parent, "device: %s registered\n",
>> +			misc->name);
>
> You have loads of debugging in this driver still, is it really needed?
>
>> +	}
>> +
>> +	ret = devm_add_action(parent, gna_devm_deregister_misc_dev, misc);
>
> Why do you need this?

I'd like to avoid having gna_probe's fail path at all.

>
>
> thanks,
>
> greg k-h