Re: [Intel-wired-lan] [PATCH net-next v7 1/8] net: ethtool: pass a pointer to parameters to get/set_rxfh ethtool ops

[Ahmed Zaki <ahmed.zaki@intel.com>]

On 2023-12-06 05:57, Alexander Lobakin wrote:
> From: Ahmed Zaki <ahmed.zaki@intel.com>
> Date: Tue,  5 Dec 2023 16:00:42 -0700
> 
> Duplicating my comment from the internal review:
> 
>> The get/set_rxfh ethtool ops currently takes the rxfh (RSS) parameters
>> as direct function arguments. This will force us to change the API (and
>> all drivers' functions) every time some new parameters are added.
> 
> [...]
> 
>> diff --git a/include/uapi/linux/ethtool.h b/include/uapi/linux/ethtool.h
>> index f7fba0dc87e5..f6e229e465b1 100644
>> --- a/include/uapi/linux/ethtool.h
>> +++ b/include/uapi/linux/ethtool.h
>> @@ -1229,6 +1229,27 @@ struct ethtool_rxnfc {
>>   	__u32				rule_locs[];
>>   };
>>   
>> +/**
>> + * struct ethtool_rxfh_param - RXFH (RSS) parameters
>> + * @hfunc: Defines the current RSS hash function used by HW (or to be set to).
>> + *	Valid values are one of the %ETH_RSS_HASH_*.
>> + * @indir_size: On SET, the array size of the user buffer for the
>> + *	indirection table, which may be zero, or
>> + *	%ETH_RXFH_INDIR_NO_CHANGE.  On GET (read from the driver),
>> + *	the array size of the hardware indirection table.
>> + * @indir: The indirection table of size @indir_size entries.
>> + * @key_size: On SET, the array size of the user buffer for the hash key,
>> + *	which may be zero.  On GET (read from the driver), the size of the
>> + *	hardware hash key.
>> + * @key: The hash key of size @key_size bytes.
>> + */
>> +struct ethtool_rxfh_param {
>> +	__u8	hfunc;
>> +	__u32   indir_size;
>> +	__u32	*indir;
>> +	__u32   key_size;
>> +	__u8	*key;
>> +};
> 1. Why is this structure needed in UAPI? Do you plan to use it somewhere
>     in userspace?
> 2. Kernel and userspace can't share pointers (as well as unsigned longs,
>     size_ts, and so on) as you may run a 32-bit application on a 64-bit
>     kernel.
> 3. Please never pass UAPI structures directly to the drivers, it's a bad
>     idea and you may end up converting all those drivers once again when
>     you'd need to to e.g. change the type of a field there. You won't be
>     able to change the type in a UAPI structure.
> 
> Thanks,
> Olek

You are right, it is not needed or planned to be used in uAPI, I will 
move this to include/linux/ethtool.h

Thanks.
_______________________________________________
Intel-wired-lan mailing list
Intel-wired-lan@osuosl.org
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan