Re: [PATCH net-next v2 3/6] tools: ynl: Add array-nest attr decoding to ynl

[Donald Hunter <donald.hunter@gmail.com>]

Jakub Kicinski <kuba@kernel.org> writes:

> On Sun, 19 Mar 2023 19:38:00 +0000 Donald Hunter wrote:
>> Add support for decoding nested arrays of scalars in netlink messages.
>
> example?

OVS_VPORT_ATTR_UPCALL_PID is a C array of u32 values. I can provide that
as an example in the commit message.

>> Signed-off-by: Donald Hunter <donald.hunter@gmail.com>
>> ---
>>  tools/net/ynl/lib/ynl.py | 6 ++++++
>>  1 file changed, 6 insertions(+)
>> 
>> diff --git a/tools/net/ynl/lib/ynl.py b/tools/net/ynl/lib/ynl.py
>> index 32536e1f9064..077ba9e8dc98 100644
>> --- a/tools/net/ynl/lib/ynl.py
>> +++ b/tools/net/ynl/lib/ynl.py
>> @@ -93,6 +93,10 @@ class NlAttr:
>>      def as_bin(self):
>>          return self.raw
>>  
>> +    def as_array(self, type):
>> +        format, _ = self.type_formats[type]
>> +        return list({ x[0] for x in struct.iter_unpack(format, self.raw) })
>
> So in terms of C this treats the payload of the attr as a packed array?
> That's not what array-nest is, array-nest wraps every entry in another
> nlattr:
> https://docs.kernel.org/next/userspace-api/netlink/genetlink-legacy.html#array-nest
>
> It's not a C array dumped into an attribute.
>
> IIRC I was intending to use 'binary' for packed arrays. Still use
> sub-type to carry the type, but main type should be 'binary'.
>
> If that sounds reasonable could you document or remind me to document
> this as the expected behavior? Sub-type appears completely undocumented
> now :S

That sounds reasonable, yes. I will also rename the method to
'as_c_array'. I think it should just be restricted to scalar subtypes,
i.e. u16, u32, etc. Do you agree?

I will update the  documentation for this.