QMI via MBIM for "FCC Authentication" fix

[Ljungdahl, Emil]

2016-04-06 11:21 skrev Aleksander Morgado:
> On Mon, Apr 4, 2016 at 7:25 PM, Emil Ljungdahl <emil at kratern.se> wrote:
>>>>>> How far did you get with that Collin?
>>>>> 
>>>>> I believe the posted patches were functional, but there were some
>>>>> outstanding questions on how to properly integrate it in
>>>>> MM/libmbim/libqmi.  Is that correct, Aleksander?
>>>> 
>>>> Yes, that's right. Anyway, if someone is up to polishing those 
>>>> patches
>>>> and integrating them in libmbim sources directly, even if not 
>>>> perfect,
>>>> I think it would be good not to forget about them. I kind of already
>>>> forgot about them myself :)
>>> 
>>> This isn't really polishing... Rather the contrary ;)
>>> 
>>> The attached patch set is a sort of working proof-of-concept.  It 
>>> needs
>>> to be cleaned up by someone understanding all the G-stuff :) And it
>>> should have a lot more error checking.  Testing for EXT_QMUX support
>>> instead of timing out would be nice for example.  And the library
>>> integration could be made nicer.  Maybe dynamically loading libmbim 
>>> from
>>> libqmi to prevent the hard dependency?
>>> 
>>> The libmbim patch is Collin's latest version with only context 
>>> changes.
>>> 
>>> The remaining two libqmi patches provide generic QMI access over 
>>> MBIM.
>>> FCC auth is only one possibility.  Many other QMI requests work too:
>>> 
>> 
>> Here is a somewhat polished version of Colins patch + a patch for 
>> mbimcli.
>> The polish includes some indentation fixes and a new json-parameter in
>> the codegen stuff to avoid changed behavior for other services using
>> "unsized-byte-array".
> 
> Hey Emil,
> 
> Looks like you polished Collin's patch in a similar way as Bjørn did,
> right? Or is there any other thing that you added and he didn't?
> Please check the qmi-over-mbim branch and see if you'd improve
> anything there.
> 
> As for the mbimcli patch, I believe Bjørn's approach of using qmi over
> mbim is the right thing to do, so that we can directly use qmicli over
> a mbim device/proxy.

Well, for a starter, it actually compiles :P
(I think Bjørn missed the json-file Collin sent in a later mail)

But, jokes aside, I also changed so that the padding behavior of calls 
to *_builder_append_byte_array() was not changed for other services than 
the new QMI service. I have no idea if it really matters, but I noticed 
there was a difference...

I have attached 2 patches based on the qmi-over-mbim branch.

/Emil
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-Added-json-file-lost-in-original-qmi-over-mbim-patch.patch
Type: text/x-diff
Size: 1102 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/libmbim-devel/attachments/20160406/537f86ed/attachment.patch>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0003-Avoid-affecting-the-padding-behavior-for-other-servi.patch
Type: text/x-diff
Size: 10802 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/libmbim-devel/attachments/20160406/537f86ed/attachment-0001.patch>