On 25-08-16 23:52, Denis Kenzior wrote:
> Hi Arend,
>
> On 08/25/2016 04:35 PM, Arend van Spriel wrote:
>> On 25-08-16 22:44, Denis Kenzior wrote:
>>> dump_wiphy_parse only assigns filter_wiphy if one of the supported
>>> NL80211 attributes is present. So for unfiltered dumps, filter_wiphy
>>> was always initialized to 0, and only interface 0 was dumped.
>>>
>>> This was introduced in commit 2d75da13fbb957e955d212555b91101cef36f0ce.
>>>
>>> Reported-by: Arend Van Spriel <[email protected]>
>>
>> Actually sent a patch for this issue a little earlier. I should have
>> Cc'ed you explicitly, I guess.
>>
>
> Whoops. I saw your regression report and didn't look to see if you
> fixed it :)
I wanted to avoid having to revert so I decided to take a closer look.
> Looking at your patch, I'm worried that
>
> - int filter_wiphy = -1;
> + int filter_wiphy;
>
> might still break things, since Johannes used:
>
> if (!cb->args[2]) {
> ...
Maybe my look was not close enough. In this branch cb->args[2] is set to
-1. So indeed that would make filter_wiphy uninitialized in subsequent
call(s). Thanks for catching that.
Regards,
Arend
> } else if (cb->args[2] > 0) {
> filter_wiphy = cb->args[2] - 1;
> }
>
> So for unfiltered dumps, filter_wiphy would not be initialized properly
> for the second & onward call of nl80211_dump_interface. Right?
>
> Regards,
> -Denis
