Re: [PATCH] cfg80211: clear cfg80211_inform_bss() from kmemleak reports

["Luis R. Rodriguez" <lrodriguez@atheros.com>]

On Thu, Sep 3, 2009 at 12:26 AM, Johannes Berg<johannes@sipsolutions.net> wrote:
> On Thu, 2009-09-03 at 02:11 -0400, Luis R. Rodriguez wrote:
>> This was giving false positives. We use eventually free this
>> through kref_put(), things are not so obvious through
>> cfg80211_bss_update().
>>
>> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
>> ---
>>  net/wireless/scan.c |    3 +++
>>  1 files changed, 3 insertions(+), 0 deletions(-)
>>
>> diff --git a/net/wireless/scan.c b/net/wireless/scan.c
>> index 19c5a9a..79f7a5d 100644
>> --- a/net/wireless/scan.c
>> +++ b/net/wireless/scan.c
>> @@ -495,6 +495,9 @@ cfg80211_inform_bss(struct wiphy *wiphy,
>>
>>       kref_init(&res->ref);
>>
>> +     /* cfg80211_bss_update() eats up res - we ensure we free it there */
>> +     kmemleak_ignore(res);
>> +
>>       res = cfg80211_bss_update(wiphy_to_dev(wiphy), res, 0);
>>       if (!res)
>>               return NULL;
>
> That's not making sense. cfg80211_bss_update() doesn't actually take a
> reference, it adds a new one for itself and then we return one to the
> caller.

What I meant is it gobbles it up and spits another thing out. When it
gobbles it up the routine then uses kref_put().

> Why can it not track this?

It probably can, just not sure if it follows kref_put(), I was under
the impression here it doesn't and because of it we were getting false
positives. Catalin, can you confirm?

> Actually it looks like we do leak one in net/mac80211/ibss.c.


  Luis