Dan Carpenter <dan.carpenter@xxxxxxxxxx> writes:
> On Mon, Jun 22, 2020 at 05:51:16PM +0300, Kalle Valo wrote:
>> Dan Carpenter <dan.carpenter@xxxxxxxxxx> writes:
>>
>> > There is a success path where "ret" isn't initialized where we never
>> > have a ATH11K_SPECTRAL_TAG_SCAN_SEARCH and then ret isn't initialized.
>> >
>> > Fixes: 9d11b7bff950 ("ath11k: add support for spectral scan")
>> > Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
>> > ---
>> > drivers/net/wireless/ath/ath11k/spectral.c | 2 +-
>> > 1 file changed, 1 insertion(+), 1 deletion(-)
>> >
>> > diff --git a/drivers/net/wireless/ath/ath11k/spectral.c b/drivers/net/wireless/ath/ath11k/spectral.c
>> > index 1c5d65bb411f..bfbf905f7507 100644
>> > --- a/drivers/net/wireless/ath/ath11k/spectral.c
>> > +++ b/drivers/net/wireless/ath/ath11k/spectral.c
>> > @@ -677,7 +677,7 @@ static int ath11k_spectral_process_data(struct ath11k *ar,
>> > u32 data_len, i;
>> > u8 sign, tag;
>> > int tlv_len, sample_sz;
>> > - int ret;
>> > + int ret = 0;
>> > bool quit = false;
>>
>> I try to avoid initialising ret variables so I would like find another
>> way. What about doing this (completely untested!) in the end of the
>> function:
>>
>> return 0;
>>
>> err:
>> kfree(fft_sample);
>> unlock:
>> spin_unlock_bh(&ar->spectral.lock);
>> return ret;
>
> I normally avoid it as well... If I were to redo this patch, I would
> probably do:
>
> ret = 0;
> err:
> kfree(fft_sample);
> unlock:
> spin_unlock_bh(&ar->spectral.lock);
> return ret;
>
> Would that be better?
Ah, my proposal was buggy. So yes, this is definitely better :) I
changed now the patch in the pending branch, please check:
https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git/commit/?h=pending&id=586f8e6779fa75ade112cb48606794e034e0ff1d
--
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
