Re: [alsa-devel] [PATCH 1/2] ASoC: codec: wm9860: avoid maybe-uninitialized warning

[Daniel Baluta]

On Mon, Apr 24, 2017 at 6:27 PM, Arnd Bergmann <arnd@xxxxxxxx> wrote:
> On Mon, Apr 24, 2017 at 3:15 PM, Daniel Baluta <daniel.baluta@xxxxxxxxx> wrote:
>> On Fri, Apr 21, 2017 at 5:46 PM, Arnd Bergmann <arnd@xxxxxxxx> wrote:
>>> On Fri, Apr 21, 2017 at 3:07 PM, Daniel Baluta <daniel.baluta@xxxxxxx> wrote:
>>>> The new PLL configuration code triggers a harmless warning:
>>>>
>>>> sound/soc/codecs/wm8960.c: In function 'wm8960_configure_clocking':
>>>> sound/soc/codecs/wm8960.c:735:3: error: 'best_freq_out' may be used
>>>> uninitialized in this function [-Werror=maybe-uninitialized]
>>>>    wm8960_set_pll(codec, freq_in, best_freq_out);
>>>>    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>> sound/soc/codecs/wm8960.c:699:12: note: 'best_freq_out' was declared
>>>> here
>>>>
>>>> Fixes: 84fdc00d519f ("ASoC: codec: wm9860: Refactor PLL out freq search")
>>>> Fixes: 303e8954af8d ("ASoC: codec: wm8960: Stop when a matching PLL freq is found")
>>>> Suggested-by: Arnd Bergmann <arnd@xxxxxxxx>
>>>> Signed-off-by: Daniel Baluta <daniel.baluta@xxxxxxx>
>>>> ---
>>>> Arnd,
>>>>
>>>> I agree that your code was more both humans and gcc anyhow
>>>> for consistency with wm8960_configure_sysclk function I preferred
>>>> to keep the "if(..) break" statements.
>>>
>>> How about changing both functions the same way then?
>>
>> I've tried but I couldn't find any solution. For clarity here is how
>> the code actually looks like.
>>
>> The git diff is a little bit misleading. Here is how wm8960_configure_pll code
>> looks like:
>>
>> https://pastebin.com/naGdVNQz
>>
>> static
>> int wm8960_configure_pll(struct snd_soc_codec *codec, int freq_in,
>> Â       Â       Â        int *sysclk_idx, int *dac_idx, int *bclk_idx)
>> {
>> Â       struct wm8960_priv *wm8960 = snd_soc_codec_get_drvdata(codec);
>> Â       int sysclk, bclk, lrclk, freq_out;
>> Â       int diff, closest, best_freq_out;
>> Â       int i, j, k;
>>
>> Â       bclk = wm8960->bclk;
>> Â       lrclk = wm8960->lrclk;
>> Â       closest = freq_in;
>>
>> Â       best_freq_out = -EINVAL;
>> Â       *sysclk_idx = *dac_idx = *bclk_idx = -1;
>>
>> Â       for (i = 0; i < ARRAY_SIZE(sysclk_divs); ++i) {
>> Â       Â       if (sysclk_divs[i] == -1)
>> Â       Â       Â       continue;
>> Â       Â       for (j = 0; j < ARRAY_SIZE(dac_divs); ++j) {
>> Â       Â       Â       sysclk = lrclk * dac_divs[j];
>> Â       Â       Â       freq_out = sysclk * sysclk_divs[i];
>>
>> Â       Â       Â       for (k = 0; k < ARRAY_SIZE(bclk_divs); ++k) {
>> Â       Â       Â       Â       if (!is_pll_freq_available(freq_in, freq_out))
>> Â       Â       Â       Â       Â       continue;
>>
>> Â       Â       Â       Â       diff = sysclk - bclk * bclk_divs[k] / 10;
>> Â       Â       Â       Â       if (diff == 0) {
>> Â       Â       Â       Â       Â       *sysclk_idx = i;
>> Â       Â       Â       Â       Â       *dac_idx = j;
>> Â       Â       Â       Â       Â       *bclk_idx = k;
>> Â       Â       Â       Â       Â       best_freq_out = freq_out;
>> Â       Â       Â       Â       Â       break;
>> Â       Â       Â       Â       }
>> Â       Â       Â       Â       if (diff > 0 && closest > diff) {
>> Â       Â       Â       Â       Â       *sysclk_idx = i;
>> Â       Â       Â       Â       Â       *dac_idx = j;
>> Â       Â       Â       Â       Â       *bclk_idx = k;
>> Â       Â       Â       Â       Â       closest = diff;
>> Â       Â       Â       Â       Â       best_freq_out = freq_out;
>> Â       Â       Â       Â       }
>> Â       Â       Â       }
>> Â       Â       Â       if (k != ARRAY_SIZE(bclk_divs))
>> Â       Â       Â       Â       break;
>> Â       Â       }
>> Â       Â       if (j != ARRAY_SIZE(dac_divs))
>> Â       Â       Â       break;
>> Â       }
>>
>> Â       return best_freq_out;
>> }
>>
>> In my opinion this is a compiler false positive. Any clue on how to rework this
>> would be welcomed :). I couldn't find any decent solution.
>
> Actually I think in this case the compiler is supposed to warn if
> best_freq_out is not initialized, as we would never set it
> in case is_pll_freq_available() returns false for all inputs or
> sysclk_divs[] is -1 for all fields.
> I'd leave the initialization then, and only replace the breaks
> with a goto (not tested):
>
>> Â       for (i = 0; i < ARRAY_SIZE(sysclk_divs); ++i) {
>> Â       Â       if (sysclk_divs[i] == -1)
>> Â       Â       Â       continue;
>> Â       Â       for (j = 0; j < ARRAY_SIZE(dac_divs); ++j) {
>> Â       Â       Â       sysclk = lrclk * dac_divs[j];
>> Â       Â       Â       freq_out = sysclk * sysclk_divs[i];
>>
>> Â       Â       Â       for (k = 0; k < ARRAY_SIZE(bclk_divs); ++k) {
>> Â       Â       Â       Â       if (!is_pll_freq_available(freq_in, freq_out))
>> Â       Â       Â       Â       Â       continue;
>>
>> Â       Â       Â       Â       diff = sysclk - bclk * bclk_divs[k] / 10;
>> Â       Â       Â       Â       if (diff == 0) {
>> Â       Â       Â       Â       Â       *sysclk_idx = i;
>> Â       Â       Â       Â       Â       *dac_idx = j;
>> Â       Â       Â       Â       Â       *bclk_idx = k;
>> Â       Â       Â       Â       Â       best_freq_out = freq_out;
>> Â       Â       Â       Â       Â       goto out;
>> Â       Â       Â       Â       }
>> Â       Â       Â       Â       if (diff > 0 && closest > diff) {
>> Â       Â       Â       Â       Â       *sysclk_idx = i;
>> Â       Â       Â       Â       Â       *dac_idx = j;
>> Â       Â       Â       Â       Â       *bclk_idx = k;
>> Â       Â       Â       Â       Â       closest = diff;
>> Â       Â       Â       Â       Â       best_freq_out = freq_out;
>> Â       Â       Â       Â       }
>> Â       Â       Â       }
>> Â       Â       }
>> Â       }
>>out:
>> Â       return best_freq_out;
>> }

Sure, this looks reasonable. I will send v2.

Daniel.