Re: pcm_meter.c issue at s16_update

[Jaroslav Kysela <perex@perex.cz>]

Dne 03. 08. 20 v 8:17 Takashi Iwai napsal(a):
> On Sun, 02 Aug 2020 19:50:44 +0200,
> Pavel Hofman wrote:
>>
>>
>> Dne 28. 07. 20 v 20:54 Pavel Hofman napsal(a):
>>>
>>> Dne 28. 07. 20 v 20:04 Pavel Hofman napsal(a):
>>>> Dne 28. 07. 20 v 19:04 Takashi Iwai napsal(a):
>>>>> Would adding atomic_add(&meter->reset, 1) in snd_pcm_meter_reset()
>>>>> help?
>>>>>
>>>> Unfortunately not.
>>>>
>>>> s16_reset is called correctly, setting s16->old = meter->now;  But at
>>>> that time meter->now is still 22751, setting s16->old to the same value.
>>>>
>>>> s16_update 1: meter->now 22751, s16->old 22751, size 0
>>>>
>>>> However, in the next update call meter->now comes from the freshly
>>>> started application pointer:
>>>>
>>>> s16_update 1: meter->now 839, s16->old 22751, size -21912
>>>>
>>>>
>>>> Of course this helps:
>>>>
>>>> -       if (size < 0)
>>>> -               size += spcm->boundary;
>>>> +       if (size < 0) {
>>>> +               size = meter->now;
>>>> +               s16->old = 0;
>>>> +       }
>>>>
>>>> But I understand this is not a solution because:
>>>>
>>>> * it will not work at reaching spcm->boundary (after thousands of hours?)
>>>> * it will cause the same problem when the stream is rewound (which is
>>>> the problem now too) - size will equal to large meter->now (length from
>>>> the beginning of the stream minus the rewound = large number).
>>>>
>>>
>>> IMHO there are two cases of the [application pointer + delay] drop
>>> compared to the previous run:
>>>
>>> * stream start, rewinding => s16->old = meter->now; size =0, i.e.
>>> skipping the samples to show
>>> * wrapping at spcm->boundary => size += spcm->boundary, i.e. showing the
>>> wrapped samples
>>>
>>> Optionally the second case could be handled just like the first case by
>>> resetting s16->old, assuming the boundary wrap occurs very infrequently.
>>
>> The following patch is tested to work OK, no CPU peaks and no meter
>> output glitches when the size < 0 condition occurs:
>>
>> diff --git a/src/pcm/pcm_meter.c b/src/pcm/pcm_meter.c
>> index 20b41876..48df5945 100644
>> --- a/src/pcm/pcm_meter.c
>> +++ b/src/pcm/pcm_meter.c
>> @@ -1098,8 +1098,15 @@ static void s16_update(snd_pcm_scope_t *scope)
>>         snd_pcm_sframes_t size;
>>         snd_pcm_uframes_t offset;
>>         size = meter->now - s16->old;
>> -       if (size < 0)
>> -               size += spcm->boundary;
>> +       if (size < 0) {
>> +               /**
>> +                * Application pointer adjusted for delay (meter->now)
>> has dropped compared
>> +                * to the previous update cycle. Either spcm->boundary
>> wraparound, pcm rewinding,
>> +                * or pcm restart without s16->old properly reset.
>> +                * In any case the safest solution is skipping this
>> conversion cycle.
>> +                */
>> +               size = 0;
>> +       }
>>         offset = s16->old % meter->buf_size;
>>         while (size > 0) {
>>                 snd_pcm_uframes_t frames = size;
>>
>>
>>
>> Please will you accept this (workaround) bugfix? If so, I would send a
>> proper patch.
> 
> It looks OK, at least this must be safe.
> So yes, I'll happily apply if you submit a proper patch.

It would be probably better to check against the boundary / 2 value to check
correctly the boundary wrap instead to drop all negative size values:

  if (size < 0) {
     if (size < -(spcm->boundary / 2))
        size += spcm->boundary;
     else
        size = 0;
  }

The "hidden" pcm restart referred in the comment should not occur, otherwise
it's another bug somewhere.

					Jaroslav

-- 
Jaroslav Kysela <perex@perex.cz>
Linux Sound Maintainer; ALSA Project; Red Hat, Inc.