[alsa-devel] [PATCH] aplay: Fix header for raw format

[Takashi Iwai]

On Mon, 03 Jul 2017 11:41:44 +0200,
Daniel Baluta wrote:
> 
> On Fri, Jun 30, 2017 at 5:20 PM, Takashi Iwai <tiwai at suse.de> wrote:
> > On Fri, 30 Jun 2017 15:11:38 +0200,
> > Daniel Baluta wrote:
> >>
> >> Hi Takashi,
> >>
> >> Thanks a lot for your feedback.
> >>
> >> On Fri, Jun 30, 2017 at 9:54 AM, Takashi Iwai <tiwai at suse.de> wrote:
> >> > On Thu, 29 Jun 2017 13:17:16 +0200,
> >> > Ion-Horia Petrisor wrote:
> >> >>
> >> >> Raw format has no header, so use 0 when calling playback_go.
> >> >
> >> > It's the value passed for the length that has been already loaded.
> >> > The program has read dta bytes for parsing the header, and it's raw
> >> > data without header.  Thus you need to pass dta there.
> >> >
> >> With the current code we assume that the raw files have at least 26 bytes.
> >> (sizeof (VocHeader).
> >>
> >> I think it doesn't matter here that dta bytes has already been loaded. These
> >> bytes useful to figure out the file header type.
> >>
> >> To keep things simple we can assume that in the case of raw files there is no
> >> header, so second parameter of playback_go (named loaded) can be set to 0.
> >
> > The loaded parameter of playback_go() doesn't mean that.  It's the
> > size that has been already read onto the audio buffer.  If you pass 0
> > there, it means that you discard the first 26 bytes samples you have
> > already read without playing.
> 
> Agree. Passing 0 discards the first 26 bytes that have already been read
> but these bytes will be re-read inside playback_go so they will be played.
> 
> playback_go( .. loaded = 0)
> ..
> 
> r = safe_read(fd, audio_buf + loaded, c). [aplay.c: 2757]

No no, safe_read won't *re*-read.  It just reads the data after the
point you've already read without seeking back.  So the first 26 bytes
will be lost by your patch.


thanks,

Takashi