Re: [RFC PATCH] audio: proper support for float samples in mixeng

[Peter Maydell]

On Mon, 3 Feb 2020 at 20:38, Zoltán Kővágó <address@hidden> wrote:
>
> On 2020-02-03 11:00, Peter Maydell wrote:
> > On Sun, 2 Feb 2020 at 19:39, Kővágó, Zoltán <address@hidden> wrote:
> >>
> >> This adds proper support for float samples in mixeng by adding a new
> >> audio format for it.
> >>
> >> Limitations: only native endianness is supported.
> >
> > Could you explain a bit more what this limitation means, please?
> > In general QEMU behaviour shouldn't depend on the endianness
> > of the host, ie we should byteswap where necessary.
>
> None of the virtual sound cards support float samples (it looks like
> most of them only support 8 and 16 bit, only hda supports 32 bit), it is
> only used for the audio backends (i.e. host side).  In
> audiodev_to_audsettings we set endianness to AUDIO_HOST_ENDIANNESS, so
> audio backends should always use native endian.
>
> So this limitation should only cause problems when an audio backend
> overrides the endian setting.  Wavcapture does it, but it does not
> support float.  Alsa, sdl, puleaudio and oss can also do it if for some
> weird reason it acquires a stream with a different endianness than
> requested.

Ah, right, I had missed that this is only used by backends; makes
sense not to worry about non-native endianness then. If you
do a respin of the patch you might add some of that into the commit
message as an explanation of why the limitation isn't a significant one.

thanks
-- PMM