[alsa-devel] [PATCH v2 5/7] ALSA: hda - Use ops instead of directly accessing chmap helpers

[Takashi Iwai]

On Thu, 03 Mar 2016 12:54:43 +0100,
Subhransu S. Prusty wrote:
> 
> On Thu, Mar 03, 2016 at 12:39:05PM +0100, Takashi Iwai wrote:
> > On Thu, 03 Mar 2016 12:29:23 +0100,
> > Vinod Koul wrote:
> > > 
> > > On Wed, Mar 02, 2016 at 05:03:10PM +0100, Takashi Iwai wrote:
> > > > On Wed, 02 Mar 2016 13:03:26 +0100,
> > > > Subhransu S. Prusty wrote:
> > > > > 
> > > > > Added more ops to the chmap object, so most of the helpers can be
> > > > > accessed through ops instead instead of directly calling them.
> > > > 
> > > > Too many instead :)
> > > > 
> > > > > With this moving the chmap helpers to core will be easy.
> > > > 
> > > > Well, although I asked about reduction of exported stuff, I don't
> > > > think it's worth to put everything in ops.  If any of them would be
> > > > overridden, it's good to have as an ops.  If not, it's fine to have
> > > > exported symbols.  These are just a few entries, not over 20.
> > > 
> > > I think the thought was that patch_hdmi and ASoC driver will use defaults
> > > whereas the patch_nvidia etc will override with their own stuff..
> > 
> > Yes, but are these new ops overridden actually?
> 
> I think the new ops added with this patch can be removed. They can be kept
> as exported APIs and not overriden.
> 
> basically these ones:
> +	int (*get_active_channels)(int ca);
> +	void (*setup_channel_mapping)(struct hdmi_chmap *chmap,
> +			hda_nid_t pin_nid, bool non_pcm, int ca,
> +			int channels, unsigned char *map,
> +			bool chmap_set);
> +	int (*channel_allocation)(struct hdac_device *hdac, int spk_alloc,
> +				int channels, bool chmap_set,
> +				bool non_pcm, unsigned char *map);
> +	struct hdmi_cea_channel_speaker_allocation *(*get_cap_from_ca)(int
> ca);
> +	int (*alsa_chmap_to_spk_mask)(unsigned char c);
> +	int (*spk_to_alsa_chmap)(int spk);

Right, I meant the comment above particularly to the ops added in this
patch 5.  The first few ops are reasonable.


Takashi