Re: [PATCH] ALSA: seq: Add tempo base unit for MIDI2 Set Tempo messages

[Jaroslav Kysela <perex@xxxxxxxx>]

On 04. 07. 24 14:35, Takashi Iwai wrote:
> MIDI2 Set Tempo message defines the tempo in 10ns unit for finer
> accuracy, while MIDI1 was defined in 1us unit.  For adapting this
> different unit, introduce "tempo_base" field to snd_seq_queue_tempo
> struct so that user-space can pass the proper tempo base unit.
>
> The accepted value is limited, it must be either 0, 10 or 1000.
>
> The protocol version is bumped to 1.0.4 along with this.
>
> The access with the older protocol version ignores the tempo-base
> value in ioctls and always treats as 1000.
>
> Signed-off-by: Takashi Iwai <tiwai@xxxxxxx>
> ---
>   include/uapi/sound/asequencer.h |  7 ++++---
>   sound/core/seq/seq_clientmgr.c  |  4 ++++
>   sound/core/seq/seq_queue.c      |  6 ++++--
>   sound/core/seq/seq_timer.c      | 21 +++++++++++++++------
>   sound/core/seq/seq_timer.h      |  4 +++-
>   5 files changed, 30 insertions(+), 12 deletions(-)
>
> diff --git a/include/uapi/sound/asequencer.h b/include/uapi/sound/asequencer.h
> index c85fdd8895d8..c7acfeb514a3 100644
> --- a/include/uapi/sound/asequencer.h
> +++ b/include/uapi/sound/asequencer.h
> @@ -10,7 +10,7 @@
>   #include <sound/asound.h>
>   
>   /** version of the sequencer */
> -#define SNDRV_SEQ_VERSION SNDRV_PROTOCOL_VERSION(1, 0, 3)
> +#define SNDRV_SEQ_VERSION SNDRV_PROTOCOL_VERSION(1, 0, 4)
>   
>   /**
>    * definition of sequencer event types
> @@ -523,11 +523,12 @@ struct snd_seq_queue_status {
>   /* queue tempo */
>   struct snd_seq_queue_tempo {
>   	int queue;			/* sequencer queue */
> -	unsigned int tempo;		/* current tempo, us/tick */
> +	unsigned int tempo;		/* current tempo, us/tick (or different time-base below) */
>   	int ppq;			/* time resolution, ticks/quarter */
>   	unsigned int skew_value;	/* queue skew */
>   	unsigned int skew_base;		/* queue skew base */
> -	char reserved[24];		/* for the future */
> +	unsigned int tempo_base;	/* tempo base in nsec unit; either 10 or 1000 */

Perhaps unsigned short value may be enough to carry this settings? The two 
bytes will be always zero for unsigned int.

					Jaroslav

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