On Mon, 19 Apr 2021 18:40:23 +0200,
David Henningsson wrote:
>
> This commit adds a new framing mode that frames all MIDI data into
> 32-byte frames with a timestamp.
>
> The main benefit is that we can get accurate timestamps even if
> userspace wakeup and processing is not immediate.
>
> Testing on a Celeron N3150 with this mode has a max jitter of 2.8 ms,
> compared to the in-kernel seq implementation which has a max jitter
> of 5 ms during idle and much worse when running scheduler stress tests
> in parallel.
>
> Signed-off-by: David Henningsson <coding@xxxxxxxx>
> ---
>
> Changes since v5: Added realtime clock and changed params struct according to
> Jaroslav's wishes.
>
> This version of the patch has been compile tested only.
Thanks for the updated patch. It looks almost good, just some minor
issues below:
> +struct snd_rawmidi_framing_tstamp {
> + /* For now, frame_type is always 0. Midi 2.0 is expected to add new
> + * types here. Applications are expected to skip unknown frame types.
> + */
> + u8 frame_type;
> + u8 length; /* number of valid bytes in data field */
> + u8 reserved[2];
> + u32 tv_nsec; /* nanoseconds */
> + u64 tv_sec; /* seconds */
> + u8 data[SNDRV_RAWMIDI_FRAMING_DATA_LENGTH];
> +} __attribute__((packed));
Use __packed instead.
> @@ -720,7 +723,18 @@ EXPORT_SYMBOL(snd_rawmidi_output_params);
> int snd_rawmidi_input_params(struct snd_rawmidi_substream *substream,
> struct snd_rawmidi_params *params)
> {
> + unsigned int framing = params->mode & SNDRV_RAWMIDI_MODE_FRAMING_MASK;
> + unsigned int clock_type = params->mode & SNDRV_RAWMIDI_MODE_CLOCK_MASK;
> +
> + if (framing == SNDRV_RAWMIDI_MODE_FRAMING_NONE && clock_type != SNDRV_RAWMIDI_MODE_CLOCK_NONE)
> + return -EINVAL;
> + else if (clock_type > SNDRV_RAWMIDI_MODE_CLOCK_MONOTONIC_RAW)
> + return -EINVAL;
> + if (framing > SNDRV_RAWMIDI_MODE_FRAMING_TSTAMP)
> + return -EINVAL;
> snd_rawmidi_drain_input(substream);
> + substream->framing = framing;
> + substream->clock_type = clock_type;
> return resize_runtime_buffer(substream->runtime, params, true);
The stale framing and clock_type are set in the case of errors, which
will confuse the later operations.
> +static int receive_with_tstamp_framing(struct snd_rawmidi_substream *substream,
> + const unsigned char *buffer, int src_count, const struct timespec64 *tstamp)
> +{
....
> + return dest_frames * frame_size;
The snd_rawmidi_receive() function is supposed to return the processed
read size, hence this would become incompatible. Though, I don't find
any code that actually checks the return value from
snd_rawmidi_receive(), but a definition is a definition...
thanks,
Takashi
