[PATCH 0/3] ALSA: firewire-lib: preparation for media clock recovery
Hi,
ALSA IEC 61883-1/6 packet streaming engine have been using pre-computed parameters ideal for nominal sampling transfer frequency (STF), to transfer packets to device since it was added 2011. As a result of user experience for a decade, it is clear that the sequence is not suitable to some actual devices.
In design of audio and music unit in IEEE 1394 bus, feedback of effective sampling transfer frequency is delivered by packets transferred from device. Reuse of the sequence of parameter in the packets results in pseudo media clock recovery to transfer packet.
In this patchset, an option is added to AMDTP domain so that each of AMDTP stream for outgoing packet has association to AMDTP stream for incoming packet, to replay the sequence of parameter. Some devices require receiving packet on-the-fly to transfer packets, so another option is added.
These options are just added, not used yet in this time. In future work, they are going to be used by drivers in ALSA firewire stack.
Takashi Sakamoto (3): ALSA: firewire-lib: add replay target to cache sequence of packet ALSA: firewire-lib: replay sequence of incoming packets for outgoing packets ALSA: firewire-lib: transfer rx packets on-the-fly when replaying
sound/firewire/amdtp-stream.c | 252 ++++++++++++++++++-- sound/firewire/amdtp-stream.h | 17 +- sound/firewire/bebob/bebob_stream.c | 2 +- sound/firewire/dice/dice-stream.c | 2 +- sound/firewire/digi00x/digi00x-stream.c | 2 +- sound/firewire/fireface/ff-stream.c | 2 +- sound/firewire/fireworks/fireworks_stream.c | 2 +- sound/firewire/motu/motu-stream.c | 2 +- sound/firewire/oxfw/oxfw-stream.c | 2 +- sound/firewire/tascam/tascam-stream.c | 2 +- 10 files changed, 259 insertions(+), 26 deletions(-)
In design of audio and music unit in IEEE 1394 bus, feedback of effective sampling transfer frequency (STF) is delivered by packets transferred from device. The devices supported by ALSA firewire stack are categorized to three groups regarding to it.
* Group 1: * Echo Audio Fireworks board module * Oxford Semiconductor OXFW971 ASIC * Digidesign Digi00x family * Tascam FireWire series * RME Fireface series
* Group 2: * BridgeCo. DM1000/DM1100/DM1500 ASICs for BeBoB solution * TC Applied Technologies DICE ASICs
* Group 3: * Mark of the Unicord FireWire series
In group 1, the effective STF is determined by the sequence of the number of events per packet. In group 2, the sequence of presentation timestamp expressed in syt field of CIP header is interpreted as well. In group 3, the presentation timestamp is expressed in source packet header (SPH) of each data block.
I note that some models doesn't take care of effective STF with large internal buffer. It's reasonable to name it as group 0:
* Group 0 * Oxford Semiconductor OXFW970 ASIC
The effective STF is known to be slightly different from nominal STF for all of devices, and to be different between the devices. Furthermore, the effective STF is known to be shifted for long-period transmission. This makes it hard for software to satisfy the effective STF when processing packets to the device.
The effective STF is deterministic as a result of analyzing the batch of packet transferred from the device. For the analysis, caching the sequence of parameter in the packet is required.
This commit adds an option so that AMDTP domain structure takes AMDTP stream structure to cache the sequence of parameters in packet transferred from the device. The parameters are offset ticks of syt field against the cycle to receive the packet and the number of data blocks per packet.
Signed-off-by: Takashi Sakamoto o-takashi@sakamocchi.jp --- sound/firewire/amdtp-stream.c | 82 ++++++++++++++++++++- sound/firewire/amdtp-stream.h | 12 ++- sound/firewire/bebob/bebob_stream.c | 2 +- sound/firewire/dice/dice-stream.c | 2 +- sound/firewire/digi00x/digi00x-stream.c | 2 +- sound/firewire/fireface/ff-stream.c | 2 +- sound/firewire/fireworks/fireworks_stream.c | 2 +- sound/firewire/motu/motu-stream.c | 2 +- sound/firewire/oxfw/oxfw-stream.c | 2 +- sound/firewire/tascam/tascam-stream.c | 2 +- 10 files changed, 98 insertions(+), 12 deletions(-)
diff --git a/sound/firewire/amdtp-stream.c b/sound/firewire/amdtp-stream.c index 68ffbc33f692..860942ffb1f1 100644 --- a/sound/firewire/amdtp-stream.c +++ b/sound/firewire/amdtp-stream.c @@ -52,6 +52,7 @@ #define CIP_FDF_NO_DATA 0xff #define CIP_SYT_MASK 0x0000ffff #define CIP_SYT_NO_INFO 0xffff +#define CIP_SYT_CYCLE_MODULUS 16 #define CIP_NO_DATA ((CIP_FDF_NO_DATA << CIP_FDF_SHIFT) | CIP_SYT_NO_INFO)
#define CIP_HEADER_SIZE (sizeof(__be32) * CIP_HEADER_QUADLETS) @@ -473,6 +474,52 @@ static void pool_ideal_syt_offsets(struct amdtp_stream *s, struct seq_desc *desc s->ctx_data.rx.syt_offset_state = state; }
+static unsigned int compute_syt_offset(unsigned int syt, unsigned int cycle, + unsigned int transfer_delay) +{ + unsigned int cycle_lo = (cycle % CYCLES_PER_SECOND) & 0x0f; + unsigned int syt_cycle_lo = (syt & 0xf000) >> 12; + unsigned int syt_offset; + + // Round up. + if (syt_cycle_lo < cycle_lo) + syt_cycle_lo += CIP_SYT_CYCLE_MODULUS; + syt_cycle_lo -= cycle_lo; + + // Subtract transfer delay so that the synchronization offset is not so large + // at transmission. + syt_offset = syt_cycle_lo * TICKS_PER_CYCLE + (syt & 0x0fff); + if (syt_offset < transfer_delay) + syt_offset += CIP_SYT_CYCLE_MODULUS * TICKS_PER_CYCLE; + + return syt_offset - transfer_delay; +} + +static void cache_seq(struct amdtp_stream *s, const struct pkt_desc *descs, unsigned int desc_count) +{ + const unsigned int transfer_delay = s->transfer_delay; + const unsigned int cache_size = s->ctx_data.tx.cache.size; + struct seq_desc *cache = s->ctx_data.tx.cache.descs; + unsigned int cache_tail = s->ctx_data.tx.cache.tail; + bool aware_syt = !(s->flags & CIP_UNAWARE_SYT); + int i; + + for (i = 0; i < desc_count; ++i) { + struct seq_desc *dst = cache + cache_tail; + const struct pkt_desc *src = descs + i; + + if (aware_syt && src->syt != CIP_SYT_NO_INFO) + dst->syt_offset = compute_syt_offset(src->syt, src->cycle, transfer_delay); + else + dst->syt_offset = CIP_SYT_NO_INFO; + dst->data_blocks = src->data_blocks; + + cache_tail = (cache_tail + 1) % cache_size; + } + + s->ctx_data.tx.cache.tail = cache_tail; +} + static void pool_ideal_seq_descs(struct amdtp_stream *s, unsigned int count) { struct seq_desc *descs = s->ctx_data.rx.seq.descs; @@ -1107,7 +1154,12 @@ static void process_tx_packets(struct fw_iso_context *context, u32 tstamp, size_ return; } } else { + struct amdtp_domain *d = s->domain; + process_ctx_payloads(s, s->pkt_descs, desc_count); + + if (d->replay.enable) + cache_seq(s, s->pkt_descs, desc_count); }
for (i = 0; i < packets; ++i) { @@ -1463,6 +1515,18 @@ static int amdtp_stream_start(struct amdtp_stream *s, int channel, int speed, s->ctx_data.tx.max_ctx_payload_length = max_ctx_payload_size; s->ctx_data.tx.ctx_header_size = ctx_header_size; s->ctx_data.tx.event_starts = false; + + if (s->domain->replay.enable) { + // struct fw_iso_context.drop_overflow_headers is false therefore it's + // possible to cache much unexpectedly. + s->ctx_data.tx.cache.size = max_t(unsigned int, s->syt_interval * 2, + queue_size * 3 / 2); + s->ctx_data.tx.cache.tail = 0; + s->ctx_data.tx.cache.descs = kcalloc(s->ctx_data.tx.cache.size, + sizeof(*s->ctx_data.tx.cache.descs), GFP_KERNEL); + if (!s->ctx_data.tx.cache.descs) + goto err_context; + } } else { static const struct { unsigned int data_block; @@ -1543,8 +1607,12 @@ static int amdtp_stream_start(struct amdtp_stream *s, int channel, int speed, err_pkt_descs: kfree(s->pkt_descs); err_context: - if (s->direction == AMDTP_OUT_STREAM) + if (s->direction == AMDTP_OUT_STREAM) { kfree(s->ctx_data.rx.seq.descs); + } else { + if (s->domain->replay.enable) + kfree(s->ctx_data.tx.cache.descs); + } fw_iso_context_destroy(s->context); s->context = ERR_PTR(-1); err_buffer: @@ -1655,8 +1723,12 @@ static void amdtp_stream_stop(struct amdtp_stream *s) iso_packets_buffer_destroy(&s->buffer, s->unit); kfree(s->pkt_descs);
- if (s->direction == AMDTP_OUT_STREAM) + if (s->direction == AMDTP_OUT_STREAM) { kfree(s->ctx_data.rx.seq.descs); + } else { + if (s->domain->replay.enable) + kfree(s->ctx_data.tx.cache.descs); + }
mutex_unlock(&s->mutex); } @@ -1735,8 +1807,10 @@ EXPORT_SYMBOL_GPL(amdtp_domain_add_stream); * @d: the AMDTP domain. * @tx_init_skip_cycles: the number of cycles to skip processing packets at initial stage of IR * contexts. + * @replay_seq: whether to replay the sequence of packet in IR context for the sequence of packet in + * IT context. */ -int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles) +int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, bool replay_seq) { unsigned int events_per_buffer = d->events_per_buffer; unsigned int events_per_period = d->events_per_period; @@ -1744,6 +1818,8 @@ int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles) struct amdtp_stream *s; int err;
+ d->replay.enable = replay_seq; + // Select an IT context as IRQ target. list_for_each_entry(s, &d->streams, list) { if (s->direction == AMDTP_OUT_STREAM) diff --git a/sound/firewire/amdtp-stream.h b/sound/firewire/amdtp-stream.h index 34294776f9e8..ddfb885b6113 100644 --- a/sound/firewire/amdtp-stream.h +++ b/sound/firewire/amdtp-stream.h @@ -141,6 +141,12 @@ struct amdtp_stream {
// The device starts multiplexing events to the packet. bool event_starts; + + struct { + struct seq_desc *descs; + unsigned int size; + unsigned int tail; + } cache; } tx; struct { // To generate CIP header. @@ -292,6 +298,10 @@ struct amdtp_domain { unsigned int tx_start; unsigned int rx_start; } processing_cycle; + + struct { + bool enable; + } replay; };
int amdtp_domain_init(struct amdtp_domain *d); @@ -300,7 +310,7 @@ void amdtp_domain_destroy(struct amdtp_domain *d); int amdtp_domain_add_stream(struct amdtp_domain *d, struct amdtp_stream *s, int channel, int speed);
-int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles); +int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, bool replay_seq); void amdtp_domain_stop(struct amdtp_domain *d);
static inline int amdtp_domain_set_events_per_period(struct amdtp_domain *d, diff --git a/sound/firewire/bebob/bebob_stream.c b/sound/firewire/bebob/bebob_stream.c index 91306da1bafe..fb776f871133 100644 --- a/sound/firewire/bebob/bebob_stream.c +++ b/sound/firewire/bebob/bebob_stream.c @@ -652,7 +652,7 @@ int snd_bebob_stream_start_duplex(struct snd_bebob *bebob) // MEMO: In the early stage of packet streaming, the device transfers NODATA packets. // After several hundred cycles, it begins to multiplex event into the packet with // syt information. - err = amdtp_domain_start(&bebob->domain, tx_init_skip_cycles); + err = amdtp_domain_start(&bebob->domain, tx_init_skip_cycles, false); if (err < 0) goto error;
diff --git a/sound/firewire/dice/dice-stream.c b/sound/firewire/dice/dice-stream.c index a9a0fe9635dd..064f28f1822b 100644 --- a/sound/firewire/dice/dice-stream.c +++ b/sound/firewire/dice/dice-stream.c @@ -459,7 +459,7 @@ int snd_dice_stream_start_duplex(struct snd_dice *dice) goto error; }
- err = amdtp_domain_start(&dice->domain, 0); + err = amdtp_domain_start(&dice->domain, 0, false); if (err < 0) goto error;
diff --git a/sound/firewire/digi00x/digi00x-stream.c b/sound/firewire/digi00x/digi00x-stream.c index f11aaff2e248..5daba75a5bf3 100644 --- a/sound/firewire/digi00x/digi00x-stream.c +++ b/sound/firewire/digi00x/digi00x-stream.c @@ -375,7 +375,7 @@ int snd_dg00x_stream_start_duplex(struct snd_dg00x *dg00x) if (err < 0) goto error;
- err = amdtp_domain_start(&dg00x->domain, 0); + err = amdtp_domain_start(&dg00x->domain, 0, false); if (err < 0) goto error;
diff --git a/sound/firewire/fireface/ff-stream.c b/sound/firewire/fireface/ff-stream.c index 53a21fb95add..12779e7caf22 100644 --- a/sound/firewire/fireface/ff-stream.c +++ b/sound/firewire/fireface/ff-stream.c @@ -199,7 +199,7 @@ int snd_ff_stream_start_duplex(struct snd_ff *ff, unsigned int rate) if (err < 0) goto error;
- err = amdtp_domain_start(&ff->domain, 0); + err = amdtp_domain_start(&ff->domain, 0, false); if (err < 0) goto error;
diff --git a/sound/firewire/fireworks/fireworks_stream.c b/sound/firewire/fireworks/fireworks_stream.c index 858cd6085c1f..0106fa6d1eaf 100644 --- a/sound/firewire/fireworks/fireworks_stream.c +++ b/sound/firewire/fireworks/fireworks_stream.c @@ -272,7 +272,7 @@ int snd_efw_stream_start_duplex(struct snd_efw *efw) if (err < 0) goto error;
- err = amdtp_domain_start(&efw->domain, 0); + err = amdtp_domain_start(&efw->domain, 0, false); if (err < 0) goto error;
diff --git a/sound/firewire/motu/motu-stream.c b/sound/firewire/motu/motu-stream.c index 925241ae2551..5af9d7487cdc 100644 --- a/sound/firewire/motu/motu-stream.c +++ b/sound/firewire/motu/motu-stream.c @@ -260,7 +260,7 @@ int snd_motu_stream_start_duplex(struct snd_motu *motu) if (err < 0) goto stop_streams;
- err = amdtp_domain_start(&motu->domain, 0); + err = amdtp_domain_start(&motu->domain, 0, false); if (err < 0) goto stop_streams;
diff --git a/sound/firewire/oxfw/oxfw-stream.c b/sound/firewire/oxfw/oxfw-stream.c index 4121d95e161f..a6e97a37f129 100644 --- a/sound/firewire/oxfw/oxfw-stream.c +++ b/sound/firewire/oxfw/oxfw-stream.c @@ -354,7 +354,7 @@ int snd_oxfw_stream_start_duplex(struct snd_oxfw *oxfw) } }
- err = amdtp_domain_start(&oxfw->domain, 0); + err = amdtp_domain_start(&oxfw->domain, 0, false); if (err < 0) goto error;
diff --git a/sound/firewire/tascam/tascam-stream.c b/sound/firewire/tascam/tascam-stream.c index 296ecf5f6ddc..eb4c7c47f8e9 100644 --- a/sound/firewire/tascam/tascam-stream.c +++ b/sound/firewire/tascam/tascam-stream.c @@ -473,7 +473,7 @@ int snd_tscm_stream_start_duplex(struct snd_tscm *tscm, unsigned int rate) if (err < 0) goto error;
- err = amdtp_domain_start(&tscm->domain, 0); + err = amdtp_domain_start(&tscm->domain, 0, false); if (err < 0) return err;
ALSA IEC 61883-1/6 packet streaming engine uses pre-computed parameters ideal for nominal sampling transfer frequency (STF) to transfer packets to device since it was added 2011. As a result of user experience for a decade, it is clear that the sequence is not suitable to some actual devices. It takes the devices to generate noise, and causes any type of discontinuity in the series of packet transferred from the device. It's required for the engine to transfer packets according to effective STF.
The effective STF is given by media clock recovered by the sequence of packet transferred from the target device. In the previous commit, the sequence is already cached. The media clock recovery can be achieved by analyzing the sequence.
In technological world, many ideas are proposed for media clock recovery. However, the small part of them could be actually adopted in our case since floating point arithmetic is not mostly available in Linux kernel land.
This commit adopts the simple way from them; sequence replay, which means that the sequence of parameters from incoming packet is used as is to transfer outgoing packets. The media clock is not computed internally, but the sequence of outgoing packet superficially looks to be generated by the media clock.
The association between source and destination is decided when starting AMDTP domain. When the target device supports a pair of isochronous packet streams, the tx stream is source and the rx stream is destination. When it supports two pair of streams, each of tx stream is associated to corresponding rx stream in its order. When it supports less number of tx streams than rx streams, the fist tx stream is selected for all of rx streams. When it supports more tx streams than rx streams, the first tx packet is associated to the rx stream.
As I noted in previous commit, the sequence of parameters from incoming packet is different between devices, time to time. It is worse idea to replay the sequence of parameters from a device for the sequence of packet to the other devices even if they are in the same category of device.
Signed-off-by: Takashi Sakamoto o-takashi@sakamocchi.jp --- sound/firewire/amdtp-stream.c | 153 ++++++++++++++++++++++++++++++---- sound/firewire/amdtp-stream.h | 3 + 2 files changed, 142 insertions(+), 14 deletions(-)
diff --git a/sound/firewire/amdtp-stream.c b/sound/firewire/amdtp-stream.c index 860942ffb1f1..47ea03370858 100644 --- a/sound/firewire/amdtp-stream.c +++ b/sound/firewire/amdtp-stream.c @@ -495,6 +495,22 @@ static unsigned int compute_syt_offset(unsigned int syt, unsigned int cycle, return syt_offset - transfer_delay; }
+// Both of the producer and consumer of the queue runs in the same clock of IEEE 1394 bus. +// Additionally, the sequence of tx packets is severely checked against any discontinuity +// before filling entries in the queue. The calculation is safe even if it looks fragile by +// overrun. +static unsigned int calculate_cached_cycle_count(struct amdtp_stream *s, unsigned int head) +{ + const unsigned int cache_size = s->ctx_data.tx.cache.size; + unsigned int cycles = s->ctx_data.tx.cache.tail; + + if (cycles < head) + cycles += cache_size; + cycles -= head; + + return cycles; +} + static void cache_seq(struct amdtp_stream *s, const struct pkt_desc *descs, unsigned int desc_count) { const unsigned int transfer_delay = s->transfer_delay; @@ -536,6 +552,37 @@ static void pool_ideal_seq_descs(struct amdtp_stream *s, unsigned int count) s->ctx_data.rx.seq.tail = (seq_tail + count) % seq_size; }
+static void pool_replayed_seq(struct amdtp_stream *s, unsigned int count) +{ + struct amdtp_stream *target = s->ctx_data.rx.replay_target; + const struct seq_desc *cache = target->ctx_data.tx.cache.descs; + const unsigned int cache_size = target->ctx_data.tx.cache.size; + unsigned int cache_head = s->ctx_data.rx.cache_head; + struct seq_desc *descs = s->ctx_data.rx.seq.descs; + const unsigned int seq_size = s->ctx_data.rx.seq.size; + unsigned int seq_tail = s->ctx_data.rx.seq.tail; + int i; + + for (i = 0; i < count; ++i) { + descs[seq_tail] = cache[cache_head]; + seq_tail = (seq_tail + 1) % seq_size; + cache_head = (cache_head + 1) % cache_size; + } + + s->ctx_data.rx.seq.tail = seq_tail; + s->ctx_data.rx.cache_head = cache_head; +} + +static void pool_seq_descs(struct amdtp_stream *s, unsigned int count) +{ + struct amdtp_domain *d = s->domain; + + if (!d->replay.enable || !s->ctx_data.rx.replay_target) + pool_ideal_seq_descs(s, count); + else + pool_replayed_seq(s, count); +} + static void update_pcm_pointers(struct amdtp_stream *s, struct snd_pcm_substream *pcm, unsigned int frames) @@ -1004,7 +1051,7 @@ static void process_rx_packets(struct fw_iso_context *context, u32 tstamp, size_ // Calculate the number of packets in buffer and check XRUN. packets = header_length / sizeof(*ctx_header);
- pool_ideal_seq_descs(s, packets); + pool_seq_descs(s, packets);
generate_pkt_descs(s, ctx_header, packets);
@@ -1392,28 +1439,54 @@ static void irq_target_callback_skip(struct fw_iso_context *context, u32 tstamp, { struct amdtp_stream *s = private_data; struct amdtp_domain *d = s->domain; - unsigned int cycle; + bool ready_to_start;
skip_rx_packets(context, tstamp, header_length, header, private_data); process_ctxs_in_domain(d);
+ if (d->replay.enable) { + unsigned int rx_count = 0; + unsigned int rx_ready_count = 0; + struct amdtp_stream *rx; + + list_for_each_entry(rx, &d->streams, list) { + struct amdtp_stream *tx; + unsigned int cached_cycles; + + if (rx->direction != AMDTP_OUT_STREAM) + continue; + ++rx_count; + + tx = rx->ctx_data.rx.replay_target; + cached_cycles = calculate_cached_cycle_count(tx, 0); + if (cached_cycles > tx->ctx_data.tx.cache.size / 2) + ++rx_ready_count; + } + + ready_to_start = (rx_count == rx_ready_count); + } else { + ready_to_start = true; + } + // Decide the cycle count to begin processing content of packet in IT contexts. All of IT // contexts are expected to start and get callback when reaching here. - cycle = s->next_cycle; - list_for_each_entry(s, &d->streams, list) { - if (s->direction != AMDTP_OUT_STREAM) - continue; + if (ready_to_start) { + unsigned int cycle = s->next_cycle; + list_for_each_entry(s, &d->streams, list) { + if (s->direction != AMDTP_OUT_STREAM) + continue;
- if (compare_ohci_cycle_count(s->next_cycle, cycle) > 0) - cycle = s->next_cycle; + if (compare_ohci_cycle_count(s->next_cycle, cycle) > 0) + cycle = s->next_cycle;
- if (s == d->irq_target) - s->context->callback.sc = irq_target_callback_intermediately; - else - s->context->callback.sc = process_rx_packets_intermediately; - } + if (s == d->irq_target) + s->context->callback.sc = irq_target_callback_intermediately; + else + s->context->callback.sc = process_rx_packets_intermediately; + }
- d->processing_cycle.rx_start = cycle; + d->processing_cycle.rx_start = cycle; + } }
// This is executed one time. For in-stream, first packet has come. For out-stream, prepared to @@ -1802,6 +1875,53 @@ int amdtp_domain_add_stream(struct amdtp_domain *d, struct amdtp_stream *s, } EXPORT_SYMBOL_GPL(amdtp_domain_add_stream);
+// Make the reference from rx stream to tx stream for sequence replay. When the number of tx streams +// is less than the number of rx streams, the first tx stream is selected. +static int make_association(struct amdtp_domain *d) +{ + unsigned int dst_index = 0; + struct amdtp_stream *rx; + + // Make association to replay target. + list_for_each_entry(rx, &d->streams, list) { + if (rx->direction == AMDTP_OUT_STREAM) { + unsigned int src_index = 0; + struct amdtp_stream *tx = NULL; + struct amdtp_stream *s; + + list_for_each_entry(s, &d->streams, list) { + if (s->direction == AMDTP_IN_STREAM) { + if (dst_index == src_index) { + tx = s; + break; + } + + ++src_index; + } + } + if (!tx) { + // Select the first entry. + list_for_each_entry(s, &d->streams, list) { + if (s->direction == AMDTP_IN_STREAM) { + tx = s; + break; + } + } + // No target is available to replay sequence. + if (!tx) + return -EINVAL; + } + + rx->ctx_data.rx.replay_target = tx; + rx->ctx_data.rx.cache_head = 0; + + ++dst_index; + } + } + + return 0; +} + /** * amdtp_domain_start - start sending packets for isoc context in the domain. * @d: the AMDTP domain. @@ -1818,6 +1938,11 @@ int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, struct amdtp_stream *s; int err;
+ if (replay_seq) { + err = make_association(d); + if (err < 0) + return err; + } d->replay.enable = replay_seq;
// Select an IT context as IRQ target. diff --git a/sound/firewire/amdtp-stream.h b/sound/firewire/amdtp-stream.h index ddfb885b6113..61b6b5ae8b3b 100644 --- a/sound/firewire/amdtp-stream.h +++ b/sound/firewire/amdtp-stream.h @@ -166,6 +166,9 @@ struct amdtp_stream { unsigned int data_block_state; unsigned int syt_offset_state; unsigned int last_syt_offset; + + struct amdtp_stream *replay_target; + unsigned int cache_head; } rx; } ctx_data;
Models in below series start transmission of packet after receiving the sequence of packets:
* Digidesign Digi00x family * RME Fireface series
Additionally, models in Tascam FireWire series start multiplexing PCM frames into packets enough after receiving packets. It's required to transfer packets on-the-fly for the above models according to nominal sampling transfer frequency before starting sequence replay.
This commit allows drivers to decide whether the engine transfers packet on-the-fly or not.
Signed-off-by: Takashi Sakamoto o-takashi@sakamocchi.jp --- sound/firewire/amdtp-stream.c | 27 +++++++++++++++++---- sound/firewire/amdtp-stream.h | 6 +++-- sound/firewire/bebob/bebob_stream.c | 2 +- sound/firewire/dice/dice-stream.c | 2 +- sound/firewire/digi00x/digi00x-stream.c | 2 +- sound/firewire/fireface/ff-stream.c | 2 +- sound/firewire/fireworks/fireworks_stream.c | 2 +- sound/firewire/motu/motu-stream.c | 2 +- sound/firewire/oxfw/oxfw-stream.c | 2 +- sound/firewire/tascam/tascam-stream.c | 2 +- 10 files changed, 34 insertions(+), 15 deletions(-)
diff --git a/sound/firewire/amdtp-stream.c b/sound/firewire/amdtp-stream.c index 47ea03370858..d0e9b417c019 100644 --- a/sound/firewire/amdtp-stream.c +++ b/sound/firewire/amdtp-stream.c @@ -577,10 +577,23 @@ static void pool_seq_descs(struct amdtp_stream *s, unsigned int count) { struct amdtp_domain *d = s->domain;
- if (!d->replay.enable || !s->ctx_data.rx.replay_target) + if (!d->replay.enable || !s->ctx_data.rx.replay_target) { pool_ideal_seq_descs(s, count); - else - pool_replayed_seq(s, count); + } else { + if (!d->replay.on_the_fly) { + pool_replayed_seq(s, count); + } else { + struct amdtp_stream *tx = s->ctx_data.rx.replay_target; + const unsigned int cache_size = tx->ctx_data.tx.cache.size; + const unsigned int cache_head = s->ctx_data.rx.cache_head; + unsigned int cached_cycles = calculate_cached_cycle_count(tx, cache_head); + + if (cached_cycles > count && cached_cycles > cache_size / 2) + pool_replayed_seq(s, count); + else + pool_ideal_seq_descs(s, count); + } + } }
static void update_pcm_pointers(struct amdtp_stream *s, @@ -1444,7 +1457,7 @@ static void irq_target_callback_skip(struct fw_iso_context *context, u32 tstamp, skip_rx_packets(context, tstamp, header_length, header, private_data); process_ctxs_in_domain(d);
- if (d->replay.enable) { + if (d->replay.enable && !d->replay.on_the_fly) { unsigned int rx_count = 0; unsigned int rx_ready_count = 0; struct amdtp_stream *rx; @@ -1929,8 +1942,11 @@ static int make_association(struct amdtp_domain *d) * contexts. * @replay_seq: whether to replay the sequence of packet in IR context for the sequence of packet in * IT context. + * @replay_on_the_fly: transfer rx packets according to nominal frequency, then begin to replay + * according to arrival of events in tx packets. */ -int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, bool replay_seq) +int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, bool replay_seq, + bool replay_on_the_fly) { unsigned int events_per_buffer = d->events_per_buffer; unsigned int events_per_period = d->events_per_period; @@ -1944,6 +1960,7 @@ int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, return err; } d->replay.enable = replay_seq; + d->replay.on_the_fly = replay_on_the_fly;
// Select an IT context as IRQ target. list_for_each_entry(s, &d->streams, list) { diff --git a/sound/firewire/amdtp-stream.h b/sound/firewire/amdtp-stream.h index 61b6b5ae8b3b..b25592d5f6af 100644 --- a/sound/firewire/amdtp-stream.h +++ b/sound/firewire/amdtp-stream.h @@ -303,7 +303,8 @@ struct amdtp_domain { } processing_cycle;
struct { - bool enable; + bool enable:1; + bool on_the_fly:1; } replay; };
@@ -313,7 +314,8 @@ void amdtp_domain_destroy(struct amdtp_domain *d); int amdtp_domain_add_stream(struct amdtp_domain *d, struct amdtp_stream *s, int channel, int speed);
-int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, bool replay_seq); +int amdtp_domain_start(struct amdtp_domain *d, unsigned int tx_init_skip_cycles, bool replay_seq, + bool replay_on_the_fly); void amdtp_domain_stop(struct amdtp_domain *d);
static inline int amdtp_domain_set_events_per_period(struct amdtp_domain *d, diff --git a/sound/firewire/bebob/bebob_stream.c b/sound/firewire/bebob/bebob_stream.c index fb776f871133..47773ca97e46 100644 --- a/sound/firewire/bebob/bebob_stream.c +++ b/sound/firewire/bebob/bebob_stream.c @@ -652,7 +652,7 @@ int snd_bebob_stream_start_duplex(struct snd_bebob *bebob) // MEMO: In the early stage of packet streaming, the device transfers NODATA packets. // After several hundred cycles, it begins to multiplex event into the packet with // syt information. - err = amdtp_domain_start(&bebob->domain, tx_init_skip_cycles, false); + err = amdtp_domain_start(&bebob->domain, tx_init_skip_cycles, false, false); if (err < 0) goto error;
diff --git a/sound/firewire/dice/dice-stream.c b/sound/firewire/dice/dice-stream.c index 064f28f1822b..0fb8b4ae6a0a 100644 --- a/sound/firewire/dice/dice-stream.c +++ b/sound/firewire/dice/dice-stream.c @@ -459,7 +459,7 @@ int snd_dice_stream_start_duplex(struct snd_dice *dice) goto error; }
- err = amdtp_domain_start(&dice->domain, 0, false); + err = amdtp_domain_start(&dice->domain, 0, false, false); if (err < 0) goto error;
diff --git a/sound/firewire/digi00x/digi00x-stream.c b/sound/firewire/digi00x/digi00x-stream.c index 5daba75a5bf3..2019f6533477 100644 --- a/sound/firewire/digi00x/digi00x-stream.c +++ b/sound/firewire/digi00x/digi00x-stream.c @@ -375,7 +375,7 @@ int snd_dg00x_stream_start_duplex(struct snd_dg00x *dg00x) if (err < 0) goto error;
- err = amdtp_domain_start(&dg00x->domain, 0, false); + err = amdtp_domain_start(&dg00x->domain, 0, false, false); if (err < 0) goto error;
diff --git a/sound/firewire/fireface/ff-stream.c b/sound/firewire/fireface/ff-stream.c index 12779e7caf22..97c356f2ac04 100644 --- a/sound/firewire/fireface/ff-stream.c +++ b/sound/firewire/fireface/ff-stream.c @@ -199,7 +199,7 @@ int snd_ff_stream_start_duplex(struct snd_ff *ff, unsigned int rate) if (err < 0) goto error;
- err = amdtp_domain_start(&ff->domain, 0, false); + err = amdtp_domain_start(&ff->domain, 0, false, false); if (err < 0) goto error;
diff --git a/sound/firewire/fireworks/fireworks_stream.c b/sound/firewire/fireworks/fireworks_stream.c index 0106fa6d1eaf..6fc117c3a068 100644 --- a/sound/firewire/fireworks/fireworks_stream.c +++ b/sound/firewire/fireworks/fireworks_stream.c @@ -272,7 +272,7 @@ int snd_efw_stream_start_duplex(struct snd_efw *efw) if (err < 0) goto error;
- err = amdtp_domain_start(&efw->domain, 0, false); + err = amdtp_domain_start(&efw->domain, 0, false, false); if (err < 0) goto error;
diff --git a/sound/firewire/motu/motu-stream.c b/sound/firewire/motu/motu-stream.c index 5af9d7487cdc..5d8d067f366d 100644 --- a/sound/firewire/motu/motu-stream.c +++ b/sound/firewire/motu/motu-stream.c @@ -260,7 +260,7 @@ int snd_motu_stream_start_duplex(struct snd_motu *motu) if (err < 0) goto stop_streams;
- err = amdtp_domain_start(&motu->domain, 0, false); + err = amdtp_domain_start(&motu->domain, 0, false, false); if (err < 0) goto stop_streams;
diff --git a/sound/firewire/oxfw/oxfw-stream.c b/sound/firewire/oxfw/oxfw-stream.c index a6e97a37f129..9792d4b4373c 100644 --- a/sound/firewire/oxfw/oxfw-stream.c +++ b/sound/firewire/oxfw/oxfw-stream.c @@ -354,7 +354,7 @@ int snd_oxfw_stream_start_duplex(struct snd_oxfw *oxfw) } }
- err = amdtp_domain_start(&oxfw->domain, 0, false); + err = amdtp_domain_start(&oxfw->domain, 0, false, false); if (err < 0) goto error;
diff --git a/sound/firewire/tascam/tascam-stream.c b/sound/firewire/tascam/tascam-stream.c index eb4c7c47f8e9..4811b60e5823 100644 --- a/sound/firewire/tascam/tascam-stream.c +++ b/sound/firewire/tascam/tascam-stream.c @@ -473,7 +473,7 @@ int snd_tscm_stream_start_duplex(struct snd_tscm *tscm, unsigned int rate) if (err < 0) goto error;
- err = amdtp_domain_start(&tscm->domain, 0, false); + err = amdtp_domain_start(&tscm->domain, 0, false, false); if (err < 0) return err;
On Thu, 27 May 2021 14:26:08 +0200, Takashi Sakamoto wrote:
Hi,
ALSA IEC 61883-1/6 packet streaming engine have been using pre-computed parameters ideal for nominal sampling transfer frequency (STF), to transfer packets to device since it was added 2011. As a result of user experience for a decade, it is clear that the sequence is not suitable to some actual devices.
In design of audio and music unit in IEEE 1394 bus, feedback of effective sampling transfer frequency is delivered by packets transferred from device. Reuse of the sequence of parameter in the packets results in pseudo media clock recovery to transfer packet.
In this patchset, an option is added to AMDTP domain so that each of AMDTP stream for outgoing packet has association to AMDTP stream for incoming packet, to replay the sequence of parameter. Some devices require receiving packet on-the-fly to transfer packets, so another option is added.
These options are just added, not used yet in this time. In future work, they are going to be used by drivers in ALSA firewire stack.
Takashi Sakamoto (3): ALSA: firewire-lib: add replay target to cache sequence of packet ALSA: firewire-lib: replay sequence of incoming packets for outgoing packets ALSA: firewire-lib: transfer rx packets on-the-fly when replaying
Applied all three patches now. Thanks.
Takashi
participants (2)
-
Takashi Iwai -
Takashi Sakamoto