linux/sound/firewire/amdtp-stream.c

1079 lines
29 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* Audio and Music Data Transmission Protocol (IEC 61883-6) streams
* with Common Isochronous Packet (IEC 61883-1) headers
*
* Copyright (c) Clemens Ladisch <clemens@ladisch.de>
*/
#include <linux/device.h>
#include <linux/err.h>
#include <linux/firewire.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <sound/pcm.h>
#include <sound/pcm_params.h>
#include "amdtp-stream.h"
#define TICKS_PER_CYCLE 3072
#define CYCLES_PER_SECOND 8000
#define TICKS_PER_SECOND (TICKS_PER_CYCLE * CYCLES_PER_SECOND)
ALSA: firewire-lib: add tracepoints to dump a part of isochronous packet data When audio and music units have some quirks in their sequence of packet, it's really hard for non-owners to identify the quirks. Although developers need dumps for sequence of packets, it's difficult for users who have no knowledges and no equipments for this purpose. This commit adds tracepoints for this situation. When users encounter the issue, they can dump a part of packet data via Linux tracing framework as long as using drivers in ALSA firewire stack. Additionally, tracepoints for outgoing packets will be our help to check and debug packet processing of ALSA firewire stack. This commit newly adds 'snd_firewire_lib' subsystem with 'in_packet' and 'out_packet' events. In the events, some attributes of packets and the index of packet managed by this module are recorded per packet. This is an usage: $ trace-cmd record -e snd_firewire_lib:out_packet \ -e snd_firewire_lib:in_packet /sys/kernel/tracing/events/snd_firewire_lib/out_packet/filter /sys/kernel/tracing/events/snd_firewire_lib/in_packet/filter Hit Ctrl^C to stop recording ^C $ trace-cmd report trace.dat ... 23647.033934: in_packet: 01 4073 ffc0 ffc1 00 000f0040 9001b2d1 122 44 23647.033936: in_packet: 01 4074 ffc0 ffc1 00 000f0048 9001c83b 122 45 23647.033937: in_packet: 01 4075 ffc0 ffc1 00 000f0050 9001ffff 002 46 23647.033938: in_packet: 01 4076 ffc0 ffc1 00 000f0050 9001e1a6 122 47 23647.035426: out_packet: 01 4123 ffc1 ffc0 01 010f00d0 9001fb40 122 17 23647.035428: out_packet: 01 4124 ffc1 ffc0 01 010f00d8 9001ffff 002 18 23647.035429: out_packet: 01 4125 ffc1 ffc0 01 010f00d8 900114aa 122 19 23647.035430: out_packet: 01 4126 ffc1 ffc0 01 010f00e0 90012a15 122 20 (Here, some common fields are omitted so that a line to be within 80 characters.) ... One line represent one packet. The legend for the last nine fields is: - The second of cycle scheduled for the packet - The count of cycle scheduled for the packet - The ID of node as source (hex) - Some devices transfer packets with invalid source node ID in their CIP header. - The ID of node as destination (hex) - The value is not in CIP header of packets. - The value of isochronous channel - The first quadlet of CIP header (hex) - The second quadlet of CIP header (hex) - The number of included quadlets - The index of packet in a buffer maintained by this module This is an example to parse these lines from text file by Python3 script: \#!/usr/bin/env python3 import sys def parse_ts(second, cycle, syt): offset = syt & 0xfff syt >>= 12 if cycle & 0x0f > syt: cycle += 0x10 cycle &= 0x1ff0 cycle |= syt second += cycle // 8000 cycle %= 8000 # In CYCLE_TIMER of 1394 OHCI, second is represented in 8 bit. second %= 128 return (second, cycle, offset) def calc_ts(second, cycle, offset): ts = offset ts += cycle * 3072 # In DMA descriptor of 1394 OHCI, second is represented in 3 bit. ts += (second % 8) * 8000 * 3072 return ts def subtract_ts(minuend, subtrahend): # In DMA descriptor of 1394 OHCI, second is represented in 3 bit. if minuend < subtrahend: minuend += 8 * 8000 * 3072 return minuend - subtrahend if len(sys.argv) != 2: print('At least, one argument is required for packet dump.') sys.exit() filename = sys.argv[1] data = [] prev = 0 with open(filename, 'r') as f: for line in f: pos = line.find('packet:') if pos < 0: continue pos += len('packet:') line = line[pos:].strip() fields = line.split(' ') datum = [] datum.append(fields[8]) syt = int(fields[6][4:], 16) # Empty packet in IEC 61883-1, or NODATA in IEC 61883-6 if syt == 0xffff: data_blocks = 0 else: payload_size = int(fields[7], 10) data_block_size = int(fields[5][2:4], 16) data_blocks = (payload_size - 2) / data_block_size datum.append(data_blocks) second = int(fields[0], 10) cycle = int(fields[1], 10) start = (second << 25) | (cycle << 12) datum.append('0x{0:08x}'.format(start)) start = calc_ts(second, cycle, 0) datum.append("0x" + fields[5]) datum.append("0x" + fields[6]) if syt == 0xffff: second = 0 cycle = 0 tick = 0 else: second, cycle, tick = parse_ts(second, cycle, syt) ts = calc_ts(second, cycle, tick) datum.append(start) datum.append(ts) if ts == 0: datum.append(0) datum.append(0) else: # Usual case, or a case over 8 seconds. if ts > start or start > 7 * 8000 * 3072: datum.append(subtract_ts(ts, start)) if ts > prev or start > 7 * 8000 * 3072: gap = subtract_ts(ts, prev) datum.append(gap) else: datum.append('backward') else: datum.append('invalid') prev = ts data.append(datum) sys.exit() The data variable includes array with these elements: - The index of the packet - The number of data blocks in the packet - The value of cycle count (hex) - The value of CIP header 1 (hex) - The value of CIP header 2 (hex) - The value of cycle count (tick) - The value of calculated presentation timestamp (tick) - The offset between the cycle count and presentation timestamp - The elapsed ticks from the previous presentation timestamp Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2016-05-09 20:12:46 +08:00
/* Always support Linux tracing subsystem. */
#define CREATE_TRACE_POINTS
#include "amdtp-stream-trace.h"
#define TRANSFER_DELAY_TICKS 0x2e00 /* 479.17 microseconds */
/* isochronous header parameters */
#define ISO_DATA_LENGTH_SHIFT 16
2017-03-31 21:06:07 +08:00
#define TAG_NO_CIP_HEADER 0
#define TAG_CIP 1
/* common isochronous packet header parameters */
#define CIP_EOH_SHIFT 31
#define CIP_EOH (1u << CIP_EOH_SHIFT)
#define CIP_EOH_MASK 0x80000000
#define CIP_SID_SHIFT 24
#define CIP_SID_MASK 0x3f000000
#define CIP_DBS_MASK 0x00ff0000
#define CIP_DBS_SHIFT 16
#define CIP_SPH_MASK 0x00000400
#define CIP_SPH_SHIFT 10
#define CIP_DBC_MASK 0x000000ff
#define CIP_FMT_SHIFT 24
#define CIP_FMT_MASK 0x3f000000
#define CIP_FDF_MASK 0x00ff0000
#define CIP_FDF_SHIFT 16
#define CIP_SYT_MASK 0x0000ffff
#define CIP_SYT_NO_INFO 0xffff
/* Audio and Music transfer protocol specific parameters */
#define CIP_FMT_AM 0x10
#define AMDTP_FDF_NO_DATA 0xff
/* TODO: make these configurable */
#define INTERRUPT_INTERVAL 16
#define QUEUE_LENGTH 48
// For iso header, tstamp and 2 CIP header.
#define IR_CTX_HEADER_SIZE_CIP 16
// For iso header and tstamp.
#define IR_CTX_HEADER_SIZE_NO_CIP 8
#define HEADER_TSTAMP_MASK 0x0000ffff
#define IT_PKT_HEADER_SIZE_CIP 8 // For 2 CIP header.
#define IT_PKT_HEADER_SIZE_NO_CIP 0 // Nothing.
static void pcm_period_tasklet(unsigned long data);
/**
* amdtp_stream_init - initialize an AMDTP stream structure
* @s: the AMDTP stream to initialize
* @unit: the target of the stream
* @dir: the direction of stream
* @flags: the packet transmission method to use
* @fmt: the value of fmt field in CIP header
* @process_data_blocks: callback handler to process data blocks
* @protocol_size: the size to allocate newly for protocol
*/
int amdtp_stream_init(struct amdtp_stream *s, struct fw_unit *unit,
enum amdtp_stream_direction dir, enum cip_flags flags,
unsigned int fmt,
amdtp_stream_process_data_blocks_t process_data_blocks,
unsigned int protocol_size)
{
if (process_data_blocks == NULL)
return -EINVAL;
s->protocol = kzalloc(protocol_size, GFP_KERNEL);
if (!s->protocol)
return -ENOMEM;
s->unit = unit;
s->direction = dir;
s->flags = flags;
s->context = ERR_PTR(-1);
mutex_init(&s->mutex);
tasklet_init(&s->period_tasklet, pcm_period_tasklet, (unsigned long)s);
s->packet_index = 0;
init_waitqueue_head(&s->callback_wait);
s->callbacked = false;
s->fmt = fmt;
s->process_data_blocks = process_data_blocks;
return 0;
}
EXPORT_SYMBOL(amdtp_stream_init);
/**
* amdtp_stream_destroy - free stream resources
* @s: the AMDTP stream to destroy
*/
void amdtp_stream_destroy(struct amdtp_stream *s)
{
/* Not initialized. */
if (s->protocol == NULL)
return;
WARN_ON(amdtp_stream_running(s));
kfree(s->protocol);
mutex_destroy(&s->mutex);
}
EXPORT_SYMBOL(amdtp_stream_destroy);
const unsigned int amdtp_syt_intervals[CIP_SFC_COUNT] = {
[CIP_SFC_32000] = 8,
[CIP_SFC_44100] = 8,
[CIP_SFC_48000] = 8,
[CIP_SFC_88200] = 16,
[CIP_SFC_96000] = 16,
[CIP_SFC_176400] = 32,
[CIP_SFC_192000] = 32,
};
EXPORT_SYMBOL(amdtp_syt_intervals);
const unsigned int amdtp_rate_table[CIP_SFC_COUNT] = {
[CIP_SFC_32000] = 32000,
[CIP_SFC_44100] = 44100,
[CIP_SFC_48000] = 48000,
[CIP_SFC_88200] = 88200,
[CIP_SFC_96000] = 96000,
[CIP_SFC_176400] = 176400,
[CIP_SFC_192000] = 192000,
};
EXPORT_SYMBOL(amdtp_rate_table);
static int apply_constraint_to_size(struct snd_pcm_hw_params *params,
struct snd_pcm_hw_rule *rule)
{
struct snd_interval *s = hw_param_interval(params, rule->var);
const struct snd_interval *r =
hw_param_interval_c(params, SNDRV_PCM_HW_PARAM_RATE);
struct snd_interval t = {0};
unsigned int step = 0;
int i;
for (i = 0; i < CIP_SFC_COUNT; ++i) {
if (snd_interval_test(r, amdtp_rate_table[i]))
step = max(step, amdtp_syt_intervals[i]);
}
t.min = roundup(s->min, step);
t.max = rounddown(s->max, step);
t.integer = 1;
return snd_interval_refine(s, &t);
}
/**
* amdtp_stream_add_pcm_hw_constraints - add hw constraints for PCM substream
* @s: the AMDTP stream, which must be initialized.
* @runtime: the PCM substream runtime
*/
int amdtp_stream_add_pcm_hw_constraints(struct amdtp_stream *s,
struct snd_pcm_runtime *runtime)
{
struct snd_pcm_hardware *hw = &runtime->hw;
int err;
hw->info = SNDRV_PCM_INFO_BATCH |
SNDRV_PCM_INFO_BLOCK_TRANSFER |
SNDRV_PCM_INFO_INTERLEAVED |
SNDRV_PCM_INFO_JOINT_DUPLEX |
SNDRV_PCM_INFO_MMAP |
SNDRV_PCM_INFO_MMAP_VALID;
/* SNDRV_PCM_INFO_BATCH */
hw->periods_min = 2;
hw->periods_max = UINT_MAX;
/* bytes for a frame */
hw->period_bytes_min = 4 * hw->channels_max;
/* Just to prevent from allocating much pages. */
hw->period_bytes_max = hw->period_bytes_min * 2048;
hw->buffer_bytes_max = hw->period_bytes_max * hw->periods_min;
/*
* Currently firewire-lib processes 16 packets in one software
* interrupt callback. This equals to 2msec but actually the
* interval of the interrupts has a jitter.
* Additionally, even if adding a constraint to fit period size to
* 2msec, actual calculated frames per period doesn't equal to 2msec,
* depending on sampling rate.
* Anyway, the interval to call snd_pcm_period_elapsed() cannot 2msec.
* Here let us use 5msec for safe period interrupt.
*/
err = snd_pcm_hw_constraint_minmax(runtime,
SNDRV_PCM_HW_PARAM_PERIOD_TIME,
5000, UINT_MAX);
if (err < 0)
goto end;
/* Non-Blocking stream has no more constraints */
if (!(s->flags & CIP_BLOCKING))
goto end;
/*
* One AMDTP packet can include some frames. In blocking mode, the
* number equals to SYT_INTERVAL. So the number is 8, 16 or 32,
* depending on its sampling rate. For accurate period interrupt, it's
* preferrable to align period/buffer sizes to current SYT_INTERVAL.
*/
err = snd_pcm_hw_rule_add(runtime, 0, SNDRV_PCM_HW_PARAM_PERIOD_SIZE,
apply_constraint_to_size, NULL,
SNDRV_PCM_HW_PARAM_PERIOD_SIZE,
SNDRV_PCM_HW_PARAM_RATE, -1);
if (err < 0)
goto end;
err = snd_pcm_hw_rule_add(runtime, 0, SNDRV_PCM_HW_PARAM_BUFFER_SIZE,
apply_constraint_to_size, NULL,
SNDRV_PCM_HW_PARAM_BUFFER_SIZE,
SNDRV_PCM_HW_PARAM_RATE, -1);
if (err < 0)
goto end;
end:
return err;
}
EXPORT_SYMBOL(amdtp_stream_add_pcm_hw_constraints);
/**
* amdtp_stream_set_parameters - set stream parameters
* @s: the AMDTP stream to configure
* @rate: the sample rate
* @data_block_quadlets: the size of a data block in quadlet unit
*
* The parameters must be set before the stream is started, and must not be
* changed while the stream is running.
*/
int amdtp_stream_set_parameters(struct amdtp_stream *s, unsigned int rate,
unsigned int data_block_quadlets)
{
unsigned int sfc;
for (sfc = 0; sfc < ARRAY_SIZE(amdtp_rate_table); ++sfc) {
if (amdtp_rate_table[sfc] == rate)
break;
}
if (sfc == ARRAY_SIZE(amdtp_rate_table))
return -EINVAL;
s->sfc = sfc;
s->data_block_quadlets = data_block_quadlets;
s->syt_interval = amdtp_syt_intervals[sfc];
// default buffering in the device.
if (s->direction == AMDTP_OUT_STREAM) {
s->ctx_data.rx.transfer_delay =
TRANSFER_DELAY_TICKS - TICKS_PER_CYCLE;
if (s->flags & CIP_BLOCKING) {
// additional buffering needed to adjust for no-data
// packets.
s->ctx_data.rx.transfer_delay +=
TICKS_PER_SECOND * s->syt_interval / rate;
}
}
return 0;
}
EXPORT_SYMBOL(amdtp_stream_set_parameters);
/**
* amdtp_stream_get_max_payload - get the stream's packet size
* @s: the AMDTP stream
*
* This function must not be called before the stream has been configured
* with amdtp_stream_set_parameters().
*/
unsigned int amdtp_stream_get_max_payload(struct amdtp_stream *s)
{
unsigned int multiplier = 1;
unsigned int cip_header_size = 0;
if (s->flags & CIP_JUMBO_PAYLOAD)
multiplier = 5;
2017-03-31 21:06:07 +08:00
if (!(s->flags & CIP_NO_HEADER))
cip_header_size = sizeof(__be32) * 2;
return cip_header_size +
s->syt_interval * s->data_block_quadlets * sizeof(__be32) * multiplier;
}
EXPORT_SYMBOL(amdtp_stream_get_max_payload);
/**
* amdtp_stream_pcm_prepare - prepare PCM device for running
* @s: the AMDTP stream
*
* This function should be called from the PCM device's .prepare callback.
*/
void amdtp_stream_pcm_prepare(struct amdtp_stream *s)
{
tasklet_kill(&s->period_tasklet);
s->pcm_buffer_pointer = 0;
s->pcm_period_pointer = 0;
}
EXPORT_SYMBOL(amdtp_stream_pcm_prepare);
static unsigned int calculate_data_blocks(struct amdtp_stream *s,
unsigned int syt)
{
unsigned int phase, data_blocks;
/* Blocking mode. */
if (s->flags & CIP_BLOCKING) {
/* This module generate empty packet for 'no data'. */
if (syt == CIP_SYT_NO_INFO)
data_blocks = 0;
else
data_blocks = s->syt_interval;
/* Non-blocking mode. */
} else {
if (!cip_sfc_is_base_44100(s->sfc)) {
// Sample_rate / 8000 is an integer, and precomputed.
data_blocks = s->ctx_data.rx.data_block_state;
} else {
phase = s->ctx_data.rx.data_block_state;
/*
* This calculates the number of data blocks per packet so that
* 1) the overall rate is correct and exactly synchronized to
* the bus clock, and
* 2) packets with a rounded-up number of blocks occur as early
* as possible in the sequence (to prevent underruns of the
* device's buffer).
*/
if (s->sfc == CIP_SFC_44100)
/* 6 6 5 6 5 6 5 ... */
data_blocks = 5 + ((phase & 1) ^
(phase == 0 || phase >= 40));
else
/* 12 11 11 11 11 ... or 23 22 22 22 22 ... */
data_blocks = 11 * (s->sfc >> 1) + (phase == 0);
if (++phase >= (80 >> (s->sfc >> 1)))
phase = 0;
s->ctx_data.rx.data_block_state = phase;
}
}
return data_blocks;
}
static unsigned int calculate_syt(struct amdtp_stream *s,
unsigned int cycle)
{
unsigned int syt_offset, phase, index, syt;
if (s->ctx_data.rx.last_syt_offset < TICKS_PER_CYCLE) {
if (!cip_sfc_is_base_44100(s->sfc))
syt_offset = s->ctx_data.rx.last_syt_offset +
s->ctx_data.rx.syt_offset_state;
else {
/*
* The time, in ticks, of the n'th SYT_INTERVAL sample is:
* n * SYT_INTERVAL * 24576000 / sample_rate
* Modulo TICKS_PER_CYCLE, the difference between successive
* elements is about 1386.23. Rounding the results of this
* formula to the SYT precision results in a sequence of
* differences that begins with:
* 1386 1386 1387 1386 1386 1386 1387 1386 1386 1386 1387 ...
* This code generates _exactly_ the same sequence.
*/
phase = s->ctx_data.rx.syt_offset_state;
index = phase % 13;
syt_offset = s->ctx_data.rx.last_syt_offset;
syt_offset += 1386 + ((index && !(index & 3)) ||
phase == 146);
if (++phase >= 147)
phase = 0;
s->ctx_data.rx.syt_offset_state = phase;
}
} else
syt_offset = s->ctx_data.rx.last_syt_offset - TICKS_PER_CYCLE;
s->ctx_data.rx.last_syt_offset = syt_offset;
if (syt_offset < TICKS_PER_CYCLE) {
syt_offset += s->ctx_data.rx.transfer_delay;
syt = (cycle + syt_offset / TICKS_PER_CYCLE) << 12;
syt += syt_offset % TICKS_PER_CYCLE;
return syt & CIP_SYT_MASK;
} else {
return CIP_SYT_NO_INFO;
}
}
static void update_pcm_pointers(struct amdtp_stream *s,
struct snd_pcm_substream *pcm,
unsigned int frames)
{
unsigned int ptr;
ptr = s->pcm_buffer_pointer + frames;
if (ptr >= pcm->runtime->buffer_size)
ptr -= pcm->runtime->buffer_size;
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-24 05:07:29 +08:00
WRITE_ONCE(s->pcm_buffer_pointer, ptr);
s->pcm_period_pointer += frames;
if (s->pcm_period_pointer >= pcm->runtime->period_size) {
s->pcm_period_pointer -= pcm->runtime->period_size;
tasklet_hi_schedule(&s->period_tasklet);
}
}
static void pcm_period_tasklet(unsigned long data)
{
struct amdtp_stream *s = (void *)data;
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-24 05:07:29 +08:00
struct snd_pcm_substream *pcm = READ_ONCE(s->pcm);
if (pcm)
snd_pcm_period_elapsed(pcm);
}
static int queue_packet(struct amdtp_stream *s, struct fw_iso_packet *params)
{
int err;
params->interrupt = IS_ALIGNED(s->packet_index + 1, INTERRUPT_INTERVAL);
params->tag = s->tag;
params->sy = 0;
err = fw_iso_context_queue(s->context, params, &s->buffer.iso_buffer,
s->buffer.packets[s->packet_index].offset);
if (err < 0) {
dev_err(&s->unit->device, "queueing error: %d\n", err);
goto end;
}
if (++s->packet_index >= QUEUE_LENGTH)
s->packet_index = 0;
end:
return err;
}
static inline int queue_out_packet(struct amdtp_stream *s,
struct fw_iso_packet *params)
{
params->skip =
!!(params->header_length == 0 && params->payload_length == 0);
return queue_packet(s, params);
}
static inline int queue_in_packet(struct amdtp_stream *s,
struct fw_iso_packet *params)
{
// Queue one packet for IR context.
params->header_length = s->ctx_data.tx.ctx_header_size;
params->payload_length = s->ctx_data.tx.max_ctx_payload_length;
params->skip = false;
return queue_packet(s, params);
}
static void generate_cip_header(struct amdtp_stream *s, __be32 cip_header[2],
unsigned int syt)
{
cip_header[0] = cpu_to_be32(READ_ONCE(s->source_node_id_field) |
(s->data_block_quadlets << CIP_DBS_SHIFT) |
((s->sph << CIP_SPH_SHIFT) & CIP_SPH_MASK) |
s->data_block_counter);
cip_header[1] = cpu_to_be32(CIP_EOH |
((s->fmt << CIP_FMT_SHIFT) & CIP_FMT_MASK) |
((s->ctx_data.rx.fdf << CIP_FDF_SHIFT) & CIP_FDF_MASK) |
(syt & CIP_SYT_MASK));
}
static void build_it_pkt_header(struct amdtp_stream *s, unsigned int cycle,
struct fw_iso_packet *params,
unsigned int data_blocks, unsigned int syt,
unsigned int index)
{
unsigned int payload_length;
__be32 *cip_header;
ALSA: firewire-lib: add helper functions as interfaces between packet streaming layer and data block processing layer ALSA PCM framework uses PCM buffer with a concept of 'period' to synchronize userspace operations to hardware for nearly-realtime processing. Each driver implements snd_pcm_period_elapsed() to tell across of the period boundary to ALSA PCM middleware. To call the function, some drivers utilize hardware interrupt handlers, the others count handled PCM frames. Drivers for sound units on IEEE 1394 bus are the latter. They use two buffers; PCM buffer and DMA buffer for IEEE 1394 isochronous packet. PCM frames are copied between these two buffers and 'amdtp_stream' structure counts the handled PCM frames. Then, snd_pcm_period_elapsed() is called if required. Essentially, packet streaming layer should not be responsible for PCM frame processing. The PCM frame processing should be handled in each data block processing layer as a result of handling data blocks. Although, PCM frame counting is a common work for all of protocols which ALSA firewire stack is going to support. This commit adds two new helper functions as interfaces between packet streaming layer to data block processing layer. In future, each data block processing layer implements these functions. The packet streaming layer calls data block processing layer per packet by calling the functions. The data block processing layer processes data blocks and PCM frames, and returns the number of processed PCM frames. Then the packet streaming layer calculates handled PCM frames and calls snd_pcm_period_elapsed(). Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2015-09-19 10:21:52 +08:00
payload_length = data_blocks * sizeof(__be32) * s->data_block_quadlets;
params->payload_length = payload_length;
if (s->flags & CIP_DBC_IS_END_EVENT) {
s->data_block_counter =
(s->data_block_counter + data_blocks) & 0xff;
}
if (!(s->flags & CIP_NO_HEADER)) {
cip_header = (__be32 *)params->header;
generate_cip_header(s, cip_header, syt);
params->header_length = 2 * sizeof(__be32);
payload_length += params->header_length;
} else {
cip_header = NULL;
}
trace_amdtp_packet(s, cycle, cip_header, payload_length, data_blocks,
index);
if (!(s->flags & CIP_DBC_IS_END_EVENT)) {
s->data_block_counter =
(s->data_block_counter + data_blocks) & 0xff;
}
2017-03-31 21:06:07 +08:00
}
static int check_cip_header(struct amdtp_stream *s, const __be32 *buf,
unsigned int payload_length,
unsigned int *data_blocks, unsigned int *dbc,
unsigned int *syt)
{
u32 cip_header[2];
unsigned int sph;
unsigned int fmt;
unsigned int fdf;
bool lost;
cip_header[0] = be32_to_cpu(buf[0]);
cip_header[1] = be32_to_cpu(buf[1]);
/*
* This module supports 'Two-quadlet CIP header with SYT field'.
* For convenience, also check FMT field is AM824 or not.
*/
if ((((cip_header[0] & CIP_EOH_MASK) == CIP_EOH) ||
((cip_header[1] & CIP_EOH_MASK) != CIP_EOH)) &&
(!(s->flags & CIP_HEADER_WITHOUT_EOH))) {
dev_info_ratelimited(&s->unit->device,
"Invalid CIP header for AMDTP: %08X:%08X\n",
cip_header[0], cip_header[1]);
return -EAGAIN;
}
/* Check valid protocol or not. */
sph = (cip_header[0] & CIP_SPH_MASK) >> CIP_SPH_SHIFT;
fmt = (cip_header[1] & CIP_FMT_MASK) >> CIP_FMT_SHIFT;
if (sph != s->sph || fmt != s->fmt) {
dev_info_ratelimited(&s->unit->device,
"Detect unexpected protocol: %08x %08x\n",
cip_header[0], cip_header[1]);
return -EAGAIN;
}
/* Calculate data blocks */
fdf = (cip_header[1] & CIP_FDF_MASK) >> CIP_FDF_SHIFT;
if (payload_length < sizeof(__be32) * 2 ||
(fmt == CIP_FMT_AM && fdf == AMDTP_FDF_NO_DATA)) {
*data_blocks = 0;
} else {
unsigned int data_block_quadlets =
(cip_header[0] & CIP_DBS_MASK) >> CIP_DBS_SHIFT;
/* avoid division by zero */
if (data_block_quadlets == 0) {
dev_err(&s->unit->device,
"Detect invalid value in dbs field: %08X\n",
cip_header[0]);
return -EPROTO;
}
if (s->flags & CIP_WRONG_DBS)
data_block_quadlets = s->data_block_quadlets;
*data_blocks = (payload_length / sizeof(__be32) - 2) /
data_block_quadlets;
}
/* Check data block counter continuity */
*dbc = cip_header[0] & CIP_DBC_MASK;
if (*data_blocks == 0 && (s->flags & CIP_EMPTY_HAS_WRONG_DBC) &&
s->data_block_counter != UINT_MAX)
*dbc = s->data_block_counter;
ALSA: fireworks/firewire-lib: add support for recent firmware quirk Fireworks uses TSB43CB43(IceLynx-Micro) as its IEC 61883-1/6 interface. This chip includes ARM7 core, and loads and runs program. The firmware is stored in on-board memory and loaded every powering-on from it. Echo Audio ships several versions of firmwares for each model. These firmwares have each quirk and the quirk changes a sequence of packets. As long as I investigated, AudioFire2/AudioFire4/AudioFirePre8 have a quirk to transfer a first packet with 0x02 in its dbc field. This causes ALSA Fireworks driver to detect discontinuity. In this case, firmware version 5.7.0, 5.7.3 and 5.8.0 are used. Payload CIP CIP quadlets header1 header2 02 00050002 90ffffff <- 42 0005000a 90013000 42 00050012 90014400 42 0005001a 90015800 02 0005001a 90ffffff 42 00050022 90019000 42 0005002a 9001a400 42 00050032 9001b800 02 00050032 90ffffff 42 0005003a 9001d000 42 00050042 9001e400 42 0005004a 9001f800 02 0005004a 90ffffff (AudioFire2 with firmware version 5.7.) $ dmesg snd-fireworks fw1.0: Detect discontinuity of CIP: 00 02 These models, AudioFire8 (since Jul 2009 ) and Gibson Robot Interface Pack series uses the same ARM binary as their firmware. Thus, this quirk may be observed among them. This commit adds a new member for AMDTP structure. This member represents the value of dbc field in a first AMDTP packet. Drivers can set it with a preferred value according to model's quirk. Tested-by: Johannes Oertei <johannes.oertel@uni-due.de> Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Cc: <stable@vger.kernel.org> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2015-08-05 08:21:05 +08:00
if (((s->flags & CIP_SKIP_DBC_ZERO_CHECK) &&
*dbc == s->ctx_data.tx.first_dbc) ||
ALSA: fireworks/firewire-lib: add support for recent firmware quirk Fireworks uses TSB43CB43(IceLynx-Micro) as its IEC 61883-1/6 interface. This chip includes ARM7 core, and loads and runs program. The firmware is stored in on-board memory and loaded every powering-on from it. Echo Audio ships several versions of firmwares for each model. These firmwares have each quirk and the quirk changes a sequence of packets. As long as I investigated, AudioFire2/AudioFire4/AudioFirePre8 have a quirk to transfer a first packet with 0x02 in its dbc field. This causes ALSA Fireworks driver to detect discontinuity. In this case, firmware version 5.7.0, 5.7.3 and 5.8.0 are used. Payload CIP CIP quadlets header1 header2 02 00050002 90ffffff <- 42 0005000a 90013000 42 00050012 90014400 42 0005001a 90015800 02 0005001a 90ffffff 42 00050022 90019000 42 0005002a 9001a400 42 00050032 9001b800 02 00050032 90ffffff 42 0005003a 9001d000 42 00050042 9001e400 42 0005004a 9001f800 02 0005004a 90ffffff (AudioFire2 with firmware version 5.7.) $ dmesg snd-fireworks fw1.0: Detect discontinuity of CIP: 00 02 These models, AudioFire8 (since Jul 2009 ) and Gibson Robot Interface Pack series uses the same ARM binary as their firmware. Thus, this quirk may be observed among them. This commit adds a new member for AMDTP structure. This member represents the value of dbc field in a first AMDTP packet. Drivers can set it with a preferred value according to model's quirk. Tested-by: Johannes Oertei <johannes.oertel@uni-due.de> Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Cc: <stable@vger.kernel.org> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2015-08-05 08:21:05 +08:00
s->data_block_counter == UINT_MAX) {
lost = false;
} else if (!(s->flags & CIP_DBC_IS_END_EVENT)) {
lost = *dbc != s->data_block_counter;
} else {
unsigned int dbc_interval;
if (*data_blocks > 0 && s->ctx_data.tx.dbc_interval > 0)
dbc_interval = s->ctx_data.tx.dbc_interval;
else
dbc_interval = *data_blocks;
lost = *dbc != ((s->data_block_counter + dbc_interval) & 0xff);
}
if (lost) {
dev_err(&s->unit->device,
"Detect discontinuity of CIP: %02X %02X\n",
s->data_block_counter, *dbc);
return -EIO;
}
*syt = cip_header[1] & CIP_SYT_MASK;
return 0;
}
static int parse_ir_ctx_header(struct amdtp_stream *s, unsigned int cycle,
const __be32 *ctx_header,
unsigned int *payload_length,
unsigned int *data_blocks, unsigned int *syt,
unsigned int index)
{
unsigned int dbc;
const __be32 *cip_header;
int err;
*payload_length = be32_to_cpu(ctx_header[0]) >> ISO_DATA_LENGTH_SHIFT;
if (*payload_length > s->ctx_data.tx.ctx_header_size +
s->ctx_data.tx.max_ctx_payload_length) {
dev_err(&s->unit->device,
"Detect jumbo payload: %04x %04x\n",
*payload_length, s->ctx_data.tx.max_ctx_payload_length);
return -EIO;
}
if (!(s->flags & CIP_NO_HEADER)) {
cip_header = ctx_header + 2;
err = check_cip_header(s, cip_header, *payload_length,
data_blocks, &dbc, syt);
if (err < 0)
return err;
} else {
cip_header = NULL;
err = 0;
*data_blocks = *payload_length / sizeof(__be32) /
s->data_block_quadlets;
*syt = 0;
if (s->data_block_counter != UINT_MAX)
dbc = s->data_block_counter;
else
dbc = 0;
}
s->data_block_counter = dbc;
trace_amdtp_packet(s, cycle, cip_header, *payload_length, *data_blocks,
index);
return err;
}
// In CYCLE_TIMER register of IEEE 1394, 7 bits are used to represent second. On
// the other hand, in DMA descriptors of 1394 OHCI, 3 bits are used to represent
// it. Thus, via Linux firewire subsystem, we can get the 3 bits for second.
static inline u32 compute_cycle_count(__be32 ctx_header_tstamp)
{
u32 tstamp = be32_to_cpu(ctx_header_tstamp) & HEADER_TSTAMP_MASK;
return (((tstamp >> 13) & 0x07) * 8000) + (tstamp & 0x1fff);
}
static inline u32 increment_cycle_count(u32 cycle, unsigned int addend)
{
cycle += addend;
if (cycle >= 8 * CYCLES_PER_SECOND)
cycle -= 8 * CYCLES_PER_SECOND;
return cycle;
}
// Align to actual cycle count for the packet which is going to be scheduled.
// This module queued the same number of isochronous cycle as QUEUE_LENGTH to
// skip isochronous cycle, therefore it's OK to just increment the cycle by
// QUEUE_LENGTH for scheduled cycle.
static inline u32 compute_it_cycle(const __be32 ctx_header_tstamp)
{
u32 cycle = compute_cycle_count(ctx_header_tstamp);
return increment_cycle_count(cycle, QUEUE_LENGTH);
}
static inline void cancel_stream(struct amdtp_stream *s)
{
s->packet_index = -1;
if (in_interrupt())
amdtp_stream_pcm_abort(s);
WRITE_ONCE(s->pcm_buffer_pointer, SNDRV_PCM_POS_XRUN);
}
static void out_stream_callback(struct fw_iso_context *context, u32 tstamp,
size_t header_length, void *header,
void *private_data)
{
struct amdtp_stream *s = private_data;
const __be32 *ctx_header = header;
unsigned int i, packets = header_length / sizeof(*ctx_header);
if (s->packet_index < 0)
return;
for (i = 0; i < packets; ++i) {
u32 cycle;
unsigned int syt;
unsigned int data_block;
__be32 *buffer;
unsigned int pcm_frames;
struct {
struct fw_iso_packet params;
__be32 header[IT_PKT_HEADER_SIZE_CIP / sizeof(__be32)];
} template = { {0}, {0} };
struct snd_pcm_substream *pcm;
cycle = compute_it_cycle(*ctx_header);
syt = calculate_syt(s, cycle);
data_block = calculate_data_blocks(s, syt);
buffer = s->buffer.packets[s->packet_index].buffer;
pcm_frames = s->process_data_blocks(s, buffer, data_block, &syt);
build_it_pkt_header(s, cycle, &template.params, data_block, syt,
i);
if (queue_out_packet(s, &template.params) < 0) {
cancel_stream(s);
return;
}
pcm = READ_ONCE(s->pcm);
if (pcm && pcm_frames > 0)
update_pcm_pointers(s, pcm, pcm_frames);
++ctx_header;
}
fw_iso_context_queue_flush(s->context);
}
static void in_stream_callback(struct fw_iso_context *context, u32 tstamp,
size_t header_length, void *header,
void *private_data)
{
struct amdtp_stream *s = private_data;
unsigned int i, packets;
__be32 *ctx_header = header;
if (s->packet_index < 0)
return;
// The number of packets in buffer.
packets = header_length / s->ctx_data.tx.ctx_header_size;
for (i = 0; i < packets; i++) {
u32 cycle;
unsigned int payload_length;
unsigned int data_blocks;
unsigned int syt;
__be32 *buffer;
unsigned int pcm_frames = 0;
struct fw_iso_packet params = {0};
struct snd_pcm_substream *pcm;
int err;
cycle = compute_cycle_count(ctx_header[1]);
err = parse_ir_ctx_header(s, cycle, ctx_header, &payload_length,
&data_blocks, &syt, i);
if (err < 0 && err != -EAGAIN)
break;
if (err >= 0) {
buffer = s->buffer.packets[s->packet_index].buffer;
pcm_frames = s->process_data_blocks(s, buffer,
data_blocks, &syt);
if (!(s->flags & CIP_DBC_IS_END_EVENT)) {
s->data_block_counter += data_blocks;
s->data_block_counter &= 0xff;
}
}
if (queue_in_packet(s, &params) < 0)
break;
pcm = READ_ONCE(s->pcm);
if (pcm && pcm_frames > 0)
update_pcm_pointers(s, pcm, pcm_frames);
ctx_header += s->ctx_data.tx.ctx_header_size / sizeof(*ctx_header);
}
ALSA: firewire-lib: handle IT/IR contexts in each software interrupt context In clause 6.3 of IEC 61883-6:2000, there's an explanation about processing of presentation timestamp. In the clause, we can see "If a function block receives a CIP, processes it and subsequently re-transmits it, then the SYT of the outgoing CIP shall be the sum of the incoming SYT and the processing delay." ALSA firewire stack has an implementation to partly satisfy this specification. Developers assumed the stack to perform as an Audio function block[1]. Following to the assumption, current implementation of ALSA firewire stack use one software interrupt context to handle both of in/out packets. In most case, this is processed in 1394 OHCI IR context independently of the opposite context. Thus, this implementation uses longer CPU time in the software interrupt context. This is not better for whole system. Against the assumption, I confirmed that each ASIC for IEC 61883-1/6 doesn't necessarily expect it to the stack. Thus, current implementation of ALSA firewire stack includes over-engineering. This commit purges the implementation. As a result, packets of one direction are handled in one software interrupt context and spends minimum CPU time. [1] [alsa-devel] [PATCH 0/8] [RFC] new driver for Echo Audio's Fireworks based devices http://mailman.alsa-project.org/pipermail/alsa-devel/2013-June/062660.html Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2016-05-09 22:15:53 +08:00
/* Queueing error or detecting invalid payload. */
if (i < packets) {
cancel_stream(s);
return;
}
fw_iso_context_queue_flush(s->context);
}
/* this is executed one time */
static void amdtp_stream_first_callback(struct fw_iso_context *context,
u32 tstamp, size_t header_length,
void *header, void *private_data)
{
struct amdtp_stream *s = private_data;
const __be32 *ctx_header = header;
u32 cycle;
/*
* For in-stream, first packet has come.
* For out-stream, prepared to transmit first packet
*/
s->callbacked = true;
wake_up(&s->callback_wait);
if (s->direction == AMDTP_IN_STREAM) {
cycle = compute_cycle_count(ctx_header[1]);
context->callback.sc = in_stream_callback;
} else {
cycle = compute_it_cycle(*ctx_header);
context->callback.sc = out_stream_callback;
}
s->start_cycle = cycle;
context->callback.sc(context, tstamp, header_length, header, s);
}
/**
* amdtp_stream_start - start transferring packets
* @s: the AMDTP stream to start
* @channel: the isochronous channel on the bus
* @speed: firewire speed code
*
* The stream cannot be started until it has been configured with
* amdtp_stream_set_parameters() and it must be started before any PCM or MIDI
* device can be started.
*/
int amdtp_stream_start(struct amdtp_stream *s, int channel, int speed)
{
static const struct {
unsigned int data_block;
unsigned int syt_offset;
} *entry, initial_state[] = {
[CIP_SFC_32000] = { 4, 3072 },
[CIP_SFC_48000] = { 6, 1024 },
[CIP_SFC_96000] = { 12, 1024 },
[CIP_SFC_192000] = { 24, 1024 },
[CIP_SFC_44100] = { 0, 67 },
[CIP_SFC_88200] = { 0, 67 },
[CIP_SFC_176400] = { 0, 67 },
};
unsigned int ctx_header_size;
unsigned int max_ctx_payload_size;
enum dma_data_direction dir;
int type, tag, err;
mutex_lock(&s->mutex);
if (WARN_ON(amdtp_stream_running(s) ||
(s->data_block_quadlets < 1))) {
err = -EBADFD;
goto err_unlock;
}
if (s->direction == AMDTP_IN_STREAM) {
ALSA: bebob/firewire-lib: Add a quirk for discontinuity at bus reset Normal BeBoB firmware has a quirk. When receiving bus reset, it transmits packets with discontinuous value in dbc field. This causes two situation, one is to abort streaming by firewire-lib as a result of detecting the discontinuity. Another is to call driver's .update() because of bus reset. These two is generated independently. (The former depends on isochronous stream and the latter depends on IEEE1394 bus driver.) When BeBoB driver works with XRUN-recoverable applications, this situation looks like stream_start_duplex() call followed by stream_update_duplex() call because applications will call snd_pcm_prepare() immediately at XRUN. To update connections and streams at first, this commit use completion. When queueing error occurs, stream_start_duplex() is forced to wait maximum 1000msec. During this, when .update() is called, the completion is waken and stream_start_duplex() is processed without breaking connections. At bus reset, stream_start_duplex() shouldn't break/establish connections and stream_update_duplex() should update connections because a caller of fw_iso_resources_allocate() is responsible for calling fw_iso_resources_update() on bus reset. This commit also adds a flag, which has an effect to skip checking continuity for first packet. This flag is useful for BeBoB quirk to start handling packets during streaming. Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2014-04-25 21:45:16 +08:00
s->data_block_counter = UINT_MAX;
} else {
entry = &initial_state[s->sfc];
ALSA: bebob/firewire-lib: Add a quirk for discontinuity at bus reset Normal BeBoB firmware has a quirk. When receiving bus reset, it transmits packets with discontinuous value in dbc field. This causes two situation, one is to abort streaming by firewire-lib as a result of detecting the discontinuity. Another is to call driver's .update() because of bus reset. These two is generated independently. (The former depends on isochronous stream and the latter depends on IEEE1394 bus driver.) When BeBoB driver works with XRUN-recoverable applications, this situation looks like stream_start_duplex() call followed by stream_update_duplex() call because applications will call snd_pcm_prepare() immediately at XRUN. To update connections and streams at first, this commit use completion. When queueing error occurs, stream_start_duplex() is forced to wait maximum 1000msec. During this, when .update() is called, the completion is waken and stream_start_duplex() is processed without breaking connections. At bus reset, stream_start_duplex() shouldn't break/establish connections and stream_update_duplex() should update connections because a caller of fw_iso_resources_allocate() is responsible for calling fw_iso_resources_update() on bus reset. This commit also adds a flag, which has an effect to skip checking continuity for first packet. This flag is useful for BeBoB quirk to start handling packets during streaming. Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2014-04-25 21:45:16 +08:00
s->data_block_counter = 0;
s->ctx_data.rx.data_block_state = entry->data_block;
s->ctx_data.rx.syt_offset_state = entry->syt_offset;
s->ctx_data.rx.last_syt_offset = TICKS_PER_CYCLE;
}
/* initialize packet buffer */
if (s->direction == AMDTP_IN_STREAM) {
dir = DMA_FROM_DEVICE;
type = FW_ISO_CONTEXT_RECEIVE;
if (!(s->flags & CIP_NO_HEADER))
ctx_header_size = IR_CTX_HEADER_SIZE_CIP;
else
ctx_header_size = IR_CTX_HEADER_SIZE_NO_CIP;
max_ctx_payload_size = amdtp_stream_get_max_payload(s) -
ctx_header_size;
} else {
dir = DMA_TO_DEVICE;
type = FW_ISO_CONTEXT_TRANSMIT;
ctx_header_size = 0; // No effect for IT context.
max_ctx_payload_size = amdtp_stream_get_max_payload(s);
if (!(s->flags & CIP_NO_HEADER))
max_ctx_payload_size -= IT_PKT_HEADER_SIZE_CIP;
}
err = iso_packets_buffer_init(&s->buffer, s->unit, QUEUE_LENGTH,
max_ctx_payload_size, dir);
if (err < 0)
goto err_unlock;
s->context = fw_iso_context_create(fw_parent_device(s->unit)->card,
type, channel, speed, ctx_header_size,
amdtp_stream_first_callback, s);
if (IS_ERR(s->context)) {
err = PTR_ERR(s->context);
if (err == -EBUSY)
dev_err(&s->unit->device,
"no free stream on this controller\n");
goto err_buffer;
}
amdtp_stream_update(s);
if (s->direction == AMDTP_IN_STREAM) {
s->ctx_data.tx.max_ctx_payload_length = max_ctx_payload_size;
s->ctx_data.tx.ctx_header_size = ctx_header_size;
}
ALSA: dice: fix kernel NULL pointer dereference due to invalid calculation for array index At a commit f91c9d7610a ('ALSA: firewire-lib: cache maximum length of payload to reduce function calls'), maximum size of payload for tx isochronous packet is cached to reduce the number of function calls. This cache was programmed to updated at a first callback of ohci1394 IR context. However, the maximum size is required to queueing packets before starting the isochronous context. As a result, the cached value is reused to queue packets in next time to starting the isochronous context. Then the cache is updated in a first callback of the isochronous context. This can cause kernel NULL pointer dereference in a below call graph: (sound/firewire/amdtp-stream.c) amdtp_stream_start() ->queue_in_packet() ->queue_packet() (drivers/firewire/core-iso.c) ->fw_iso_context_queue() ->struct fw_card_driver.queue_iso() (drivers/firewire/ohci.c) = ohci_queue_iso() ->queue_iso_packet_per_buffer() buffer->pages[page] The issued dereference occurs in a case that: - target unit supports different stream formats for sampling transmission frequency. - maximum length of payload for tx stream in a first trial is bigger than the length in a second trial. In this case, correct number of pages are allocated for DMA and the 'pages' array has enough elements, while index of the element is wrongly calculated according to the old value of length of payload in a call of 'queue_in_packet()'. Then it causes the issue. This commit fixes the critical bug. This affects all of drivers in ALSA firewire stack in Linux kernel v4.12 or later. [12665.302360] BUG: unable to handle kernel NULL pointer dereference at 0000000000000030 [12665.302415] IP: ohci_queue_iso+0x47c/0x800 [firewire_ohci] [12665.302439] PGD 0 [12665.302440] P4D 0 [12665.302450] [12665.302470] Oops: 0000 [#1] SMP PTI [12665.302487] Modules linked in: ... [12665.303096] CPU: 1 PID: 12760 Comm: jackd Tainted: P OE 4.13.0-38-generic #43-Ubuntu [12665.303154] Hardware name: /DH77DF, BIOS KCH7710H.86A.0069.2012.0224.1825 02/24/2012 [12665.303215] task: ffff9ce87da2ae80 task.stack: ffffb5b8823d0000 [12665.303258] RIP: 0010:ohci_queue_iso+0x47c/0x800 [firewire_ohci] [12665.303301] RSP: 0018:ffffb5b8823d3ab8 EFLAGS: 00010086 [12665.303337] RAX: ffff9ce4f4876930 RBX: 0000000000000008 RCX: ffff9ce88a3955e0 [12665.303384] RDX: 0000000000000000 RSI: 0000000034877f00 RDI: 0000000000000000 [12665.303427] RBP: ffffb5b8823d3b68 R08: ffff9ce8ccb390a0 R09: ffff9ce877639ab0 [12665.303475] R10: 0000000000000108 R11: 0000000000000000 R12: 0000000000000003 [12665.303513] R13: 0000000000000000 R14: ffff9ce4f4876950 R15: 0000000000000000 [12665.303554] FS: 00007f2ec467f8c0(0000) GS:ffff9ce8df280000(0000) knlGS:0000000000000000 [12665.303600] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [12665.303633] CR2: 0000000000000030 CR3: 00000002dcf90004 CR4: 00000000000606e0 [12665.303674] Call Trace: [12665.303698] fw_iso_context_queue+0x18/0x20 [firewire_core] [12665.303735] queue_packet+0x88/0xe0 [snd_firewire_lib] [12665.303770] amdtp_stream_start+0x19b/0x270 [snd_firewire_lib] [12665.303811] start_streams+0x276/0x3c0 [snd_dice] [12665.303840] snd_dice_stream_start_duplex+0x1bf/0x480 [snd_dice] [12665.303882] ? vma_gap_callbacks_rotate+0x1e/0x30 [12665.303914] ? __rb_insert_augmented+0xab/0x240 [12665.303936] capture_prepare+0x3c/0x70 [snd_dice] [12665.303961] snd_pcm_do_prepare+0x1d/0x30 [snd_pcm] [12665.303985] snd_pcm_action_single+0x3b/0x90 [snd_pcm] [12665.304009] snd_pcm_action_nonatomic+0x68/0x70 [snd_pcm] [12665.304035] snd_pcm_prepare+0x68/0x90 [snd_pcm] [12665.304058] snd_pcm_common_ioctl1+0x4c0/0x940 [snd_pcm] [12665.304083] snd_pcm_capture_ioctl1+0x19b/0x250 [snd_pcm] [12665.304108] snd_pcm_capture_ioctl+0x27/0x40 [snd_pcm] [12665.304131] do_vfs_ioctl+0xa8/0x630 [12665.304148] ? entry_SYSCALL_64_after_hwframe+0xe9/0x139 [12665.304172] ? entry_SYSCALL_64_after_hwframe+0xe2/0x139 [12665.304195] ? entry_SYSCALL_64_after_hwframe+0xdb/0x139 [12665.304218] ? entry_SYSCALL_64_after_hwframe+0xd4/0x139 [12665.304242] ? entry_SYSCALL_64_after_hwframe+0xcd/0x139 [12665.304265] ? entry_SYSCALL_64_after_hwframe+0xc6/0x139 [12665.304288] ? entry_SYSCALL_64_after_hwframe+0xbf/0x139 [12665.304312] ? entry_SYSCALL_64_after_hwframe+0xb8/0x139 [12665.304335] ? entry_SYSCALL_64_after_hwframe+0xb1/0x139 [12665.304358] SyS_ioctl+0x79/0x90 [12665.304374] ? entry_SYSCALL_64_after_hwframe+0x72/0x139 [12665.304397] entry_SYSCALL_64_fastpath+0x24/0xab [12665.304417] RIP: 0033:0x7f2ec3750ef7 [12665.304433] RSP: 002b:00007fff99e31388 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 [12665.304465] RAX: ffffffffffffffda RBX: 00007fff99e312f0 RCX: 00007f2ec3750ef7 [12665.304494] RDX: 0000000000000000 RSI: 0000000000004140 RDI: 0000000000000007 [12665.304522] RBP: 0000556ebc63fd60 R08: 0000556ebc640560 R09: 0000000000000000 [12665.304553] R10: 0000000000000001 R11: 0000000000000246 R12: 0000556ebc63fcf0 [12665.304584] R13: 0000000000000000 R14: 0000000000000007 R15: 0000000000000000 [12665.304612] Code: 01 00 00 44 89 eb 45 31 ed 45 31 db 66 41 89 1e 66 41 89 5e 0c 66 45 89 5e 0e 49 8b 49 08 49 63 d4 4d 85 c0 49 63 ff 48 8b 14 d1 <48> 8b 72 30 41 8d 14 37 41 89 56 04 48 63 d3 0f 84 ce 00 00 00 [12665.304713] RIP: ohci_queue_iso+0x47c/0x800 [firewire_ohci] RSP: ffffb5b8823d3ab8 [12665.304743] CR2: 0000000000000030 [12665.317701] ---[ end trace 9d55b056dd52a19f ]--- Fixes: f91c9d7610a ('ALSA: firewire-lib: cache maximum length of payload to reduce function calls') Cc: <stable@vger.kernel.org> # v4.12+ Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-04-29 14:01:46 +08:00
2017-03-31 21:06:07 +08:00
if (s->flags & CIP_NO_HEADER)
s->tag = TAG_NO_CIP_HEADER;
else
s->tag = TAG_CIP;
s->packet_index = 0;
do {
struct fw_iso_packet params;
if (s->direction == AMDTP_IN_STREAM) {
err = queue_in_packet(s, &params);
} else {
params.header_length = 0;
params.payload_length = 0;
err = queue_out_packet(s, &params);
}
if (err < 0)
goto err_context;
} while (s->packet_index > 0);
/* NOTE: TAG1 matches CIP. This just affects in stream. */
tag = FW_ISO_CONTEXT_MATCH_TAG1;
2017-03-31 21:06:07 +08:00
if ((s->flags & CIP_EMPTY_WITH_TAG0) || (s->flags & CIP_NO_HEADER))
tag |= FW_ISO_CONTEXT_MATCH_TAG0;
s->callbacked = false;
err = fw_iso_context_start(s->context, -1, 0, tag);
if (err < 0)
goto err_context;
mutex_unlock(&s->mutex);
return 0;
err_context:
fw_iso_context_destroy(s->context);
s->context = ERR_PTR(-1);
err_buffer:
iso_packets_buffer_destroy(&s->buffer, s->unit);
err_unlock:
mutex_unlock(&s->mutex);
return err;
}
EXPORT_SYMBOL(amdtp_stream_start);
/**
* amdtp_stream_pcm_pointer - get the PCM buffer position
* @s: the AMDTP stream that transports the PCM data
*
* Returns the current buffer position, in frames.
*/
unsigned long amdtp_stream_pcm_pointer(struct amdtp_stream *s)
{
ALSA: firewire-lib: permit to flush queued packets only in process context for better PCM period granularity These three commits were merged to improve PCM pointer granularity. commit 76fb87894828 ("ALSA: firewire-lib: taskletize the snd_pcm_period_elapsed() call") commit e9148dddc3c7 ("ALSA: firewire-lib: flush completed packets when reading PCM position") commit 92b862c7d685 ("ALSA: firewire-lib: optimize packet flushing") The point of them is to handle queued packets not only in software IRQ context of IR/IT contexts, but also in process context. As a result of handling packets, period tasklet is scheduled when acrossing PCM period boundary. This is to prevent recursive call of 'struct snd_pcm_ops.pointer()' in the same context. When the pointer callback is executed in the process context, it's better to avoid the second callback in the software IRQ context. The software IRQ context runs immediately after scheduled in the process context because few packets are queued yet. For the aim, 'pointer_flush' is used, however it causes a race condition between the process context and software IRQ context of IR/IT contexts. Practically, this race is not so critical because it influences process context to skip flushing queued packet and to get worse granularity of PCM pointer. The race condition is quite rare but it should be improved for stable service. The similar effect can be achieved by using 'in_interrupt()' macro. This commit obsoletes 'pointer_flush' with it. Acked-by: Clemens Ladisch <clemens@ladisch.de> Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2016-05-12 01:17:39 +08:00
/*
* This function is called in software IRQ context of period_tasklet or
* process context.
*
* When the software IRQ context was scheduled by software IRQ context
* of IR/IT contexts, queued packets were already handled. Therefore,
* no need to flush the queue in buffer anymore.
*
* When the process context reach here, some packets will be already
* queued in the buffer. These packets should be handled immediately
* to keep better granularity of PCM pointer.
*
* Later, the process context will sometimes schedules software IRQ
* context of the period_tasklet. Then, no need to flush the queue by
* the same reason as described for IR/IT contexts.
*/
if (!in_interrupt() && amdtp_stream_running(s))
fw_iso_context_flush_completions(s->context);
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-24 05:07:29 +08:00
return READ_ONCE(s->pcm_buffer_pointer);
}
EXPORT_SYMBOL(amdtp_stream_pcm_pointer);
2017-06-07 08:38:05 +08:00
/**
* amdtp_stream_pcm_ack - acknowledge queued PCM frames
* @s: the AMDTP stream that transfers the PCM frames
*
* Returns zero always.
*/
int amdtp_stream_pcm_ack(struct amdtp_stream *s)
{
/*
* Process isochronous packets for recent isochronous cycle to handle
* queued PCM frames.
*/
if (amdtp_stream_running(s))
fw_iso_context_flush_completions(s->context);
return 0;
}
EXPORT_SYMBOL(amdtp_stream_pcm_ack);
/**
* amdtp_stream_update - update the stream after a bus reset
* @s: the AMDTP stream
*/
void amdtp_stream_update(struct amdtp_stream *s)
{
/* Precomputing. */
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-24 05:07:29 +08:00
WRITE_ONCE(s->source_node_id_field,
(fw_parent_device(s->unit)->card->node_id << CIP_SID_SHIFT) & CIP_SID_MASK);
}
EXPORT_SYMBOL(amdtp_stream_update);
/**
* amdtp_stream_stop - stop sending packets
* @s: the AMDTP stream to stop
*
* All PCM and MIDI devices of the stream must be stopped before the stream
* itself can be stopped.
*/
void amdtp_stream_stop(struct amdtp_stream *s)
{
mutex_lock(&s->mutex);
if (!amdtp_stream_running(s)) {
mutex_unlock(&s->mutex);
return;
}
tasklet_kill(&s->period_tasklet);
fw_iso_context_stop(s->context);
fw_iso_context_destroy(s->context);
s->context = ERR_PTR(-1);
iso_packets_buffer_destroy(&s->buffer, s->unit);
s->callbacked = false;
mutex_unlock(&s->mutex);
}
EXPORT_SYMBOL(amdtp_stream_stop);
/**
* amdtp_stream_pcm_abort - abort the running PCM device
* @s: the AMDTP stream about to be stopped
*
* If the isochronous stream needs to be stopped asynchronously, call this
* function first to stop the PCM device.
*/
void amdtp_stream_pcm_abort(struct amdtp_stream *s)
{
struct snd_pcm_substream *pcm;
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-24 05:07:29 +08:00
pcm = READ_ONCE(s->pcm);
if (pcm)
snd_pcm_stop_xrun(pcm);
}
EXPORT_SYMBOL(amdtp_stream_pcm_abort);