linux/sound/usb/line6/capture.c

293 lines
7.3 KiB
C
Raw Normal View History

/*
* Line 6 Linux USB driver
*
* Copyright (C) 2004-2010 Markus Grabner (grabner@icg.tugraz.at)
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation, version 2.
*
*/
#include <linux/slab.h>
#include <sound/core.h>
#include <sound/pcm.h>
#include <sound/pcm_params.h>
#include "capture.h"
#include "driver.h"
#include "pcm.h"
/*
Find a free URB and submit it.
must be called in line6pcm->in.lock context
*/
static int submit_audio_in_urb(struct snd_line6_pcm *line6pcm)
{
int index;
int i, urb_size;
int ret;
struct urb *urb_in;
index = find_first_zero_bit(&line6pcm->in.active_urbs,
line6pcm->line6->iso_buffers);
if (index < 0 || index >= line6pcm->line6->iso_buffers) {
dev_err(line6pcm->line6->ifcdev, "no free URB found\n");
return -EINVAL;
}
urb_in = line6pcm->in.urbs[index];
urb_size = 0;
for (i = 0; i < LINE6_ISO_PACKETS; ++i) {
struct usb_iso_packet_descriptor *fin =
&urb_in->iso_frame_desc[i];
fin->offset = urb_size;
fin->length = line6pcm->max_packet_size_in;
urb_size += line6pcm->max_packet_size_in;
}
urb_in->transfer_buffer =
line6pcm->in.buffer +
index * LINE6_ISO_PACKETS * line6pcm->max_packet_size_in;
urb_in->transfer_buffer_length = urb_size;
urb_in->context = line6pcm;
ret = usb_submit_urb(urb_in, GFP_ATOMIC);
if (ret == 0)
set_bit(index, &line6pcm->in.active_urbs);
else
dev_err(line6pcm->line6->ifcdev,
"URB in #%d submission failed (%d)\n", index, ret);
return 0;
}
/*
Submit all currently available capture URBs.
ALSA: line6: Reorganize PCM stream handling The current code deals with the stream start / stop solely via line6_pcm_acquire() and line6_pcm_release(). This was (supposedly) intended to avoid the races, but it doesn't work as expected. The concurrent acquire and release calls can be performed without proper protections, thus this might result in memory corruption. Furthermore, we can't take a mutex to protect the whole function because it can be called from the PCM trigger callback that is an atomic context. Also spinlock isn't appropriate because the function allocates with kmalloc with GFP_KERNEL. That is, these function just lead to singular problems. This is an attempt to reduce the existing races. First off, separate both the stream buffer management and the stream URB management. The former is protected via a newly introduced state_mutex while the latter is protected via each line6_pcm_stream lock. Secondly, the stream state are now managed in opened and running bit flags of each line6_pcm_stream. Not only this a bit clearer than previous combined bit flags, this also gives a better abstraction. These rewrites allows us to make common hw_params and hw_free callbacks for both playback and capture directions. For the monitor and impulse operations, still line6_pcm_acquire() and line6_pcm_release() are used. They call internally the corresponding functions for both playback and capture streams with proper lock or mutex. Unlike the previous versions, these function don't take the bit masks but the only single type value. Also they are supposed to be applied only as duplex operations. Tested-by: Chris Rorvick <chris@rorvick.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2015-01-27 22:24:09 +08:00
must be called in line6pcm->in.lock context
*/
int line6_submit_audio_in_all_urbs(struct snd_line6_pcm *line6pcm)
{
int ret = 0, i;
for (i = 0; i < line6pcm->line6->iso_buffers; ++i) {
ret = submit_audio_in_urb(line6pcm);
if (ret < 0)
break;
}
return ret;
}
/*
Copy data into ALSA capture buffer.
*/
void line6_capture_copy(struct snd_line6_pcm *line6pcm, char *fbuf, int fsize)
{
struct snd_pcm_substream *substream =
get_substream(line6pcm, SNDRV_PCM_STREAM_CAPTURE);
struct snd_pcm_runtime *runtime = substream->runtime;
const int bytes_per_frame =
line6pcm->properties->bytes_per_channel *
line6pcm->properties->capture_hw.channels_max;
int frames = fsize / bytes_per_frame;
if (runtime == NULL)
return;
if (line6pcm->in.pos_done + frames > runtime->buffer_size) {
/*
The transferred area goes over buffer boundary,
copy two separate chunks.
*/
int len;
len = runtime->buffer_size - line6pcm->in.pos_done;
if (len > 0) {
memcpy(runtime->dma_area +
line6pcm->in.pos_done * bytes_per_frame, fbuf,
len * bytes_per_frame);
memcpy(runtime->dma_area, fbuf + len * bytes_per_frame,
(frames - len) * bytes_per_frame);
} else {
/* this is somewhat paranoid */
dev_err(line6pcm->line6->ifcdev,
"driver bug: len = %d\n", len);
}
} else {
/* copy single chunk */
memcpy(runtime->dma_area +
line6pcm->in.pos_done * bytes_per_frame, fbuf, fsize);
}
line6pcm->in.pos_done += frames;
if (line6pcm->in.pos_done >= runtime->buffer_size)
line6pcm->in.pos_done -= runtime->buffer_size;
}
void line6_capture_check_period(struct snd_line6_pcm *line6pcm, int length)
{
struct snd_pcm_substream *substream =
get_substream(line6pcm, SNDRV_PCM_STREAM_CAPTURE);
line6pcm->in.bytes += length;
if (line6pcm->in.bytes >= line6pcm->in.period) {
line6pcm->in.bytes %= line6pcm->in.period;
spin_unlock(&line6pcm->in.lock);
snd_pcm_period_elapsed(substream);
spin_lock(&line6pcm->in.lock);
}
}
/*
* Callback for completed capture URB.
*/
static void audio_in_callback(struct urb *urb)
{
int i, index, length = 0, shutdown = 0;
unsigned long flags;
struct snd_line6_pcm *line6pcm = (struct snd_line6_pcm *)urb->context;
line6pcm->in.last_frame = urb->start_frame;
/* find index of URB */
for (index = 0; index < line6pcm->line6->iso_buffers; ++index)
if (urb == line6pcm->in.urbs[index])
break;
spin_lock_irqsave(&line6pcm->in.lock, flags);
for (i = 0; i < LINE6_ISO_PACKETS; ++i) {
char *fbuf;
int fsize;
struct usb_iso_packet_descriptor *fin = &urb->iso_frame_desc[i];
if (fin->status == -EXDEV) {
shutdown = 1;
break;
}
fbuf = urb->transfer_buffer + fin->offset;
fsize = fin->actual_length;
if (fsize > line6pcm->max_packet_size_in) {
dev_err(line6pcm->line6->ifcdev,
"driver and/or device bug: packet too large (%d > %d)\n",
fsize, line6pcm->max_packet_size_in);
}
length += fsize;
BUILD_BUG_ON_MSG(LINE6_ISO_PACKETS != 1,
"The following code assumes LINE6_ISO_PACKETS == 1");
/* TODO:
* Also, if iso_buffers != 2, the prev frame is almost random at
* playback side.
* This needs to be redesigned. It should be "stable", but we may
* experience sync problems on such high-speed configs.
*/
line6pcm->prev_fbuf = fbuf;
line6pcm->prev_fsize = fsize /
(line6pcm->properties->bytes_per_channel *
line6pcm->properties->capture_hw.channels_max);
ALSA: line6: Reorganize PCM stream handling The current code deals with the stream start / stop solely via line6_pcm_acquire() and line6_pcm_release(). This was (supposedly) intended to avoid the races, but it doesn't work as expected. The concurrent acquire and release calls can be performed without proper protections, thus this might result in memory corruption. Furthermore, we can't take a mutex to protect the whole function because it can be called from the PCM trigger callback that is an atomic context. Also spinlock isn't appropriate because the function allocates with kmalloc with GFP_KERNEL. That is, these function just lead to singular problems. This is an attempt to reduce the existing races. First off, separate both the stream buffer management and the stream URB management. The former is protected via a newly introduced state_mutex while the latter is protected via each line6_pcm_stream lock. Secondly, the stream state are now managed in opened and running bit flags of each line6_pcm_stream. Not only this a bit clearer than previous combined bit flags, this also gives a better abstraction. These rewrites allows us to make common hw_params and hw_free callbacks for both playback and capture directions. For the monitor and impulse operations, still line6_pcm_acquire() and line6_pcm_release() are used. They call internally the corresponding functions for both playback and capture streams with proper lock or mutex. Unlike the previous versions, these function don't take the bit masks but the only single type value. Also they are supposed to be applied only as duplex operations. Tested-by: Chris Rorvick <chris@rorvick.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2015-01-27 22:24:09 +08:00
if (!test_bit(LINE6_STREAM_IMPULSE, &line6pcm->in.running) &&
test_bit(LINE6_STREAM_PCM, &line6pcm->in.running) &&
fsize > 0)
line6_capture_copy(line6pcm, fbuf, fsize);
}
clear_bit(index, &line6pcm->in.active_urbs);
if (test_and_clear_bit(index, &line6pcm->in.unlink_urbs))
shutdown = 1;
if (!shutdown) {
submit_audio_in_urb(line6pcm);
ALSA: line6: Reorganize PCM stream handling The current code deals with the stream start / stop solely via line6_pcm_acquire() and line6_pcm_release(). This was (supposedly) intended to avoid the races, but it doesn't work as expected. The concurrent acquire and release calls can be performed without proper protections, thus this might result in memory corruption. Furthermore, we can't take a mutex to protect the whole function because it can be called from the PCM trigger callback that is an atomic context. Also spinlock isn't appropriate because the function allocates with kmalloc with GFP_KERNEL. That is, these function just lead to singular problems. This is an attempt to reduce the existing races. First off, separate both the stream buffer management and the stream URB management. The former is protected via a newly introduced state_mutex while the latter is protected via each line6_pcm_stream lock. Secondly, the stream state are now managed in opened and running bit flags of each line6_pcm_stream. Not only this a bit clearer than previous combined bit flags, this also gives a better abstraction. These rewrites allows us to make common hw_params and hw_free callbacks for both playback and capture directions. For the monitor and impulse operations, still line6_pcm_acquire() and line6_pcm_release() are used. They call internally the corresponding functions for both playback and capture streams with proper lock or mutex. Unlike the previous versions, these function don't take the bit masks but the only single type value. Also they are supposed to be applied only as duplex operations. Tested-by: Chris Rorvick <chris@rorvick.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2015-01-27 22:24:09 +08:00
if (!test_bit(LINE6_STREAM_IMPULSE, &line6pcm->in.running) &&
test_bit(LINE6_STREAM_PCM, &line6pcm->in.running))
line6_capture_check_period(line6pcm, length);
}
spin_unlock_irqrestore(&line6pcm->in.lock, flags);
}
/* open capture callback */
static int snd_line6_capture_open(struct snd_pcm_substream *substream)
{
int err;
struct snd_pcm_runtime *runtime = substream->runtime;
struct snd_line6_pcm *line6pcm = snd_pcm_substream_chip(substream);
err = snd_pcm_hw_constraint_ratdens(runtime, 0,
SNDRV_PCM_HW_PARAM_RATE,
&line6pcm->properties->rates);
if (err < 0)
return err;
runtime->hw = line6pcm->properties->capture_hw;
return 0;
}
/* close capture callback */
static int snd_line6_capture_close(struct snd_pcm_substream *substream)
{
return 0;
}
/* capture operators */
struct snd_pcm_ops snd_line6_capture_ops = {
.open = snd_line6_capture_open,
.close = snd_line6_capture_close,
.ioctl = snd_pcm_lib_ioctl,
ALSA: line6: Reorganize PCM stream handling The current code deals with the stream start / stop solely via line6_pcm_acquire() and line6_pcm_release(). This was (supposedly) intended to avoid the races, but it doesn't work as expected. The concurrent acquire and release calls can be performed without proper protections, thus this might result in memory corruption. Furthermore, we can't take a mutex to protect the whole function because it can be called from the PCM trigger callback that is an atomic context. Also spinlock isn't appropriate because the function allocates with kmalloc with GFP_KERNEL. That is, these function just lead to singular problems. This is an attempt to reduce the existing races. First off, separate both the stream buffer management and the stream URB management. The former is protected via a newly introduced state_mutex while the latter is protected via each line6_pcm_stream lock. Secondly, the stream state are now managed in opened and running bit flags of each line6_pcm_stream. Not only this a bit clearer than previous combined bit flags, this also gives a better abstraction. These rewrites allows us to make common hw_params and hw_free callbacks for both playback and capture directions. For the monitor and impulse operations, still line6_pcm_acquire() and line6_pcm_release() are used. They call internally the corresponding functions for both playback and capture streams with proper lock or mutex. Unlike the previous versions, these function don't take the bit masks but the only single type value. Also they are supposed to be applied only as duplex operations. Tested-by: Chris Rorvick <chris@rorvick.com> Signed-off-by: Takashi Iwai <tiwai@suse.de>
2015-01-27 22:24:09 +08:00
.hw_params = snd_line6_hw_params,
.hw_free = snd_line6_hw_free,
.prepare = snd_line6_prepare,
.trigger = snd_line6_trigger,
.pointer = snd_line6_pointer,
};
int line6_create_audio_in_urbs(struct snd_line6_pcm *line6pcm)
{
struct usb_line6 *line6 = line6pcm->line6;
int i;
line6pcm->in.urbs = kzalloc(
sizeof(struct urb *) * line6->iso_buffers, GFP_KERNEL);
if (line6pcm->in.urbs == NULL)
return -ENOMEM;
/* create audio URBs and fill in constant values: */
for (i = 0; i < line6->iso_buffers; ++i) {
struct urb *urb;
/* URB for audio in: */
urb = line6pcm->in.urbs[i] =
usb_alloc_urb(LINE6_ISO_PACKETS, GFP_KERNEL);
if (urb == NULL)
return -ENOMEM;
urb->dev = line6->usbdev;
urb->pipe =
usb_rcvisocpipe(line6->usbdev,
line6->properties->ep_audio_r &
USB_ENDPOINT_NUMBER_MASK);
urb->transfer_flags = URB_ISO_ASAP;
urb->start_frame = -1;
urb->number_of_packets = LINE6_ISO_PACKETS;
urb->interval = LINE6_ISO_INTERVAL;
urb->error_count = 0;
urb->complete = audio_in_callback;
}
return 0;
}