Just for consistency, use unsigned char for iface and altsetting in
allover places. Also rearrange the field positions of
snd_usb_endpiont and tidy up with some comments.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-35-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Follow the inclusive terminology, just replace sync_master/sync_slave
with sync_source/sync_sink. It's also a bit clearer from its meaning,
too.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-34-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
There are two places calculating the next packet size for the playback
stream in the exactly same way. Provide the single helper for this
purpose and use it from both places gracefully.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-32-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Minor code refactoring to consolidate the URB deactivation code in
endpoint.c. A slight behavior change is that the error handling in
snd_usb_endpoint_start() leaves EP_FLAG_STOPPING now. This should be
synced with the later PCM sync_stop callback.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-30-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The endpoint objects may be started/stopped concurrently by different
substreams in the case of implicit feedback mode, while the current
code handles the reference counter without any protection.
This patch changes the refcount to atomic_t for avoiding the
inconsistency. We need no reference_t here as the refcount goes only
up to 2.
Also the name "use_count" is renamed to "running" since this is about
actually the running status, not the open refcount.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-29-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The audioformat is referred in many places but most of usages are
read-only. Let's add const prefix in the possible places.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-28-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The implicit feedback mode uses a ring buffer for storing the received
packet sizes from the feedback source, and the code has a slight flaw;
when a playback stream stalls by some reason and the URBs aren't
processed, the next_packet FIFO might become empty, but the driver
can't distinguish whether it's empty or full because it's managed with
read_poss and write_pos.
This patch addresses those by changing the next_packet array
management. Instead of keeping read and write positions, now the head
position and the queued amount are kept. It's easier to understand
about the emptiness. Also, the URB active flag is now cleared before
calling queue_pending_output_urbs() for avoiding (theoretically)
possible inconsistency.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-27-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This is an intensive surgery for the endpoint and stream management
for achieving more robust and clean code.
The goals of this patch are:
- More clear endpoint resource changes
- The interface altsetting control in a single place
Below are brief description of the whole changes.
First off, most of the endpoint operations are moved into endpoint.c,
so that the snd_usb_endpoint object is only referred in other places.
The endpoint object is acquired and released via the new functions
snd_usb_endpoint_open() and snd_usb_endpoint_close() that are called
at PCM hw_params and hw_free callbacks, respectively. Those are
ref-counted and EPs can manage the multiple opens.
The open callback receives the audioformat and hw_params arguments,
and those are used for initializing the EP parameters; especially the
endpoint, interface and altset numbers are read from there, as well as
the PCM parameters like the format, rate and channels. Those are
stored in snd_usb_endpoint object. If it's the secondary open, the
function checks whether the given parameters are compatible with the
already opened EP setup, too.
The coupling with a sync EP (including an implicit feedback sync) is
done by the sole snd_usb_endpoint_set_sync() call.
The configuration of each endpoint is done in a single shot via
snd_usb_endpoint_configure() call. This is the place where most of
PCM configurations are done. A few flags and special handling in the
snd_usb_substream are dropped along with this change.
A significant difference wrt the configuration from the previous code
is the order of USB host interface setups. Now the interface is
always disabled at beginning and (re-)enabled at the last step of
snd_usb_endpoint_configure(), in order to be compliant with the
standard UAC2/3. For UAC1, the interface is set before the parameter
setups since there seem devices that require it (e.g. Yamaha THR10),
just like how it was done in the previous driver code.
The start/stop are almost same as before, also single-shots. The URB
callbacks need to be set via snd_usb_endpoint_set_callback() like the
previous code at the trigger phase, too.
Finally, the flag for the re-setup is set at the device suspend
through the full EP list, instead of PCM trigger. This catches the
overlooked cases where the PCM hasn't been running yet but the device
needs the full setup after resume.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-26-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The prepare_data_urb and retire_data_urb fields of the endpoint object
are set dynamically at PCM trigger start/stop. Those are evaluated in
the endpoint handler, but there can be a race, especially if two
different PCM substreams are handling the same endpoint for the
implicit feedback case. Also, the data_subs field of the endpoint is
set and accessed dynamically, too, which has the same risk.
As a slight improvement for the concurrency, this patch introduces the
function to set the callbacks and the data in a shot with the memory
barrier. In the reader side, it's also fetched with the memory
barrier.
There is still a room of race if prepare and retire callbacks are set
during executing the URB completion. But such an inconsistency may
happen only for the implicit fb source, i.e. it's only about the
capture stream. And luckily, the capture stream never sets the
prepare callback, hence the problem doesn't happen practically.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-23-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
start_endpoints() may leave the data endpoint running if an error
happens at starting the sync endpoint. We should stop both streams
properly, instead.
While we're at it, move the debug prints into the endpoint.c that is a
more suitable place.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-22-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Currently snd_usb_endpoint objects are created at first when the
substream is opened and tries to assign the endpoints corresponding to
the matching audioformat. But since basically the all endpoints have
been already parsed and the information have been obtained, we may
create the endpoint objects statically at the init phase. It's easier
to manage for the implicit fb case, for example.
This patch changes the endpoint object management and lets the parser
to create the all endpoint objects.
This change shouldn't bring any functional changes.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-15-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In the current code, there is no check at the stream open time whether
the endpoint is being already used by others. In the normal
operations, this shouldn't happen, but in the case of the implicit
feedback mode, it's a common problem with the full duplex operation,
because the capture stream is always opened by the playback stream as
an implicit sync source.
Although we recently introduced the check of such a conflict of
parameters at the PCM hw_params time, it doesn't give any hint at the
hw_params itself and just gives the error. This isn't quite
comfortable, and it caused problems on many applications.
This patch attempts to make the parameter handling easier by
introducing the strict hw constraint matching with the counterpart
stream that is being used. That said, when an implicit feedback
playback stream is running before a capture stream is opened, the
capture stream carries the PCM hw-constraint to allow only the same
sample rate, format, periods and period frames as the running playback
stream. If not opened or there is no conflict of endpoints, the
behavior remains as same as before.
Note that this kind of "weak link" should work for most cases, but
this is no concrete solution; e.g. if an application changes the hw
params multiple times while another stream is opened, this would lead
to inconsistencies.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-11-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
There are a few rooms for improvements wrt the debug prints:
- The EP debug print is shown only at starting, not at stopping
- The EP debug print contains useless object addresses
- Some helpers show the urb and the EP object addresses, too
This patch addresses those shortcomings.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-8-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Factor out the code to obtain snd_usb_endpoint object matching with
the given endpoint. It'll be used in the later patch to add the
implicit feedback hw-constraint.
No functional change by this patch itself.
Tested-by: Keith Milner <kamilner@superlative.org>
Tested-by: Dylan Robinson <dylan_robinson@motu.com>
Link: https://lore.kernel.org/r/20201123085347.19667-6-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This became wide and scattered updates all over the sound tree as
diffstat shows: lots of (still ongoing) refactoring works in ASoC,
fixes and cleanups caught by static analysis, inclusive term
conversions as well as lots of new drivers. Below are highlights:
ASoC core:
* API cleanups and conversions to the unified mute_stream() call
* Simplify I/O helper functions
* Use helper macros to retrieve RTD from substreams
ASoC drivers:
* Lots of fixes and cleanups in Intel ASoC drivers
* Lots of new stuff: Freescale MQS and i.MX6sx, Intel KeemBay I2S,
Maxim MAX98360A and MAX98373 SoundWire, various Mediatek boards,
nVidia Tegra 186 and 210, RealTek RL6231, Samsung Midas and Aries
boards, TI J721e EVM
ALSA core:
* Minor code refacotring for SG-buffer handling
HD-audio:
* Generalization of mute-LED handling with LED classdev
* Intel silent stream support for HDMI
* Device-specific fixes: CA0132, Loongson-3
Others:
* Usual USB- and HD-audio quirks for various devices
* Fixes for echoaudio DMA position handling
* Various documents and trivial fixes for sparse warnings
* Conversion to adapt inclusive terms
-----BEGIN PGP SIGNATURE-----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=sy7n
-----END PGP SIGNATURE-----
Merge tag 'sound-5.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Pull sound updates from Takashi Iwai:
"This became wide and scattered updates all over the sound tree as
diffstat shows: lots of (still ongoing) refactoring works in ASoC,
fixes and cleanups caught by static analysis, inclusive term
conversions as well as lots of new drivers. Below are highlights:
ASoC core:
- API cleanups and conversions to the unified mute_stream() call
- Simplify I/O helper functions
- Use helper macros to retrieve RTD from substreams
ASoC drivers:
- Lots of fixes and cleanups in Intel ASoC drivers
- Lots of new stuff: Freescale MQS and i.MX6sx, Intel KeemBay I2S,
Maxim MAX98360A and MAX98373 SoundWire, various Mediatek boards,
nVidia Tegra 186 and 210, RealTek RL6231, Samsung Midas and Aries
boards, TI J721e EVM
ALSA core:
- Minor code refacotring for SG-buffer handling
HD-audio:
- Generalization of mute-LED handling with LED classdev
- Intel silent stream support for HDMI
- Device-specific fixes: CA0132, Loongson-3
Others:
- Usual USB- and HD-audio quirks for various devices
- Fixes for echoaudio DMA position handling
- Various documents and trivial fixes for sparse warnings
- Conversion to adopt inclusive terms"
* tag 'sound-5.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (479 commits)
ALSA: pci: delete repeated words in comments
ALSA: isa: delete repeated words in comments
ALSA: hda/tegra: Add 100us dma stop delay
ALSA: hda: Add dma stop delay variable
ASoC: hda/tegra: Set buffer alignment to 128 bytes
ALSA: seq: oss: Serialize ioctls
ALSA: hda/hdmi: Add quirk to force connectivity
ALSA: usb-audio: add startech usb audio dock name
ALSA: usb-audio: Add support for Lenovo ThinkStation P620
Revert "ALSA: hda: call runtime_allow() for all hda controllers"
ALSA: hda/ca0132 - Fix AE-5 microphone selection commands.
ALSA: hda/ca0132 - Add new quirk ID for Recon3D.
ALSA: hda/ca0132 - Fix ZxR Headphone gain control get value.
ALSA: hda/realtek: Add alc269/alc662 pin-tables for Loongson-3 laptops
ALSA: docs: fix typo
ALSA: doc: use correct config variable name
ASoC: core: Two step component registration
ASoC: core: Simplify snd_soc_component_initialize declaration
ASoC: core: Relocate and expose snd_soc_component_initialize
ASoC: sh: Replace 'select' DMADEVICES 'with depends on'
...
Using uninitialized_var() is dangerous as it papers over real bugs[1]
(or can in the future), and suppresses unrelated compiler warnings
(e.g. "unused variable"). If the compiler thinks it is uninitialized,
either simply initialize the variable or make compiler changes.
In preparation for removing[2] the[3] macro[4], remove all remaining
needless uses with the following script:
git grep '\buninitialized_var\b' | cut -d: -f1 | sort -u | \
xargs perl -pi -e \
's/\buninitialized_var\(([^\)]+)\)/\1/g;
s:\s*/\* (GCC be quiet|to make compiler happy) \*/$::g;'
drivers/video/fbdev/riva/riva_hw.c was manually tweaked to avoid
pathological white-space.
No outstanding warnings were found building allmodconfig with GCC 9.3.0
for x86_64, i386, arm64, arm, powerpc, powerpc64le, s390x, mips, sparc64,
alpha, and m68k.
[1] https://lore.kernel.org/lkml/20200603174714.192027-1-glider@google.com/
[2] https://lore.kernel.org/lkml/CA+55aFw+Vbj0i=1TGqCR5vQkCzWJ0QxK6CernOU6eedsudAixw@mail.gmail.com/
[3] https://lore.kernel.org/lkml/CA+55aFwgbgqhbp1fkxvRKEpzyR5J8n1vKT1VZdz9knmPuXhOeg@mail.gmail.com/
[4] https://lore.kernel.org/lkml/CA+55aFz2500WfbKXAx8s67wrm9=yVJu65TpLgN_ybYNv0VEOKA@mail.gmail.com/
Reviewed-by: Leon Romanovsky <leonro@mellanox.com> # drivers/infiniband and mlx4/mlx5
Acked-by: Jason Gunthorpe <jgg@mellanox.com> # IB
Acked-by: Kalle Valo <kvalo@codeaurora.org> # wireless drivers
Reviewed-by: Chao Yu <yuchao0@huawei.com> # erofs
Signed-off-by: Kees Cook <keescook@chromium.org>
Commit f0bd62b640 ("ALSA: usb-audio: Improve frames size computation")
introduced a regression for devices which have playback endpoints with
bInterval > 1. Fix this by taking ep->datainterval into account.
Note that frame and fps are actually mean packet and packets per second
in the code introduces by the mentioned commit. This will be fixed in a
follow-up patch.
Fixes: f0bd62b640 ("ALSA: usb-audio: Improve frames size computation")
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=208353
Signed-off-by: Alexander Tsoy <alexander@tsoy.me>
Link: https://lore.kernel.org/r/20200629025934.154288-1-alexander@tsoy.me
Signed-off-by: Takashi Iwai <tiwai@suse.de>
For USB sound devices using implicit feedback the endpoint used for
this feedback should be able to be opened twice, once for required
feedback and second time for audio data. This way these devices can be
put in duplex audio mode. Since this only works if the settings of the
endpoint don't change a check is included for this.
This fixes bug 207023 ("MOTU M2 regression on duplex audio") and
should also fix bug 103751 ("M-Audio Fast Track Ultra usb audio device
will not operate full-duplex")
Fixes: c249177944 ("ALSA: usb-audio: add implicit fb quirk for MOTU M Series")
Signed-off-by: Erwin Burema <e.burema@gmail.com>
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=207023
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=103751
Link: https://lore.kernel.org/r/2410739.SCZni40SNb@alpha-wolf
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The linked list entry from FIFO is peeked at
queue_pending_output_urbs() but the actual element pop-out is
performed outside the spinlock, and it's potentially racy.
Do delete the link at the right place inside the spinlock.
Fixes: 8fdff6a319 ("ALSA: snd-usb: implement new endpoint streaming model")
Link: https://lore.kernel.org/r/20200424074016.14301-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
For computation of the the next frame size current value of fs/fps and
accumulated fractional parts of fs/fps are used, where values are stored
in Q16.16 format. This is quite natural for computing frame size for
asynchronous endpoints driven by explicit feedback, since in this case
fs/fps is a value provided by the feedback endpoint and it's already in
the Q format. If an error is accumulated over time, the device can
adjust fs/fps value to prevent buffer overruns/underruns.
But for synchronous endpoints the accuracy provided by these computations
is not enough. Due to accumulated error the driver periodically produces
frames with incorrect size (+/- 1 audio sample).
This patch fixes this issue by implementing a different algorithm for
frame size computation. It is based on accumulating of the remainders
from division fs/fps and it doesn't accumulate errors over time. This
new method is enabled for synchronous and adaptive playback endpoints.
Signed-off-by: Alexander Tsoy <alexander@tsoy.me>
Link: https://lore.kernel.org/r/20200424022449.14972-1-alexander@tsoy.me
Signed-off-by: Takashi Iwai <tiwai@suse.de>
While output urb's snd_complete_urb() is executing, calling
prepare_outbound_urb() may cause endpoint stopped before
prepare_outbound_urb() returns and result in next urb submitted
to stopped endpoint. usb-audio driver cannot re-use it afterwards as
the urb is still hold by usb stack.
This change checks EP_FLAG_RUNNING flag after prepare_outbound_urb() again
to let snd_complete_urb() know the endpoint already stopped and does not
submit next urb. Below kind of error will be fixed:
[ 213.153103] usb 1-2: timeout: still 1 active urbs on EP #1
[ 213.164121] usb 1-2: cannot submit urb 0, error -16: unknown error
Signed-off-by: Henry Lin <henryl@nvidia.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191113021420.13377-1-henryl@nvidia.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Based on 1 normalized pattern(s):
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 either version 2 of the license or at
your option any later version this program is distributed in the
hope that it will be useful but without any warranty without even
the implied warranty of merchantability or fitness for a particular
purpose see the gnu general public license for more details you
should have received a copy of the gnu general public license along
with this program if not write to the free software foundation inc
59 temple place suite 330 boston ma 02111 1307 usa
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-or-later
has been chosen to replace the boilerplate/reference in 1334 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Allison Randal <allison@lohutok.net>
Reviewed-by: Richard Fontana <rfontana@redhat.com>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190527070033.113240726@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Pointer 'urb' is being assigned but is never used hence it is
redundant and can be removed.
Cleans up clang warning:
warning: variable 'urb' set but not used [-Wunused-but-set-variable]
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Testing EP_FLAG_RUNNING in snd_complete_urb() before running the completion
logic allows us to save a few cpu cycles by returning early, skipping the
pending urb in case the stream was stopped; the stop logic handles the urb
and sets the completion callbacks to NULL.
Signed-off-by: Ioan-Adrian Ratiu <adi@adirat.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Commit 16200948d8 ("ALSA: usb-audio: Fix race at stopping the stream") was
incomplete causing another more severe kernel panic, so it got reverted.
This fixes both the original problem and its fallout kernel race/crash.
The original fix is to move the endpoint member NULL clearing logic inside
wait_clear_urbs() so the irq triggering the urb completion doesn't call
retire_capture/playback_urb() after the NULL clearing and generate a panic.
However this creates a new race between snd_usb_endpoint_start()'s call
to wait_clear_urbs() and the irq urb completion handler which again calls
retire_capture/playback_urb() leading to a new NULL dereference.
We keep the EP deactivation code in snd_usb_endpoint_start() because
removing it will break the EP reference counting (see [1] [2] for info),
however we don't need the "can_sleep" mechanism anymore because a new
function was introduced (snd_usb_endpoint_sync_pending_stop()) which
synchronizes pending stops and gets called inside the pcm prepare callback.
It also makes sense to remove can_sleep because it was also removed from
deactivate_urbs() signature in [3] so we benefit from more simplification.
[1] commit 015618b90 ("ALSA: snd-usb: Fix URB cancellation at stream start")
[2] commit e9ba389c5 ("ALSA: usb-audio: Fix scheduling-while-atomic bug in PCM capture stream")
[3] commit ccc1696d5 ("ALSA: usb-audio: simplify endpoint deactivation code")
Fixes: f8114f8583 ("Revert "ALSA: usb-audio: Fix race at stopping the stream"")
Signed-off-by: Ioan-Adrian Ratiu <adi@adirat.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This reverts commit 16200948d8.
The commit was intended to cover the race condition, but it introduced
yet another regression for devices with the implicit feedback, leading
to a kernel panic due to NULL-dereference in an irq context.
As the race condition that was addressed by the commit is very rare
and the regression is much worse, let's revert the commit for rc1, and
fix the issue properly in a later patch.
Fixes: 16200948d8 ("ALSA: usb-audio: Fix race at stopping the stream")
Reported-by: Ioan-Adrian Ratiu <adi@adirat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[Problem]
In some USB DACs, a terrible pop noise comes to be heard
at the start of DSD playback (in the following situations).
- play first DSD track
- change from PCM track to DSD track
- change from DSD64 track to DSD128 track (and etc...)
- seek DSD track
- Fast-Forward/Rewind DSD track
[Cause]
At the start of playback, there is a little silence.
The silence bit pattern "0x69" is required on DSD mode,
but it is not like that.
[Solution]
This patch adds DSD silence pattern to the endpoint settings.
Signed-off-by: Nobutaka Okabe <nob77413@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
since commit 57e6dae108 ("ALSA: usb-audio: do not trust too-big
wMaxPacketSize values"), the expected packetsize is always limited
to nominal + 25%. It was discovered, that some devices (Android audio
accessory) have a much higher jitter in used packetsizes than 25%
which would result in BABBLE condition and dropping of packets.
A better solution is so assume the jitter to be the nominal packetsize:
-one nearly empty packet followed by a almost 150% sized one.
V2: changed to assume max frequency is +50 of nominal packetsize.
Signed-off-by: Andreas Pape <apape@de.adit-jv.com>
Signed-off-by: Jiada Wang <jiada_wang@mentor.com>
Acked-by: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
We've got a kernel crash report showing like:
Unable to handle kernel NULL pointer dereference at virtual address 00000008 pgd = a1d7c000
[00000008] *pgd=31c93831, *pte=00000000, *ppte=00000000
Internal error: Oops: 17 [#1] PREEMPT SMP ARM
CPU: 0 PID: 250 Comm: dbus-daemon Not tainted 3.14.51-03479-gf50bdf4 #1
task: a3ae61c0 ti: a08c8000 task.ti: a08c8000
PC is at retire_capture_urb+0x10/0x1f4 [snd_usb_audio]
LR is at snd_complete_urb+0x140/0x1f0 [snd_usb_audio]
pc : [<7f0eb22c>] lr : [<7f0e57fc>] psr: 200e0193
sp : a08c9c98 ip : a08c9ce8 fp : a08c9ce4
r10: 0000000a r9 : 00000102 r8 : 94cb3000
r7 : 94cb3000 r6 : 94d0f000 r5 : 94d0e8e8 r4 : 94d0e000
r3 : 7f0eb21c r2 : 00000000 r1 : 94cb3000 r0 : 00000000
Flags: nzCv IRQs off FIQs on Mode SVC_32 ISA ARM Segment user
Control: 10c5387d Table: 31d7c04a DAC: 00000015
Process dbus-daemon (pid: 250, stack limit = 0xa08c8238)
Stack: (0xa08c9c98 to 0xa08ca000)
...
Backtrace:
[<7f0eb21c>] (retire_capture_urb [snd_usb_audio]) from [<7f0e57fc>] (snd_complete_urb+0x140/0x1f0 [snd_usb_audio])
[<7f0e56bc>] (snd_complete_urb [snd_usb_audio]) from [<80371118>] (__usb_hcd_giveback_urb+0x78/0xf4)
[<803710a0>] (__usb_hcd_giveback_urb) from [<80371514>] (usb_giveback_urb_bh+0x8c/0xc0)
[<80371488>] (usb_giveback_urb_bh) from [<80028e3c>] (tasklet_hi_action+0xc4/0x148)
[<80028d78>] (tasklet_hi_action) from [<80028358>] (__do_softirq+0x190/0x380)
[<800281c8>] (__do_softirq) from [<80028858>] (irq_exit+0x8c/0xfc)
[<800287cc>] (irq_exit) from [<8000ea88>] (handle_IRQ+0x8c/0xc8)
[<8000e9fc>] (handle_IRQ) from [<800085e8>] (gic_handle_irq+0xbc/0xf8)
[<8000852c>] (gic_handle_irq) from [<80509044>] (__irq_svc+0x44/0x78)
[<80508820>] (_raw_spin_unlock_irq) from [<8004b880>] (finish_task_switch+0x5c/0x100)
[<8004b824>] (finish_task_switch) from [<805052f0>] (__schedule+0x48c/0x6d8)
[<80504e64>] (__schedule) from [<805055d4>] (schedule+0x98/0x9c)
[<8050553c>] (schedule) from [<800116c8>] (do_work_pending+0x30/0xd0)
[<80011698>] (do_work_pending) from [<8000e160>] (work_pending+0xc/0x20)
Code: e1a0c00d e92ddff0 e24cb004 e24dd024 (e5902008)
Kernel panic - not syncing: Fatal exception in interrupt
There is a race between retire_capture_urb() and stop_endpoints().
The latter is called at stopping the stream and it sets some endpoint
fields to NULL. But its call is asynchronous, thus the pending
complete callback might get called after these NULL clears, and it
leads the NULL dereference like the above.
The fix is to move the NULL clearance after the synchronization,
i.e. wait_clear_urbs(). This is called at prepare and hw_free
callbacks, so it's assured to be called before the restart of the
stream or the release of the stream.
Also, while we're at it, put the EP_FLAG_RUNNING flag check at the
beginning of snd_complete_urb() to skip the pending complete after the
stream is stopped.
Fixes: b2eb950de2 ("ALSA: usb-audio: stop both data and sync...")
Reported-by: Jiada Wang <jiada_wang@mentor.com>
Reported-by: Mark Craske <Mark_Craske@mentor.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Users of devices affected by the Tenor feedback data error report
buffer underruns, even with the +/- 0x1.0000 quirk applied.
Compensating the error with 0xf000 instead seems to reliably fix
that issue.
See
https://sourceforge.net/p/alsa/mailman/message/35230259/
Reported-and-tested-by: Norman Nolte <norman.nolte@gmx.net>
Reported-and-tested-by: Thomas Gresens <T.Gresens@intershop.de>
Signed-off-by: Daniel Mack <daniel@zonque.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The quirk seems to be necessary not only for TEAC UD-H01 devices, but to
more that are based on the Tenor 8802TL chipset. Devices built by T+A
are affected too, and they apparently all use the same USB PID:PID.
Extend the quirky handling for that device as well, and rename the
quirks flag.
Reported-and-tested-by: Thomas Gresens <T.Gresens@intershop.de>
Signed-off-by: Daniel Mack <daniel@zonque.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
That's a quirk, after all, so move it where to all the other quirks
live.
Signed-off-by: Daniel Mack <daniel@zonque.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Add some sanity check codes before actually accessing the endpoint via
get_endpoint() in order to avoid the invalid access through a
malformed USB descriptor. Mostly just checking bNumEndpoints, but in
one place (snd_microii_spdif_default_get()), the validity of iface and
altsetting index is checked as well.
Bugzilla: https://bugzilla.suse.com/show_bug.cgi?id=971125
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
For the Zoom R16/24 (tx_length_quirk set), when calculating the maximum
sample frequency, consideration must be made for the fact that four bytes
of the packet contain a length descriptor and consequently must not be
counted as part of the audio data.
This is corroborated by the wMaxPacketSize for this device, which is 108
bytes according for the USB playback endpoint descriptor. The frame size
is 8 bytes (2 channels of 4 bytes each), and the 108 bytes thus work out
as 13 * 8 + 4, i.e. corresponding to 13 frames plus the additional 4 byte
length descriptor.
Signed-off-by: Ricard Wanderlof <ricardw@axis.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The Zoom R16/24 have a nonstandard playback format where each isochronous
packet contains a length descriptor in the first four bytes. (Curiously,
capture data does not contain this and requires no quirk.)
The quirk involves adding the extra length descriptor whenever outgoing
isochronous packets are generated, both in pcm.c (outgoing audio) and
endpoint.c (silent data).
In order to make the quirk as unintrusive as possible, for
pcm.c:prepare_playback_urb(), the isochronous packet descriptors are
initially set up in the same way no matter if the quirk is enabled or not.
Once it is time to actually copy the data into the outgoing packet buffer
(together with the added length descriptors) the isochronous descriptors
are adjusted in order take the increased payload length into account.
For endpoint.c:prepare_silent_urb() it makes more sense to modify the
actual function, partly because the function is less complex to start with
and partly because it is not as time-critical as prepare_playback_urb()
(whose bulk is run with interrupts disabled), so the (minute) additional
time spent in the non-quirk case is motivated by the simplicity of having
a single function for all cases.
The quirk is controlled by the new tx_length_quirk member in struct
snd_usb_substream and struct snd_usb_audio, which is conveyed to pcm.c
and endpoint.c from quirks.c in a similar manner to the txfr_quirk member
in the same structs.
In contrast to txfr_quirk however, the quirk is enabled directly in
quirks.c:create_standard_audio_quirk() by checking the USB ID in that
function. Another option would be to introduce a new
QUIRK_AUDIO_ZOOM_INTERFACE or somesuch, which would have made the quirk
very plain to see in the quirk table, but it was felt that the additional
code needed to implement it this way would just make the implementation
more complex with no real gain.
Tested with a Zoom R16, both by doing capture and playback separately
using arecord and aplay (8 channel capture and 2 channel playback,
respectively), as well as capture and playback together using Ardour, as
well as Audacity and Qtractor together with jackd.
The R24 is reportedly compatible with the R16 when used as an audio
interface. Both devices share the same USB ID and have the same number of
inputs (8) and outputs (2). Therefore "R16/24" is mentioned throughout the
patch.
Regression tested using an Edirol UA-5 in both class compliant (16-bit)
and "advanced" (24 bit, forces the use of quirks) modes.
Signed-off-by: Ricard Wanderlof <ricardw@axis.com>
Tested-by: Panu Matilainen <pmatilai@laiskiainen.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Rounding must take place before multiplication with the frame size, since
each packet contains a whole number of frames.
We must also properly consider the data interval, as a larger data
interval will result in larger packets, which, depending on the sampling
frequency, can result in packet sizes that are less than integral
multiples of the packet size for a lower data interval.
Detailed explanation and rationale:
The code before this commit had the following expression on line 613 to
calculate the maximum isochronous packet size:
maxsize = ((ep->freqmax + 0xffff) * (frame_bits >> 3))
>> (16 - ep->datainterval);
Here, ep->freqmax is the maximum assumed sample frequency, calculated from the
nominal sample frequency plus 25%. It is ultimately derived from ep->freqn,
which is in the units of frames per packet, from get_usb_full_speed_rate()
or usb_high_speed_rate(), as applicable, in Q16.16 format.
The expression essentially adds the Q16.16 equivalent of 0.999... (i.e.
the largest number less than one) to the sample rate, in order to get a
rate whose integer part is rounded up from the fractional value. The
multiplication with (frame_bits >> 3) yields the number of bytes in a
packet, and the (16 >> ep->datainterval) then converts it from Q16.16 back
to an integer, taking into consideration the bDataInterval field of the
endpoint descriptor (which describes how often isochronous packets are
transmitted relative to the (micro)frame rate (125us or 1ms, for USB high
speed and full speed, respectively)). For this discussion we will initially
assume a bDataInterval of 0, so the second line of the expression just
converts the Q16.16 value to an integer.
In order to illustrate the problem, we will set frame_bits 64, which
corresponds to a frame size of 8 bytes.
The problem here is twofold. First, the rounding operation consists
of the addition of 0x0.ffff and subsequent conversion to integer, but as the
expression stands, the conversion to integer is done after multiplication
with the frame size, rather than before. This results in the resulting
maxsize becoming too large.
Let's take an example. We have a sample rate of 96 kHz, so our ep->freqn is
0xc0000 (see usb_high_speed_rate()). Add 25% (line 612) and we get 0xf0000.
The calculated maxsize is then ((0xf0000 + 0x0ffff) * 8) >> 16 = 127 .
However, if we do the number of bytes calculation in a less obscure way it's
more apparent what the true corresponding packet size is: we get
ceil(96000 * 1.25 / 8000) * 8 = 120, where 1.25 is the 25% from line 612,
and the 8000 is the number of isochronous packets per second on a high
speed USB connection (125 us microframe interval).
This is fixed by performing the complete rounding operation prior to
multiplication with the frame rate.
The second problem is that when considering the ep->datainterval, this
must be done before rounding, in order to take the advantage of the fact
that if the number of bytes per packet is not an integer, the resulting
rounded-up integer is not necessarily a factor of two when the data
interval is increased by the same factor.
For instance, assuming a freqency of 41 kHz, the resulting
bytes-per-packet value for USB high speed is 41 kHz / 8000 = 5.125, or
0x52000 in Q16.16 format. With a data interval of 1 (ep->datainterval = 0),
this means that 6 frames per packet are needed, whereas with a data
interval of 2 we need 10.25, i.e. 11 frames needed.
Rephrasing the maxsize expression to:
maxsize = (((ep->freqmax << ep->datainterval) + 0xffff) >> 16) *
(frame_bits >> 3);
for the above 96 kHz example we instead get
((0xf0000 + 0xffff) >> 16) * 8 = 120 which is the correct value.
We can also do the calculation with a non-integer sample rate which is when
rounding comes into effect: say we have 44.1 kHz (resulting ep->freqn =
0x58333, and resulting ep->freqmax 0x58333 * 1.25 = 0x6e3ff (rounded down)):
Original maxsize = ((0x6e3ff + 0xffff) * 8) << 16 = 63 (63.124.. rounded down)
True maxsize = ceil(44100 * 1.25 / 8000) * 8 = 7 * 8 = 56
New maxsize = ((0x6e3ff + 0xffff) >> 16) * 8 = 7 * 8 = 56
This is also corroborated by the wMaxPacketSize check on line 616. Assume
that wMaxPacketSize = 104, with ep->maxpacksize then having the same value.
As 104 < 127, we get maxsize = 104. ep->freqmax is then recalculated to
(104 / 8) << 16 = 0xd0000 . Putting that rate into the original maxsize
calculation yields a maxsize of ((0xd0000 + 0xffff) * 8) >> 16 = 111
(with decimals 111.99988). Clearly, we should get back the 104 here,
which we would with the new expression: ((0xd0000 + 0xffff) >> 16) * 8 = 104 .
(The error has not been a problem because it only results in maxsize being
a bit too big which just wastes a couple of bytes, either as a result of
the first maxsize calculation, or because the resulting calculation will
hit the wMaxPacketSize value before the packet is too big, resulting in
fixing the size to wMaxPacketSize even though the packet is actually not
too long.)
Tested with an Edirol UA-5 both at 44.1 kHz and 96 kHz.
Signed-off-by: Ricard Wanderlof <ricardw@axis.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
After the recent fix of runtime PM for USB-audio driver, we got a
lockdep warning like:
=============================================
[ INFO: possible recursive locking detected ]
4.2.0-rc8+ #61 Not tainted
---------------------------------------------
pulseaudio/980 is trying to acquire lock:
(&chip->shutdown_rwsem){.+.+.+}, at: [<ffffffffa0355dac>] snd_usb_autoresume+0x1d/0x52 [snd_usb_audio]
but task is already holding lock:
(&chip->shutdown_rwsem){.+.+.+}, at: [<ffffffffa0355dac>] snd_usb_autoresume+0x1d/0x52 [snd_usb_audio]
This comes from snd_usb_autoresume() invoking down_read() and it's
used in a nested way. Although it's basically safe, per se (as these
are read locks), it's better to reduce such spurious warnings.
The read lock is needed to guarantee the execution of "shutdown"
(cleanup at disconnection) task after all concurrent tasks are
finished. This can be implemented in another better way.
Also, the current check of chip->in_pm isn't good enough for
protecting the racy execution of multiple auto-resumes.
This patch rewrites the logic of snd_usb_autoresume() & co; namely,
- The recursive call of autopm is avoided by the new refcount,
chip->active. The chip->in_pm flag is removed accordingly.
- Instead of rwsem, another refcount, chip->usage_count, is introduced
for tracking the period to delay the shutdown procedure. At
the last clear of this refcount, wake_up() to the shutdown waiter is
called.
- The shutdown flag is replaced with shutdown atomic count; this is
for reducing the lock.
- Two new helpers are introduced to simplify the management of these
refcounts; snd_usb_lock_shutdown() increases the usage_count, checks
the shutdown state, and does autoresume. snd_usb_unlock_shutdown()
does the opposite. Most of mixer and other codes just need this,
and simply returns an error if it receives an error from lock.
Fixes: 9003ebb13f ('ALSA: usb-audio: Fix runtime PM unbalance')
Reported-and-tested-by: Alexnader Kuleshov <kuleshovmail@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Add a new helper function snd_pcm_stop_xrun() to the standard sequnce
lock/snd_pcm_stop(XRUN)/unlock by a single call, and replace the
existing open codes with this helper.
The function checks the PCM running state to prevent setting the wrong
state, too, for more safety.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The usb-audio driver detects XRUN at its complete callback, but the
actual code to trigger PCM XRUN is commented out because it caused
deadlock in the past. This patch revives the PCM trigger properly.
It resulted in more than just enabling snd_pcm_stop(), but it had to
deduce the PCM substream with proper NULL checks and holds the stream
lock around the call.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Some functions in mixer.c and endpoint.c receive list_head instead of
the object itself. This is not obvious and rather error-prone. Let's
pass the proper object directly instead.
The functions in midi.c still receive list_head and this can't be
changed since the object definition isn't exposed to the outside of
midi.c, so left as is.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When a USB-audio device is disconnected while PCM is still running, we
still see some race: the disconnect callback calls
snd_usb_endpoint_free() that calls release_urbs() and then kfree()
while a PCM stream would be closed at the same time and calls
stop_endpoints() that leads to wait_clear_urbs(). That is, the EP
object might be deallocated while a PCM stream is syncing with
wait_clear_urbs() with the same EP.
Basically calling multiple wait_clear_urbs() would work fine, also
calling wait_clear_urbs() and release_urbs() would work, too, as
wait_clear_urbs() just reads some fields in ep. The problem is the
succeeding kfree() in snd_pcm_endpoint_free().
This patch moves out the EP deallocation into the later point, the
destructor callback. At this stage, all PCMs must have been already
closed, so it's safe to free the objects.
Reported-by: Alan Stern <stern@rowland.harvard.edu>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The TEAC UD-H01 firmware sends wrong feedback frequency values, thus
causing the PC to send the samples at a wrong rate, which results in
clicks and crackles in the output.
Add a workaround to detect and fix the corruption.
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
[mick37@gmx.de: use sender->udh01_fb_quirk rather than
ep->udh01_fb_quirk in snd_usb_handle_sync_urb()]
Reported-and-tested-by: Mick <mick37@gmx.de>
Reported-and-tested-by: Andrea Messa <andr.messa@tiscali.it>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Convert with dev_err() and co from snd_printk(), etc.
As there are too deep indirections (e.g. ep->chip->dev->dev),
a few new local macros, usb_audio_err() & co, are introduced.
Also, the device numbers in some messages are dropped, as they are
shown in the prefix automatically.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Takashi Iwai
Randy Dunlap
Linus Torvalds
Xu Wang
Kees Cook
Alexander Tsoy
Erwin Burema
Henry Lin
Thomas Gleixner
Colin Ian King
Ioan-Adrian Ratiu
Nobutaka Okabe
Andreas Pape
Daniel Mack
Ricard Wanderlof
Clemens Ladisch