linux.git/drivers/usb/gadget/function/u_audio.c, branch v6.5

usb: gadget: u_audio: don't let userspace block driver unbind

2023-03-09T13:56:32+00:00

In the unbind callback for f_uac1 and f_uac2, a call to snd_card_free()
via g_audio_cleanup() will disconnect the card and then wait for all
resources to be released, which happens when the refcount falls to zero.
Since userspace can keep the refcount incremented by not closing the
relevant file descriptor, the call to unbind may block indefinitely.
This can cause a deadlock during reboot, as evidenced by the following
blocked task observed on my machine:

  task:reboot  state:D stack:0   pid:2827  ppid:569    flags:0x0000000c
  Call trace:
   __switch_to+0xc8/0x140
   __schedule+0x2f0/0x7c0
   schedule+0x60/0xd0
   schedule_timeout+0x180/0x1d4
   wait_for_completion+0x78/0x180
   snd_card_free+0x90/0xa0
   g_audio_cleanup+0x2c/0x64
   afunc_unbind+0x28/0x60
   ...
   kernel_restart+0x4c/0xac
   __do_sys_reboot+0xcc/0x1ec
   __arm64_sys_reboot+0x28/0x30
   invoke_syscall+0x4c/0x110
   ...

The issue can also be observed by opening the card with arecord and
then stopping the process through the shell before unbinding:

  # arecord -D hw:UAC2Gadget -f S32_LE -c 2 -r 48000 /dev/null
  Recording WAVE '/dev/null' : Signed 32 bit Little Endian, Rate 48000 Hz, Stereo
  ^Z[1]+  Stopped                    arecord -D hw:UAC2Gadget -f S32_LE -c 2 -r 48000 /dev/null
  # echo gadget.0 > /sys/bus/gadget/drivers/configfs-gadget/unbind
  (observe that the unbind command never finishes)

Fix the problem by using snd_card_free_when_closed() instead, which will
still disconnect the card as desired, but defer the task of freeing the
resources to the core once userspace closes its file descriptor.

Fixes: 132fcb460839 ("usb: gadget: Add Audio Class 2.0 Driver")
Cc: stable@vger.kernel.org
Signed-off-by: Alvin Šipraga 
Reviewed-by: Ruslan Bilovol 
Reviewed-by: John Keeping 
Link: https://lore.kernel.org/r/20230302163648.3349669-1-alvin@pqrs.dk
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: clean up some inconsistent indenting

2022-04-21T17:03:07+00:00

Eliminate the follow smatch warning:

drivers/usb/gadget/function/u_audio.c:1182 g_audio_setup() warn:
inconsistent indenting.

Reported-by: Abaci Robot 
Signed-off-by: Jiapeng Chong 
Link: https://lore.kernel.org/r/20220407062410.89214-1-jiapeng.chong@linux.alibaba.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: Add suspend call

2022-01-26T13:06:09+00:00

Add exported method u_audio_suspend which sets stream status to
inactive and sends notifications. The method does not free any
resources.

Signed-off-by: Pavel Hofman 
Link: https://lore.kernel.org/r/20220121155308.48794-9-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: Rate ctl notifies about current srate (0=stopped)

2022-01-26T13:06:09+00:00

The Playback/Capture ctl currently reports rate value set by USB
control selector UAC2_CS_CONTROL_SAM_FREQ (fixed for UAC1). When the
stops playback/capture, the reported value does not change.  The gadget
side has no information whether the host has started/stopped
capture/playback.

This patch sets the value reported by the respective rate ctl to zero
when the host side has stopped playback/capture. Also, it calls
snd_ctl_notify when start/stop  occurs, so that a subscribed client can
act appropriately.

Tests have confirmed that USB hosts change UAC2_CS_CONTROL_SAM_FREQ
before switching altsetting to activate playback/capture, resulting in
correct order (params->c/p_srate is set to requested rate before
u_audio_start_capture/playback is called).

The gadget rate notifications are used by user-space audio gadget
controller gaudio_ctl https://github.com/pavhofman/gaudio_ctl.

Signed-off-by: Pavel Hofman 
Link: https://lore.kernel.org/r/20220121155308.48794-8-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: Add capture/playback srate getter

2022-01-26T13:06:08+00:00

UAC1/UAC2 functions will need to query u_audio about the currently set
srate. Add the getter functions.

Signed-off-by: Pavel Hofman 
Link: https://lore.kernel.org/r/20220121155308.48794-5-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: Move dynamic srate from params to rtd

2022-01-26T13:06:08+00:00

Parameters uac_params.p_srate/c_srate are dynamic now and are not part
of parametric configuration anymore. Move them to the
runtime struct uac_rtd_params for each stream.

Suggested-by: John Keeping 
Signed-off-by: Pavel Hofman 
Link: https://lore.kernel.org/r/20220121155308.48794-4-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: Support multiple sampling rates

2022-01-26T13:06:08+00:00

Implement support for multiple sampling rates in u_audio part of the
audio gadget. The currently configured rates are exposed through
read-only amixer controls 'Capture Rate' and 'Playback Rate'.

Signed-off-by: Julian Scheel 
Signed-off-by: Pavel Hofman 
Link: https://lore.kernel.org/r/20220121155308.48794-3-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: Subdevice 0 for capture ctls

2022-01-06T14:31:02+00:00

Both capture and playback alsa devices use subdevice 0. Yet capture-side
ctls are defined for subdevice 1. The patch sets subdevice 0 for them.

Fixes: 02de698ca812 ("usb: gadget: u_audio: add bi-directional volume and mute support")
Signed-off-by: Pavel Hofman 
Link: https://lore.kernel.org/r/20220105104643.90125-1-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio: fix calculations for small bInterval

2022-01-06T14:28:45+00:00

If bInterval is 1, then p_interval is 8000 and p_interval_mil is 8E9,
which is too big for a 32-bit value.  While the storage is indeed
64-bit, this value is used as the divisor in do_div() which will
truncate it into a uint32_t leading to incorrect calculated values.

Switch back to keeping the base value in struct snd_uac_chip which fits
easily into an int, meaning that the division can be done in two steps
with the divisor fitting safely into a uint32_t on both steps.

Fixes: 6fec018a7e70 ("usb: gadget: u_audio.c: Adding Playback Pitch ctl for sync playback")
Tested-by: Pavel Hofman 
Signed-off-by: John Keeping 
Link: https://lore.kernel.org/r/20220104183243.718258-1-john@metanate.com
Signed-off-by: Greg Kroah-Hartman

usb: gadget: u_audio.c: Adding Playback Pitch ctl for sync playback

2021-10-13T11:59:43+00:00

EP IN is hard-coded as ASYNC both in f_uac1 and f_uac2 but u_audio sends
steady number of audio frames in each USB packet, without any control.

This patch adds 'Playback Pitch 1000000' ctl analogous to the existing
'Capture Pitch 1000000' ctl. The calculation of playback req->length in
u_audio_iso_complete respects the Playback Pitch ctl value to 1ppm now.

Max. value for Playback Pitch is configured by the existing parameter
uac2_opts->fb_max, used also for the Capture Pitch.

Since the EP IN packet size can be increased by uac2_opts->fb_max now,
maxPacketSize for the playback direction is calculated by the same
algorithm as for the async capture direction in
f_uac2.c:set_ep_max_packet_size.

Signed-off-by: Pavel Hofman 
Link: https://lore.kernel.org/r/20211013073934.36476-1-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman