Due to the wrong "return" in the loop, a capture substream won't be
released at disconnection properly if the device is capture only and has
no playback substream.  This caused Oops occasionally at the device
reconnection.

Reported-by: Kim Minhyoung <minhyoung.kim@lge.com>
Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The Audio Class v2 support code in 2.6.35 added checks for the
bInterfaceProtocol field.  However, there are devices (usually those
detected by vendor-specific quirks) that do not have one of the
predefined values in this field, which made the driver reject them.

To fix this regression, restore the old behaviour, i.e., assume that
a device with an unknown bInterfaceProtocol field (other than
UAC_VERSION_2) has more or less UAC-v1-compatible descriptors.

[compile warning fixes by tiwai]

Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Cc: Daniel Mack <daniel@caiaq.de>
Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

For devices with more than one control interface, let's assume the first
one contains the audio controls. Unfortunately, there is no field in any
of the descriptors to tell us whether a control interface is for audio
or MIDI controls, so a better check is not easy to implement.

On a composite device with audio and MIDI functions, for example, the
code currently overwrites chip->ctrl_intf, causing operations on the
control interface to fail if they are issued after the device probe.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Gcc complains that ret might be used uninitialized:

sound/usb/format.c: In function ‘snd_usb_parse_audio_format’:
sound/usb/format.c:354: warning: ‘ret’ may be used uninitialized in this function
sound/usb/format.c:354: note: ‘ret’ was declared here
sound/usb/format.c:414: warning: ‘ret’ may be used uninitialized in this function
sound/usb/format.c:414: note: ‘ret’ was declared here

I suppose it could be uninitialized if there is ever a UAC_VERSION_3
released. Anyway this patch is worthwhile if only to silence the gcc
warning.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Match usb ids in usb/quirks-table.h for some Hauppage HVR-950Q models
and for the HVR850 model to those ids at the end of au0828-cards.c

Thanks to nhJm449 for pointing out the problem.

Signed-off-by: John S Gruber <JohnSGruber@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>

It is not advisable to print a warning when a device does not support
setting the sample rate because this is perfectly valid for devices with
a single rate or where rates are implicitly changed by selecting another
alternate setting.

Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Replaced the forgotten cval->mixer->ctrlif.

Signed-off-by: Takashi Iwai <tiwai@suse.de>

As the control interface is now carried in struct snd_usb_audio, we can
simplify the API a little and also drop the private ctrlif field from
struct usb_mixer_interface.

Also remove a left-over function prototype in pcm.h.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Also add a list of open topics.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Get rid of the last occurances of _v1 suffixes, and move the version
number right after the "uac" string. Now things are consitent again.

Sorry for the forth and back, but it just looks much nicer this way.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Some programs like Skype trying to set capture volume automatically.
Normally it will tray, carefully step by step lover or higher, set the volume.
In real word it work not really well, because devises and vendors lie about
real audio settings.
For example most Logitech webcams have 6400 or 3500 steps for capture volume.
They do not tell that actual resolution is 384. So we have only 7 or 18 real
steps. In this patch I set real resolution only for tested devices.

Signed-off-by: Alexey Fisher <bug-track@fisher-privat.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Stanse found that in snd_usb_parse_audio_endpoints, there is a
dangling pointer dereference. When snd_usb_parse_audio_format fails,
fp is freed, and continue invoked. On the next loop, there is
"fp && fp->altsetting == 1 && fp->channels == 1" test, but fp is set
from the last iteration (but is bogus) and thus ilegally dereferenced.

Set fp to NULL before "continue".

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Acked-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

For RANGE requests, we should only query as much bytes as we're in fact
interested in.

For CUR requests, we shouldn't confuse the firmware with an overlong
request but just ask for 2 bytes.

This might need fixing in the future as it's not entirely clear when to
dispatch 1-byte, 2-byte and 4-byte request blocks. For now, we assume
everything is coded in 16bit - this works for all firmware
implementations I've seen.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Reported-by: Alex Lee <alexlee188@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

A device may report its supported sample rates in ranges rather than in
discrete triplets. The code used to only parse the MIN field instead of
properly paying attention to the MAX and RES values.

Also, handle RES values of 1 correctly and announce a continous sample
rate range in this case.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Reported-by: Alex Lee <alexlee188@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Control messages directed to an interface must have the interface number
set in the lower 8 bits of wIndex. This wasn't done correctly for some
clock and mixer messages.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Reported-by: Alex Lee <alexlee188@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The UAC2 clock selectors are fortunately compatible with UAC1 audio
selector units, so we can simply reuse the same approach to get all the
linked units.

Requests to this control need a different CS value though.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Use a struct to parse the audio units, and return usable descriptors
for all types. There's no need to limit the result set, except for some
kind of sanity check.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The bits to enable them are always 0 for UAC1 devices, so no additional
checks are required.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Move more definitions from private enums to appropriate header files.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Audio devices which comply to the UAC2 standard can export complex clock
topologies in its descriptors and set up links between them.

The entities that are defined are

 - clock sources, which define the end-leafs.
 - clock selectors, which act as switch to select one out of many
   possible clocks sources.
 - clock multipliers, which have an input clock source, and act as clock
   source again. They can be used to derive one clock from another.

All sample rate changes, clock validity queries and the like must go to
clock source elements, while clock selectors and multipliers can be used
as terminal clock source.

The following patch adds a parser for these elements and functions to
iterate over the tree and find the leaf nodes (clock sources).

The samplerate set functions were moved to the new clock.c file.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

So far, UAC2 controls are marked read-only if any of the channels are
marked read-only in the descriptors. Change this behaviour and

 - mark them writeable unless all channels are read-only
 - store the read-only mask in usb_mixer_elem_info and
 - check the mask again in set_cur_mix_value(), and bail out for
   write-protected channels.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Introduce two new static inline functions for a more readable parsing
of UAC2 bmaControls.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Acked-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Mark Hills <mark@pogo.org.uk>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Do not explicity set the default input mode. Use the hardware default
of mode 0 ('Control vinyl'), which is now available.

This reverts commit e3ca4c99.

Acked-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Mark Hills <mark@pogo.org.uk>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

After removing code, only one case remains. So use an 'if' instead.

Acked-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Mark Hills <mark@pogo.org.uk>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

This feature was undocumented on early A4DJ units. It is indicated
by lighting both the 'line' and 'phono' lamps at the same time.
Newer units document this and the newer Windows drivers enable this
for all units, so restore the functionality.

This patch simplifies the code and changes the mode mapping to match
the A8DJ, favouring simpler code and consistency over keeping the
existing mapping.

Both 'Control vinyl' and 'Phono' input modes enable the hardware
preamp. The difference is the input impedance.

This reverts commit 9a9527ed.

Acked-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Mark Hills <mark@pogo.org.uk>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Fix a small off-by-one bug which causes the feature unit to announce a
wrong number of channels. This leads to illegal requests sent to the
firmware eventually.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

This request is again handled differently in comparison to UAC1.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

UAC2 devices have their information about pitch control stored in a
different field. Parse it, and emulate the bits for a v1 device.

A new struct uac2_iso_endpoint_descriptor is added.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

-1 is not a good return value as it means -EPERM, "not permitted".
Choose -ENOTSUPP instead, which is what the code really wants to tell
its callers.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

The decoding/encoding is based on own reverse-engineering. Both control and
data ports are handled. Writing to control port supports SysEx events only,
as this is the only type of messages that MPD16 recognizes.

Signed-off-by: Krzysztof Foltman <wdev@foltman.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

This is needed before the USB merge.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

For more clearance what the functions actually do,

  usb_buffer_alloc() is renamed to usb_alloc_coherent()
  usb_buffer_free()  is renamed to usb_free_coherent()

They should only be used in code which really needs DMA coherency.

All call sites have been changed accordingly, except for staging
drivers.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Pedro Ribeiro <pedrib@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

For both UAC1 and UAC2, interrupt endpoint messages are now parsed with
structs rather that with anonymous buffer array accesses.

For UAC2, only CUR interrupt notifications are supported for now.

snd_usb_mixer_status_complete() was renamed to
snd_usb_mixer_interrupt().

Fixed one indentation flaw on the way.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Cc: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>