Merge tag 'sound-4.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound

Pull sound updates from Takashi Iwai:
 "After a heavy storm by syzkaller in 4.5 cycle, we have relatively few
  changes in the core at this time while a lot of changes are found in
  the driver side, unsurprisingly.  Below are some highlights:

  ALSA core:
   - A few more hardening in ALSA timer codes
   - An extension of sequencer API for advertising the card / pid
   - Small fixes in compress-offload and jack layers

  HD-audio:
   - Dynamic PCM assignment in HDMI/DP codec; preparation for upcoming
     DP-MST support
   - Lots of code refactoring for sharing with ASoC SKL driver
   - Regression fixes for Intel HDMI/DP
   - Fixups for CX20724 codec, Lenovo AiO

  USB-audio:
   - Add quirk_alias option to make quirk debugging easier
   - Fixes for possible Oops by malformed firmware

  Firewire:
   - Add support for FW-1804 in tascam driver
   - Improvements / changes in card registration, multi stream handling,
     etc for DICE
   - Lots of code refactoring

  ASoC:
   - Enhancements of still ongoing topology API
   - Lots of commits for Intel Skylake support including HDMI support
   - A few Intel Atom driver updates for recent devices
   - Lots of improvements to the Renesas drivers
   - Capture support for Qualcomm drivers
   - Support for TI DaVinci DRA7xxx devices
   - New machine drivers for Freescale systems with Cirrus CODECs,
     Mediatek systems with RT5650 CODECs
   - New CPU drivers for Allwinner S/PDIF controllers
   - New CODEC drivers for Maxim MAX9867 and MAX98926 and Realtek RT5514"

* tag 'sound-4.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (291 commits)
  ALSA: hda - Fix mutex deadlock at HDMI/DP hotplug
  ALSA: ctl: change return value in compatibility layer so that it's the same value in core implementation
  ALSA: mixart: silence an uninitialized variable warning
  ALSA: usb-audio: Add sanity checks for endpoint accesses
  ALSA: usb-audio: Minor code cleanup in create_fixed_stream_quirk()
  ALSA: usb-audio: Fix NULL dereference in create_fixed_stream_quirk()
  ALSA: hda - Limit i915 HDMI binding only for HSW and later
  ALSA: hda - Fix unconditional GPIO toggle via automute
  ALSA: mixart: silence unitialized variable warnings
  ALSA: hda - Fixes double fault in nvhdmi_chmap_cea_alloc_validate_get_type
  ALSA: intel8x0: Add clock quirk entry for AD1981B on IBM ThinkPad X41.
  ALSA: hda - Add new GPU codec ID 0x10de0082 to snd-hda
  ASoC: rsnd: add simplified module explanation
  ASoC: hdac_hdmi: Add broxton device ID
  ASoC: Intel: Bxtn: Add Broxton PCI ID
  ASoC: Intel: Skylake: Move Skylake dsp ops & loader ops
  ASoC: Intel: add dmabuffer to common sst_dsp
  ASoC: Intel: Skylake: Unstatify skl_dsp_enable_core
  ASoC: Intel: Skylake: Fix whitepsace issues
  ASoC: Intel: Skylake: Move module id defines
  ...

1 files changed, 74 insertions, 0 deletions

diff --git a/Documentation/sound/alsa/HD-Audio-DP-MST-audio.txt b/Documentation/sound/alsa/HD-Audio-DP-MST-audio.txt
new file mode 100644
index 000000000000..82744ac3513d
--- /dev/null
+++ b/Documentation/sound/alsa/HD-Audio-DP-MST-audio.txt
@@ -0,0 +1,74 @@
+To support DP MST audio, HD Audio hdmi codec driver introduces virtual pin
+and dynamic pcm assignment.
+
+Virtual pin is an extension of per_pin. The most difference of DP MST
+from legacy is that DP MST introduces device entry. Each pin can contain
+several device entries. Each device entry behaves as a pin.
+
+As each pin may contain several device entries and each codec may contain
+several pins, if we use one pcm per per_pin, there will be many PCMs.
+The new solution is to create a few PCMs and to dynamically bind pcm to
+per_pin. Driver uses spec->dyn_pcm_assign flag to indicate whether to use
+the new solution.
+
+PCM
+===
+To be added
+
+
+Jack
+====
+
+Presume:
+ - MST must be dyn_pcm_assign, and it is acomp (for Intel scenario);
+ - NON-MST may or may not be dyn_pcm_assign, it can be acomp or !acomp;
+
+So there are the following scenarios:
+ a. MST (&& dyn_pcm_assign && acomp)
+ b. NON-MST && dyn_pcm_assign && acomp
+ c. NON-MST && !dyn_pcm_assign && !acomp
+
+Below discussion will ignore MST and NON-MST difference as it doesn't
+impact on jack handling too much.
+
+Driver uses struct hdmi_pcm pcm[] array in hdmi_spec and snd_jack is
+a member of hdmi_pcm. Each pin has one struct hdmi_pcm * pcm pointer.
+
+For !dyn_pcm_assign, per_pin->pcm will assigned to spec->pcm[n] statically.
+
+For dyn_pcm_assign, per_pin->pcm will assigned to spec->pcm[n]
+when monitor is hotplugged.
+
+
+Build Jack
+----------
+
+- dyn_pcm_assign
+Will not use hda_jack but use snd_jack in spec->pcm_rec[pcm_idx].jack directly.
+
+- !dyn_pcm_assign
+Use hda_jack and assign spec->pcm_rec[pcm_idx].jack = jack->jack statically.
+
+
+Unsolicited Event Enabling
+--------------------------
+Enable unsolicited event if !acomp.
+
+
+Monitor Hotplug Event Handling
+------------------------------
+- acomp
+pin_eld_notify() -> check_presence_and_report() -> hdmi_present_sense() ->
+sync_eld_via_acomp().
+Use directly snd_jack_report() on spec->pcm_rec[pcm_idx].jack for
+both dyn_pcm_assign and !dyn_pcm_assign
+
+- !acomp
+Hdmi_unsol_event() -> hdmi_intrinsic_event() -> check_presence_and_report() ->
+hdmi_present_sense() -> hdmi_prepsent_sense_via_verbs()
+Use directly snd_jack_report() on spec->pcm_rec[pcm_idx].jack for dyn_pcm_assign.
+Use hda_jack mechanism to handle jack events.
+
+
+Others to be added later
+========================