If the system enters S3 during a playback, codec power needs to
be turned OFF during suspend and restored during resume. With
this patch the AFG node is set to D3 and codec power is turned
OFF during controller suspend call.

During resume, the codec power is left in ON state if the
playback was in progress while suspending.

Also setting power state for AFG node is optimized. With this the
loop with timeout is removed and codec_read is used instead.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Use dev_pm ops feature instead of soc pm as core assumes system
is capable of direct complete. Register with complete callback
instead of resume to synchronize with Jack notification from
display driver. This ensures correct Jack notification to user
space.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Broxton HDMI codec is similar to Skylake so add the device ID
Signed-off-by: NJeeja KP <jeeja.kp@intel.com>
Signed-off-by: NGuruprasadX Pawse <guruprasadx.pawse@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Audio infoframe used incorrect buffer, so fix it.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The codec registers are reset during S3. So need to reconfigure
all pins and DP1.2 feature again after resume from S3. Also
reprogram the required registers if the S3 was triggered during
playback.

In suspended state ELD notify callback is not processed, So add
ELD check for all pins as well.

Also turn the codec power domain OFF which is kept ON during
controller resequencing and codec reenumeration.
Signed-off-by: NJeeja KP <jeeja.kp@intel.com>
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Powering off codec immediately after sending D3 verb may not set
the node to D3 state. So wait for a confirmation response before
shutting down codec.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The register are reset during the D3 to D0 transition. So
reconfigure them.

Also as all DAIs are now supported, remove the dai id check in prepare.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The codec power is turned OFF in the first explicit call to
pm_runtime_suspend to keep the i915 refcount balanced. During regular
operation, the power is turned ON/OFF in runtime PM handlers.
Signed-off-by: NRamesh Babu <ramesh.babu@intel.com>
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

In dai startup, driver was checking for ELD and would fail if no
monitor is connected. This causes userland like PA, CRAS to be
unhappy as they scan the device list at bootup.

So move the ELD check to hw_params and fail if valid ELD is not
found.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

It's possible for hw_params to be called two times. So add NULL
check to prevent memory leak.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Commit 4a3478de ("ASoC: hdac_hdmi: Add jack reporting")
leads to the following Smatch complaint:

sound/soc/codecs/hdac_hdmi.c:971 hdac_hdmi_present_sense()
         warn: variable dereferenced before check 'edev' (see line 967)

As explained by Subhransu S. Prusty:

"This NULL check is not required, can be removed.

edev object is expected to be present when the present_sense is
called. This is called from places where this edev object is
already available."

So remove the evdev NULL check.
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NFabio Estevam <fabio.estevam@nxp.com>
Acked-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

For DP audio support, infoframe needs to be different. Based on
get_conn_type result we pack either HDMI or DP infoframe.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Last patch added infrastructure to render over all the ports,
PCM<->cvt<- pin mapping and user selection of controls. But we
still have restriction of playback on the default port alone, so
remove that.

This patch removes the hardcoding of cvt<->pin map from the dai.
Cvt and pin for a dai are now derived from the already stored pcm
list of device opened.  We query connection list of a pin from
codec to validate the cvt<->pin map.

If connection list returns zero, then monitor is not connected so
fail playback.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NJeeja KP <jeeja.kp@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Jack is created based on pcm devices enumerated, so we will
create Jack as "HDMI/DP, pcm=x Jack". This style is expected by
current usermode like PulseAudio and CRAS.

This patch exports an API which can be used to register Jack
based on PCM. This API also establishes the map between PCM and
cvt. Further cvt to pin mapping is established with the help of
usermode selection based on the board topology.

During device probe as the PCMs may not be registered, initial
pin sense don't report jack events. So, first time jack reporting
is done during user selection of mux control.
Signed-off-by: NJeeja KP <jeeja.kp@intel.com>
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

In skylake, HDMI codec enumerates 3 pins and 3 cvts. Stream can
be routed from any cvt to any pin based on the connection list
queried from the pin.

This patch removes the static modelling of widget/route and
creates it dynamically based on the codec widgets enumerated.

Mux widgets are added to represent the map between a cvt and pin.
Ideally the mux widgets should be created based on the connection
list queried from the pin widget. But due to HW behavior, if an
external display is not connected on a port, querying the
connection list returns zero. So create mux widgets to map all
the cvt to all pins. At runtime, playback support on a pin can be
verified based on the connection list query.

Few function in driver have additional arguments now to support
this.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

After enabling all pins/cvts, Skylake HDMI codec enumerates 3
converters.  Three independent streams can be supported with 3
cvts. This patch removes the static dai creation and creates dais
based on the number of cvts queried.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Skylake supports 3 pin and 3 converter widgets. But by default
only one converter and pin widget are enabled. In skylake
platform the DP port is on a different port which is not enabled
by default. To enable playback on DP port, enable all pin and
converter widget by sending a vendor VERB for a vendor widget to
set required bits.

As we are enabling the DP support enable the DP1.2 feature as well.

Enabling DP1.2 and all widget changes are copied from patch_hdmi.c.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Uses the drm ELD core framework to apply rate and channel

Also compute the format to be set based on ELD.

Even though the channel constraint is based on ELD, infoframe
is set with stereo only. Multichannel support will be added
later.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

This patch uses i915 component framework to register for hotplug
notification.

In the hotplug notification, driver reads pin sense and ELD by
sending PIN_SENSE and ELD verbs over HDA bus. Once it identifies
valid pin sense and valid ELD, store the ELD into the
corresponding pin map buffer.

Also read the monitor present sense during resume and ignore the
ELD notify from graphics during PM as is done in legacy hda,
commit 8ae743e8 ("ALSA: hda - Skip ELD notification during
system suspend")
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

A recent rework removed the only user of the hdac_hdmi_query_pin_connlist
function, so we now get a warning when building the hdac_hdmi driver:

hdac_hdmi.c:313:12: warning: 'hdac_hdmi_query_pin_connlist' defined but not used [-Wunused-function]

This removes the function, which makes the file build cleanly again.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Fixes: 15b91447 ("ASoC: hdac_hdmi: Use list to add pins and converters")
Signed-off-by: NMark Brown <broonie@kernel.org>

Future platforms may have a different set of pins/converters.
So use lists to add pins and converters based on enumeration.

Also it may be required to connect any converter to any pin
dynamically as per different use cases (for example DP is
connected to pin 6 on skylake board). So this will help in
dynamically select and route.

Fix the dai map as well to use the pin/cvt from list. Not
enabling all dai maps for now.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

It is possible that some pin widget may return with no converter
connected. So don't throw error if none are found to be connected.
Instead print a warning and continue.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

commit 3c83ac23 ("ASoC: hdac_hdmi: check error return") fixes
the static checker warning reported by Dan Carpenter:

	sound/soc/codecs/hdac_hdmi.c:416 hdac_hdmi_parse_and_map_nid()
	warn: unsigned 'hdac->num_nodes' is never less than zero.

But it doesn't fix the issue completely.

It's also a failure if no sub nodes found for an afg node. So modify
the return condition appropriately.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Use dev_to_hdac_dev() and to_ehdac_device() instead of open-coding.
Signed-off-by: NGeliang Tang <geliangtang@163.com>
Reviewed-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

As hdac->num_nodes is unsigned we can not check if
snd_hdac_get_sub_nodes() has returned error or success. Lets have a
temporary int to check the error value.
Signed-off-by: NSudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

kzalloc() can return NULL if it fails, and then we will be dereferencing
a NULL pointer.
Signed-off-by: NSudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Use the component framework to keep the display on till the
playback in progress.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

This patch uses hdmi framework in video to fill audio infoframe.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The DAI ops are used for triggering HDMI streams and configuring
the parameters
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Power up/down the AFG node during runtime resume/suspend.
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

This adds HDA based HDMI driver to be used in platforms like SKL
and onwards

Register the hdmi driver with hda bus and register dais.

Also parse the widget and initialize identified pin and converter
widgets.

For simplification, currently only one pin and one converter
widget are enabled on board, as well as limit the rates supported
to simples ones and not based on ELD.  This things will come
eventually once basic support for this is merged
Signed-off-by: NSubhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
Signed-off-by: NMark Brown <broonie@kernel.org>