Change merged from Kernel 3.10 is not
compatible with Kernel 3.18.
Make changes as per kernel 3.18 to fix the
crash.
Bug 1653770
Change-Id: I8b9e416f9dd4a905a465f6bd06732a2ec782995a
Signed-off-by: Uday Gupta <udayg@nvidia.com>
Reviewed-on: http://git-master/r/755197
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Tested-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Error in app loading skips loading of other apps and
disables further use of ADSP. This prevents use of
other successfully loaded apps which could have worked
fine.
Avoid skipping of remaining apps when loading fails for
one app. Failed app will anyway throw error whenever
user attempts to use it. This also allows adding entries
for optional apps.
Check if app was loaded when calling app_init().
Bug 200100724
Change-Id: I70a818df6fdeb54f94700e6c5f10a1dd1b269b72
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/736736
(cherry picked from commit 5387c9989420525b6ce7eaafd2ddb913f496740d)
Reviewed-on: http://git-master/r/740608
Reviewed-on: http://git-master/r/750726
Tested-by: Ravindra Lokhande <rlokhande@nvidia.com>
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Capture using last ADMA plugin was failing due to incorrect
condition in for loop. Fix it by changing condition.
Increase max number of ADSP ADMA plugins to 10 to support
additional use cases.
Bug 200056741
Change-Id: Ifbc45ee2ad8093f67d357543fa2ffe3597cf77e8
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/735456
(cherry picked from commit d233ba3302b9cda9f8d5e7ef57d005a646c7047c)
Reviewed-on: http://git-master/r/740607
Reviewed-on: http://git-master/r/750725
Tested-by: Ravindra Lokhande <rlokhande@nvidia.com>
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
This reverts commit e88ffa5fc211e68b888027ede341e264717a57d6.
If system is suspended before autosuspend time expires,
runtime_suspend does not get called and ADSP does not get
suspended. This breaks audio playback after resume due to
incorrect state of ADSP after system resume.
Originally autosuspend delay was added to avoid frequent
OS suspend and resume. No functionality issues were/are
seen even without autosuspend delay. So removing
autosuspend delay to avoid this issue.
Remove unnecessary debug print.
Bug 200090996
Change-Id: I9bfb5056e74690cf29a56273a825705cbecc0ac1
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/731620
(cherry picked from commit 53d141ffa25c4e89625bfef8d299da452d8859a4)
Reviewed-on: http://git-master/r/740606
Reviewed-on: http://git-master/r/750724
Tested-by: Ravindra Lokhande <rlokhande@nvidia.com>
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Add pm runtime callbacks for ADSP OS start/suspend.
Stop ADSP OS if app load fails.
Call to runtime_suspend after stream stop doesn't happen
when audio adsp driver is under APE PD because ADSP os
driver holds APE PD count during nvadsp_os_start() which
prevents APE PD from calling suspend. So do not add audio
driver to APE PD.
Add validity check for prtd in message handler.
Bug 200087374
Change-Id: Ie813e834f4f71fee5bbd4699cbc887e6fe8397f8
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Signed-off-by: Nitin Kumbhar <nkumbhar@nvidia.com>
Reviewed-on: http://git-master/r/720215
(cherry picked from commit f49832b8b51b657dc4e3367268788641a9ed4892)
Reviewed-on: http://git-master/r/715488
Reviewed-on: http://git-master/r/750721
Tested-by: Ravindra Lokhande <rlokhande@nvidia.com>
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
snd_kcontrol_chip(kcontrol) no more returns pointer to dapm_widget_list
from kernel 3.18 onwards, which is used to get pointer to soc_platform,
instead now use snd_soc_dapm_kcontrol_dapm(kcontrol).
Change-Id: I85457b4edb0d163b77cc9362df6bd0ef2ad39213
Signed-off-by: Ravindra Lokhande <rlokhande@nvidia.com>
Reviewed-on: http://git-master/r/738921
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
Tested-by: Bharat Nihalani <bnihalani@nvidia.com>
- removed snd_soc_codec_set_cache_io function
- use snd_soc_kcontrol_codec instead of snd_kcontrol_chip
- use codec->component.val_bytes instead of codec->val_bytes
- use devm_ioremap_resource instead of devm_request_and_ioremap
- snd_soc_dapm_mux_update_power prototype is changed
Change-Id: Ieb699a0e8a12b341c6823337ef2deb0d99292240
Signed-off-by: Ravindra Lokhande <rlokhande@nvidia.com>
Set the plugin params only if ADSP is booted and plugin
is loaded
Bug 200075850
Change-Id: Ib174afe270357ae3c21e50b9e0e85348024ec681
Signed-off-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Reviewed-on: http://git-master/r/715944
There is a scenario where SPK/HP routing change triggers
hw_params callback and DAPM widget events for ADSP stream
while stream is in paused/running state. Call to hw_params,
can re-assign a different ADMA channel than what was
assigned at start of playback. This happens when there
are multiple ADMA apps running on ADSP (eg. 3 in case of
offload + speaker protection). This change of ADMA channel
while stream is in running/paused state is unsupported and
unnecessary.
Avoid re-assignment of ADMA channel and other possible
issues by allocating ADMA channel in ADMA app init and
releasing it in app deinit.
Bug 200082413
Change-Id: I869c2a1c9e92e74e5ab2c6b9893ebd3e664def91
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/713112
Reviewed-by: Uday Gupta <udayg@nvidia.com>
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
With multiple APM's present, plugin addition
or removal does not work.
Fix it by checking only for ADSP_APP when
trying to find the APM of which the plugin
is part of.
Bug 1605750
Change-Id: I457b04a5f1ead9ac682f734b0d145f5250c411a9
Signed-off-by: Uday Gupta <udayg@nvidia.com>
Reviewed-on: http://git-master/r/714834
Reviewed-by: Omar Nemri <onemri@nvidia.com>
Tested-by: Omar Nemri <onemri@nvidia.com>
Reviewed-by: Stephen Holmes <sholmes@nvidia.com>
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Tested-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Do not send message to inactive state transition and flush.
APM moves to inactive state on its own.
Convert sample rate from index to value when sending to user.
Bug 200082504
Change-Id: I9ee1c4a995940882e7cf163ab90fc59ffd0804df
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/710381
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Stephen Holmes <sholmes@nvidia.com>
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
1. Fix ADSP crash due to reorder of mixer commands
2. Add support to track the PCM paths constructed, have a
record of FE/BE connections made by the user.
3. Support ADSP pipeline breakage, linkage, modifications
4. Plugin params can be set at any time
This change depends on http://git-master/r/#/c/709103/
functionality wise.
Bug 1605750
Bug 1605316
Bug 200075850
Change-Id: I24af6177f8ae21e7a352a58056a7267a129c3846
Signed-off-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Reviewed-on: http://git-master/r/710448
Actmon logic takes time to increase ADSP clock in case of sudden
load. This causes initial glitches in audio. Request higher ADSP
clock at stream start to avoid this. After initial boost, actmon
controls ADSP frequency based on load.
Higher ADSP clock is requested only when using ADSP speaker
protection path.
Bug 200077167
Change-Id: I8d2eae7f1c7e789a1d2012b2fd9b7cf27d6ac851
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/710727
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Uday Gupta <udayg@nvidia.com>
Reviewed-by: Puneet Saxena <puneets@nvidia.com>
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
Update all gain parameters when updating gain. Not doing so
causes mixer to take other parameters in undefined state
causing distortion.
Cache gain values and use that value when setting in hw params.
This avoids overwriting user programmed gain.
Bug 200078605
Change-Id: I2ff9c096445ccd928a7c1c482bf418f5564175c3
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/696351
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Rahul Mittal <rmittal@nvidia.com>
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
I2S_CYA register is a spare register reserved for any issue
in design. I2S4 has "i2s4a" and "i2s4b" pins in the pinmux.
In order to select i2s4b pin for I2S4 instance, we need to
enable CYA register
Bug 1602439
Change-Id: I42c401a301efb648e1141159367038a084718a93
Signed-off-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Reviewed-on: http://git-master/r/677018
(cherry picked from commit ecae3a30ba4bd9333872f1aff6e6215169fd16d0)
Reviewed-on: http://git-master/r/679674
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Justin Kim (SW-TEGRA) <juskim@nvidia.com>
Reviewed-by: Uday Gupta <udayg@nvidia.com>
set params format for the plugin is as follows
@byte_format: <param1>,..<paramN>
@int_format: <plugin_method>,<#params>,<param1>,..<paramN>
Default parameter type for plugins is @byte_format
Bug 1587869
Change-Id: Ie1b8fc4c9229e8b149ea0ff7a985e9e99ab26fc1
Signed-off-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Reviewed-on: http://git-master/r/676566
ADMAIF Global can get disabled before ADSP disables
its ADMA channels. This is not recommended as per
programming guidelines. So set ADMAIF Global Enable
bit in probe and keept is set always.
Bug 200069840
Change-Id: I73b38c4554b33c1af9d9027333ad12c55892ed7c
Signed-off-by: Viraj Karandikar <vkarandikar@nvidia.com>
Reviewed-on: http://git-master/r/668983
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Reviewed-by: Uday Gupta <udayg@nvidia.com>
Reviewed-by: Dara Ramesh <dramesh@nvidia.com>
Set default register values for AHUB modules in regmap. This will
ensure after runtime suspend/resume AHUB register POR values
does not get reset to 0.
Bug 200039212
Change-Id: I38e4c04721450b7511404c0db2911b314b68a880
Signed-off-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-on: http://git-master/r/603339
Make local data and funcs static.
Bug 200032218
Change-Id: Ib2270a19df65b53f1b547efd3f4bc7bcfc0f15a4
Signed-off-by: Sachin Nikam <snikam@nvidia.com>
Reviewed-on: http://git-master/r/600190
1. Add APE power domain support to xbar and adsp driver.
2. Also add runtime PM support in adsp audio driver to ensure APE PM
domain works properly.
3. Add code in xbar system suspend callback to support calling of all
xbar child driver system suspend. This is required to properly
support APE PM domain.
Bug 200039212
Change-Id: Ieb937b05e39d4e980872745e3b6580bca3eb6956
Signed-off-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-on: http://git-master/r/555433
Renaming MSGQ_MSG_SIZE to MSGQ_MSG_WSIZE as this macro returns the
size in word rather in bytes.
Bug 200025742
Bug 1526538
Change-Id: I7f1dd858fa9c8fc779d80afda95b66d66d7bd554
Signed-off-by: Ajay Nandakumar <anandakumarm@nvidia.com>
Reviewed-on: http://git-master/r/558844
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
remove admaif clk get apis from admaif driver.
admaif module is under ahub clock and no separate
clock for this module.
Change-Id: I5354aff4aba5c1d3c5c759704e35e53803e67242
Signed-off-by: Dara Ramesh <dramesh@nvidia.com>
Reviewed-on: http://git-master/r/555497
Reviewed-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Set idle_bias_off for all t210 xbar modules to ensure module runtime
suspend/resume works when module is idle.
Ensure regcache is synced back to hardware during runtime resume so
that register content does not get lost if it is written during
runtime suspend state.
Add suspend APIs for all module to mark regcache dirty while device
goes into suspend to ensure register values does not get lost across
system suspend/resume.
Change-Id: I2828beeed859df4f8084dd70bbcde5ed62f2525c
Signed-off-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-on: http://git-master/r/555028
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
GVS: Gerrit_Virtual_Submit
Instead of using duplicate clock of "d_audio" for all AHUB modules
get/put runtime pm reference of the parent module from runtime_pm
suspend/resume routine of all AHUB modules. This will ensure AHUB
xbar is up before other drivers tries to access any register. Also
it will ensure both d_audio and APE clocks are enabled when needed.
Bug 200042312
Change-Id: I0346728f15b135bb619de40fbd3fc440a5505940
Signed-off-by: Sumit Bhattacharya <sumitb@nvidia.com>
Reviewed-on: http://git-master/r/554863
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
GVS: Gerrit_Virtual_Submit
When using RBTREE cache, there can be allocations the first time
a register is accessed. This can cause an attempt to schedule while
atomic in the case that the regmap is using a spinlock. This can be
resolved by using a flat cache.
Bug 200041820
Change-Id: Id69592cd5fadbb5ad9ccfdbb1f184733a332512c
Signed-off-by: Arun Shamanna Lakshmi <aruns@nvidia.com>
Reviewed-on: http://git-master/r/552940
