Support BT audio offload on HDA machines #5123

brentlu · 2024-07-29T03:33:30Z

SOF platform driver walks through HNLT endpoints and looks for endpoint which device type is BT sideband (0). Once a BT sideband endpoint is found, the SSP port number will be used to compose link mask and passed to machine driver to create BE link for BT audio offload.

bardliao · 2024-07-29T05:58:07Z

sound/soc/intel/boards/skl_hda_dsp_generic.c

@@ -150,6 +152,28 @@ static int skl_hda_fill_card_info(struct snd_soc_card *card,
 }
 }

+ if (!ctx->bt_offload_present) {
+ /* remove last link since bt audio offload is not supported */
+ num_links--;


You define #define BT_DAI_COUNT 1, INHO, it is better to use num_links -= BT_DAI_COUNT in case BT_DAI_COUNT is changed one day.

ujfalusi · 2024-07-30T07:20:15Z

include/sound/soc-acpi.h

@@ -165,6 +167,12 @@ struct snd_soc_acpi_link_adr {
 */
 #define SND_SOC_ACPI_TPLG_INTEL_CODEC_NAME BIT(4)

+/*
+ * when set the BT offload name suffix '-bt' will be appended to topology file


the code appends -bt-ssp%d tot he topology filename...

ujfalusi · 2024-07-30T07:22:12Z

include/sound/soc-acpi.h

+ * when set the BT offload name suffix '-bt' will be appended to topology file
+ * name if the feature is implemented in ACPI NHLT table
+ */
+#define SND_SOC_ACPI_TPLG_INTEL_BT_OFFLOAD BIT(5)


Do we have a documented rule on what order these appends should be done, I believe they must be always <bit(0)><bit(1)>...<bit(last)> ?

ujfalusi · 2024-07-30T07:26:29Z

sound/soc/sof/intel/hda.c

@@ -444,6 +444,10 @@ static int mclk_id_override = -1;
 module_param_named(mclk_id, mclk_id_override, int, 0444);
 MODULE_PARM_DESC(mclk_id, "SOF SSP mclk_id");

+static int bt_link_mask_override = 0x0;


No need to initialize to 0

ujfalusi · 2024-07-30T07:27:47Z

sound/soc/sof/intel/hda.c

+ mach->mach_params.bt_link_mask = check_nhlt_ssp_mask(sdev, NHLT_DEVICE_BT);
+
+ /* allow for module parameter override */
+ if (bt_link_mask_override != 0x00) {


if (!bt_link_mask_override) { ?

ujfalusi · 2024-07-30T07:28:36Z

sound/soc/sof/intel/hda.c

+ /* allow for module parameter override */
+ if (bt_link_mask_override != 0x00) {
+ dev_dbg(sdev->dev,
+ "overriding bt link mask detected in NHLT tables 0x%x by kernel param 0x%x\n",


New code should use %#x for hexadecimal numbers.

ujfalusi · 2024-07-30T08:31:23Z

sound/soc/sof/intel/hda.c

 if (ssp_mask)
- dev_info(sdev->dev, "NHLT_DEVICE_I2S detected, ssp_mask %#x\n", ssp_mask);
+ dev_info(sdev->dev, "NHLT device %d detected, ssp_mask %#x\n",


I would be hard pressed to guess based on the number what device is detected..

ujfalusi · 2024-07-30T08:36:30Z

sound/soc/intel/boards/skl_hda_dsp_common.c

@@ -132,6 +137,14 @@ struct snd_soc_dai_link skl_hda_be_dai_links[HDA_DSP_MAX_BE_DAI_LINKS] = {
 .no_pcm = 1,
 SND_SOC_DAILINK_REG(dmic16k, dmic_codec, platform),
 },
+ {
+ .name = "SSP%d-BT",


It is a novel way, but I would rather not set this string as initial name.

sound/soc/intel/boards/skl_hda_dsp_generic.c

ujfalusi · 2024-07-30T08:47:30Z

sound/soc/intel/boards/skl_hda_dsp_generic.c

+ return -ENOMEM;
+
+ bt_link->cpus->dai_name = devm_kasprintf(dev, GFP_KERNEL,
+ bt_link->cpus->dai_name,


The dai_name comes from topology and in a form of a magic string also?

The name is from here:

SND_SOC_DAILINK_DEF(bt_offload_pin,
DAILINK_COMP_ARRAY(COMP_CPU("SSP%d Pin")));

ujfalusi · 2024-07-30T08:49:22Z

sound/soc/intel/common/soc-acpi-intel-hda-match.c

@@ -14,7 +14,8 @@ struct snd_soc_acpi_mach snd_soc_acpi_intel_hda_machines[] = {
 /* .id is not used in this file */
 .drv_name = "skl_hda_dsp_generic",
 .sof_tplg_filename = "sof-hda-generic", /* the tplg suffix is added at run time */
- .tplg_quirk_mask = SND_SOC_ACPI_TPLG_INTEL_DMIC_NUMBER,
+ .tplg_quirk_mask = SND_SOC_ACPI_TPLG_INTEL_DMIC_NUMBER |
+ SND_SOC_ACPI_TPLG_INTEL_BT_OFFLOAD,


and this is going to break a lot of system presumably since we don not have any -bt-ssp topology out there..

I removed all tplg name fixup code and let OSV to decide so we don't break all HDA PC...

I removed all tplg name fixup code and let OSV to decide so we don't break all HDA PC...

The problem is that we don't know who the OSV is, and we don't know how to distribute the required topology. So while the changes look ok at a high-level, who's going to use all this?

I don't want to merge something that creates a headache for support teams. We don't have a plan that makes any sense at the moment.

Sorry I should provide the background story instead of just shooting the patches.

Currently there is a Chrome reference design brox which enables HDA external codec. Since BT audio offload was enabled for I2S Chromebooks, we want this feature could be also supported for HDA Chromebooks as well.

For the kernel driver, we will backport these two patches to Chrome tree and update the module option since Chromebook does not support NHLT table. And for topology, we will backport the patch to cavs2.5 branch and request Google to release the topology.

plbossart

Looks mostly good now @brentlu, that's an acceptable solution that allows for new functionality without breaking older setups. Thanks for taking this initiative.

sound/soc/sof/intel/hda.c

plbossart · 2024-08-01T07:08:26Z

sound/soc/intel/boards/skl_hda_dsp_common.c

+ .name = NULL, /* initialized in driver probe function */
+ .id = 8,
+ .dpcm_playback = 1,
+ .dpcm_capture = 1,


I forgot if those flags still exist or have been removed.

not yet removed. still doing some cleanups
https://patchwork.kernel.org/project/alsa-devel/list/?series=874792

Add an new variable bt_link_mask to snd_soc_acpi_mach_params structure. SSP port mask of BT offload found in NHLT table will be sent to machine driver to setup BE dai link with correct SSP port number. This patch only detects and enables the BT dailink. The functionality will only be unlocked with a topology file that makes a reference to that BT dailink. For backwards-compatibility reasons, this topology will not be used by default. Chromebooks and Linux users willing to experiment shall use the tplg_name kernel parameter to force the use of an enhanced topology. Signed-off-by: Brent Lu <brent.lu@intel.com>

Add BT offload BE link to dai link array if the BT offload link mask is valid (only one bit set). Signed-off-by: Brent Lu <brent.lu@intel.com>

plbossart · 2024-08-06T07:48:51Z

@ujfalusi @bardliao @kv2019i can you take a look?

brentlu requested review from bardliao, lgirdwood, plbossart, ranj063, kv2019i, dbaluta and ujfalusi as code owners July 29, 2024 03:33

This was referenced Jul 29, 2024

[STABLE-v2.2] topology1: sof-hda-generic: support BT offload pipelines thesofproject/sof#9336

Merged

topology2: sof-hda-generic: support BT offload pipelines thesofproject/sof#9337

Closed

bardliao reviewed Jul 29, 2024

View reviewed changes

brentlu force-pushed the hda-mach-2 branch 2 times, most recently from d197f34 to 6960b82 Compare July 29, 2024 06:34

bardliao previously approved these changes Jul 29, 2024

View reviewed changes

ujfalusi reviewed Jul 30, 2024

View reviewed changes

brentlu dismissed bardliao’s stale review via 6f78e60 July 31, 2024 00:03

brentlu force-pushed the hda-mach-2 branch 4 times, most recently from 146498b to 5ed91b9 Compare July 31, 2024 00:38

plbossart reviewed Aug 1, 2024

View reviewed changes

brentlu force-pushed the hda-mach-2 branch from 5ed91b9 to bb4a82f Compare August 2, 2024 00:54

brentlu added 2 commits August 2, 2024 16:34

ASoC: Intel: skl_hda_dsp_generic: support BT audio offload

bb4a82f

Add BT offload BE link to dai link array if the BT offload link mask is valid (only one bit set). Signed-off-by: Brent Lu <brent.lu@intel.com>

plbossart approved these changes Aug 6, 2024

View reviewed changes

plbossart requested review from ujfalusi and bardliao August 6, 2024 07:48

bardliao approved these changes Aug 6, 2024

View reviewed changes

plbossart merged commit bc47b82 into thesofproject:topic/sof-dev Aug 7, 2024
11 of 14 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Support BT audio offload on HDA machines #5123

Support BT audio offload on HDA machines #5123

brentlu commented Jul 29, 2024 •

edited

Loading

bardliao Jul 29, 2024

brentlu Jul 29, 2024

ujfalusi Jul 30, 2024

ujfalusi Jul 30, 2024

ujfalusi Jul 30, 2024

brentlu Jul 31, 2024

ujfalusi Jul 30, 2024

brentlu Jul 31, 2024

ujfalusi Jul 30, 2024

brentlu Jul 31, 2024

ujfalusi Jul 30, 2024

brentlu Jul 31, 2024

ujfalusi Jul 30, 2024

brentlu Jul 31, 2024

ujfalusi Jul 30, 2024

brentlu Jul 31, 2024

ujfalusi Jul 30, 2024

brentlu Jul 31, 2024

plbossart Jul 31, 2024

brentlu Aug 1, 2024

plbossart left a comment

plbossart Aug 1, 2024

brentlu Aug 2, 2024

plbossart commented Aug 6, 2024

Support BT audio offload on HDA machines #5123

Support BT audio offload on HDA machines #5123

Conversation

brentlu commented Jul 29, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

plbossart left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

plbossart commented Aug 6, 2024

brentlu commented Jul 29, 2024 •

edited

Loading