Skip to contentSkip to conversation

ASoC: ti: davinci-mcasp: Fix the DIT mode and OMAP4 support

PATCH Superseded Peter Ujfalusi 4 participants 12 messages Jul 4, 2021 – Jul 5, 2021

Subsystems (4)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover ASoC: ti: davinci-mcasp: Fix the DIT mode and OMAP4 support

Hi,

it has been on my todo list for several years to support McASP on OMAP4 devices. For Galaxy Nexus we had an omap-mcasp driver (which was mostly a stripped down davinci-mcasp driver) to support what was needed on that specific phone + it's dock for S/PDIF (48KHz, 16bit, stereo).

Not many (if any) device available to test the DIT mode of McASP. I have used BeagleBone White (McASP1 AXR3 can be routed to a pin) to get the S/PDIF mode working then PandaES for OMAP4 support (on PandaES the gpio_121 is not used and the signal is routed to expansion J6 pin14)

In theory the McASP in OMAP5 should be working after this series, but the OMAP5 TRM is not public and I do not have one to check the addresses and see if there is a way to test it on omap5-uevm.

Mark, Tony:
The ASoC and dts patches can go via separate tree I felt that it is better if they are together, at least initially.

Nikolaus: fyi, this might be useful for Pyra?

Regards,
Péter
---
Peter Ujfalusi (5):
ASoC: ti: davinci-mcasp: Fix DIT mode support
ASoC: dt-bindings: davinci-mcasp: Add compatible string for OMAP4
ASoC: ti: davinci-mcasp: Add support for the OMAP4 version of McASP
ARM: dts: omap4-l4-abe: Correct sidle modes for McASP
ARM: dts: omap4-l4-abe: Add McASP configuration

 .../bindings/sound/davinci-mcasp-audio.txt    |   1 +
 arch/arm/boot/dts/omap4-l4-abe.dtsi           |  39 ++--
 include/linux/platform_data/davinci_asp.h     |   1 +
 sound/soc/ti/Kconfig                          |   1 +
 sound/soc/ti/davinci-mcasp.c                  | 168 +++++++++++++++---
 5 files changed, 168 insertions(+), 42 deletions(-)
Signature · 1 line
2.32.0
#1 lore raw

PATCH 1/5 ASoC: ti: davinci-mcasp: Fix DIT mode support

The DIT mode support has not been tested due to lack of platform where it can be tested.
To be able to use the McASP on OMAP4/5 (only supporting DIT mode) we need to have DIT mode working in the McASP driver on a know platform. After hacking around (on BBW, mcasp1.axr1 can be routed out for this) it appeared that DIT mode is broken.

This patch fixes it up and 16/24 bit audio works along with passthrough, but I have only tested with DTS example and test files.

---
 sound/soc/ti/davinci-mcasp.c | 142 ++++++++++++++++++++++++++++++-----
 1 file changed, 122 insertions(+), 20 deletions(-)
sound/soc/ti/davinci-mcasp.c +122 -20
index 017a5a5e56cd..e56a22c28467 100644
@@ -83,6 +83,8 @@ struct davinci_mcasp {
 	struct snd_pcm_substream *substreams[2];
 	unsigned int dai_fmt;
 
+	u32 iec958_status;
+
 	/* Audio can not be enabled due to missing parameter(s) */
 	bool	missing_audio_param;
 
@@ -757,6 +759,9 @@ static int davinci_mcasp_set_tdm_slot(struct snd_soc_dai *dai,
 {
 	struct davinci_mcasp *mcasp = snd_soc_dai_get_drvdata(dai);
 
+	if (mcasp->op_mode == DAVINCI_MCASP_DIT_MODE)
+		return 0;
+
 	dev_dbg(mcasp->dev,
 		 "%s() tx_mask 0x%08x rx_mask 0x%08x slots %d width %d\n",
 		 __func__, tx_mask, rx_mask, slots, slot_width);
@@ -827,6 +832,20 @@ static int davinci_config_channel_size(struct davinci_mcasp *mcasp,
 		mcasp_mod_bits(mcasp, DAVINCI_MCASP_RXFMT_REG, RXROT(rx_rotate),
 			       RXROT(7));
 		mcasp_set_reg(mcasp, DAVINCI_MCASP_RXMASK_REG, mask);
+	} else {
+		/*
+		 * according to the TRM it should be TXROT=0, this one works:
+		 * 16 bit to 23-8 (TXROT=6, rotate 24 bits)
+		 * 24 bit to 23-0 (TXROT=0, rotate 0 bits)
+		 *
+		 * TXROT = 0 only works with 24bit samples
+		 */
+		tx_rotate = (sample_width / 4 + 2) & 0x7;
+
+		mcasp_mod_bits(mcasp, DAVINCI_MCASP_TXFMT_REG, TXROT(tx_rotate),
+			       TXROT(7));
+		mcasp_mod_bits(mcasp, DAVINCI_MCASP_TXFMT_REG, TXSSZ(15),
+			       TXSSZ(0x0F));
 	}
 
 	mcasp_set_reg(mcasp, DAVINCI_MCASP_TXMASK_REG, mask);
@@ -841,7 +860,7 @@ static int mcasp_common_hw_param(struct davinci_mcasp *mcasp, int stream,
 	int i;
 	u8 tx_ser = 0;
 	u8 rx_ser = 0;
-	u8 slots = mcasp->tdm_slots;
+	u8 slots = mcasp->op_mode == DAVINCI_MCASP_DIT_MODE ? 384 : mcasp->tdm_slots;
 	u8 max_active_serializers = (channels + slots - 1) / slots;
 	u8 max_rx_serializers, max_tx_serializers;
 	int active_serializers, numevt;
@@ -1031,16 +1050,18 @@ static int mcasp_i2s_hw_param(struct davinci_mcasp *mcasp, int stream,
 static int mcasp_dit_hw_param(struct davinci_mcasp *mcasp,
 			      unsigned int rate)
 {
-	u32 cs_value = 0;
-	u8 *cs_bytes = (u8*) &cs_value;
+	u8 *cs_bytes = (u8 *)&mcasp->iec958_status;
 
-	/* Set the TX format : 24 bit right rotation, 32 bit slot, Pad 0
-	   and LSB first */
-	mcasp_set_bits(mcasp, DAVINCI_MCASP_TXFMT_REG, TXROT(6) | TXSSZ(15));
+	if (!mcasp->dat_port)
+		mcasp_set_bits(mcasp, DAVINCI_MCASP_TXFMT_REG, TXSEL);
+	else
+		mcasp_clr_bits(mcasp, DAVINCI_MCASP_TXFMT_REG, TXSEL);
 
 	/* Set TX frame synch : DIT Mode, 1 bit width, internal, rising edge */
 	mcasp_set_reg(mcasp, DAVINCI_MCASP_TXFMCTL_REG, AFSXE | FSXMOD(0x180));
 
+	mcasp_set_reg(mcasp, DAVINCI_MCASP_TXMASK_REG, 0xFFFF);
+
 	/* Set the TX tdm : for all the slots */
 	mcasp_set_reg(mcasp, DAVINCI_MCASP_TXTDM_REG, 0xFFFFFFFF);
 
@@ -1049,16 +1070,8 @@ static int mcasp_dit_hw_param(struct davinci_mcasp *mcasp,
 
 	mcasp_clr_bits(mcasp, DAVINCI_MCASP_XEVTCTL_REG, TXDATADMADIS);
 
-	/* Only 44100 and 48000 are valid, both have the same setting */
-	mcasp_set_bits(mcasp, DAVINCI_MCASP_AHCLKXCTL_REG, AHCLKXDIV(3));
-
-	/* Enable the DIT */
-	mcasp_set_bits(mcasp, DAVINCI_MCASP_TXDITCTL_REG, DITEN);
-
 	/* Set S/PDIF channel status bits */
-	cs_bytes[0] = IEC958_AES0_CON_NOT_COPYRIGHT;
-	cs_bytes[1] = IEC958_AES1_CON_PCM_CODER;
-
+	cs_bytes[3] &= ~IEC958_AES3_CON_FS;
 	switch (rate) {
 	case 22050:
 		cs_bytes[3] |= IEC958_AES3_CON_FS_22050;
@@ -1088,12 +1101,15 @@ static int mcasp_dit_hw_param(struct davinci_mcasp *mcasp,
 		cs_bytes[3] |= IEC958_AES3_CON_FS_192000;
 		break;
 	default:
-		printk(KERN_WARNING "unsupported sampling rate: %d\n", rate);
+		dev_err(mcasp->dev, "unsupported sampling rate: %d\n", rate);
 		return -EINVAL;
 	}
 
-	mcasp_set_reg(mcasp, DAVINCI_MCASP_DITCSRA_REG, cs_value);
-	mcasp_set_reg(mcasp, DAVINCI_MCASP_DITCSRB_REG, cs_value);
+	mcasp_set_reg(mcasp, DAVINCI_MCASP_DITCSRA_REG, mcasp->iec958_status);
+	mcasp_set_reg(mcasp, DAVINCI_MCASP_DITCSRB_REG, mcasp->iec958_status);
+
+	/* Enable the DIT */
+	mcasp_set_bits(mcasp, DAVINCI_MCASP_TXDITCTL_REG, DITEN);
 
 	return 0;
 }
@@ -1237,12 +1253,18 @@ static int davinci_mcasp_hw_params(struct snd_pcm_substream *substream,
 		int slots = mcasp->tdm_slots;
 		int rate = params_rate(params);
 		int sbits = params_width(params);
+		unsigned int bclk_target;
 
 		if (mcasp->slot_width)
 			sbits = mcasp->slot_width;
 
+		if (mcasp->op_mode == DAVINCI_MCASP_IIS_MODE)
+			bclk_target = rate * sbits * slots;
+		else
+			bclk_target = rate * 128;
+
 		davinci_mcasp_calc_clk_div(mcasp, mcasp->sysclk_freq,
-					   rate * sbits * slots, true);
+					   bclk_target, true);
 	}
 
 	ret = mcasp_common_hw_param(mcasp, substream->stream,
@@ -1598,6 +1620,77 @@ static const struct snd_soc_dai_ops davinci_mcasp_dai_ops = {
 	.set_tdm_slot	= davinci_mcasp_set_tdm_slot,
 };
 
+static int davinci_mcasp_iec958_info(struct snd_kcontrol *kcontrol,
+				     struct snd_ctl_elem_info *uinfo)
+{
+	uinfo->type = SNDRV_CTL_ELEM_TYPE_IEC958;
+	uinfo->count = 1;
+
+	return 0;
+}
+
+static int davinci_mcasp_iec958_get(struct snd_kcontrol *kcontrol,
+				    struct snd_ctl_elem_value *uctl)
+{
+	struct snd_soc_dai *cpu_dai = snd_kcontrol_chip(kcontrol);
+	struct davinci_mcasp *mcasp = snd_soc_dai_get_drvdata(cpu_dai);
+
+	memcpy(uctl->value.iec958.status, &mcasp->iec958_status,
+	       sizeof(mcasp->iec958_status));
+
+	return 0;
+}
+
+static int davinci_mcasp_iec958_put(struct snd_kcontrol *kcontrol,
+				    struct snd_ctl_elem_value *uctl)
+{
+	struct snd_soc_dai *cpu_dai = snd_kcontrol_chip(kcontrol);
+	struct davinci_mcasp *mcasp = snd_soc_dai_get_drvdata(cpu_dai);
+
+	memcpy(&mcasp->iec958_status, uctl->value.iec958.status,
+	       sizeof(mcasp->iec958_status));
+
+	return 0;
+}
+
+static int davinci_mcasp_iec958_con_mask_get(struct snd_kcontrol *kcontrol,
+					     struct snd_ctl_elem_value *ucontrol)
+{
+	struct snd_soc_dai *cpu_dai = snd_kcontrol_chip(kcontrol);
+	struct davinci_mcasp *mcasp = snd_soc_dai_get_drvdata(cpu_dai);
+
+	memset(ucontrol->value.iec958.status, 0xff, sizeof(mcasp->iec958_status));
+	return 0;
+}
+
+static const struct snd_kcontrol_new davinci_mcasp_iec958_ctls[] = {
+	{
+		.access = (SNDRV_CTL_ELEM_ACCESS_READWRITE |
+			   SNDRV_CTL_ELEM_ACCESS_VOLATILE),
+		.iface = SNDRV_CTL_ELEM_IFACE_PCM,
+		.name = SNDRV_CTL_NAME_IEC958("", PLAYBACK, DEFAULT),
+		.info = davinci_mcasp_iec958_info,
+		.get = davinci_mcasp_iec958_get,
+		.put = davinci_mcasp_iec958_put,
+	}, {
+		.access = SNDRV_CTL_ELEM_ACCESS_READ,
+		.iface = SNDRV_CTL_ELEM_IFACE_MIXER,
+		.name = SNDRV_CTL_NAME_IEC958("", PLAYBACK, CON_MASK),
+		.info = davinci_mcasp_iec958_info,
+		.get = davinci_mcasp_iec958_con_mask_get,
+	},
+};
+
+static void davinci_mcasp_init_iec958_status(struct davinci_mcasp *mcasp)
+{
+	unsigned char *cs = (u8 *)&mcasp->iec958_status;
+
+	cs[0] = IEC958_AES0_CON_NOT_COPYRIGHT | IEC958_AES0_CON_EMPHASIS_NONE;
+	cs[1] = IEC958_AES1_CON_PCM_CODER;
+	cs[2] = IEC958_AES2_CON_SOURCE_UNSPEC | IEC958_AES2_CON_CHANNEL_UNSPEC;
+	cs[3] = IEC958_AES3_CON_CLOCK_1000PPM;
+}
+
 static int davinci_mcasp_dai_probe(struct snd_soc_dai *dai)
 {
 	struct davinci_mcasp *mcasp = snd_soc_dai_get_drvdata(dai);
@@ -1605,6 +1698,12 @@ static int davinci_mcasp_dai_probe(struct snd_soc_dai *dai)
 	dai->playback_dma_data = &mcasp->dma_data[SNDRV_PCM_STREAM_PLAYBACK];
 	dai->capture_dma_data = &mcasp->dma_data[SNDRV_PCM_STREAM_CAPTURE];
 
+	if (mcasp->op_mode == DAVINCI_MCASP_DIT_MODE) {
+		davinci_mcasp_init_iec958_status(mcasp);
+		snd_soc_add_dai_controls(dai, davinci_mcasp_iec958_ctls,
+					 ARRAY_SIZE(davinci_mcasp_iec958_ctls));
+	}
+
 	return 0;
 }
 
@@ -1651,7 +1750,8 @@ static struct snd_soc_dai_driver davinci_mcasp_dai[] = {
 			.channels_min	= 1,
 			.channels_max	= 384,
 			.rates		= DAVINCI_MCASP_RATES,
-			.formats	= DAVINCI_MCASP_PCM_FMTS,
+			.formats	= SNDRV_PCM_FMTBIT_S16_LE |
+					  SNDRV_PCM_FMTBIT_S24_LE,
 		},
 		.ops 		= &davinci_mcasp_dai_ops,
 	},
@@ -1871,6 +1971,8 @@ static int davinci_mcasp_get_config(struct davinci_mcasp *mcasp,
 		} else {
 			mcasp->tdm_slots = pdata->tdm_slots;
 		}
+	} else {
+		mcasp->tdm_slots = 32;
 	}
 
 	mcasp->num_serializer = pdata->num_serializer;
Signature · 1 line
2.32.0
#2 lore raw

Re: [PATCH 1/5] ASoC: ti: davinci-mcasp: Fix DIT mode support

I love your patch! Perhaps something to improve:

Hi Peter,

I love your patch! Perhaps something to improve:

[auto build test WARNING on asoc/for-next]
[also build test WARNING on robh/for-next omap/for-next linus/master v5.13 next-20210701] [If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Peter-Ujfalusi/ASoC-ti-davinci-mcasp-Fix-the-DIT-mode-and-OMAP4-support/20210704-230253
base:   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git for-next
config: riscv-randconfig-r021-20210705 (attached as .config)
compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project 3f9bf9f42a9043e20c6d2a74dd4f47a90a7e2b41)

reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# install riscv cross compiling tool for clang build
# apt-get install binutils-riscv64-linux-gnu
# https://github.com/0day-ci/linux/commit/085675d5332885fb340fcf97f4ed826259c5fb0d
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Peter-Ujfalusi/ASoC-ti-davinci-mcasp-Fix-the-DIT-mode-and-OMAP4-support/20210704-230253 git checkout 085675d5332885fb340fcf97f4ed826259c5fb0d
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=riscv

If you fix the issue, kindly add following tag as appropriate

All warnings (new ones prefixed by >>):

>> sound/soc/ti/davinci-mcasp.c:863:56: warning: implicit conversion from 'int' to 'u8' (aka 'unsigned char') changes value from 384 to 128 [-Wconstant-conversion]
           u8 slots = mcasp->op_mode == DAVINCI_MCASP_DIT_MODE ? 384 : mcasp->tdm_slots;
              ~~~~~                                              ^~~

1 warning generated.

Kconfig warnings: (for reference only)
WARNING: unmet direct dependencies detected for LOCKDEP
Depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT && (FRAME_POINTER || MIPS || PPC || S390 || MICROBLAZE || ARM || ARC || X86) Selected by
- LOCK_STAT && DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
- DEBUG_LOCK_ALLOC && DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT

vim +863 sound/soc/ti/davinci-mcasp.c

855
856 static int mcasp_common_hw_param(struct davinci_mcasp *mcasp, int stream,
857 int period_words, int channels)
858 {
859 struct snd_dmaengine_dai_dma_data *dma_data = &mcasp->dma_data[stream];
860 int i;
861 u8 tx_ser = 0;
862 u8 rx_ser = 0;

 > 863		u8 slots = mcasp->op_mode == DAVINCI_MCASP_DIT_MODE ? 384 : mcasp->tdm_slots;

864 u8 max_active_serializers = (channels + slots - 1) / slots;
865 u8 max_rx_serializers, max_tx_serializers;
866 int active_serializers, numevt;
867 u32 reg;
868 /* Default configuration */
869 if (mcasp->version < MCASP_VERSION_3)
870 mcasp_set_bits(mcasp, DAVINCI_MCASP_PWREMUMGT_REG, MCASP_SOFT);
871
872 if (stream == SNDRV_PCM_STREAM_PLAYBACK) {
873 mcasp_set_reg(mcasp, DAVINCI_MCASP_TXSTAT_REG, 0xFFFFFFFF);
874 mcasp_clr_bits(mcasp, DAVINCI_MCASP_XEVTCTL_REG, TXDATADMADIS);
875 max_tx_serializers = max_active_serializers;
876 max_rx_serializers =
877 mcasp->active_serializers[SNDRV_PCM_STREAM_CAPTURE];
878 } else {
879 mcasp_set_reg(mcasp, DAVINCI_MCASP_RXSTAT_REG, 0xFFFFFFFF);
880 mcasp_clr_bits(mcasp, DAVINCI_MCASP_REVTCTL_REG, RXDATADMADIS);
881 max_tx_serializers =
882 mcasp->active_serializers[SNDRV_PCM_STREAM_PLAYBACK];
883 max_rx_serializers = max_active_serializers;
884 }
885
886 for (i = 0; i < mcasp->num_serializer; i++) {
887 mcasp_set_bits(mcasp, DAVINCI_MCASP_XRSRCTL_REG(i),
888 mcasp->serial_dir[i]);
889 if (mcasp->serial_dir[i] == TX_MODE &&
890 tx_ser < max_tx_serializers) {
891 mcasp_mod_bits(mcasp, DAVINCI_MCASP_XRSRCTL_REG(i),
892 mcasp->dismod, DISMOD_MASK);
893 set_bit(PIN_BIT_AXR(i), &mcasp->pdir);
894 tx_ser++;
895 } else if (mcasp->serial_dir[i] == RX_MODE &&
896 rx_ser < max_rx_serializers) {
897 clear_bit(PIN_BIT_AXR(i), &mcasp->pdir);
898 rx_ser++;
899 } else {
900 /* Inactive or unused pin, set it to inactive */
901 mcasp_mod_bits(mcasp, DAVINCI_MCASP_XRSRCTL_REG(i),
902 SRMOD_INACTIVE, SRMOD_MASK);
903 /* If unused, set DISMOD for the pin */
904 if (mcasp->serial_dir[i] != INACTIVE_MODE)
905 mcasp_mod_bits(mcasp,
906 DAVINCI_MCASP_XRSRCTL_REG(i),
907 mcasp->dismod, DISMOD_MASK);
908 clear_bit(PIN_BIT_AXR(i), &mcasp->pdir);
909 }
910 }
911
912 if (stream == SNDRV_PCM_STREAM_PLAYBACK) {
913 active_serializers = tx_ser;
914 numevt = mcasp->txnumevt;
915 reg = mcasp->fifo_base + MCASP_WFIFOCTL_OFFSET;
916 } else {
917 active_serializers = rx_ser;
918 numevt = mcasp->rxnumevt;
919 reg = mcasp->fifo_base + MCASP_RFIFOCTL_OFFSET;
920 }
921
922 if (active_serializers < max_active_serializers) {
923 dev_warn(mcasp->dev, "stream has more channels (%d) than are "
924 "enabled in mcasp (%d)\n", channels,
925 active_serializers * slots);
926 return -EINVAL;
927 }
928
929 /* AFIFO is not in use */
930 if (!numevt) {
931 /* Configure the burst size for platform drivers */
932 if (active_serializers > 1) {
933 /*
934 * If more than one serializers are in use we have one
935 * DMA request to provide data for all serializers.
936 * For example if three serializers are enabled the DMA
937 * need to transfer three words per DMA request.
938 */
939 dma_data->maxburst = active_serializers;
940 } else {
941 dma_data->maxburst = 0;
942 }
943
944 goto out;
945 }
946
947 if (period_words % active_serializers) {
948 dev_err(mcasp->dev, "Invalid combination of period words and "
949 "active serializers: %d, %d\n", period_words,
950 active_serializers);
951 return -EINVAL;
952 }
953
954 /*
955 * Calculate the optimal AFIFO depth for platform side:
956 * The number of words for numevt need to be in steps of active
957 * serializers.
958 */
959 numevt = (numevt / active_serializers) * active_serializers;
960
961 while (period_words % numevt && numevt > 0)
962 numevt -= active_serializers;
963 if (numevt <= 0)
964 numevt = active_serializers;
965
966 mcasp_mod_bits(mcasp, reg, active_serializers, NUMDMA_MASK);
967 mcasp_mod_bits(mcasp, reg, NUMEVT(numevt), NUMEVT_MASK);
968
969 /* Configure the burst size for platform drivers */
970 if (numevt == 1)
971 numevt = 0;
972 dma_data->maxburst = numevt;
973
974 out:
975 mcasp->active_serializers[stream] = active_serializers;
976
977 return 0;
978 }
979

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

#3 lore raw

Re: [PATCH 1/5] ASoC: ti: davinci-mcasp: Fix DIT mode support

In reply to Peter Ujfalusi

Hi,

Quoted 60 lines · Peter Ujfalusi

The DIT mode support has not been tested due to lack of platform where it can be tested.
To be able to use the McASP on OMAP4/5 (only supporting DIT mode) we need to have DIT mode working in the McASP driver on a know platform. After hacking around (on BBW, mcasp1.axr1 can be routed out for this) it appeared that DIT mode is broken.

This patch fixes it up and 16/24 bit audio works along with passthrough, but I have only tested with DTS example and test files.

---
 sound/soc/ti/davinci-mcasp.c | 142 ++++++++++++++++++++++++++++++-----
 1 file changed, 122 insertions(+), 20 deletions(-)
sound/soc/ti/davinci-mcasp.c +19 -0
index 017a5a5e56cd..e56a22c28467 100644
@@ -83,6 +83,8 @@ struct davinci_mcasp {
 	struct snd_pcm_substream *substreams[2];
 	unsigned int dai_fmt;
 
+	u32 iec958_status;
+
 	/* Audio can not be enabled due to missing parameter(s) */
 	bool	missing_audio_param;
 
@@ -757,6 +759,9 @@ static int davinci_mcasp_set_tdm_slot(struct snd_soc_dai *dai,
 {
 	struct davinci_mcasp *mcasp = snd_soc_dai_get_drvdata(dai);
 
+	if (mcasp->op_mode == DAVINCI_MCASP_DIT_MODE)
+		return 0;
+
 	dev_dbg(mcasp->dev,
 		 "%s() tx_mask 0x%08x rx_mask 0x%08x slots %d width %d\n",
 		 __func__, tx_mask, rx_mask, slots, slot_width);
@@ -827,6 +832,20 @@ static int davinci_config_channel_size(struct davinci_mcasp *mcasp,
 		mcasp_mod_bits(mcasp, DAVINCI_MCASP_RXFMT_REG, RXROT(rx_rotate),
 			       RXROT(7));
 		mcasp_set_reg(mcasp, DAVINCI_MCASP_RXMASK_REG, mask);
+	} else {
+		/*
+		 * according to the TRM it should be TXROT=0, this one works:
+		 * 16 bit to 23-8 (TXROT=6, rotate 24 bits)
+		 * 24 bit to 23-0 (TXROT=0, rotate 0 bits)
+		 *
+		 * TXROT = 0 only works with 24bit samples
+		 */
+		tx_rotate = (sample_width / 4 + 2) & 0x7;
+
+		mcasp_mod_bits(mcasp, DAVINCI_MCASP_TXFMT_REG, TXROT(tx_rotate),
+			       TXROT(7));
+		mcasp_mod_bits(mcasp, DAVINCI_MCASP_TXFMT_REG, TXSSZ(15),
+			       TXSSZ(0x0F));
 	}
 
 	mcasp_set_reg(mcasp, DAVINCI_MCASP_TXMASK_REG, mask);

sound/soc/ti/davinci-mcasp.c static int mcasp_common_hw_param(struct davinci_mcasp *mcasp, int stream, · line 863

Quoted 5 lines
sound/soc/ti/davinci-mcasp.c +1 -1
@@ -841,7 +860,7 @@ static int mcasp_common_hw_param(struct davinci_mcasp *mcasp, int stream,
 	int i;
 	u8 tx_ser = 0;
 	u8 rx_ser = 0;
-	u8 slots = mcasp->tdm_slots;
+	u8 slots = mcasp->op_mode == DAVINCI_MCASP_DIT_MODE ? 384 : mcasp->tdm_slots;

This is wrong.
384 will not fit to u8...
I wonder why I don't saw this...

Signature · 1 line
Péter
maintainer
#4 lore raw

Re: [PATCH 1/5] ASoC: ti: davinci-mcasp: Fix DIT mode support

Quoted 7 lines · Péter Ujfalusi

sound/soc/ti/davinci-mcasp.c static int mcasp_common_hw_param(struct davinci_mcasp *mcasp, int stream, · line 863

Quoted 1 line · Peter Ujfalusi
sound/soc/ti/davinci-mcasp.c +1 -0
+	u8 slots = mcasp->op_mode == DAVINCI_MCASP_DIT_MODE ? 384 : mcasp->tdm_slots;

This is wrong.
384 will not fit to u8...
I wonder why I don't saw this...

Or the compiler for that matter...

#5 lore raw

PATCH 2/5 ASoC: dt-bindings: davinci-mcasp: Add compatible string for OMAP4

In reply to Peter Ujfalusi

OMAP4 has one McASP instance with single serializer and supporting only DIT mode.
According to the TRM the DAT port needs to be accessed as specific offset compared to other devices where access to any part of the DAT region is valid.
To handle this constraint we need to introduce a new compatiple string for OMAP4.

---
 Documentation/devicetree/bindings/sound/davinci-mcasp-audio.txt | 1 +
 1 file changed, 1 insertion(+)
Documentation/devicetree/bindings/sound/davinci-mcasp-audio.txt +1 -0
index c483dcec01f8..bd863bd69501 100644
@@ -6,6 +6,7 @@ Required properties:
 	"ti,da830-mcasp-audio"	: for both DA830 & DA850 platforms
 	"ti,am33xx-mcasp-audio"	: for AM33xx platforms (AM33xx, AM43xx, TI81xx)
 	"ti,dra7-mcasp-audio"	: for DRA7xx platforms
+	"ti,omap4-mcasp-audio"	: for OMAP4
 
 - reg : Should contain reg specifiers for the entries in the reg-names property.
 - reg-names : Should contain:
Signature · 1 line
2.32.0
#6 lore raw

PATCH 3/5 ASoC: ti: davinci-mcasp: Add support for the OMAP4 version of McASP

In reply to Peter Ujfalusi

From: Peter Ujfalusi <[email protected]>

There is a single McASP on OMAP4 (and OMAP5) which is configured to only support DIT playback mode on a single serializer.

Add 0x200 offset to DAT port address as the TRM suggests it.

---
 include/linux/platform_data/davinci_asp.h |  1 +
 sound/soc/ti/Kconfig                      |  1 +
 sound/soc/ti/davinci-mcasp.c              | 26 ++++++++++++++++++++---
 3 files changed, 25 insertions(+), 3 deletions(-)
include/linux/platform_data/davinci_asp.h +1 -0
index 5d1fb0d78a22..76b13ef67562 100644
@@ -96,6 +96,7 @@ enum {
 	MCASP_VERSION_2,	/* DA8xx/OMAPL1x */
 	MCASP_VERSION_3,        /* TI81xx/AM33xx */
 	MCASP_VERSION_4,	/* DRA7xxx */
+	MCASP_VERSION_OMAP,	/* OMAP4/5 */
 };
 
 enum mcbsp_clk_input_pin {
sound/soc/ti/Kconfig +1 -0
index 698d7bc84dcf..1d9fe3fca193 100644
@@ -35,6 +35,7 @@ config SND_SOC_DAVINCI_MCASP
 	  various Texas Instruments SoCs like:
 	  - daVinci devices
 	  - Sitara line of SoCs (AM335x, AM438x, etc)
+	  - OMAP4
 	  - DRA7x devices
 	  - Keystone devices
 	  - K3 devices (am654, j721e)
sound/soc/ti/davinci-mcasp.c +23 -3
index e56a22c28467..f453728d2afc 100644
@@ -1788,6 +1788,12 @@ static struct davinci_mcasp_pdata dra7_mcasp_pdata = {
 	.version = MCASP_VERSION_4,
 };
 
+static struct davinci_mcasp_pdata omap_mcasp_pdata = {
+	.tx_dma_offset = 0x200,
+	.rx_dma_offset = 0,
+	.version = MCASP_VERSION_OMAP,
+};
+
 static const struct of_device_id mcasp_dt_ids[] = {
 	{
 		.compatible = "ti,dm646x-mcasp-audio",
@@ -1805,6 +1811,10 @@ static const struct of_device_id mcasp_dt_ids[] = {
 		.compatible = "ti,dra7-mcasp-audio",
 		.data = &dra7_mcasp_pdata,
 	},
+	{
+		.compatible = "ti,omap4-mcasp-audio",
+		.data = &omap_mcasp_pdata,
+	},
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, mcasp_dt_ids);
@@ -2344,10 +2354,17 @@ static int davinci_mcasp_probe(struct platform_device *pdev)
 
 	dma_data = &mcasp->dma_data[SNDRV_PCM_STREAM_PLAYBACK];
 	dma_data->filter_data = "tx";
-	if (dat)
+	if (dat) {
 		dma_data->addr = dat->start;
-	else
+		/*
+		 * According to the TRM there should be 0x200 offset added to
+		 * the DAT port address
+		 */
+		if (mcasp->version == MCASP_VERSION_OMAP)
+			dma_data->addr += davinci_mcasp_txdma_offset(mcasp->pdata);
+	} else {
 		dma_data->addr = mem->start + davinci_mcasp_txdma_offset(mcasp->pdata);
+	}
 
 
 	/* RX is not valid in DIT mode */
@@ -2412,7 +2429,10 @@ static int davinci_mcasp_probe(struct platform_device *pdev)
 		ret = edma_pcm_platform_register(&pdev->dev);
 		break;
 	case PCM_SDMA:
-		ret = sdma_pcm_platform_register(&pdev->dev, "tx", "rx");
+		if (mcasp->op_mode == DAVINCI_MCASP_IIS_MODE)
+			ret = sdma_pcm_platform_register(&pdev->dev, "tx", "rx");
+		else
+			ret = sdma_pcm_platform_register(&pdev->dev, "tx", NULL);
 		break;
 	case PCM_UDMA:
 		ret = udma_pcm_platform_register(&pdev->dev);
Signature · 1 line
2.32.0
#7 lore raw

PATCH 4/5 ARM: dts: omap4-l4-abe: Correct sidle modes for McASP

In reply to Peter Ujfalusi

McASP only supports Force-idle, No-idle and Smart-idle modes

---
 arch/arm/boot/dts/omap4-l4-abe.dtsi | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
arch/arm/boot/dts/omap4-l4-abe.dtsi +1 -2
index a9573d441dea..8287fdaa526e 100644
@@ -192,8 +192,7 @@ target-module@28000 {			/* 0x40128000, ap 8 08.0 */
 			reg-names = "rev", "sysc";
 			ti,sysc-sidle = <SYSC_IDLE_FORCE>,
 					<SYSC_IDLE_NO>,
-					<SYSC_IDLE_SMART>,
-					<SYSC_IDLE_SMART_WKUP>;
+					<SYSC_IDLE_SMART>;
 			/* Domains (V, P, C): iva, abe_pwrdm, abe_clkdm */
 			clocks = <&abe_clkctrl OMAP4_MCASP_CLKCTRL 0>;
 			clock-names = "fck";
Signature · 1 line
2.32.0
#8 lore raw

PATCH 5/5 ARM: dts: omap4-l4-abe: Add McASP configuration

In reply to Peter Ujfalusi

OMAP4 has a single McASP instance with single serializer and locked for DIT mode.
To be able to enable the support the following fixes needed:
- Add the DAT port ranges to the target module's ranges
- SIDLE mode must be disabled as it is not working with McASP most likely module integration issue with McASP

We can already fill in the op-mode and serial-dir for McASP as it only supports this configuration, but keep the module disabled as there is no known device available where it is used.

---
 arch/arm/boot/dts/omap4-l4-abe.dtsi | 38 +++++++++++++++--------------
 1 file changed, 20 insertions(+), 18 deletions(-)
arch/arm/boot/dts/omap4-l4-abe.dtsi +20 -18
index 8287fdaa526e..a8d66240d17d 100644
@@ -186,36 +186,38 @@ mcbsp3: mcbsp@0 {
 		};
 
 		target-module@28000 {			/* 0x40128000, ap 8 08.0 */
+							/* 0x4012a000, ap 10 0a.0 */
 			compatible = "ti,sysc-mcasp", "ti,sysc";
 			reg = <0x28000 0x4>,
 			      <0x28004 0x4>;
 			reg-names = "rev", "sysc";
 			ti,sysc-sidle = <SYSC_IDLE_FORCE>,
-					<SYSC_IDLE_NO>,
-					<SYSC_IDLE_SMART>;
+					<SYSC_IDLE_NO>;
 			/* Domains (V, P, C): iva, abe_pwrdm, abe_clkdm */
 			clocks = <&abe_clkctrl OMAP4_MCASP_CLKCTRL 0>;
 			clock-names = "fck";
 			#address-cells = <1>;
 			#size-cells = <1>;
 			ranges = <0x0 0x28000 0x1000>,
-				 <0x49028000 0x49028000 0x1000>;
-
-			/*
-			 * Child device unsupported by davinci-mcasp. At least
-			 * RX path is disabled for omap4, and only DIT mode
-			 * works with no I2S. See also old Android kernel
-			 * omap-mcasp driver for more information.
-			 */
-		};
-
-		target-module@2a000 {			/* 0x4012a000, ap 10 0a.0 */
-			compatible = "ti,sysc";
-			status = "disabled";
-			#address-cells = <1>;
-			#size-cells = <1>;
-			ranges = <0x0 0x2a000 0x1000>,
+				 <0x49028000 0x49028000 0x1000>,
+				 <0x2000 0x2a000 0x1000>,
 				 <0x4902a000 0x4902a000 0x1000>;
+
+			mcasp0: mcasp@0 {
+				compatible = "ti,omap4-mcasp-audio";
+				reg = <0x0 0x2000>,
+				      <0x4902a000 0x1000>;	/* L3 data port */
+				reg-names = "mpu","dat";
+				interrupts = <GIC_SPI 109 IRQ_TYPE_LEVEL_HIGH>;
+				interrupt-names = "tx";
+				dmas = <&sdma 8>;
+				dma-names = "tx";
+				clocks = <&abe_clkctrl OMAP4_MCASP_CLKCTRL 0>;
+				clock-names = "fck";
+				op-mode = <1>;	/* MCASP_DIT_MODE */
+				serial-dir = < 1 >; /* 1 TX serializers */
+				status = "disabled";
+			};
 		};
 
 		target-module@2e000 {			/* 0x4012e000, ap 12 0c.0 */
Signature · 1 line
2.32.0
#9 lore raw
In reply to Peter Ujfalusi

Hi Péter,

Quoted 18 lines

Am 04.07.2021 um 17:04 schrieb Peter Ujfalusi <[email protected]>:

Hi,

it has been on my todo list for several years to support McASP on OMAP4 devices. For Galaxy Nexus we had an omap-mcasp driver (which was mostly a stripped down davinci-mcasp driver) to support what was needed on that specific phone + it's dock for S/PDIF (48KHz, 16bit, stereo).

Not many (if any) device available to test the DIT mode of McASP. I have used BeagleBone White (McASP1 AXR3 can be routed to a pin) to get the S/PDIF mode working then PandaES for OMAP4 support (on PandaES the gpio_121 is not used and the signal is routed to expansion J6 pin14)

In theory the McASP in OMAP5 should be working after this series, but the OMAP5 TRM is not public and I do not have one to check the addresses and see if there is a way to test it on omap5-uevm.

Quoted 5 lines

Mark, Tony:
The ASoC and dts patches can go via separate tree I felt that it is better if they are together, at least initially.

Nikolaus: fyi, this might be useful for Pyra?

Yes, definitively.

I just wonder how compatible it is with the "old" omap-mcasp driver we carry along for the Pyra:

https://git.goldelico.com/?p=letux-kernel.git;a=blob;f=sound/soc/ti/omap-mcasp.c;h=24d839a24f34ec02456176744d3690016592b5a3;hb=7789ab635e56a2fd9fe04ca2e9ccfeaccb8e0f52

Does it now have additional functionality or robustness? And does it need changes of the OMAP5 device tree?

And also important seems to be the ABE/AESS integration where we so far failed to make it work again on more recent kernels. Our most recent patch set is here:

https://git.goldelico.com/?p=letux-kernel.git;a=shortlog;h=refs/heads/letux/aess-v5

Anyways I am happy to see a little progress on the omap4/5 audio subsystem.

BR and thanks,
Nikolaus

Quoted 21 lines

Regards,
Péter
---
Peter Ujfalusi (5):
ASoC: ti: davinci-mcasp: Fix DIT mode support
ASoC: dt-bindings: davinci-mcasp: Add compatible string for OMAP4
ASoC: ti: davinci-mcasp: Add support for the OMAP4 version of McASP
ARM: dts: omap4-l4-abe: Correct sidle modes for McASP
ARM: dts: omap4-l4-abe: Add McASP configuration

.../bindings/sound/davinci-mcasp-audio.txt    |   1 +
arch/arm/boot/dts/omap4-l4-abe.dtsi           |  39 ++--
include/linux/platform_data/davinci_asp.h     |   1 +
sound/soc/ti/Kconfig                          |   1 +
sound/soc/ti/davinci-mcasp.c                  | 168 +++++++++++++++---
5 files changed, 168 insertions(+), 42 deletions(-)
Signature · 1 line
2.32.0
#10 lore raw

Hi Nikolaus,

Quoted 26 lines · H. Nikolaus Schaller

Hi Péter,

Quoted 18 lines

Am 04.07.2021 um 17:04 schrieb Peter Ujfalusi <[email protected]>:

Hi,

it has been on my todo list for several years to support McASP on OMAP4 devices. For Galaxy Nexus we had an omap-mcasp driver (which was mostly a stripped down davinci-mcasp driver) to support what was needed on that specific phone + it's dock for S/PDIF (48KHz, 16bit, stereo).

Not many (if any) device available to test the DIT mode of McASP. I have used BeagleBone White (McASP1 AXR3 can be routed to a pin) to get the S/PDIF mode working then PandaES for OMAP4 support (on PandaES the gpio_121 is not used and the signal is routed to expansion J6 pin14)

In theory the McASP in OMAP5 should be working after this series, but the OMAP5 TRM is not public and I do not have one to check the addresses and see if there is a way to test it on omap5-uevm.

Quoted 5 lines

Mark, Tony:
The ASoC and dts patches can go via separate tree I felt that it is better if they are together, at least initially.

Nikolaus: fyi, this might be useful for Pyra?

Quoted 8 lines

Yes, definitively.

I just wonder how compatible it is with the "old" omap-mcasp driver we carry along for the Pyra:

https://git.goldelico.com/?p=letux-kernel.git;a=blob;f=sound/soc/ti/omap-mcasp.c;h=24d839a24f34ec02456176744d3690016592b5a3;hb=7789ab635e56a2fd9fe04ca2e9ccfeaccb8e0f52

Does it now have additional functionality or robustness?

The davinci-mcasp with series does work on OMAP4. I could not made the omap-mcasp working. I have picked the driver from pyra in hopes that it is working so I will ahve a stable base.
Have you tested it?

the omap-mcasp supports only 48KHz and 16bit, no passthrough, no 24bit, all is hardwired for the Galaxy Nexus setup, afaik.

The davinci-mcasp should be better all around, I think the PM side is also more robust.

The other point is that I would not accept the omap-mcasp driver upstream. The IP is the same so the driver should be also...

Quoted 1 line

And does it need changes of the OMAP5 device tree?

In a similar way as the dts patches in this series. fwiw:
https://github.com/omap-audio/linux-audio/commit/2f0a632db0b52321f1fc3623d088a741aeb72ef7
is the patch I have for PandaES to get S/PDIF working.

Quoted 2 lines

And also important seems to be the ABE/AESS integration where we so far failed to make it work again on more recent kernels.

The McASP is not routed via ABE/AESS, it has been always handled outside of the firmware, so it is just a simple link outside.

Quoted 3 lines

Our most recent patch set is here:

https://git.goldelico.com/?p=letux-kernel.git;a=shortlog;h=refs/heads/letux/aess-v5

Yes, I'm following it ;)

Quoted 1 line

Anyways I am happy to see a little progress on the omap4/5 audio subsystem.

It is in maintenance mode as there is not much of an interest outside of a small group of people, but I try to help to get it better for Pyra.

Quoted 27 lines

BR and thanks,
Nikolaus

Quoted 21 lines

Regards,
Péter
---
Peter Ujfalusi (5):
ASoC: ti: davinci-mcasp: Fix DIT mode support
ASoC: dt-bindings: davinci-mcasp: Add compatible string for OMAP4
ASoC: ti: davinci-mcasp: Add support for the OMAP4 version of McASP
ARM: dts: omap4-l4-abe: Correct sidle modes for McASP
ARM: dts: omap4-l4-abe: Add McASP configuration

.../bindings/sound/davinci-mcasp-audio.txt    |   1 +
arch/arm/boot/dts/omap4-l4-abe.dtsi           |  39 ++--
include/linux/platform_data/davinci_asp.h     |   1 +
sound/soc/ti/Kconfig                          |   1 +
sound/soc/ti/davinci-mcasp.c                  | 168 +++++++++++++++---
5 files changed, 168 insertions(+), 42 deletions(-)
Signature · 1 line
2.32.0
Signature · 1 line
Péter
#11 lore raw
Quoted 12 lines

Am 05.07.2021 um 16:10 schrieb Péter Ujfalusi <[email protected]>:

Hi Nikolaus,

Quoted 6 lines

Yes, definitively.

I just wonder how compatible it is with the "old" omap-mcasp driver we carry along for the Pyra:

Quoted 8 lines
Quoted 3 lines

https://git.goldelico.com/?p=letux-kernel.git;a=blob;f=sound/soc/ti/omap-mcasp.c;h=24d839a24f34ec02456176744d3690016592b5a3;hb=7789ab635e56a2fd9fe04ca2e9ccfeaccb8e0f52

Does it now have additional functionality or robustness?

The davinci-mcasp with series does work on OMAP4. I could not made the omap-mcasp working. I have picked the driver from pyra in hopes that it is working so I will ahve a stable base.
Have you tested it?

AFAIR it was not tested. e have the sound system working (with some glitches) but that goes through McPDM I think.

Quoted 9 lines

the omap-mcasp supports only 48KHz and 16bit, no passthrough, no 24bit, all is hardwired for the Galaxy Nexus setup, afaik.

The davinci-mcasp should be better all around, I think the PM side is also more robust.

The other point is that I would not accept the omap-mcasp driver upstream. The IP is the same so the driver should be also...

:)

Quoted 7 lines
Quoted 2 lines

And does it need changes of the OMAP5 device tree?

In a similar way as the dts patches in this series. fwiw:
https://github.com/omap-audio/linux-audio/commit/2f0a632db0b52321f1fc3623d088a741aeb72ef7
is the patch I have for PandaES to get S/PDIF working.

If I find a little time it should not be difficult to replace your new driver and fix the DTS in our ABE/AESS branch.

Quoted 6 lines
Quoted 3 lines

And also important seems to be the ABE/AESS integration where we so far failed to make it work again on more recent kernels.

The McASP is not routed via ABE/AESS, it has been always handled outside of the firmware, so it is just a simple link outside.

Ah, ok. Makes it easier to test :)

Quoted 11 lines
Quoted 4 lines

Our most recent patch set is here:

https://git.goldelico.com/?p=letux-kernel.git;a=shortlog;h=refs/heads/letux/aess-v5

Yes, I'm following it ;)

Quoted 1 line

Anyways I am happy to see a little progress on the omap4/5 audio subsystem.

It is in maintenance mode as there is not much of an interest outside of a small group of people, but I try to help to get it better for Pyra.

And a lot of public information has been lost (OMAPpedia). But there are some backups:

https://web.archive.org/web/20181111114134/http://omappedia.org/wiki/Audio_Drive_Arch

BR and thanks,
Nikolaus