ALSA: hda - Update CA0132 codec to load DSP firmware binary

This patch adds the code needed to fetch the DSP binary image from the local
firmware install location and transfer it over to the chip using the new DSP
loader bus ops.
Actual DSP effect controls, parameters and mixers are to be included later.

- Add calls to new DSP loader system to transfer firmware to the hardware.
- Add chip read/write routines, DSP I/O, SCP packet format helper functions and
  transfer DMA management.
- Add guard around DSP download to ensure loader config switch is enabled.

The general scheme for downloading the DSP is as follows:
1) If DSP firmware loader is enabled, ca0132_download_dsp() is called
to start the process.
2) Driver requests DSP image using request_firmware().
3) Driver sets up the streaming DMA for DSP image download
with dspload_image() and dspxfr_image(), which in turn calls the DSP loader
op snd_hda_codec_load_dsp_prepare() to ready the system.
4) DSP image will consist of 1 or more segments, each transferred in sequence
by a call to dspxfr_one_seg() and snd_hda_codec_load_dsp_trigger().
5) Once complete, the loader state is cleaned up with
snd_hda_codec_load_dsp_cleanup().

Signed-off-by: Ian Minett <ian_minett@creativelabs.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

sound/pci/hda/patch_ca0132.c

@@ -48,9 +48,6 @@
 #define WIDGET_CHIP_CTRL      0x15
 #define WIDGET_DSP_CTRL       0x16
 
-#define WUH_MEM_CONNID        10
-#define DSP_MEM_CONNID        16
-
 #define MEM_CONNID_MICIN1     3
 #define MEM_CONNID_MICIN2     5
 #define MEM_CONNID_MICOUT1    12
@@ -62,6 +59,10 @@
 #define SCP_SET    0
 #define SCP_GET    1
 
+#define EFX_FILE   "ctefx.bin"
+
+MODULE_FIRMWARE(EFX_FILE);
+
 enum hda_cmd_vendor_io {
 	/* for DspIO node */
 	VENDOR_DSPIO_SCP_WRITE_DATA_LOW      = 0x000,
@@ -320,192 +321,1736 @@ static int _add_switch(struct hda_codec *codec, hda_nid_t nid, const char *pfx,
 	return snd_hda_ctl_add(codec, nid, snd_ctl_new1(&knew, codec));
 }
 
-static int _add_volume(struct hda_codec *codec, hda_nid_t nid, const char *pfx,
-		       int chan, int dir)
+static int _add_volume(struct hda_codec *codec, hda_nid_t nid, const char *pfx,
+		       int chan, int dir)
+{
+	char namestr[44];
+	int type = dir ? HDA_INPUT : HDA_OUTPUT;
+	struct snd_kcontrol_new knew =
+		HDA_CODEC_VOLUME_MONO(namestr, nid, chan, 0, type);
+	if ((query_amp_caps(codec, nid, type) & AC_AMPCAP_NUM_STEPS) == 0) {
+		snd_printdd("Skipping '%s %s Volume' (no amp on node 0x%x)\n", pfx, dirstr[dir], nid);
+		return 0;
+	}
+	sprintf(namestr, "%s %s Volume", pfx, dirstr[dir]);
+	return snd_hda_ctl_add(codec, nid, snd_ctl_new1(&knew, codec));
+}
+
+#define add_out_switch(codec, nid, pfx) _add_switch(codec, nid, pfx, 3, 0)
+#define add_out_volume(codec, nid, pfx) _add_volume(codec, nid, pfx, 3, 0)
+#define add_in_switch(codec, nid, pfx) _add_switch(codec, nid, pfx, 3, 1)
+#define add_in_volume(codec, nid, pfx) _add_volume(codec, nid, pfx, 3, 1)
+#define add_mono_switch(codec, nid, pfx, chan) \
+	_add_switch(codec, nid, pfx, chan, 0)
+#define add_mono_volume(codec, nid, pfx, chan) \
+	_add_volume(codec, nid, pfx, chan, 0)
+#define add_in_mono_switch(codec, nid, pfx, chan) \
+	_add_switch(codec, nid, pfx, chan, 1)
+#define add_in_mono_volume(codec, nid, pfx, chan) \
+	_add_volume(codec, nid, pfx, chan, 1)
+
+enum dsp_download_state {
+	DSP_DOWNLOAD_FAILED = -1,
+	DSP_DOWNLOAD_INIT   = 0,
+	DSP_DOWNLOADING     = 1,
+	DSP_DOWNLOADED      = 2
+};
+
+struct hda_stream_format {
+	unsigned int   sample_rate;
+	unsigned short valid_bits_per_sample;
+	unsigned short container_size;
+	unsigned short number_channels;
+};
+
+/* retrieve parameters from hda format */
+#define get_hdafmt_chs(fmt)	(fmt & 0xf)
+#define get_hdafmt_bits(fmt)	((fmt >> 4) & 0x7)
+#define get_hdafmt_rate(fmt)	((fmt >> 8) & 0x7f)
+#define get_hdafmt_type(fmt)	((fmt >> 15) & 0x1)
+
+/*
+ * CA0132 specific
+ */
+
+struct ca0132_spec {
+	struct auto_pin_cfg autocfg;
+	struct hda_multi_out multiout;
+	hda_nid_t out_pins[AUTO_CFG_MAX_OUTS];
+	hda_nid_t dacs[AUTO_CFG_MAX_OUTS];
+	hda_nid_t hp_dac;
+	hda_nid_t input_pins[AUTO_PIN_LAST];
+	hda_nid_t adcs[AUTO_PIN_LAST];
+	hda_nid_t dig_out;
+	hda_nid_t dig_in;
+	unsigned int num_inputs;
+	long curr_hp_switch;
+	long curr_hp_volume[2];
+	long curr_speaker_switch;
+	const char *input_labels[AUTO_PIN_LAST];
+	struct hda_pcm pcm_rec[2]; /* PCM information */
+
+	/* chip access */
+	struct mutex chipio_mutex; /* chip access mutex */
+	u32 curr_chip_addx;
+
+	/* DSP download related */
+	enum dsp_download_state dsp_state;
+	unsigned int dsp_stream_id;
+	unsigned int wait_scp;
+	unsigned int wait_scp_header;
+	unsigned int wait_num_data;
+	unsigned int scp_resp_header;
+	unsigned int scp_resp_data[4];
+	unsigned int scp_resp_count;
+};
+
+/*
+ * CA0132 codec access
+ */
+unsigned int codec_send_command(struct hda_codec *codec, hda_nid_t nid,
+		unsigned int verb, unsigned int parm, unsigned int *res)
+{
+	unsigned int response;
+	response = snd_hda_codec_read(codec, nid, 0, verb, parm);
+	*res = response;
+
+	return ((response == -1) ? -1 : 0);
+}
+
+static int codec_set_converter_format(struct hda_codec *codec, hda_nid_t nid,
+		unsigned short converter_format, unsigned int *res)
+{
+	return codec_send_command(codec, nid, VENDOR_CHIPIO_STREAM_FORMAT,
+				converter_format & 0xffff, res);
+}
+
+static int codec_set_converter_stream_channel(struct hda_codec *codec,
+				hda_nid_t nid, unsigned char stream,
+				unsigned char channel, unsigned int *res)
+{
+	unsigned char converter_stream_channel = 0;
+
+	converter_stream_channel = (stream << 4) | (channel & 0x0f);
+	return codec_send_command(codec, nid, AC_VERB_SET_CHANNEL_STREAMID,
+				converter_stream_channel, res);
+}
+
+/* Chip access helper function */
+static int chipio_send(struct hda_codec *codec,
+		       unsigned int reg,
+		       unsigned int data)
+{
+	unsigned int res;
+	int retry = 50;
+
+	/* send bits of data specified by reg */
+	do {
+		res = snd_hda_codec_read(codec, WIDGET_CHIP_CTRL, 0,
+					 reg, data);
+		if (res == VENDOR_STATUS_CHIPIO_OK)
+			return 0;
+	} while (--retry);
+	return -EIO;
+}
+
+/*
+ * Write chip address through the vendor widget -- NOT protected by the Mutex!
+ */
+static int chipio_write_address(struct hda_codec *codec,
+				unsigned int chip_addx)
+{
+	int res;
+
+	/* send low 16 bits of the address */
+	res = chipio_send(codec, VENDOR_CHIPIO_ADDRESS_LOW,
+			  chip_addx & 0xffff);
+
+	if (res != -EIO) {
+		/* send high 16 bits of the address */
+		res = chipio_send(codec, VENDOR_CHIPIO_ADDRESS_HIGH,
+				  chip_addx >> 16);
+	}
+
+	return res;
+}
+
+static int chipio_write_addx(struct hda_codec *codec, u32 chip_addx)
+{
+	struct ca0132_spec *spec = codec->spec;
+	int status;
+
+	if (spec->curr_chip_addx == chip_addx)
+		return 0;
+
+	/* send low 16 bits of the address */
+	status = chipio_send(codec, VENDOR_CHIPIO_ADDRESS_LOW,
+			  chip_addx & 0xffff);
+
+	if (status < 0)
+		return status;
+
+	/* send high 16 bits of the address */
+	status = chipio_send(codec, VENDOR_CHIPIO_ADDRESS_HIGH,
+			  chip_addx >> 16);
+
+	spec->curr_chip_addx = (status < 0) ? ~0UL : chip_addx;
+
+	return status;
+}
+
+/*
+ * Write data through the vendor widget -- NOT protected by the Mutex!
+ */
+
+static int chipio_write_data(struct hda_codec *codec, unsigned int data)
+{
+	int res;
+
+	/* send low 16 bits of the data */
+	res = chipio_send(codec, VENDOR_CHIPIO_DATA_LOW, data & 0xffff);
+
+	if (res != -EIO) {
+		/* send high 16 bits of the data */
+		res = chipio_send(codec, VENDOR_CHIPIO_DATA_HIGH,
+				  data >> 16);
+	}
+
+	return res;
+}
+
+static int chipio_write_data_multiple(struct hda_codec *codec,
+				      const u32 *data,
+				      unsigned int count)
+{
+	int status = 0;
+
+	if (data == NULL) {
+		snd_printdd(KERN_ERR "chipio_write_data null ptr");
+		return -EINVAL;
+	}
+
+	while ((count-- != 0) && (status == 0))
+		status = chipio_write_data(codec, *data++);
+
+	return status;
+}
+
+
+/*
+ * Read data through the vendor widget -- NOT protected by the Mutex!
+ */
+static int chipio_read_data(struct hda_codec *codec, unsigned int *data)
+{
+	int res;
+
+	/* post read */
+	res = chipio_send(codec, VENDOR_CHIPIO_HIC_POST_READ, 0);
+
+	if (res != -EIO) {
+		/* read status */
+		res = chipio_send(codec, VENDOR_CHIPIO_STATUS, 0);
+	}
+
+	if (res != -EIO) {
+		/* read data */
+		*data = snd_hda_codec_read(codec, WIDGET_CHIP_CTRL, 0,
+					   VENDOR_CHIPIO_HIC_READ_DATA,
+					   0);
+	}
+
+	return res;
+}
+
+/*
+ * Write given value to the given address through the chip I/O widget.
+ * protected by the Mutex
+ */
+static int chipio_write(struct hda_codec *codec,
+		unsigned int chip_addx, const unsigned int data)
+{
+	struct ca0132_spec *spec = codec->spec;
+	int err;
+
+	mutex_lock(&spec->chipio_mutex);
+
+	/* write the address, and if successful proceed to write data */
+	err = chipio_write_address(codec, chip_addx);
+	if (err < 0)
+		goto exit;
+
+	err = chipio_write_data(codec, data);
+	if (err < 0)
+		goto exit;
+
+exit:
+	mutex_unlock(&spec->chipio_mutex);
+	return err;
+}
+
+static int chipio_write_multiple(struct hda_codec *codec,
+				 u32 chip_addx,
+				 const u32 *data,
+				 unsigned int count)
+{
+	struct ca0132_spec *spec = codec->spec;
+	int status;
+
+	mutex_lock(&spec->chipio_mutex);
+	status = chipio_write_addx(codec, chip_addx);
+	if (status < 0)
+		goto error;
+
+	status = chipio_write_data_multiple(codec, data, count);
+error:
+	mutex_unlock(&spec->chipio_mutex);
+
+	return status;
+}
+
+/*
+ * Read the given address through the chip I/O widget
+ * protected by the Mutex
+ */
+static int chipio_read(struct hda_codec *codec,
+		unsigned int chip_addx, unsigned int *data)
+{
+	struct ca0132_spec *spec = codec->spec;
+	int err;
+
+	mutex_lock(&spec->chipio_mutex);
+
+	/* write the address, and if successful proceed to write data */
+	err = chipio_write_address(codec, chip_addx);
+	if (err < 0)
+		goto exit;
+
+	err = chipio_read_data(codec, data);
+	if (err < 0)
+		goto exit;
+
+exit:
+	mutex_unlock(&spec->chipio_mutex);
+	return err;
+}
+
+static void chipio_set_control_flag(struct hda_codec *codec,
+				    enum control_flag_id flag_id,
+				    bool flag_state)
+{
+	unsigned int val;
+	unsigned int flag_bit;
+
+	flag_bit = (flag_state ? 1 : 0);
+	val = (flag_bit << 7) | (flag_id);
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_FLAG_SET, val);
+}
+
+static void chipio_set_control_param(struct hda_codec *codec,
+		enum control_param_id param_id, int param_val)
+{
+	struct ca0132_spec *spec = codec->spec;
+	int val;
+
+	if ((param_id < 32) && (param_val < 8)) {
+		val = (param_val << 5) | (param_id);
+		snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+				    VENDOR_CHIPIO_PARAM_SET, val);
+	} else {
+		mutex_lock(&spec->chipio_mutex);
+		if (chipio_send(codec, VENDOR_CHIPIO_STATUS, 0) == 0) {
+			snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+					    VENDOR_CHIPIO_PARAM_EX_ID_SET,
+					    param_id);
+			snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+					    VENDOR_CHIPIO_PARAM_EX_VALUE_SET,
+					    param_val);
+		}
+		mutex_unlock(&spec->chipio_mutex);
+	}
+}
+
+static void chipio_set_conn_rate(struct hda_codec *codec,
+				int connid, enum ca0132_sample_rate rate)
+{
+	chipio_set_control_param(codec, CONTROL_PARAM_CONN_POINT_ID, connid);
+	chipio_set_control_param(codec, CONTROL_PARAM_CONN_POINT_SAMPLE_RATE,
+				 rate);
+}
+
+static void chipio_enable_clocks(struct hda_codec *codec)
+{
+	struct ca0132_spec *spec = codec->spec;
+
+	mutex_lock(&spec->chipio_mutex);
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_8051_ADDRESS_LOW, 0);
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_PLL_PMU_WRITE, 0xff);
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_8051_ADDRESS_LOW, 5);
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_PLL_PMU_WRITE, 0x0b);
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_8051_ADDRESS_LOW, 6);
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_PLL_PMU_WRITE, 0xff);
+	mutex_unlock(&spec->chipio_mutex);
+}
+
+/*
+ * CA0132 DSP IO stuffs
+ */
+static int dspio_send(struct hda_codec *codec, unsigned int reg,
+		      unsigned int data)
+{
+	unsigned int res;
+	int retry = 50;
+
+	/* send bits of data specified by reg to dsp */
+	do {
+		res = snd_hda_codec_read(codec, WIDGET_DSP_CTRL, 0, reg, data);
+		if ((res >= 0) && (res != VENDOR_STATUS_DSPIO_BUSY))
+			return res;
+	} while (--retry);
+
+	return -EIO;
+}
+
+static void dspio_write_wait(struct hda_codec *codec)
+{
+	int cur_val, prv_val;
+	int retry = 50;
+
+	cur_val = 0;
+	do {
+		prv_val = cur_val;
+		msleep(20);
+		dspio_send(codec, VENDOR_DSPIO_SCP_POST_COUNT_QUERY, 1);
+		dspio_send(codec, VENDOR_DSPIO_STATUS, 0);
+		cur_val = snd_hda_codec_read(codec, WIDGET_DSP_CTRL, 0,
+					   VENDOR_DSPIO_SCP_READ_COUNT, 0);
+	} while (cur_val && (cur_val == prv_val) && --retry);
+}
+
+static int dspio_write(struct hda_codec *codec, unsigned int scp_data)
+{
+	struct ca0132_spec *spec = codec->spec;
+	int status;
+
+	dspio_write_wait(codec);
+
+	mutex_lock(&spec->chipio_mutex);
+	status = dspio_send(codec, VENDOR_DSPIO_SCP_WRITE_DATA_LOW,
+			    scp_data & 0xffff);
+	if (status < 0)
+		goto error;
+
+	status = dspio_send(codec, VENDOR_DSPIO_SCP_WRITE_DATA_HIGH,
+				    scp_data >> 16);
+	if (status < 0)
+		goto error;
+
+	/* OK, now check if the write itself has executed*/
+	status = snd_hda_codec_read(codec, WIDGET_DSP_CTRL, 0,
+				    VENDOR_DSPIO_STATUS, 0);
+error:
+	mutex_unlock(&spec->chipio_mutex);
+
+	return (status == VENDOR_STATUS_DSPIO_SCP_COMMAND_QUEUE_FULL) ?
+			-EIO : 0;
+}
+
+static int dspio_write_multiple(struct hda_codec *codec,
+				unsigned int *buffer, unsigned int size)
+{
+	int status = 0;
+	unsigned int count;
+
+	if ((buffer == NULL))
+		return -EINVAL;
+
+	count = 0;
+	while (count < size) {
+		status = dspio_write(codec, *buffer++);
+		if (status != 0)
+			break;
+		count++;
+	}
+
+	return status;
+}
+
+static inline unsigned int
+make_scp_header(unsigned int target_id, unsigned int source_id,
+		unsigned int get_flag, unsigned int req,
+		unsigned int device_flag, unsigned int resp_flag,
+		unsigned int error_flag, unsigned int data_size)
+{
+	unsigned int header = 0;
+
+	header = (data_size & 0x1f) << 27;
+	header |= (error_flag & 0x01) << 26;
+	header |= (resp_flag & 0x01) << 25;
+	header |= (device_flag & 0x01) << 24;
+	header |= (req & 0x7f) << 17;
+	header |= (get_flag & 0x01) << 16;
+	header |= (source_id & 0xff) << 8;
+	header |= target_id & 0xff;
+
+	return header;
+}
+
+static inline void
+extract_scp_header(unsigned int header,
+		   unsigned int *target_id, unsigned int *source_id,
+		   unsigned int *get_flag, unsigned int *req,
+		   unsigned int *device_flag, unsigned int *resp_flag,
+		   unsigned int *error_flag, unsigned int *data_size)
+{
+	if (data_size)
+		*data_size = (header >> 27) & 0x1f;
+	if (error_flag)
+		*error_flag = (header >> 26) & 0x01;
+	if (resp_flag)
+		*resp_flag = (header >> 25) & 0x01;
+	if (device_flag)
+		*device_flag = (header >> 24) & 0x01;
+	if (req)
+		*req = (header >> 17) & 0x7f;
+	if (get_flag)
+		*get_flag = (header >> 16) & 0x01;
+	if (source_id)
+		*source_id = (header >> 8) & 0xff;
+	if (target_id)
+		*target_id = header & 0xff;
+}
+
+#define SCP_MAX_DATA_WORDS  (16)
+
+/* Structure to contain any SCP message */
+struct scp_msg {
+	unsigned int hdr;
+	unsigned int data[SCP_MAX_DATA_WORDS];
+};
+
+static int dspio_send_scp_message(struct hda_codec *codec,
+				  unsigned char *send_buf,
+				  unsigned int send_buf_size,
+				  unsigned char *return_buf,
+				  unsigned int return_buf_size,
+				  unsigned int *bytes_returned)
+{
+	struct ca0132_spec *spec = codec->spec;
+	int retry;
+	int status = -1;
+	unsigned int scp_send_size = 0;
+	unsigned int total_size;
+	bool waiting_for_resp = false;
+	unsigned int header;
+	struct scp_msg *ret_msg;
+	unsigned int resp_src_id, resp_target_id;
+	unsigned int data_size, src_id, target_id, get_flag, device_flag;
+
+	if (bytes_returned)
+		*bytes_returned = 0;
+
+	/* get scp header from buffer */
+	header = *((unsigned int *)send_buf);
+	extract_scp_header(header, &target_id, &src_id, &get_flag, NULL,
+			   &device_flag, NULL, NULL, &data_size);
+	scp_send_size = data_size + 1;
+	total_size = (scp_send_size * 4);
+
+	if (send_buf_size < total_size)
+		return -EINVAL;
+
+	if (get_flag || device_flag) {
+		if (!return_buf || return_buf_size < 4 || !bytes_returned)
+			return -EINVAL;
+
+		spec->wait_scp_header = *((unsigned int *)send_buf);
+
+		/* swap source id with target id */
+		resp_target_id = src_id;
+		resp_src_id = target_id;
+		spec->wait_scp_header &= 0xffff0000;
+		spec->wait_scp_header |= (resp_src_id << 8) | (resp_target_id);
+		spec->wait_num_data = return_buf_size/sizeof(unsigned int) - 1;
+		spec->wait_scp = 1;
+		waiting_for_resp = true;
+	}
+
+	status = dspio_write_multiple(codec, (unsigned int *)send_buf,
+				      scp_send_size);
+	if (status < 0) {
+		spec->wait_scp = 0;
+		return status;
+	}
+
+	if (waiting_for_resp) {
+		memset(return_buf, 0, return_buf_size);
+		retry = 50;
+		do {
+			msleep(20);
+		} while (spec->wait_scp && (--retry != 0));
+		waiting_for_resp = false;
+		if (retry != 0) {
+			ret_msg = (struct scp_msg *)return_buf;
+			memcpy(&ret_msg->hdr, &spec->scp_resp_header, 4);
+			memcpy(&ret_msg->data, spec->scp_resp_data,
+			       spec->wait_num_data);
+			*bytes_returned = (spec->scp_resp_count + 1) * 4;
+			status = 0;
+		} else {
+			status = -EIO;
+		}
+		spec->wait_scp = 0;
+	}
+
+	return status;
+}
+
+static int dspio_scp(struct hda_codec *codec,
+		int mod_id, int req, int dir, void *data, unsigned int len,
+		void *reply, unsigned int *reply_len)
+{
+	int status = 0;
+	struct scp_msg scp_send, scp_reply;
+	unsigned int ret_bytes, send_size, ret_size;
+	unsigned int send_get_flag, reply_resp_flag, reply_error_flag;
+	unsigned int reply_data_size;
+
+	memset(&scp_send, 0, sizeof(scp_send));
+	memset(&scp_reply, 0, sizeof(scp_reply));
+
+	if ((len != 0 && data == NULL) || (len > SCP_MAX_DATA_WORDS))
+		return -EINVAL;
+
+	if (dir == SCP_GET && reply == NULL) {
+		snd_printdd(KERN_ERR "dspio_scp get but has no buffer");
+		return -EINVAL;
+	}
+
+	if (reply != NULL && (reply_len == NULL || (*reply_len == 0))) {
+		snd_printdd(KERN_ERR "dspio_scp bad resp buf len parms");
+		return -EINVAL;
+	}
+
+	scp_send.hdr = make_scp_header(mod_id, 0x20, (dir == SCP_GET), req,
+				       0, 0, 0, len/sizeof(unsigned int));
+	if (data != NULL && len > 0) {
+		len = min((unsigned int)(sizeof(scp_send.data)), len);
+		memcpy(scp_send.data, data, len);
+	}
+
+	ret_bytes = 0;
+	send_size = sizeof(unsigned int) + len;
+	status = dspio_send_scp_message(codec, (unsigned char *)&scp_send,
+					send_size, (unsigned char *)&scp_reply,
+					sizeof(scp_reply), &ret_bytes);
+
+	if (status < 0) {
+		snd_printdd(KERN_ERR "dspio_scp: send scp msg failed");
+		return status;
+	}
+
+	/* extract send and reply headers members */
+	extract_scp_header(scp_send.hdr, NULL, NULL, &send_get_flag,
+			   NULL, NULL, NULL, NULL, NULL);
+	extract_scp_header(scp_reply.hdr, NULL, NULL, NULL, NULL, NULL,
+			   &reply_resp_flag, &reply_error_flag,
+			   &reply_data_size);
+
+	if (!send_get_flag)
+		return 0;
+
+	if (reply_resp_flag && !reply_error_flag) {
+		ret_size = (ret_bytes - sizeof(scp_reply.hdr))
+					/ sizeof(unsigned int);
+
+		if (*reply_len < ret_size*sizeof(unsigned int)) {
+			snd_printdd(KERN_ERR "reply too long for buf");
+			return -EINVAL;
+		} else if (ret_size != reply_data_size) {
+			snd_printdd(KERN_ERR "RetLen and HdrLen .NE.");
+			return -EINVAL;
+		} else {
+			*reply_len = ret_size*sizeof(unsigned int);
+			memcpy(reply, scp_reply.data, *reply_len);
+		}
+	} else {
+		snd_printdd(KERN_ERR "reply ill-formed or errflag set");
+		return -EIO;
+	}
+
+	return status;
+}
+
+static int dspio_alloc_dma_chan(struct hda_codec *codec, unsigned int *dma_chan)
+{
+	int status = 0;
+	unsigned int size = sizeof(dma_chan);
+
+	snd_printdd(KERN_INFO "     dspio_alloc_dma_chan() -- begin");
+	status = dspio_scp(codec, MASTERCONTROL, MASTERCONTROL_ALLOC_DMA_CHAN,
+			SCP_GET, NULL, 0, dma_chan, &size);
+
+	if (status < 0) {
+		snd_printdd(KERN_INFO "dspio_alloc_dma_chan: SCP Failed");
+		return status;
+	}
+
+	if ((*dma_chan + 1) == 0) {
+		snd_printdd(KERN_INFO "no free dma channels to allocate");
+		return -EBUSY;
+	}
+
+	snd_printdd("dspio_alloc_dma_chan: chan=%d\n", *dma_chan);
+	snd_printdd(KERN_INFO "     dspio_alloc_dma_chan() -- complete");
+
+	return status;
+}
+
+static int dspio_free_dma_chan(struct hda_codec *codec, unsigned int dma_chan)
+{
+	int status = 0;
+	unsigned int dummy = 0;
+
+	snd_printdd(KERN_INFO "     dspio_free_dma_chan() -- begin");
+	snd_printdd("dspio_free_dma_chan: chan=%d\n", dma_chan);
+
+	status = dspio_scp(codec, MASTERCONTROL, MASTERCONTROL_ALLOC_DMA_CHAN,
+			   SCP_SET, &dma_chan, sizeof(dma_chan), NULL, &dummy);
+
+	if (status < 0) {
+		snd_printdd(KERN_INFO "dspio_free_dma_chan: SCP Failed");
+		return status;
+	}
+
+	snd_printdd(KERN_INFO "     dspio_free_dma_chan() -- complete");
+
+	return status;
+}
+
+/*
+ * CA0132 DSP access stuffs
+ */
+static int dsp_set_run_state(struct hda_codec *codec)
+{
+	unsigned int dbg_ctrl_reg;
+	unsigned int halt_state;
+	int err;
+
+	err = chipio_read(codec, DSP_DBGCNTL_INST_OFFSET, &dbg_ctrl_reg);
+	if (err < 0)
+		return err;
+
+	halt_state = (dbg_ctrl_reg & DSP_DBGCNTL_STATE_MASK) >>
+		      DSP_DBGCNTL_STATE_LOBIT;
+
+	if (halt_state != 0) {
+		dbg_ctrl_reg &= ~((halt_state << DSP_DBGCNTL_SS_LOBIT) &
+				  DSP_DBGCNTL_SS_MASK);
+		err = chipio_write(codec, DSP_DBGCNTL_INST_OFFSET,
+				   dbg_ctrl_reg);
+		if (err < 0)
+			return err;
+
+		dbg_ctrl_reg |= (halt_state << DSP_DBGCNTL_EXEC_LOBIT) &
+				DSP_DBGCNTL_EXEC_MASK;
+		err = chipio_write(codec, DSP_DBGCNTL_INST_OFFSET,
+				   dbg_ctrl_reg);
+		if (err < 0)
+			return err;
+	}
+
+	return 0;
+}
+
+static int dsp_reset(struct hda_codec *codec)
+{
+	unsigned int res;
+	int retry = 20;
+
+	snd_printdd("dsp_reset\n");
+	do {
+		res = dspio_send(codec, VENDOR_DSPIO_DSP_INIT, 0);
+		retry--;
+	} while (res == -EIO && retry);
+
+	if (!retry) {
+		snd_printdd("dsp_reset timeout\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static unsigned int dsp_chip_to_dsp_addx(unsigned int chip_addx,
+					bool *code, bool *yram)
+{
+	*code = *yram = false;
+
+	if (UC_RANGE(chip_addx, 1)) {
+		*code = true;
+		return UC_OFF(chip_addx);
+	} else if (X_RANGE_ALL(chip_addx, 1)) {
+		return X_OFF(chip_addx);
+	} else if (Y_RANGE_ALL(chip_addx, 1)) {
+		*yram = true;
+		return Y_OFF(chip_addx);
+	}
+
+	return (unsigned int)INVALID_CHIP_ADDRESS;
+}
+
+static bool dsp_is_dma_active(struct hda_codec *codec, unsigned int dma_chan)
+{
+	unsigned int dma_chnlstart_reg;
+
+	chipio_read(codec, DSPDMAC_CHNLSTART_INST_OFFSET, &dma_chnlstart_reg);
+
+	return ((dma_chnlstart_reg & (1 <<
+			(DSPDMAC_CHNLSTART_EN_LOBIT + dma_chan))) != 0);
+}
+
+static int dsp_dma_setup_common(struct hda_codec *codec,
+				unsigned int chip_addx,
+				unsigned int dma_chan,
+				unsigned int port_map_mask,
+				bool ovly)
+{
+	int status = 0;
+	unsigned int chnl_prop;
+	unsigned int dsp_addx;
+	unsigned int active;
+	bool code, yram;
+
+	snd_printdd(KERN_INFO "-- dsp_dma_setup_common() -- Begin ---------");
+
+	if (dma_chan >= DSPDMAC_DMA_CFG_CHANNEL_COUNT) {
+		snd_printdd(KERN_ERR "dma chan num invalid");
+		return -EINVAL;
+	}
+
+	if (dsp_is_dma_active(codec, dma_chan)) {
+		snd_printdd(KERN_ERR "dma already active");
+		return -EBUSY;
+	}
+
+	dsp_addx = dsp_chip_to_dsp_addx(chip_addx, &code, &yram);
+
+	if (dsp_addx == INVALID_CHIP_ADDRESS) {
+		snd_printdd(KERN_ERR "invalid chip addr");
+		return -ENXIO;
+	}
+
+	chnl_prop = DSPDMAC_CHNLPROP_AC_MASK;
+	active = 0;
+
+	snd_printdd(KERN_INFO "   dsp_dma_setup_common()    start reg pgm");
+
+	if (ovly) {
+		status = chipio_read(codec, DSPDMAC_CHNLPROP_INST_OFFSET,
+				     &chnl_prop);
+
+		if (status < 0) {
+			snd_printdd(KERN_ERR "read CHNLPROP Reg fail");
+			return status;
+		}
+		snd_printdd(KERN_INFO "dsp_dma_setup_common() Read CHNLPROP");
+	}
+
+	if (!code)
+		chnl_prop &= ~(1 << (DSPDMAC_CHNLPROP_MSPCE_LOBIT + dma_chan));
+	else
+		chnl_prop |=  (1 << (DSPDMAC_CHNLPROP_MSPCE_LOBIT + dma_chan));
+
+	chnl_prop &= ~(1 << (DSPDMAC_CHNLPROP_DCON_LOBIT + dma_chan));
+
+	status = chipio_write(codec, DSPDMAC_CHNLPROP_INST_OFFSET, chnl_prop);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write CHNLPROP Reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "   dsp_dma_setup_common()    Write CHNLPROP");
+
+	if (ovly) {
+		status = chipio_read(codec, DSPDMAC_ACTIVE_INST_OFFSET,
+				     &active);
+
+		if (status < 0) {
+			snd_printdd(KERN_ERR "read ACTIVE Reg fail");
+			return status;
+		}
+		snd_printdd(KERN_INFO "dsp_dma_setup_common() Read ACTIVE");
+	}
+
+	active &= (~(1 << (DSPDMAC_ACTIVE_AAR_LOBIT + dma_chan))) &
+		DSPDMAC_ACTIVE_AAR_MASK;
+
+	status = chipio_write(codec, DSPDMAC_ACTIVE_INST_OFFSET, active);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write ACTIVE Reg fail");
+		return status;
+	}
+
+	snd_printdd(KERN_INFO "   dsp_dma_setup_common()    Write ACTIVE");
+
+	status = chipio_write(codec, DSPDMAC_AUDCHSEL_INST_OFFSET(dma_chan),
+			      port_map_mask);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write AUDCHSEL Reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "   dsp_dma_setup_common()    Write AUDCHSEL");
+
+	status = chipio_write(codec, DSPDMAC_IRQCNT_INST_OFFSET(dma_chan),
+			DSPDMAC_IRQCNT_BICNT_MASK | DSPDMAC_IRQCNT_CICNT_MASK);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write IRQCNT Reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "   dsp_dma_setup_common()    Write IRQCNT");
+
+	snd_printdd(
+		   "ChipA=0x%x,DspA=0x%x,dmaCh=%u, "
+		   "CHSEL=0x%x,CHPROP=0x%x,Active=0x%x\n",
+		   chip_addx, dsp_addx, dma_chan,
+		   port_map_mask, chnl_prop, active);
+
+	snd_printdd(KERN_INFO "-- dsp_dma_setup_common() -- Complete ------");
+
+	return 0;
+}
+
+static int dsp_dma_setup(struct hda_codec *codec,
+			unsigned int chip_addx,
+			unsigned int count,
+			unsigned int dma_chan)
+{
+	int status = 0;
+	bool code, yram;
+	unsigned int dsp_addx;
+	unsigned int addr_field;
+	unsigned int incr_field;
+	unsigned int base_cnt;
+	unsigned int cur_cnt;
+	unsigned int dma_cfg = 0;
+	unsigned int adr_ofs = 0;
+	unsigned int xfr_cnt = 0;
+	const unsigned int max_dma_count = 1 << (DSPDMAC_XFRCNT_BCNT_HIBIT -
+						DSPDMAC_XFRCNT_BCNT_LOBIT + 1);
+
+	snd_printdd(KERN_INFO "-- dsp_dma_setup() -- Begin ---------");
+
+	if (count > max_dma_count) {
+		snd_printdd(KERN_ERR "count too big");
+		return -EINVAL;
+	}
+
+	dsp_addx = dsp_chip_to_dsp_addx(chip_addx, &code, &yram);
+	if (dsp_addx == INVALID_CHIP_ADDRESS) {
+		snd_printdd(KERN_ERR "invalid chip addr");
+		return -ENXIO;
+	}
+
+	snd_printdd(KERN_INFO "   dsp_dma_setup()    start reg pgm");
+
+	addr_field = dsp_addx << DSPDMAC_DMACFG_DBADR_LOBIT;
+	incr_field   = 0;
+
+	if (!code) {
+		addr_field <<= 1;
+		if (yram)
+			addr_field |= (1 << DSPDMAC_DMACFG_DBADR_LOBIT);
+
+		incr_field  = (1 << DSPDMAC_DMACFG_AINCR_LOBIT);
+	}
+
+	dma_cfg = addr_field + incr_field;
+	status = chipio_write(codec, DSPDMAC_DMACFG_INST_OFFSET(dma_chan),
+				dma_cfg);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write DMACFG Reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "   dsp_dma_setup()    Write DMACFG");
+
+	adr_ofs = (count - 1) << (DSPDMAC_DSPADROFS_BOFS_LOBIT +
+							(code ? 0 : 1));
+
+	status = chipio_write(codec, DSPDMAC_DSPADROFS_INST_OFFSET(dma_chan),
+				adr_ofs);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write DSPADROFS Reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "   dsp_dma_setup()    Write DSPADROFS");
+
+	base_cnt = (count - 1) << DSPDMAC_XFRCNT_BCNT_LOBIT;
+
+	cur_cnt  = (count - 1) << DSPDMAC_XFRCNT_CCNT_LOBIT;
+
+	xfr_cnt = base_cnt | cur_cnt;
+
+	status = chipio_write(codec,
+				DSPDMAC_XFRCNT_INST_OFFSET(dma_chan), xfr_cnt);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write XFRCNT Reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "   dsp_dma_setup()    Write XFRCNT");
+
+	snd_printdd(
+		   "ChipA=0x%x, cnt=0x%x, DMACFG=0x%x, "
+		   "ADROFS=0x%x, XFRCNT=0x%x\n",
+		   chip_addx, count, dma_cfg, adr_ofs, xfr_cnt);
+
+	snd_printdd(KERN_INFO "-- dsp_dma_setup() -- Complete ---------");
+
+	return 0;
+}
+
+static int dsp_dma_start(struct hda_codec *codec,
+			 unsigned int dma_chan, bool ovly)
+{
+	unsigned int reg = 0;
+	int status = 0;
+
+	snd_printdd(KERN_INFO "-- dsp_dma_start() -- Begin ---------");
+
+	if (ovly) {
+		status = chipio_read(codec,
+				     DSPDMAC_CHNLSTART_INST_OFFSET, &reg);
+
+		if (status < 0) {
+			snd_printdd(KERN_ERR "read CHNLSTART reg fail");
+			return status;
+		}
+		snd_printdd(KERN_INFO "-- dsp_dma_start()    Read CHNLSTART");
+
+		reg &= ~(DSPDMAC_CHNLSTART_EN_MASK |
+				DSPDMAC_CHNLSTART_DIS_MASK);
+	}
+
+	status = chipio_write(codec, DSPDMAC_CHNLSTART_INST_OFFSET,
+			reg | (1 << (dma_chan + DSPDMAC_CHNLSTART_EN_LOBIT)));
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write CHNLSTART reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "-- dsp_dma_start() -- Complete ---------");
+
+	return status;
+}
+
+static int dsp_dma_stop(struct hda_codec *codec,
+			unsigned int dma_chan, bool ovly)
+{
+	unsigned int reg = 0;
+	int status = 0;
+
+	snd_printdd(KERN_INFO "-- dsp_dma_stop() -- Begin ---------");
+
+	if (ovly) {
+		status = chipio_read(codec,
+				     DSPDMAC_CHNLSTART_INST_OFFSET, &reg);
+
+		if (status < 0) {
+			snd_printdd(KERN_ERR "read CHNLSTART reg fail");
+			return status;
+		}
+		snd_printdd(KERN_INFO "-- dsp_dma_stop()    Read CHNLSTART");
+		reg &= ~(DSPDMAC_CHNLSTART_EN_MASK |
+				DSPDMAC_CHNLSTART_DIS_MASK);
+	}
+
+	status = chipio_write(codec, DSPDMAC_CHNLSTART_INST_OFFSET,
+			reg | (1 << (dma_chan + DSPDMAC_CHNLSTART_DIS_LOBIT)));
+	if (status < 0) {
+		snd_printdd(KERN_ERR "write CHNLSTART reg fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "-- dsp_dma_stop() -- Complete ---------");
+
+	return status;
+}
+
+static int dsp_allocate_router_ports(struct hda_codec *codec,
+				     unsigned int num_chans,
+				     unsigned int ports_per_channel,
+				     unsigned int start_device,
+				     unsigned int *port_map)
+{
+	int status = 0;
+	int res;
+	u8 val;
+
+	status = chipio_send(codec, VENDOR_CHIPIO_STATUS, 0);
+	if (status < 0)
+		return status;
+
+	val = start_device << 6;
+	val |= (ports_per_channel - 1) << 4;
+	val |= num_chans - 1;
+
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_PORT_ALLOC_CONFIG_SET,
+			    val);
+
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_PORT_ALLOC_SET,
+			    MEM_CONNID_DSP);
+
+	status = chipio_send(codec, VENDOR_CHIPIO_STATUS, 0);
+	if (status < 0)
+		return status;
+
+	res = snd_hda_codec_read(codec, WIDGET_CHIP_CTRL, 0,
+				VENDOR_CHIPIO_PORT_ALLOC_GET, 0);
+
+	*port_map = res;
+
+	return (res < 0) ? res : 0;
+}
+
+static int dsp_free_router_ports(struct hda_codec *codec)
+{
+	int status = 0;
+
+	status = chipio_send(codec, VENDOR_CHIPIO_STATUS, 0);
+	if (status < 0)
+		return status;
+
+	snd_hda_codec_write(codec, WIDGET_CHIP_CTRL, 0,
+			    VENDOR_CHIPIO_PORT_FREE_SET,
+			    MEM_CONNID_DSP);
+
+	status = chipio_send(codec, VENDOR_CHIPIO_STATUS, 0);
+
+	return status;
+}
+
+static int dsp_allocate_ports(struct hda_codec *codec,
+			unsigned int num_chans,
+			unsigned int rate_multi, unsigned int *port_map)
+{
+	int status;
+
+	snd_printdd(KERN_INFO "     dsp_allocate_ports() -- begin");
+
+	if ((rate_multi != 1) && (rate_multi != 2) && (rate_multi != 4)) {
+		snd_printdd(KERN_ERR "bad rate multiple");
+		return -EINVAL;
+	}
+
+	status = dsp_allocate_router_ports(codec, num_chans,
+					   rate_multi, 0, port_map);
+
+	snd_printdd(KERN_INFO "     dsp_allocate_ports() -- complete");
+
+	return status;
+}
+
+static int dsp_free_ports(struct hda_codec *codec)
+{
+	int status;
+
+	snd_printdd(KERN_INFO "     dsp_free_ports() -- begin");
+
+	status = dsp_free_router_ports(codec);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "free router ports fail");
+		return status;
+	}
+	snd_printdd(KERN_INFO "     dsp_free_ports() -- complete");
+
+	return status;
+}
+
+static int dsp_allocate_ports_format(struct hda_codec *codec,
+			const unsigned short fmt,
+			unsigned int *port_map)
+{
+	int status;
+	unsigned int num_chans;
+
+	unsigned int sample_rate_div = ((get_hdafmt_rate(fmt) >> 0) & 3) + 1;
+	unsigned int sample_rate_mul = ((get_hdafmt_rate(fmt) >> 3) & 3) + 1;
+	unsigned int rate_multi = sample_rate_mul / sample_rate_div;
+
+	if ((rate_multi != 1) && (rate_multi != 2) && (rate_multi != 4)) {
+		snd_printdd(KERN_ERR "bad rate multiple");
+		return -EINVAL;
+	}
+
+	num_chans = get_hdafmt_chs(fmt) + 1;
+
+	status = dsp_allocate_ports(codec, num_chans, rate_multi, port_map);
+
+	return status;
+}
+
+/*
+ *  HDA DMA engine stuffs for DSP code download
+ */
+struct dma_engine {
+	struct hda_codec *codec;
+	unsigned short m_converter_format;
+	struct snd_dma_buffer *dmab;
+	unsigned int buf_size;
+};
+
+
+enum dma_state {
+	DMA_STATE_STOP  = 0,
+	DMA_STATE_RUN   = 1
+};
+
+static int dma_convert_to_hda_format(
+		struct hda_stream_format *stream_format,
+		unsigned short *hda_format)
+{
+	unsigned int format_val;
+
+	format_val = snd_hda_calc_stream_format(
+				stream_format->sample_rate,
+				stream_format->number_channels,
+				SNDRV_PCM_FORMAT_S32_LE,
+				stream_format->container_size, 0);
+
+	if (hda_format)
+		*hda_format = (unsigned short)format_val;
+
+	return 0;
+}
+
+static int dma_reset(struct dma_engine *dma)
+{
+	struct hda_codec *codec = dma->codec;
+	struct ca0132_spec *spec = codec->spec;
+	int status;
+
+	if (dma->dmab)
+		snd_hda_codec_load_dsp_cleanup(codec, dma->dmab);
+
+	status = snd_hda_codec_load_dsp_prepare(codec,
+			dma->m_converter_format,
+			dma->buf_size,
+			dma->dmab);
+	if (status < 0)
+		return status;
+	spec->dsp_stream_id = status;
+	return 0;
+}
+
+static int dma_set_state(struct dma_engine *dma, enum dma_state state)
 {
-	char namestr[44];
-	int type = dir ? HDA_INPUT : HDA_OUTPUT;
-	struct snd_kcontrol_new knew =
-		HDA_CODEC_VOLUME_MONO(namestr, nid, chan, 0, type);
-	if ((query_amp_caps(codec, nid, type) & AC_AMPCAP_NUM_STEPS) == 0) {
-		snd_printdd("Skipping '%s %s Volume' (no amp on node 0x%x)\n", pfx, dirstr[dir], nid);
+	bool cmd;
+
+	snd_printdd("dma_set_state state=%d\n", state);
+
+	switch (state) {
+	case DMA_STATE_STOP:
+		cmd = false;
+		break;
+	case DMA_STATE_RUN:
+		cmd = true;
+		break;
+	default:
 		return 0;
 	}
-	sprintf(namestr, "%s %s Volume", pfx, dirstr[dir]);
-	return snd_hda_ctl_add(codec, nid, snd_ctl_new1(&knew, codec));
+
+	snd_hda_codec_load_dsp_trigger(dma->codec, cmd);
+	return 0;
 }
 
-#define add_out_switch(codec, nid, pfx) _add_switch(codec, nid, pfx, 3, 0)
-#define add_out_volume(codec, nid, pfx) _add_volume(codec, nid, pfx, 3, 0)
-#define add_in_switch(codec, nid, pfx) _add_switch(codec, nid, pfx, 3, 1)
-#define add_in_volume(codec, nid, pfx) _add_volume(codec, nid, pfx, 3, 1)
-#define add_mono_switch(codec, nid, pfx, chan) \
-	_add_switch(codec, nid, pfx, chan, 0)
-#define add_mono_volume(codec, nid, pfx, chan) \
-	_add_volume(codec, nid, pfx, chan, 0)
-#define add_in_mono_switch(codec, nid, pfx, chan) \
-	_add_switch(codec, nid, pfx, chan, 1)
-#define add_in_mono_volume(codec, nid, pfx, chan) \
-	_add_volume(codec, nid, pfx, chan, 1)
+static unsigned int dma_get_buffer_size(struct dma_engine *dma)
+{
+	return dma->dmab->bytes;
+}
 
+static unsigned char *dma_get_buffer_addr(struct dma_engine *dma)
+{
+	return dma->dmab->area;
+}
 
-/*
- * CA0132 specific
- */
+static int dma_xfer(struct dma_engine *dma,
+		const unsigned int *data,
+		unsigned int count)
+{
+	memcpy(dma->dmab->area, data, count);
+	return 0;
+}
 
-struct ca0132_spec {
-	struct auto_pin_cfg autocfg;
-	struct hda_multi_out multiout;
-	hda_nid_t out_pins[AUTO_CFG_MAX_OUTS];
-	hda_nid_t dacs[AUTO_CFG_MAX_OUTS];
-	hda_nid_t hp_dac;
-	hda_nid_t input_pins[AUTO_PIN_LAST];
-	hda_nid_t adcs[AUTO_PIN_LAST];
-	hda_nid_t dig_out;
-	hda_nid_t dig_in;
-	unsigned int num_inputs;
-	long curr_hp_switch;
-	long curr_hp_volume[2];
-	long curr_speaker_switch;
-	struct mutex chipio_mutex;
-	const char *input_labels[AUTO_PIN_LAST];
-	struct hda_pcm pcm_rec[2]; /* PCM information */
-};
+static void dma_get_converter_format(
+		struct dma_engine *dma,
+		unsigned short *format)
+{
+	if (format)
+		*format = dma->m_converter_format;
+}
 
-/* Chip access helper function */
-static int chipio_send(struct hda_codec *codec,
-		       unsigned int reg,
-		       unsigned int data)
+static unsigned int dma_get_stream_id(struct dma_engine *dma)
 {
-	unsigned int res;
-	int retry = 50;
+	struct ca0132_spec *spec = dma->codec->spec;
 
-	/* send bits of data specified by reg */
-	do {
-		res = snd_hda_codec_read(codec, WIDGET_CHIP_CTRL, 0,
-					 reg, data);
-		if (res == VENDOR_STATUS_CHIPIO_OK)
-			return 0;
-	} while (--retry);
-	return -EIO;
+	return spec->dsp_stream_id;
 }
 
-/*
- * Write chip address through the vendor widget -- NOT protected by the Mutex!
- */
-static int chipio_write_address(struct hda_codec *codec,
-				unsigned int chip_addx)
+struct dsp_image_seg {
+	u32 magic;
+	u32 chip_addr;
+	u32 count;
+	u32 data[0];
+};
+
+static const u32 g_magic_value = 0x4c46584d;
+static const u32 g_chip_addr_magic_value = 0xFFFFFF01;
+
+static bool is_valid(const struct dsp_image_seg *p)
 {
-	int res;
+	return p->magic == g_magic_value;
+}
 
-	/* send low 16 bits of the address */
-	res = chipio_send(codec, VENDOR_CHIPIO_ADDRESS_LOW,
-			  chip_addx & 0xffff);
+static bool is_hci_prog_list_seg(const struct dsp_image_seg *p)
+{
+	return g_chip_addr_magic_value == p->chip_addr;
+}
 
-	if (res != -EIO) {
-		/* send high 16 bits of the address */
-		res = chipio_send(codec, VENDOR_CHIPIO_ADDRESS_HIGH,
-				  chip_addx >> 16);
-	}
+static bool is_last(const struct dsp_image_seg *p)
+{
+	return p->count == 0;
+}
 
-	return res;
+static size_t dsp_sizeof(const struct dsp_image_seg *p)
+{
+	return sizeof(*p) + p->count*sizeof(u32);
+}
+
+static const struct dsp_image_seg *get_next_seg_ptr(
+				const struct dsp_image_seg *p)
+{
+	return (struct dsp_image_seg *)((unsigned char *)(p) + dsp_sizeof(p));
 }
 
 /*
- * Write data through the vendor widget -- NOT protected by the Mutex!
+ * CA0132 chip DSP transfer stuffs.  For DSP download.
  */
+#define INVALID_DMA_CHANNEL (~0UL)
 
-static int chipio_write_data(struct hda_codec *codec, unsigned int data)
+static int dspxfr_hci_write(struct hda_codec *codec,
+			const struct dsp_image_seg *fls)
 {
-	int res;
+	int status;
+	const u32 *data;
+	unsigned int count;
 
-	/* send low 16 bits of the data */
-	res = chipio_send(codec, VENDOR_CHIPIO_DATA_LOW, data & 0xffff);
-
-	if (res != -EIO) {
-		/* send high 16 bits of the data */
-		res = chipio_send(codec, VENDOR_CHIPIO_DATA_HIGH,
-				  data >> 16);
+	if (fls == NULL || fls->chip_addr != g_chip_addr_magic_value) {
+		snd_printdd(KERN_ERR "hci_write invalid params");
+		return -EINVAL;
 	}
 
-	return res;
+	count = fls->count;
+	data = (u32 *)(fls->data);
+	while (count >= 2) {
+		status = chipio_write(codec, data[0], data[1]);
+		if (status < 0) {
+			snd_printdd(KERN_ERR "hci_write chipio failed");
+			return status;
+		}
+		count -= 2;
+		data  += 2;
+	}
+	return 0;
 }
 
-/*
- * Read data through the vendor widget -- NOT protected by the Mutex!
- */
-static int chipio_read_data(struct hda_codec *codec, unsigned int *data)
+static int dspxfr_one_seg(struct hda_codec *codec,
+			const struct dsp_image_seg *fls,
+			unsigned int reloc,
+			struct dma_engine *dma_engine,
+			unsigned int dma_chan,
+			unsigned int port_map_mask,
+			bool ovly)
 {
-	int res;
+	int status;
+	bool comm_dma_setup_done = false;
+	const unsigned int *data;
+	unsigned int chip_addx;
+	unsigned int words_to_write;
+	unsigned int buffer_size_words;
+	unsigned char *buffer_addx;
+	unsigned short hda_format;
+	unsigned int sample_rate_div;
+	unsigned int sample_rate_mul;
+	unsigned int num_chans;
+	unsigned int hda_frame_size_words;
+	unsigned int remainder_words;
+	const u32 *data_remainder;
+	u32 chip_addx_remainder;
+	unsigned int run_size_words;
+	const struct dsp_image_seg *hci_write = NULL;
+	int retry;
+
+	if (fls == NULL)
+		return -EINVAL;
+	if (is_hci_prog_list_seg(fls)) {
+		hci_write = fls;
+		fls = get_next_seg_ptr(fls);
+	}
 
-	/* post read */
-	res = chipio_send(codec, VENDOR_CHIPIO_HIC_POST_READ, 0);
+	if (hci_write && (!fls || is_last(fls))) {
+		snd_printdd("hci_write\n");
+		return dspxfr_hci_write(codec, hci_write);
+	}
 
-	if (res != -EIO) {
-		/* read status */
-		res = chipio_send(codec, VENDOR_CHIPIO_STATUS, 0);
+	if (fls == NULL || dma_engine == NULL || port_map_mask == 0) {
+		snd_printdd("Invalid Params\n");
+		return -EINVAL;
 	}
 
-	if (res != -EIO) {
-		/* read data */
-		*data = snd_hda_codec_read(codec, WIDGET_CHIP_CTRL, 0,
-					   VENDOR_CHIPIO_HIC_READ_DATA,
-					   0);
+	data = fls->data;
+	chip_addx = fls->chip_addr,
+	words_to_write = fls->count;
+
+	if (!words_to_write)
+		return hci_write ? dspxfr_hci_write(codec, hci_write) : 0;
+	if (reloc)
+		chip_addx = (chip_addx & (0xFFFF0000 << 2)) + (reloc << 2);
+
+	if (!UC_RANGE(chip_addx, words_to_write) &&
+	    !X_RANGE_ALL(chip_addx, words_to_write) &&
+	    !Y_RANGE_ALL(chip_addx, words_to_write)) {
+		snd_printdd("Invalid chip_addx Params\n");
+		return -EINVAL;
 	}
 
-	return res;
+	buffer_size_words = (unsigned int)dma_get_buffer_size(dma_engine) /
+					sizeof(u32);
+
+	buffer_addx = dma_get_buffer_addr(dma_engine);
+
+	if (buffer_addx == NULL) {
+		snd_printdd(KERN_ERR "dma_engine buffer NULL\n");
+		return -EINVAL;
+	}
+
+	dma_get_converter_format(dma_engine, &hda_format);
+	sample_rate_div = ((get_hdafmt_rate(hda_format) >> 0) & 3) + 1;
+	sample_rate_mul = ((get_hdafmt_rate(hda_format) >> 3) & 3) + 1;
+	num_chans = get_hdafmt_chs(hda_format) + 1;
+
+	hda_frame_size_words = ((sample_rate_div == 0) ? 0 :
+			(num_chans * sample_rate_mul / sample_rate_div));
+
+	buffer_size_words = min(buffer_size_words,
+				(unsigned int)(UC_RANGE(chip_addx, 1) ?
+				65536 : 32768));
+	buffer_size_words -= buffer_size_words % hda_frame_size_words;
+	snd_printdd(
+		   "chpadr=0x%08x frmsz=%u nchan=%u "
+		   "rate_mul=%u div=%u bufsz=%u\n",
+		   chip_addx, hda_frame_size_words, num_chans,
+		   sample_rate_mul, sample_rate_div, buffer_size_words);
+
+	if ((buffer_addx == NULL) || (hda_frame_size_words == 0) ||
+	    (buffer_size_words < hda_frame_size_words)) {
+		snd_printdd(KERN_ERR "dspxfr_one_seg:failed\n");
+		return -EINVAL;
+	}
+
+	remainder_words = words_to_write % hda_frame_size_words;
+	data_remainder = data;
+	chip_addx_remainder = chip_addx;
+
+	data += remainder_words;
+	chip_addx += remainder_words*sizeof(u32);
+	words_to_write -= remainder_words;
+
+	while (words_to_write != 0) {
+		run_size_words = min(buffer_size_words, words_to_write);
+		snd_printdd("dspxfr (seg loop)cnt=%u rs=%u remainder=%u\n",
+			    words_to_write, run_size_words, remainder_words);
+		dma_xfer(dma_engine, data, run_size_words*sizeof(u32));
+		if (!comm_dma_setup_done) {
+			status = dsp_dma_stop(codec, dma_chan, ovly);
+			if (status < 0)
+				return -EIO;
+			status = dsp_dma_setup_common(codec, chip_addx,
+						dma_chan, port_map_mask, ovly);
+			if (status < 0)
+				return status;
+			comm_dma_setup_done = true;
+		}
+
+		status = dsp_dma_setup(codec, chip_addx,
+						run_size_words, dma_chan);
+		if (status < 0)
+			return status;
+		status = dsp_dma_start(codec, dma_chan, ovly);
+		if (status < 0)
+			return status;
+		if (!dsp_is_dma_active(codec, dma_chan)) {
+			snd_printdd(KERN_ERR "dspxfr:DMA did not start");
+			return -EIO;
+		}
+		status = dma_set_state(dma_engine, DMA_STATE_RUN);
+		if (status < 0)
+			return status;
+		if (remainder_words != 0) {
+			status = chipio_write_multiple(codec,
+						chip_addx_remainder,
+						data_remainder,
+						remainder_words);
+			remainder_words = 0;
+		}
+		if (hci_write) {
+			status = dspxfr_hci_write(codec, hci_write);
+			hci_write = NULL;
+		}
+		retry = 5000;
+		while (dsp_is_dma_active(codec, dma_chan)) {
+			if (--retry <= 0)
+				break;
+		}
+		snd_printdd(KERN_INFO "+++++ DMA complete");
+		dma_set_state(dma_engine, DMA_STATE_STOP);
+		dma_reset(dma_engine);
+
+		if (status < 0)
+			return status;
+
+		data += run_size_words;
+		chip_addx += run_size_words*sizeof(u32);
+		words_to_write -= run_size_words;
+	}
+
+	if (remainder_words != 0) {
+		status = chipio_write_multiple(codec, chip_addx_remainder,
+					data_remainder, remainder_words);
+	}
+
+	return status;
 }
 
-/*
- * Write given value to the given address through the chip I/O widget.
- * protected by the Mutex
- */
-static int chipio_write(struct hda_codec *codec,
-		unsigned int chip_addx, const unsigned int data)
+static int dspxfr_image(struct hda_codec *codec,
+			const struct dsp_image_seg *fls_data,
+			unsigned int reloc, struct hda_stream_format *format,
+			bool ovly)
 {
 	struct ca0132_spec *spec = codec->spec;
-	int err;
+	int status;
+	unsigned short hda_format = 0;
+	unsigned int response;
+	unsigned char stream_id = 0;
+	struct dma_engine *dma_engine;
+	unsigned int dma_chan;
+	unsigned int port_map_mask;
+
+	if (fls_data == NULL)
+		return -EINVAL;
+
+	dma_engine = kzalloc(sizeof(*dma_engine), GFP_KERNEL);
+	if (!dma_engine) {
+		status = -ENOMEM;
+		goto exit;
+	}
+	memset((void *)dma_engine, 0, sizeof(*dma_engine));
 
-	mutex_lock(&spec->chipio_mutex);
+	dma_engine->dmab = kzalloc(sizeof(*dma_engine->dmab), GFP_KERNEL);
+	if (!dma_engine->dmab) {
+		status = -ENOMEM;
+		goto exit;
+	}
 
-	/* write the address, and if successful proceed to write data */
-	err = chipio_write_address(codec, chip_addx);
-	if (err < 0)
+	dma_engine->codec = codec;
+	dma_convert_to_hda_format(format, &hda_format);
+	dma_engine->m_converter_format = hda_format;
+	dma_engine->buf_size = (ovly ? DSP_DMA_WRITE_BUFLEN_OVLY :
+			DSP_DMA_WRITE_BUFLEN_INIT) * 2;
+
+	dma_chan = 0;
+
+	status = codec_set_converter_format(codec, WIDGET_CHIP_CTRL,
+					hda_format, &response);
+
+	if (status < 0) {
+		snd_printdd(KERN_ERR "set converter format fail");
+		goto exit;
+	}
+
+	status = snd_hda_codec_load_dsp_prepare(codec,
+				dma_engine->m_converter_format,
+				dma_engine->buf_size,
+				dma_engine->dmab);
+	if (status < 0)
 		goto exit;
+	spec->dsp_stream_id = status;
+
+	if (ovly) {
+		status = dspio_alloc_dma_chan(codec, &dma_chan);
+		if (status < 0) {
+			snd_printdd(KERN_ERR "alloc dmachan fail");
+			dma_chan = (unsigned int)INVALID_DMA_CHANNEL;
+			goto exit;
+		}
+	}
 
-	err = chipio_write_data(codec, data);
-	if (err < 0)
+	port_map_mask = 0;
+	status = dsp_allocate_ports_format(codec, hda_format,
+					&port_map_mask);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "alloc ports fail");
+		goto exit;
+	}
+
+	stream_id = dma_get_stream_id(dma_engine);
+	status = codec_set_converter_stream_channel(codec,
+			WIDGET_CHIP_CTRL, stream_id, 0, &response);
+	if (status < 0) {
+		snd_printdd(KERN_ERR "set stream chan fail");
+		goto exit;
+	}
+
+	while ((fls_data != NULL) && !is_last(fls_data)) {
+		if (!is_valid(fls_data)) {
+			snd_printdd(KERN_ERR "FLS check fail");
+			status = -EINVAL;
+			goto exit;
+		}
+		status = dspxfr_one_seg(codec, fls_data, reloc,
+					dma_engine, dma_chan,
+					port_map_mask, ovly);
+		if (status < 0)
+			break;
+
+		if (is_hci_prog_list_seg(fls_data))
+			fls_data = get_next_seg_ptr(fls_data);
+
+		if ((fls_data != NULL) && !is_last(fls_data))
+			fls_data = get_next_seg_ptr(fls_data);
+	}
+
+	if (port_map_mask != 0)
+		status = dsp_free_ports(codec);
+
+	if (status < 0)
 		goto exit;
 
+	status = codec_set_converter_stream_channel(codec,
+				WIDGET_CHIP_CTRL, 0, 0, &response);
+
 exit:
-	mutex_unlock(&spec->chipio_mutex);
-	return err;
+	if (ovly && (dma_chan != INVALID_DMA_CHANNEL))
+		dspio_free_dma_chan(codec, dma_chan);
+
+	if (dma_engine->dmab)
+		snd_hda_codec_load_dsp_cleanup(codec, dma_engine->dmab);
+	kfree(dma_engine->dmab);
+	kfree(dma_engine);
+
+	return status;
 }
 
 /*
- * Read the given address through the chip I/O widget
- * protected by the Mutex
+ * CA0132 DSP download stuffs.
  */
-static int chipio_read(struct hda_codec *codec,
-		unsigned int chip_addx, unsigned int *data)
+static void dspload_post_setup(struct hda_codec *codec)
 {
-	struct ca0132_spec *spec = codec->spec;
-	int err;
+	snd_printdd(KERN_INFO "---- dspload_post_setup ------");
 
-	mutex_lock(&spec->chipio_mutex);
+	/*set DSP speaker to 2.0 configuration*/
+	chipio_write(codec, XRAM_XRAM_INST_OFFSET(0x18), 0x08080080);
+	chipio_write(codec, XRAM_XRAM_INST_OFFSET(0x19), 0x3f800000);
 
-	/* write the address, and if successful proceed to write data */
-	err = chipio_write_address(codec, chip_addx);
-	if (err < 0)
-		goto exit;
+	/*update write pointer*/
+	chipio_write(codec, XRAM_XRAM_INST_OFFSET(0x29), 0x00000002);
+}
 
-	err = chipio_read_data(codec, data);
-	if (err < 0)
-		goto exit;
+static int dspload_image(struct hda_codec *codec,
+			const struct dsp_image_seg *fls,
+			bool ovly,
+			unsigned int reloc,
+			bool autostart,
+			int router_chans)
+{
+	int status = 0;
+	struct hda_stream_format stream_format;
+
+	snd_printdd(KERN_INFO "---- dspload_image begin ------");
+	if (router_chans == 0) {
+		if (!ovly)
+			router_chans = DMA_TRANSFER_FRAME_SIZE_NWORDS;
+		else
+			router_chans = DMA_OVERLAY_FRAME_SIZE_NWORDS;
+	}
 
-exit:
-	mutex_unlock(&spec->chipio_mutex);
-	return err;
+	stream_format.sample_rate = 48000;
+	stream_format.number_channels = (unsigned short)router_chans;
+
+	while (stream_format.number_channels > 16) {
+		stream_format.sample_rate *= 2;
+		stream_format.number_channels /= 2;
+	}
+
+	stream_format.container_size = 32;
+	stream_format.valid_bits_per_sample = 32;
+
+	do {
+		snd_printdd(KERN_INFO "Ready to program DMA");
+		if (!ovly)
+			status = dsp_reset(codec);
+
+		if (status < 0)
+			break;
+
+		snd_printdd(KERN_INFO "dsp_reset() complete");
+		status = dspxfr_image(codec, fls, reloc, &stream_format, ovly);
+
+		if (status < 0)
+			break;
+
+		snd_printdd(KERN_INFO "dspxfr_image() complete");
+		if (autostart && !ovly) {
+			dspload_post_setup(codec);
+			status = dsp_set_run_state(codec);
+		}
+
+		snd_printdd(KERN_INFO "LOAD FINISHED");
+	} while (0);
+
+	return status;
+}
+
+static bool dspload_is_loaded(struct hda_codec *codec)
+{
+	unsigned int data = 0;
+	int status = 0;
+
+	status = chipio_read(codec, 0x40004, &data);
+	if ((status < 0) || (data != 1))
+		return false;
+
+	return true;
+}
+
+static bool dspload_wait_loaded(struct hda_codec *codec)
+{
+	int retry = 100;
+
+	do {
+		msleep(20);
+		if (dspload_is_loaded(codec)) {
+			pr_info("ca0132 DOWNLOAD OK :-) DSP IS RUNNING.\n");
+			return true;
+		}
+	} while (--retry);
+
+	pr_err("ca0132 DOWNLOAD FAILED!!! DSP IS NOT RUNNING.\n");
+	return false;
 }
 
 /*
@@ -979,12 +2524,68 @@ static void ca0132_exit_chip(struct hda_codec *codec)
 	/* put any chip cleanup stuffs here. */
 }
 
+static void ca0132_set_dsp_msr(struct hda_codec *codec, bool is96k)
+{
+	chipio_set_control_flag(codec, CONTROL_FLAG_DSP_96KHZ, is96k);
+	chipio_set_control_flag(codec, CONTROL_FLAG_DAC_96KHZ, is96k);
+	chipio_set_control_flag(codec, CONTROL_FLAG_SRC_RATE_96KHZ, is96k);
+	chipio_set_control_flag(codec, CONTROL_FLAG_SRC_CLOCK_196MHZ, is96k);
+	chipio_set_control_flag(codec, CONTROL_FLAG_ADC_B_96KHZ, is96k);
+	chipio_set_control_flag(codec, CONTROL_FLAG_ADC_C_96KHZ, is96k);
+
+	chipio_set_conn_rate(codec, MEM_CONNID_MICIN1, SR_16_000);
+	chipio_set_conn_rate(codec, MEM_CONNID_MICOUT1, SR_16_000);
+	chipio_set_conn_rate(codec, MEM_CONNID_WUH, SR_48_000);
+}
+
+static bool ca0132_download_dsp_images(struct hda_codec *codec)
+{
+	bool dsp_loaded = false;
+	const struct dsp_image_seg *dsp_os_image;
+	const struct firmware *fw_entry;
+
+	if (request_firmware(&fw_entry, EFX_FILE, codec->bus->card->dev) != 0)
+		return false;
+
+	dsp_os_image = (struct dsp_image_seg *)(fw_entry->data);
+	dspload_image(codec, dsp_os_image, 0, 0, true, 0);
+	dsp_loaded = dspload_wait_loaded(codec);
+
+	release_firmware(fw_entry);
+
+
+	return dsp_loaded;
+}
+
+static void ca0132_download_dsp(struct hda_codec *codec)
+{
+	struct ca0132_spec *spec = codec->spec;
+
+	spec->dsp_state = DSP_DOWNLOAD_INIT;
+
+	if (spec->dsp_state == DSP_DOWNLOAD_INIT) {
+		chipio_enable_clocks(codec);
+		spec->dsp_state = DSP_DOWNLOADING;
+		if (!ca0132_download_dsp_images(codec))
+			spec->dsp_state = DSP_DOWNLOAD_FAILED;
+		else
+			spec->dsp_state = DSP_DOWNLOADED;
+	}
+
+	if (spec->dsp_state == DSP_DOWNLOADED)
+		ca0132_set_dsp_msr(codec, true);
+}
+
 static int ca0132_init(struct hda_codec *codec)
 {
 	struct ca0132_spec *spec = codec->spec;
 	struct auto_pin_cfg *cfg = &spec->autocfg;
 	int i;
 
+#ifdef CONFIG_SND_HDA_DSP_LOADER
+	ca0132_download_dsp(codec);
+#endif
+
 	for (i = 0; i < spec->multiout.num_dacs; i++) {
 		init_output(codec, spec->out_pins[i],
 			    spec->multiout.dac_nids[i]);