Skip to contentSkip to conversation

ASoC: rt5677: Support DSP function for VAD application

PATCH Applied Oder Chiou 3 participants 3 messages Oct 6, 2014 – Oct 15, 2014

Subsystems (3)
Sound ASoC Realtek audio codecs
lore
Share & tools
lore

Oder Chiou

author maintainer
#0 lore raw

PATCH v2 1/1 ASoC: rt5677: Support DSP function for VAD application

The ALC5677 has a programmable DSP, and there is a SPI that is dadicated for DSP firmware loading. Therefore, the patch includes a SPI driver for writing the DSP firmware. The VAD(Voice Activaty Detection) has be implemented and use the DSP to recognize the key phase.

Signed-off-by Oder Chiou <[email protected]>
---
 sound/soc/codecs/Makefile     |   2 +-
 sound/soc/codecs/rt5677-spi.c | 129 +++++++++++++++++++
 sound/soc/codecs/rt5677-spi.h |  21 +++
 sound/soc/codecs/rt5677.c     | 288 +++++++++++++++++++++++++++++++++++++++++-
 sound/soc/codecs/rt5677.h     |   6 +
 5 files changed, 441 insertions(+), 5 deletions(-)
 create mode 100644 sound/soc/codecs/rt5677-spi.c
 create mode 100644 sound/soc/codecs/rt5677-spi.h
sound/soc/codecs/Makefile +1 -1
index 5dce451..4435f9f 100644
@@ -79,7 +79,7 @@ snd-soc-rt5640-objs := rt5640.o
 snd-soc-rt5645-objs := rt5645.o
 snd-soc-rt5651-objs := rt5651.o
 snd-soc-rt5670-objs := rt5670.o
-snd-soc-rt5677-objs := rt5677.o
+snd-soc-rt5677-objs := rt5677.o rt5677-spi.o
 snd-soc-sgtl5000-objs := sgtl5000.o
 snd-soc-alc5623-objs := alc5623.o
 snd-soc-alc5632-objs := alc5632.o
sound/soc/codecs/rt5677-spi.c new file +129 -0

129 lines not on this page · raw message

sound/soc/codecs/rt5677-spi.h new file +21 -0
new file mode 100644
index 0000000..7528bfd
@@ -0,0 +1,21 @@
+/*
+ * rt5677-spi.h  --  RT5677 ALSA SoC audio codec driver
+ *
+ * Copyright 2013 Realtek Semiconductor Corp.
+ * Author: Oder Chiou <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __RT5671_SPI_H__
+#define __RT5671_SPI_H__
+
+#define RT5677_SPI_BUF_LEN 240
+#define RT5677_SPI_CMD_BURST_WRITE 0x05
+
+int rt5677_spi_write(u8 *txbuf, size_t len);
+int rt5677_spi_burst_write(u32 addr, const struct firmware *fw);
+
+#endif /* __RT5677_SPI_H__ */
sound/soc/codecs/rt5677.c +284 -4

348 lines not on this page · raw message

sound/soc/codecs/rt5677.h +6 -0
index d4eb6d5..a7ea27a 100644
@@ -1506,6 +1506,9 @@
 #define RT5677_GPIO5_FUNC_GPIO			(0x0 << 9)
 #define RT5677_GPIO5_FUNC_DMIC			(0x1 << 9)
 
+#define RT5677_FIRMWARE1	"rt5677_dsp_fw1.bin"
+#define RT5677_FIRMWARE2	"rt5677_dsp_fw2.bin"
+
 /* System Clock Source */
 enum {
 	RT5677_SCLK_S_MCLK,

@@ -1545,6 +1548,8 @@ struct rt5677_priv {
 	struct snd_soc_codec *codec;
 	struct rt5677_platform_data pdata;
 	struct regmap *regmap;
+	const struct firmware *fw1, *fw2;
+	struct mutex dsp_cmd_lock;
 
 	int sysclk;
 	int sysclk_src;

@@ -1558,6 +1563,7 @@ struct rt5677_priv {
 #ifdef CONFIG_GPIOLIB
 	struct gpio_chip gpio_chip;
 #endif
+	bool dsp_vad_en;
 };
 
 #endif /* __RT5677_H__ */
Signature · 1 line
1.8.1.1.439.g50a6b54
maintainer Applied
#1 lore raw
Quoted 4 lines · Oder Chiou

The ALC5677 has a programmable DSP, and there is a SPI that is dadicated for DSP firmware loading. Therefore, the patch includes a SPI driver for writing the DSP firmware. The VAD(Voice Activaty Detection) has be implemented and use the DSP to recognize the key phase.

Applied, thanks. One small thing which I fixed up:

sound/soc/codecs/rt5677-spi.c line 125

Quoted 9 lines
sound/soc/codecs/rt5677-spi.c new file +9 -0
+static struct spi_driver rt5677_spi_driver = {
+	.driver = {
+			.name = "rt5677_spidev",
+			.bus = &spi_bus_type,
+			.owner = THIS_MODULE,
+	},
+	.probe = rt5677_spi_probe,
+};
+module_spi_driver(rt5677_spi_driver);

It is better form to remove the _spidev from the name. I fixed this, the indentation and the redundant .bus initialization.

We really should do something better to tie the two devices together though - but I'm not quite seeing the elegant thing. I'm wondering if the I2C driver should defer probe until the SPI driver is there.

#2 lore raw
In reply to Oder Chiou

Hi

Quoted 6 lines · Oder Chiou

The ALC5677 has a programmable DSP, and there is a SPI that is dadicated for DSP firmware loading. Therefore, the patch includes a SPI driver for writing the DSP firmware. The VAD(Voice Activaty Detection) has be implemented and use the DSP to recognize the key phase.

Signed-off-by Oder Chiou <[email protected]>
Quoted 9 lines
---
 sound/soc/codecs/Makefile     |   2 +-
 sound/soc/codecs/rt5677-spi.c | 129 +++++++++++++++++++
 sound/soc/codecs/rt5677-spi.h |  21 +++
 sound/soc/codecs/rt5677.c     | 288 +++++++++++++++++++++++++++++++++++++++++-
 sound/soc/codecs/rt5677.h     |   6 +
 5 files changed, 441 insertions(+), 5 deletions(-)
 create mode 100644 sound/soc/codecs/rt5677-spi.c
 create mode 100644 sound/soc/codecs/rt5677-spi.h

I am trying to compile rt5677 topic branch from Mark's repo and I get following linking problem

LD [M] sound/soc/codecs/snd-soc-rt5670.o
LD [M] sound/soc/codecs/snd-soc-rt5677.o
sound/soc/codecs/rt5677-spi.o: In function `init_module':
rt5677-spi.c:(.init.text+0x0): multiple definition of init_module' sound/soc/codecs/rt5677.o:rt5677.c:(.init.text+0x0): first defined here sound/soc/codecs/rt5677-spi.o: In function cleanup_module':
rt5677-spi.c:(.exit.text+0x0): multiple definition of `cleanup_module' sound/soc/codecs/rt5677.o:rt5677.c:(.exit.text+0x0): first defined here make[3]: *** [sound/soc/codecs/snd-soc-rt5677.o] Error 1 make[2]: *** [sound/soc/codecs] Error 2
make[1]: *** [sound/soc] Error 2
make[1]: *** Waiting for unfinished jobs....

I use CONFIG_SND_SOC_ALL_CODEC.