Driver fixes and updates to kernel 3.18.16 and 4.0.5

Hi Hias,

This is a little bit off topic, but I am now trying to access the registers on the WM5102 so that I can use the cirrus logic card as a real time audio processing box (takes input from line in, processes audio in C, passes output through line out). I am trying to do it using i2c with the help of wiring pi. Here is the code I have as well as the output when running it. Is there something additional that I have to do to get i2c working with the cirrus card that I am missing?

#include <stdio.h>

#include <wiringPiI2C.h>

int main()

{

   int fd = wiringPiI2CSetup(26);

   int test1 = wiringPiI2CReadReg8(fd,768);

   int test2 = wiringPiI2CWriteReg8(fd,768,16);

   int test3 = wiringPiI2CReadReg8(fd,768);

   printf("fd = %i, test1 = %i, test2 = %i, test3 = %i\n", fd, test1, test2, test3);

   return 0;

}

it prints: "fd = 3, test1 = 0, test2 = 0 test3 = 0"

Thanks for any help,

Ryan

Hi Ryan!

This is a little bit off topic, but I am now trying to access the registers on the WM5102 so that I can use the cirrus logic card as a real time audio processing box (takes input from line in, processes audio in C, passes output through line out). I am trying to do it using i2c with the help of wiring pi. Here is the code I have as well as the output when running it. Is there something additional that I have to do to get i2c working with the cirrus card that I am missing?

So far I didn't have much luch talking to the wm5102 via I2C. I tried it with the standard /dev/i2c interface (I2C_RDWR ioctl, 32 bits register address and 16 bit data).

Changing the devicetree overlay and hooking up the wm5102 to i2c address 0x1a didn't work either, it detected the wm5102 via I2C but couldn't apply the hardware patch. I2C reads from the wm5102 always seem to return 0x5102. Not sure what's going on here, maybe I missed something.

Even if I2C access in parallel to SPI access would work it would mess things up with the kernel driver. It uses the regmap/regcache layer and if that's out of sync with the real register values (changed behind it's back via I2C) that could lead to interesting results...

The question is: what do you want to do and do you really have to access the chip-registers directly? Most of the chip's internal registers are exposed via alsa mixer controls and for standard realtime processing you should be fine using the userspace alsa interface.

so long,

Hias

Hi Hias,

Okay, I don't have much experience using alsa. I am working on a reverberation effect to be used with guitar. I need to take the guitar signal from line input one sample at a time, put it through some C code that I have already finished, and then put the new sample back out of line out. Will that be possible using alsa? Also I'll probably want to change the sampling rate and bps. If it is possible, will I be using alsa outside of the C program I have written?

Thanks again,

Ryan

Hi Ryan!

Okay, I don't have much experience using alsa. I am working on a reverberation effect to be used with guitar. I need to take the guitar signal from line input one sample at a time, put it through some C code that I have already finished, and then put the new sample back out of line out. Will that be possible using alsa? Also I'll probably want to change the sampling rate and bps. If it is possible, will I be using alsa outside of the C program I have written?

That depends on your program

Using alsa (as it's the native linux audio layer) is just one of the possiblities. Some people prefer using jack (which then interfaces to alsa) and there are even very high level applications like PD (pure data) that allow you to create realtime effects.

Just do some googling and choose the solution that meets your requirements best.

so long,

Hias

Hi Matthias,

So far I didn't have much luch talking to the wm5102 via I2C. I tried it with the standard /dev/i2c interface (I2C_RDWR ioctl, 32 bits register address and 16 bit data).

 

Changing the devicetree overlay and hooking up the wm5102 to i2c address 0x1a didn't work either, it detected the wm5102 via I2C but couldn't apply the hardware patch. I2C reads from the wm5102 always seem to return 0x5102. Not sure what's going on here, maybe I missed something.

Not sure that it is related to this, but do you know about the incomplete implementation of the I2C protocol on RPI? Specifically, the "repeated starts read" was not supported by the i2c driver about one year ago. I found this quite painfully, after dispairing in not being able to communicate with my MPR121 chip (but working with no problems on Arduino). After getting the same bad answer from a register I found this article: https://jjackowski.wordpress.com/2013/07/13/i2c-repeated-starts-implemented-on-the-raspberry-pi/

They might be fixed this in the meantime (in the master), I don't know: I decided to refrain from doing any I2C (or even SPI) task with the Raspberry (it was a big shock for me and I think it's a shame, mainly because it is not mentioned anywhere.) But I guess there are a few patched drivers around, if one searches carefully (also mentioned in the article). If you know anything, let me know.

Cheers,

Hi Zsolt!

Not sure that it is related to this, but do you know about the incomplete implementation of the I2C protocol on RPI? Specifically, the "repeated starts read" was not supported by the i2c driver about one year ago.

Excellent info, this was exactly the part I was missing!

The wm5102 register read operation needs a repeated start, and fortunately it's possible to make it working by enabling combined transactions in the bcm2708 I2C driver

options i2c-bcm2708 combined=1

If anyone's interested here's the (now working) C code I used to read/write the wm5102 registers.

/*
* cirrus-reg - access wm5102 registers via i2c
*
* Copyright (C) 2015 Matthias Reichl <hias@horus.com>
*
* 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.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
* General Public License for more details.
*/

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/ioctl.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <linux/i2c.h>
#include <linux/i2c-dev.h>

static const char* i2cdev = "/dev/i2c-1";
static const int i2caddr = 0x1a;

int main(int argc, char** argv)
{
    int f;
    int ret = 0;
    int reg;
    int value;
    int wrmode = 1;
    // buffer for 4-byte register address plus 2 byte register data
    unsigned char buf[6];
    // I2C messages:
    // reading register needs 2 messages with repeated start,
    // writing register sends 1 message with data+value
    struct i2c_msg msgs[2];
    struct i2c_rdwr_ioctl_data rdwr_data;

    if (argc < 2 || argc > 3) {
        goto usage;
    }

    // first 4 bytes of buffer are register address in big endian format
    reg = strtol(argv[1], NULL, 0);
    buf[0] = (reg >> 24) & 0xff;
    buf[1] = (reg >> 16) & 0xff;
    buf[2] = (reg >> 8) & 0xff;
    buf[3] = reg & 0xff;

    // default setup: 1 i2c message for write
    rdwr_data.msgs = &msgs[0];
    rdwr_data.nmsgs = 1;

    // first message: write 6 bytes
    msgs[0].addr = i2caddr;
    msgs[0].flags = 0;
    msgs[0].buf = &buf[0];
    msgs[0].len = 6;

    if (argc == 3) {
        // write register case, store value in buffer
        value = strtol(argv[2], NULL, 0);
        buf[4] = (value >> 8) & 0xff;
        buf[5] = value & 0xff;
    } else {
        wrmode = 0;
        // only transfer 4 bytes in first message
        msgs[0].len = 4;
        // second message, read 2 bytes into buf+4
        msgs[1].addr = i2caddr;
        msgs[1].flags = I2C_M_RD;
        msgs[1].buf = &buf[4];
        msgs[1].len = 2;
        rdwr_data.nmsgs = 2;
    }

    f = open(i2cdev, O_RDWR);
    if (f < 0) {
        printf("cannot open i2c device %s\n", i2cdev);
        return 1;
    }

    if ((ret = ioctl(f, I2C_RDWR, &rdwr_data)) < 0) {
        perror("I2C_RDWR");
        return ret;
    }

    if (wrmode) {
        printf("%06x <= %04x\n", reg, value);
    } else {
        value = (buf[4] << 8) | buf[5];
        printf("%06x: %04x\n", reg, value);
    }
    close(f);
    return 0;

usage:
    printf("RPi Cirrus sound card direct register access via i2c\n");
    printf("(c) 2015 Matthias Reichl <hias@horus.com>\n\n");
    printf("usage: cirrus-reg REGISTER        to read a register\n");
    printf("   or: cirrus-reg REGISTER VALUE  to write to a register\n\n");
    printf("eg:\n");
    printf("cirrus-reg 0xc02\n");
    printf("cirrus-reg 0xc02 0xa101\n\n");
    printf("note: combined transactions need to be enabled in i2c-bcm2708!\n");

    return 1;
}

so long,

Hias

Hi Matthias,

This is awesome (and that you figured it out)! Just a lame question: where should I put the driver options (maybe in /etc/modules)? Sorry...

Hi Zsolt!

This is awesome (and that you figured it out)! Just a lame question: where should I put the driver options (maybe in /etc/modules)? Sorry...

Put them in a *.conf file inside /etc/modprobe.d - I choose /etc/modprobe.d/i2c-combined.conf for that setting.

The combined transfer functionality in the kernel is similar to the code in the link you posted, but supports a wider range of I2C transfers (first message may be up to 16 bytes).

so long,

Hias

Hi Zsolt!

This is awesome (and that you figured it out)! Just a lame question: where should I put the driver options (maybe in /etc/modules)? Sorry...

Put them in a *.conf file inside /etc/modprobe.d - I choose /etc/modprobe.d/i2c-combined.conf for that setting.

The combined transfer functionality in the kernel is similar to the code in the link you posted, but supports a wider range of I2C transfers (first message may be up to 16 bytes).

so long,

Hias