Kernel 3.x - testing i2c support

[Fadil Berisha]

i2c driver for imx23 is included in kernel 3.x. To add i2c support on olinuxino needed to patch kernel source with patch available here:
https://www.dropbox.com/s/v9z0q9mkmoa5vsm/0001-MXS-imx23-olinuxino-Add-i2c-support.patch

Following code define i2c support on olinuxino:

         i2c0: i2c@80058000 {
            pinctrl-names = "default";
            pinctrl-0 = <&i2c1_pins_a>;
            status = "okay";
         };

Example-1: MOD-IO2 board
MOD-IO2 has i2c interface to connect external  RELAYs and GPIOs to any boards with UEXT connector.
Test is done with kernel 3.7.0-rc3-next-20121102-dirty and utility i2c-tool available from Olimex web site.
To activate both relays, run command:

./i2c-tool -w 0 0x48 4 0x02 0xA0 0x40 0x03

here is response:

SLAVE ADDRESS: 0x48
NUMBER OF BYTES TO WRITE: 4
MEMORY ALLOCATED AT ADDRESS: 0x743008
/dev/i2c-0 OPENDED!
WRITE:SUCCESS

to deactivate relays, run:

./i2c-tool -w 0 0x48 4 0x02 0xA0 0x40 0x00

response:

SLAVE ADDRESS: 0x48
NUMBER OF BYTES TO WRITE: 4
MEMORY ALLOCATED AT ADDRESS: 0x1DAC008
/dev/i2c-0 OPENDED!
WRITE:SUCCESS

Following photo shown olinuxino-mini MOD-IO2 with activated relays: 


If device attached on i2c UEXT connector need linux driver, previous code need to adapted by end user.

Example-2: EEPROM atmel,24c128

         i2c0: i2c@80058000 {
            pinctrl-names = "default";
            pinctrl-0 = <&i2c1_pins_a>;
            status = "okay";
            
            eeprom: eeprom@51 {
               compatible = "atmel,24c128";
               reg = <0x51>;
               pagesize = <32>;
            };
         };


Acknowledgements

Fabio Estevam, Freescale; Providing assistance and guidance in development of i2c patch
Tsvetan Usunov, Olimex:  Providing hardware for testing of i2c interface

Fadil Berisha

[olimex]

Amazing! Big THANKS for your efforts, if you need any other hardware you need to test/play with just let me know

So now we have in 3.x: SPI, I2C, UART, USB, SD-card, GPIOs and only LRADC, Audio and Video left unimplemented from all features iMX233 have?

[Fabio Estevam]

LRADC is already supported in 3.7-rc4.

Regards,

Fabio Estevam

[robwoj]

Hi,

When applied to linux-3.7-rc5 /dev/i2c-0 appeared. I have some questions now.

If device attached on i2c UEXT connector need linux driver, previous code need to adapted by end user.

Example-2: EEPROM atmel,24c128

         i2c0: i2c@80058000 {
            pinctrl-names = "default";
            pinctrl-0 = <&i2c1_pins_a>;
            status = "okay";
            
            eeprom: eeprom@51 {
               compatible = "atmel,24c128";
               reg = <0x51>;
               pagesize = <32>;
            };
         };

Is changing imx23.dtsi and/or imx23-olinuxino.dts file necssary whenever I change connected devices, or only when such device need to load external module?
Let's say that I have PCF8574 device and I want to use Write/read operations only (eg. using olimex' i2c-tool or my own dedicated program). Must I change device tree files?

What is needed to get i2cdetect working? For now it says 'Error: Can't use SMBus Quick Write command on this bus'.

Thank you in advance, Robert

[Fadil Berisha]

Is changing imx23.dtsi and/or imx23-olinuxino.dts file necssary whenever I change connected devices,

1. If device does not require driver - no need to change files,
2. If device need driver - need to change only imx23-olinuxino.dts file:

         i2c0: i2c@80058000 {
            pinctrl-names = "default";
            pinctrl-0 = <&i2c1_pins_a>;
            status = "okay";
            /* Add node for slave device here, e.g. PCF8574 */
         };

Compile new DT file and append DT file on kernel. For procedure see https://github.com/koliqi/imx23-olinuxino

Fadil Berisha

[Niii]

Hi,

I've some issues with i2c-tools.
Tested with linux-next git (3.8.0-rc1) and your modifications.

Now when I scan the i2c bus I've no longer error message about SMBUS.
But, after scanning address 0x30, I've got some errors about PIO reg.

Fadil, have you try scanning with i2c-tools ?
Maybe is DMA who are not ready ? (http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-mxs.c#L217)

Code Select Expand


[79108.240000] i2c i2c-0: master_xfer[0] W, addr=0x25, len=0
[79108.250000] mxs-i2c 80058000.i2c: addr: 0x0025, len: 0, flags: 0x0, stop: 1
[79108.250000] i2c i2c-0: ioctl, cmd=0x703, arg=0x26
[79108.250000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.250000] i2c i2c-0: master_xfer[0] W, addr=0x26, len=0
[79108.260000] mxs-i2c 80058000.i2c: addr: 0x0026, len: 0, flags: 0x0, stop: 1
[79108.260000] i2c i2c-0: ioctl, cmd=0x703, arg=0x27
[79108.260000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.260000] i2c i2c-0: master_xfer[0] W, addr=0x27, len=0
[79108.270000] mxs-i2c 80058000.i2c: addr: 0x0027, len: 0, flags: 0x0, stop: 1
[79108.270000] i2c i2c-0: ioctl, cmd=0x703, arg=0x28
[79108.270000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.270000] i2c i2c-0: master_xfer[0] W, addr=0x28, len=0
[79108.280000] mxs-i2c 80058000.i2c: addr: 0x0028, len: 0, flags: 0x0, stop: 1
[79108.280000] i2c i2c-0: ioctl, cmd=0x703, arg=0x29
[79108.280000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.280000] i2c i2c-0: master_xfer[0] W, addr=0x29, len=0
[79108.290000] mxs-i2c 80058000.i2c: addr: 0x0029, len: 0, flags: 0x0, stop: 1
[79108.290000] i2c i2c-0: ioctl, cmd=0x703, arg=0x2a
[79108.290000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.290000] i2c i2c-0: master_xfer[0] W, addr=0x2a, len=0
[79108.300000] mxs-i2c 80058000.i2c: addr: 0x002a, len: 0, flags: 0x0, stop: 1
[79108.300000] i2c i2c-0: ioctl, cmd=0x703, arg=0x2b
[79108.300000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.300000] i2c i2c-0: master_xfer[0] W, addr=0x2b, len=0
[79108.310000] mxs-i2c 80058000.i2c: addr: 0x002b, len: 0, flags: 0x0, stop: 1
[79108.310000] i2c i2c-0: ioctl, cmd=0x703, arg=0x2c
[79108.310000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.310000] i2c i2c-0: master_xfer[0] W, addr=0x2c, len=0
[79108.320000] mxs-i2c 80058000.i2c: addr: 0x002c, len: 0, flags: 0x0, stop: 1
[79108.320000] i2c i2c-0: ioctl, cmd=0x703, arg=0x2d
[79108.320000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.320000] i2c i2c-0: master_xfer[0] W, addr=0x2d, len=0
[79108.330000] mxs-i2c 80058000.i2c: addr: 0x002d, len: 0, flags: 0x0, stop: 1
[79108.330000] i2c i2c-0: ioctl, cmd=0x703, arg=0x2e
[79108.330000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.330000] i2c i2c-0: master_xfer[0] W, addr=0x2e, len=0
[79108.340000] mxs-i2c 80058000.i2c: addr: 0x002e, len: 0, flags: 0x0, stop: 1
[79108.340000] i2c i2c-0: ioctl, cmd=0x703, arg=0x2f
[79108.340000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.340000] i2c i2c-0: master_xfer[0] W, addr=0x2f, len=0
[79108.350000] mxs-i2c 80058000.i2c: addr: 0x002f, len: 0, flags: 0x0, stop: 1
[79108.360000] i2c i2c-0: ioctl, cmd=0x703, arg=0x30
[79108.360000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.360000] i2c i2c-0: master_xfer[0] R, addr=0x30, len=1
[79108.370000] mxs-i2c 80058000.i2c: addr: 0x0030, len: 1, flags: 0x1, stop: 1
[79108.380000] mxs-i2c 80058000.i2c: Failed to get PIO reg. write descriptor.
[79108.380000] i2c i2c-0: ioctl, cmd=0x703, arg=0x31
[79108.380000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79108.380000] i2c i2c-0: master_xfer[0] R, addr=0x31, len=1
[79108.390000] mxs-i2c 80058000.i2c: addr: 0x0031, len: 1, flags: 0x1, stop: 1
[79109.410000] mxs-i2c 80058000.i2c: Timeout!
[79109.410000] i2c i2c-0: ioctl, cmd=0x703, arg=0x32
[79109.410000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79109.410000] i2c i2c-0: master_xfer[0] R, addr=0x32, len=1
[79109.420000] mxs-i2c 80058000.i2c: addr: 0x0032, len: 1, flags: 0x1, stop: 1
[79109.430000] mxs-i2c 80058000.i2c: Failed to get PIO reg. write descriptor.
[79109.430000] i2c i2c-0: ioctl, cmd=0x703, arg=0x33
[79109.430000] i2c i2c-0: ioctl, cmd=0x720, arg=0xbeb57c8c
[79109.430000] i2c i2c-0: master_xfer[0] R, addr=0x33, len=1
[79109.440000] mxs-i2c 80058000.i2c: addr: 0x0033, len: 1, flags: 0x1, stop: 1
[79110.460000] mxs-i2c 80058000.i2c: Timeout!


[Fadil Berisha]

Hi Niii
Test is done with kernel 3.7.0-rc3-next-20121102 and 3.7.1-stable using utility i2c-tool available from Olimex web site (found in software package of  MOD-IO2). i2c-tool working very well so I didn't had need to try another tool.

Fadil Berisha

[Niii]

Fadil,

Have you a copy of your precompiled i2c-tool binary somewhere ?
I don't have windows OS with me at this time for compilation :/.

Thanks.

[htarold]

Hi all,

I'm having trouble also with i2c.  Specifically, a read or write results in either EINVAL with a kernel error "mxs-i2c 80058000.i2c: Failed to get PIO reg. write descriptor", or ETIMEDOUT, when using "i2c-tool -w ..." or "i2c-tool -r ...".

Output from strace:
...
open("/dev/i2c-0", O_RDWR)              = 3
write(1, "/dev/i2c-0 OPENDED!\n", 20/dev/i2c-0 OPENDED!
)   = 20
ioctl(3, 0x703, 0x18)                   = 0
write(3, " '", 2)                       = -1 ETIMEDOUT (Connection timed out)
...
Or:
...
open("/dev/i2c-0", O_RDWR)              = 3
write(1, "/dev/i2c-0 OPENDED!\n", 20/dev/i2c-0 OPENDED!
)   = 20
ioctl(3, 0x703, 0x18)                   = 0
write(3, " '", 2)                       = -1 EINVAL (Invalid argument)
...

I know the device (LSM303 compass/accel) works, it has been tested with other hardware.

There appears no signal on the SCL pin when I attach a scope.

I have tried the stock kernel that comes with the Arch linux image (http://archlinuxarm.org/platforms/armv5/olinuxino) which is 3.7.2, and also 3.7.1 patched and compiled per instructions in https://github.com/koliqi/imx23-olinuxino.

I would appreciate any help or pointers anyone has.

Thanks,
-harold

[Niii]

Hi,

Can you try this patch (tested on 3.8.0-rc3) :
https://www.dropbox.com/s/3jl58z6av6yfzwl/i2c_dma_reg-imx233_fix.patch
Source : http://permalink.gmane.org/gmane.linux.drivers.i2c/13798

It fix error on PIO and imx233 registers.
But your i2c hw bus will works only with "Read byte" commands(i2cdetect -r).

I've got nothing too with quick probe on my scope, it seems to be not supported (no reference in imx233 datasheet)

Nicolas

[Fadil Berisha]

Hi,

Can you try this patch (tested on 3.8.0-rc3) :

Excellent work. No more  "Failed to get PIO reg. write descriptor" error when tested with i2cdetect. My applaud and Karma +1


Fadil Berisha

[htarold]

Hi Niii,

Thanks I will try this with 3.8.  Did you apply the new patch over the old ones or just this patch alone?

Are you saying i2c now actually works, albeit only in "read byte" mode?

Thanks,
-harold

[Niii]

Hi Harold,

Try this :
Just before compile your kernel, apply these three patchs :
1) Same as Fadil patch, but set compatibily for i2c0 to imx23 instead imx28.
https://www.dropbox.com/s/r6quykxgozuwiak/i2c_mxs-dts.patch

2) Patch from Marex with a minor fix (now INIT_COMPLETION is lowercase in driver, so hunk 4 fail with original patch).
https://www.dropbox.com/s/v07eehulrc4dv18/i2c_mxs-dma.patch

3) Patch for set correctly registers for imx23.
https://www.dropbox.com/s/pvbt9d883de0m6v/i2c_mxs-reg.patch

With Fadil, all mode works on 3.8.0-rc3.
For me, scan only works with -r option, and 50% scan bus fail at error -110.

I need to take a look at that.

Nicolas

Ps : For information i2c bus is on pins 23/24 CON1, and you need to solder jumpers for HW i2c (Micro manual section 6.11.1)

[Niii]

Hi,

I just made a patch to test debugfs, if someone wanna try :
https://www.dropbox.com/s/tdzhzmokpdqzkuw/i2c_mxs-debugfs.patch

Before a scan :

Code Select Expand


root@micro:~# cat /sys/kernel/debug/mxs-i2c/regs
MXS_I2C_DATA: 0x00000000
MXS_I2C_DEBUG0: 0x00100000
MXS_I2C_DEBUG0_CLR: 0x00100000


After a scan :

Code Select Expand


root@micro:~# cat /sys/kernel/debug/mxs-i2c/regs
MXS_I2C_DATA: 0x00000015
MXS_I2C_DEBUG0: 0x0010d000
MXS_I2C_DEBUG0_CLR: 0x0010d000



Nicolas

[htarold]

Hi Nicholas,

I have tried your suggested patches on 3.8-rc3, and I can get a response sometimes.  Alternate times it times out (errno = ETIMEDOUT = 110) as you observed.

Also if I read a larger buffer, I get:
[  363.140000] INFO: trying to register non-static key.
[  363.140000] the code is fine but needs lockdep annotation.
[  363.140000] turning off the locking correctness validator.
followed by an oops.

Is there any way to avoid the ETIMEOUT?

Thanks,
-harold