i2c_davinci i2c_davinci.1: timeout waiting for bus ready

philip lee46717

I am using the AIC3107 in our target board. It is connected with DM6467 through the I2C bus. I am using the kernel version 2.6.18. I have a problem of the I2C bus when I used a very frequent communication of I2C bus. It send out this error message.

i2c_davinci i2c_davinci.1: timeout waiting for bus ready
i2c_davinci i2c_davinci.1: timeout waiting for bus ready
i2c_davinci i2c_davinci.1: timeout waiting for bus ready
i2c_davinci i2c_davinci.1: timeout waiting for bus ready
i2c_davinci i2c_davinci.1: timeout waiting for bus ready

This error cannot be recoverd except I power cycle/reset the DM6467 again.

1. Is there any solution for this?

2. I found a patch can fix this one but it is used in DM355. I cannot convert this patch to DM6467 because it need to control the I2C-SDA and I2C-SCL pins as GPIO. Can I change it in DM6467?

over 13 years ago

0 Sudhakar Rajashekhara over 13 years ago

TI__Expert 5170 points

Such errors have not been observed with the latest kernel source from http://git.kernel.org/?p=linux/kernel/git/khilman/linux-davinci.git;a=summary. You can refer to the I2C driver here for any clue.

Thanks, Sudhakar

0 philip lee46717 over 13 years ago in reply to Sudhakar Rajashekhara

Intellectual 515 points

Hi Sudhakar,

I tried the 2.6.36 and made the changes to the i2c-davinci.c. It can compile after I made some changes for the davinci dm6467. However, when I run it in our target. It doesn't boot up. I have a question. For the DM6467, it doesn't have any pinmux for the sda and scl as gpio. Can I still apply the changes for this chip?

Best Regards,

Philip Lee

0 Prateek Bansal over 13 years ago in reply to Sudhakar Rajashekhara

TI__Expert 4510 points

Hi Philip,

Can you look at Montavista 2.6.18 specific kernel patches and check if any of those patches address i2c timeout issue you are seeing?

Hi Sudhakar,

Do we know root cause of i2c timeout on 2.6.18 kernel based on some of the patches we have seen like the one below -

https://patchwork.kernel.org/patch/62200/

2. https://patchwork.kernel.org/patch/118801

3. https://patchwork.kernel.org/patch/72354

Most of the patches above are on later kernel version. Considering we are interested in 2.6.18 linux kernel based patches, wondering if we have came across any 2.6.18 specific kernel patches to address i2c timeout error. Thanks.

Prateek

0 philip lee46717 over 13 years ago in reply to Prateek Bansal

Intellectual 515 points

Hi Prateek,

I checked the kernel patches for the Montavista 2.6.18 (up to Oct 9,2009). It doesn't have any patch related with this item. There are 2 of those are related with the i2c. Patch 2260 and 2262.

Philip

0 Sudhakar Rajashekhara over 13 years ago in reply to philip lee46717

TI__Expert 5170 points

Hi Philip,

As DM6467 is not having any GPIO pins to be configured as I2C, you cannot the use the patch as is. I have never faced such I2C issues on DM6467 even while using 2.6.18. It may be an hardware issue. Can you get hold of another DM6467 EVM and test it?

Regards, Sudhakar

0 philip lee46717 over 13 years ago in reply to Sudhakar Rajashekhara

Intellectual 515 points

Hi Sudhakar,

In general, I don't meet this problem. However, in some case, the system has busy i2c communication. It will happen. I cannot use the evm to test it because the environment is different.

Phiip Lee

0 Mingqian over 13 years ago in reply to philip lee46717

Intellectual 635 points

Probably, this is a hardware problem rather than a software bug. Poor signals on I2C bus may cause this to happen.

0 hutchman over 13 years ago

Expert 1100 points

I just stepped up to using the community dm6446 kernel 2.6.32 (Linux 255 2.6.32-rc2-davinci1) and I'm having the exact same issue.

My console fills up with:

0 Brandy Jabkiewicz over 13 years ago in reply to hutchman

Mastermind 6325 points

I am not sure if you got an answer to this but I can't help but give you this hint: In my experience with I2C, if the clk is interrupted some how the slave device will hold the line waiting instead of sending an ack until it gets the proper amount of clock cycles. We had to have a special case in the code for the timeout - if it occured we sent clock cycles until the slave released the line. It locked up more often the busier the communication line was. It took me days to figure this out.

0 Vibha Acharya over 9 years ago in reply to Brandy Jabkiewicz

Prodigy 30 points

I am working on DM365 processor and I am facing the same issue.

There is no problem on the EVM. When I run it on the custom board I am facing the same above mentioned error.

I have tried applying the above mentioned patches.

1. Could this be a hardware issue? I see that the BB bit in the DAVINCI_I2C_STR_REG is always being set to 1, even on manually writing a 0 to it.

2. Does anyone have a solution to this problem?

Thanks in advance,

Vibha

Processors

Processors forum

i2c_davinci i2c_davinci.1: timeout waiting for bus ready