topic Re: To find the root cause of i.MX6Q IPU error interrupt. in i.MX Processors

To find the root cause of i.MX6Q IPU error interrupt.

satoshishimoda — Mon, 09 Jun 2014 05:52:37 GMT

Hi community,

I have a question about i.MX6Q IPU.

Please see Table 37-35 in IMX6DQRM Rev.1.

Sometimes IPU_INT_STAT5[0] and IPU_INT_STAT10[0] is occurred on a custom board.

These interrupts seem that these are generated by input data from CSI, so we checked whether input data from camera is over the i.MX6Q CSI spec, but the input data is smaller than the maximum spec of CSI sufficiently. (camera data: 1920x1080@30fps BT.1120)

And they have already set IPUx_HSP_CLK_ROOT to large value adequately (264MHz), so we think the root cause is not performance of CSI.

In this case, could you give me a advice what should we check to find root cause?

(e.g. we should check the signal integrity of camera data, IPU output IPU_INT_STAT5[0] or IPU_INT_STAT10[0] if memory bandwitdh is not sufficient, etc..)

For your information, IPU is used to output 320x240 10fps also.

Best Regards,

Satoshi Shimoda

Re: To find the root cause of i.MX6Q IPU error interrupt.

igorpadykov — Tue, 10 Jun 2014 03:18:18 GMT

Hi Satoshi,

you are right, this is performance issue as described in link below

https://community.freescale.com/message/320553#320553

Please understand that this is specific application issue -

that is, this depends on customers configuration.

Issue may be solved by finding source of bandwidth limitation:

you can start test with SDK, then with Linux min. configuration.

Then, adding drivers one by one, you can find when error will start to appear:

this will be source of this error.

Best regards

chip

-----------------------------------------------------------------------------------------------------------------------

Note: If this post answers your question, please click the Correct Answer button. Thank you!

-----------------------------------------------------------------------------------------------------------------------

Re: To find the root cause of i.MX6Q IPU error interrupt.

satoshishimoda — Tue, 10 Jun 2014 04:15:00 GMT

Hi chipexpert,

Thank you for your reply.

Table 37-35 shows these error are occurred SMFC or IDMAC, so I guessed this is camera input performance issue.

According to your reply, are these error "IPU_INT_STAT5[0] and "IPU_INT_STAT10[0]" also occurred by not only camera input performance but also system memory bandwidth?

Best Regards,

Satoshi Shimoda

Re: To find the root cause of i.MX6Q IPU error interrupt.

igorpadykov — Tue, 10 Jun 2014 07:44:25 GMT

Hi Satoshi,

yes I agree with you.

One can try to reserve more memory to IPU tasks

https://community.freescale.com/docs/DOC-93591

or use more new 3.10.17 kernel which has better memory

management

L3.10.17_1.0.0_IMX6QDLS_BUNDLE

Best regards

chip

Re: To find the root cause of i.MX6Q IPU error interrupt.

satoshishimoda — Tue, 10 Jun 2014 10:34:22 GMT

Hi chip,

We changed the resolution of camera input from 1080p to VGA to reduce load for memory.

But these error have remained yet.

So we think these error are not performance issue.

Do you think so?

For your information, the following messages are log when the error was occurred.

=====

[pattern 1]

ERROR: v4l2 capture: mxc_v4l_dqueue() interrupt received

Segmentation fault

imx-ipuv3 imx-ipuv3.0: IPU Warning - IPU_INT_STAT_10 = 0x00000001

[pattern 2]

Segmentation fault

imx-ipuv3 imx-ipuv3.0: IPU Warning - IPU_INT_STAT_10 = 0x00000001

[pattern 3]

VPU interrupt received.

imx-ipuv3 imx-ipuv3.0: IPU Warning - IPU_INT_STAT_10 = 0x00000001

Segmentation fault

=====

We can find "Segmentation fault" in all patterns.

So we doubt it as the root cause.

Could you let me know what module output "Segmentation fault" and when output it?

Best Regards,

Satoshi Shimoda

Re: To find the root cause of i.MX6Q IPU error interrupt.

satoshishimoda — Wed, 11 Jun 2014 10:08:21 GMT

If you can, could you let me know the root cause of the above 3 patterns error?

Re: To find the root cause of i.MX6Q IPU error interrupt.

igorpadykov — Wed, 11 Jun 2014 10:55:47 GMT

Hi Satoshi,

to answer your question

it is necessary to check this issue on our side.

Could you tell steps, how this issue can be reproduced

on Freescale reference board with Freescale BSPs ?

BTW had you chance to try new 3.10.17 kernel

which has much better memory management ?

L3.10.17_1.0.0_IMX6QDLS_BUNDLE

Best regards

chip

Re: To find the root cause of i.MX6Q IPU error interrupt.

satoshishimoda — Thu, 12 Jun 2014 08:38:34 GMT

Hi chip,

Unfortunately, we cannot reproduce this issue with SABRE board because SABRE board cannot receive BT.1120.

So could you reply to our simple question one by one to investigate what is the root cause of this issue?

At first, our first simple question is "Could you let me know what module output "Segmentation fault" and when output it?"

Best Regards,

Satoshi Shimoda

Re: To find the root cause of i.MX6Q IPU error interrupt.

igorpadykov — Thu, 12 Jun 2014 11:55:53 GMT

Hi Satoshi,

regarding "what module output "Segmentation fault" and when output it" -

it outputs by IPU, when - please look below

Segmentation fault - Wikipedia, the free encyclopedia

Could you post this issue to meta-fsl-arm mailing list, so that someone familiar with

IPU "Segmentation fault" errors could try to assist you.

https://lists.yoctoproject.org/listinfo/meta-freescale

Also did you check below patch

Support BT1120 progressive mode in Mx6Q

Best regards

chip

Re: To find the root cause of i.MX6Q IPU error interrupt.

satoshishimoda — Mon, 16 Jun 2014 05:38:45 GMT

Hi chip,

Thank you for your reply.

=====

> it outputs by IPU, when - please look below

> Segmentation fault - Wikipedia, the free encyclopedia

If you know, could you let me know what source code in BSP (L3.0.35_4.1.0) output the "Segmentation fault" also?

=====

> Could you post this issue to meta-fsl-arm mailing list, so that someone familiar with

> IPU "Segmentation fault" errors could try to assist you.

> https://lists.yoctoproject.org/listinfo/meta-freescal

Sorry, I don't know what is the mailing list and don't know how to use.

=====

> Also did you check below patch

> Support BT1120 progressive mode in Mx6Q

For your information, out partner have already applied the below patch.

Support BT1120 progressive mode in Mx6Q

=====

Best Regards,

Satoshi Shimoda

Re: To find the root cause of i.MX6Q IPU error interrupt.

igorpadykov — Mon, 16 Jun 2014 16:13:56 GMT

Hi Satoshi,

when one goes on link below, there you will be prompted to enter e-mail

https://lists.yoctoproject.org/listinfo/meta-freescale

Regarding :

"If you know, could you let me know what source code in BSP (L3.0.35_4.1.0) output the

"Segmentation fault" also?"

- sorry, it is impossible to say in which on many thousands of Linux instructions

this error happens.

Actually, as you correctly noted, this case is not supported in Freescale Linux BSPs

and it is not possible to reproduce it on Freescale boards.

You can post issue on this mailing list, there

i.MX developers are working and they know many typical errors, they may help.

Alternatively you can apply to FPS

http://cache.freescale.com/files/training/doc/ftf/2014/FTF-SDS-F0251.pdf

Best regards

chip