'xtensa_read_memory' failed - MIMXRT685 - EVK

キャンセル
次の結果を表示 
表示  限定  | 次の代わりに検索 
もしかして: 

'xtensa_read_memory' failed - MIMXRT685 - EVK

966件の閲覧回数
shri00
Contributor II

Hello, 

I am currently working on a project involving DSP-ARM communication using mimxrt685-evk and encountering inconsistent memory access exceptions. Here are the details of my setup and the issue:

  • IDEs - MCUXpresso and Xtensa Xplorer, Debugger- jlink
  • Shared memory locations to store the data before and after processing: 0x20020000 and 0x20040000 (at max 512 bytes)
  • I encounter an inconsistent memory access exception: ocd.PNGIt is not observed everytime but is often. I'm pretty confident about the code on the M33 side running on MCUXpresso, but the DSP is something I'm very new to. 
  • I am not even sure how to debug it because most times it appears even before I run the code on Xtensa Xplorer
  • I suspect it is happening when I am copying memory to a buffer. 
  • Unfortunately I cannot post the whole code here but here's some outline: I am allocating memory in the main function and have an infinite loop for continuous processing. My goal is to process audio samples in input_buffer and write the processed data to output_buffer.

I'm attaching something like a pseudocode that may help convey my problem better.  

Could you please help me understand the potential causes of this exception and provide some suggestions on debugging strategies? 

It would also be really helpful if you could give some tips on how to avoid memory access issues. 

Thanks in advance!

Best regards,

Shri

ラベル(1)
0 件の賞賛
返信
6 返答(返信)

815件の閲覧回数
shri00
Contributor II

Hello Gavin,

 

Thanks for the response. Can you please elaborate on this because I could make out that there was a memory corruption but I'm seeing these statements on the cmd prompt even before I run the  code on DSP and there is no crash point or a line in code where I get this exception, the code runs normally after I hit run. Should I be focusing checking on the M33 side?

 

Thanks,

shri

 

 

0 件の賞賛
返信

751件の閲覧回数
Gavin_Jia
NXP TechSupport
NXP TechSupport

Hi @shri00 ,

This is a very strange phenomenon you describe. To be honest I haven't encountered this before.

After running DSP program in Xplorer IDE,  xtensa_read_memory starts to work.

I don't quite understand xtensa_read_memory can be used without DSP code execution.

For example, if you use IAR or MCUXPRESSO IDE to debug a CM33 program. You cannot get Xplorer IDE error.

Best regards,
Gavin

0 件の賞賛
返信

671件の閲覧回数
shri00
Contributor II
Thanks for the support, I don't get how it's happening either.
Anyway, I'll see what I can do further.

Regards,
shri
0 件の賞賛
返信

898件の閲覧回数
Gavin_Jia
NXP TechSupport
NXP TechSupport

Hi @shri00 ,

Thanks for your interest in NXP MIMXRT series!

I will work with you to resolve this issue. I have submitted this issue for internal discussion at the moment, so please give me some time. I will keep you updated here as soon as there are new developments. Thanks for your patience in advance!

Best regards,
Gavin

889件の閲覧回数
shri00
Contributor II
Thank you.
0 件の賞賛
返信

818件の閲覧回数
Gavin_Jia
NXP TechSupport
NXP TechSupport

Hi @shri00 ,

Thanks for your patience.

The following responses, from experts is:

The excause 0xd(13 ) meant LoadStorePIFDataErrorCause Synchronous PIF data error during
LoadStore access and the address is  0x40202bb8.

0x40202bb8 is not a valid address, thus exception occurs.

It is likely a memory corruption issue.

 

Best regards<
Gavin

0 件の賞賛
返信