Memory leak issue with NDK Reboot

1074316

Hi,

We have an issue in restarting NDK.
We see memory leak with NDK reboot,the memory allocation fails with few NDK restarts.

We are using a bit old versions of the NDK.

NDK 2.0
CCS v3.3.81.6
BIOS 5.33.06
CGT v6.0.19

In our application we need to restart the communication(NDK) in order to get rid of a communication error.

Currently we are using NC_NetStop(1) to restart NDK and it is working properly.

But the problem is when we restart few times (around 5 times) the NDK stops working,
and we found that the problem occurs with the memory allocation for PBM_alloc.
When we checked the address of PBM_alloc/PBM_free, we found there is a memory leak.

I would like to know what could be the proper procedure to restart the NDK
do we need to do anything else before executing NDK_NetStop(1)?

Best Regards
kummi

over 10 years ago

0 ToddMullanix over 10 years ago

TI__Guru* 96960 points

Can you go into more detail on why you think there is a PBM leak? What is the requested size of the PBM_alloc call that fails?

0 1074316 over 10 years ago in reply to ToddMullanix

Intellectual 775 points

Hi,

Thank you so much for the quick reply.
I will get back to you with the details from the customer.
It seems the memory allocation through PBM_alloc() function fails .

Meanwhile could you please let me know what is the correct procedure to restart the NDK.
Is it OK if we execute only NDK_NetStop(1)?

Best Regards
kummi

0 ToddMullanix over 10 years ago in reply to 1074316

TI__Guru* 96960 points

You just need to call NC_NetStop. There was one bug where if a task is in accept, there is a problem. So with your version, make sure you are not doing this when you call NC_NetStop.

0 1074316 over 10 years ago in reply to ToddMullanix

Intellectual 775 points

Hi,

Thank you,

Regarding the details, the requested size of the PBM_alloc is 76 Bytes
and the PKT_MAX is set to 64, out of which only 8 are freed during NDK restart.

Just in case I would like to know what was the bug you mentioned above.
As you mentioned "task is in accept", does it mean if there is a active Task
when executing NC_NetStop(1), there may be a problem?
Do we require to stop all the tasks when executing NC_NetStop(1)?
If possible could you please let us know about that bug.

Best Regards
Kummi

0 ToddMullanix over 10 years ago in reply to 1074316

TI__Guru* 96960 points

If there is an active task calling "accept", there is a problem if another task calls NC_NetStop. For the version you are using, please avoid this use case.

0 1074316 over 10 years ago in reply to ToddMullanix

Intellectual 775 points

Hi,

Thank you..Your advice is very much helpful.
Please let us know if there is anything else we need to take care of
to avoid the NDK Reboot failure issue.

Best Regards
Kummi

0 ToddMullanix over 10 years ago in reply to 1074316

TI__Guru* 96960 points

You mention there is a PBM memory leak. Have you identified where the leak is occurring?

0 1074316 over 9 years ago in reply to ToddMullanix

Intellectual 775 points

Hi,

I am sorry for the delay.
Actually the problem was solved by adding an extra hardware to reset the PHY.

Thank you for the support.

Best Regards
Kummi

Processors

Processors forum

Memory leak issue with NDK Reboot