RTOS/cc2640: CC2640: memory leak that was addressed in BLE-Stack v2.2.1 when using GAPRole_StartDiscovery

user5192739

Part Number: CC2640
Other Parts Discussed in Thread: BLE-STACK

Tool/software: TI-RTOS

I had survey it's a known issue for v2.2.1

but TI still not update SDK to fix this bug

Dose anyone have another solution?

thanks

I use the project of simple_peripheral_observer of ble_examples-ble_examples-2.2 on BLE-Stack v2.2.1

over 7 years ago

0 Sean2 over 7 years ago

TI__Mastermind 33460 points

Hello,

Thanks for reporting this, it is a known issue. Generally it has been seen when scanning.
The fix has been reported and posted on the Workarounds and known issues page here
processors.wiki.ti.com/.../CC2640_Porting_Projects

Specifically you will want to refer to
e2e.ti.com/.../1994563

0 user5192739 over 7 years ago in reply to Sean2

Prodigy 30 points

appreciate for your reply!!

I find that still nedd some setting in CCS

but I am using IAR for ARM

Do I need to change some setting?

thanks

0 user5192739 over 7 years ago in reply to Sean2

Prodigy 30 points

I alread replace Util_enqueueMsg and Util_dequeueMsg by Queue_put.
The result seem can reduce the frequency of the problem, but still will happen during 1~2 hour
Does anyone has this problem?

0 Sean2 over 7 years ago in reply to user5192739

TI__Mastermind 33460 points

Hello,

We have not observed the issue after applying the aforementioned patch.
Do you have any additional information about the issue.

Regarding toolchains, the patch is independent and should work for both IAR and CCS.

0 Mattia Fiumara over 7 years ago in reply to Sean2

Intellectual 475 points

Dear Sean,

Is this patch applied in the github examples peripheral_observer located here: github.com/.../simple_peripheral_observer ? We are having issues with scanning as well and are wondering if this patch needs to be applied in the github. Thanks for the answer.

Best regards,

Mattia

0 Sean2 over 7 years ago in reply to Mattia Fiumara

TI__Mastermind 33460 points

Hello,

The util bug has been fixed as of all BLE-Stack 3.x releases.
However, there is another known issue in Peripheral observer, it is described here
e2e.ti.com/.../520223

Can you try the patch from that thread?

0 Mattia Fiumara over 7 years ago in reply to Sean2

Intellectual 475 points

Hi Sean,

Thank you very much for your answer. I have seen that thread but I had the assumption that it would have been fixed in the newest peripheral+observer example on the github.

It seems that the contents of the callbacks are different, and that it has been fixed by allocating a new message and freeing it as suggested in the thread linked at https://e2e.ti.com/support/wireless_connectivity/bluetooth_low_energy/f/538/p/520223/1895316?tisearch=e2e-quicksearch&keymatch=GAPRole_StartDevice%20heap#1895316. How it's implemented is a different story and on github an extra free is still present in the scanning callback. This can be seen by plotting the variable 'heapmgrMemAlo', by others suggestions. The variable is consequently decreased, leading to be underflowed when plotting it:

I have made a pull request / fix regarding this bug on the github for this specific issue, you might want to take a look at it. Also please take a look at the following graph of the variable 'heapmgrMemAlo', when running the code with the patch:

Pull request can be found here: https://github.com/ti-simplelink/ble_examples/pull/29.

Best regards,

Mattia

0 Sean2 over 7 years ago in reply to Mattia Fiumara

TI__Mastermind 33460 points

Hi Mattia,

Thanks for the followup and for the pull request.
Unfortunately, we are unable to accept pull requests at this time. This is due to TI internal licensing policies.
That said, we appreciate your work looking into this, and will provide a solution on the Github, taking your findings into consideration.

To all others reading this thread, I am confirming that Mattia's findings are a known issue based on the 1.50.01.01 release of the github, and are planned to be fixed in release 1.50.01.02.

Thanks again!

0 Mattia Fiumara over 7 years ago in reply to Sean2

Intellectual 475 points

Hi Sean,

Thanks for your reply. I think I made a mistake in the pull request, as I removed the ICall_free from the processroleevent while it should be removed from the CB function. In any case I guess your developers have figured it out as well, do you have any information regarding the official release date of 1.50.01.02, so I can implement your solution? Untill then I will have to keep using my own fix and I am unsure if it is the cleanest solution.

Best regards,

Mattia

0 Sean2 over 7 years ago in reply to Mattia Fiumara

TI__Mastermind 33460 points

Hi Mattia,

I would agree that we are planning to remove from the CB function.
I do not have a firm date on the Github update, but plan to get a patch out within the coming weeks.

Please don't hesitate to reach out with any other questions.

Bluetooth®︎

Bluetooth forum

RTOS/cc2640: CC2640: memory leak that was addressed in BLE-Stack v2.2.1 when using GAPRole_StartDiscovery