CC2652R: HCI_RESET does not response using HeapMem

Hi Tobias,

Thank you for bringing this issue to our attention. I saw in the previous thread Luis said he was able to recreate it. I will verify with him on if he was able to find a root cause. If so, I will see if there is an internal ticket filed for this and the status of the ticket.

After reading through your previous post, I wanted to ask some additional questions about the behavior of the program.

- Do all HCI commands not work when using the setup detailed in your previous post or is it only HCI_Reset?

- Does the program stop or are you still able to execute other commands?

Best Regards,

Jenny

Hi Jenny,

Thanks for the quick response.

It seem to be only the HCI_RESET command, others are working fine.
After the HCI_RESET command no other commands can be executed.
I tried to run the Debugger and got following callstack:

Regards,
Tobias

Hi Tobias,

Thank you for providing a screenshot of the callstack. I have also verified with Luis that he has submitted an internal ticket regarding this issue and it will be addressed in a future fix. In the meantime, I want to see if we can create a workaround to resolve this issue you're seeing.

I had a few additional questions about the heap configurations:

- Do you see the same behavior when using the other heap configurations (since I saw you said both 0x01 and 0x81 result in the same behavior) such as:

• 0x00: Osal heap manager, static size
• 0x80: Osal heap manager, auto size
• 0x02: HeapTrack (with HeapMem), fixed size
• 0x82: HeapTrack (with HeapMem), auto size

- Is there a requirement to use HeapMem (static size or auto-size) as the heap configuration?

- I believe that when using HeapMem (and similarly HeapTrack + HeapMem), you are able to view heap metrics within the built-in CCS tool ROV. If it's possible to run the program using ROV, do you notice any strange behavior exhibited by the heap when analyzing using ROV?

Thank you for your patience!

Best Regards,

Jenny

Hi Jenny,

the miss behavior is reproduceable with following heap configurations:

• 0x01
• 0x02
• 0x81
• 0x82

So it seems not to be dependent on static or autosize.
Osal heap manager seems to be the only working heap manager.

I started with HeapMem and use HeapTrack to debug memory issues. It worked fine unless I tried to integrate the PTM.

You said there is a internal ticket for this issue, is ther a link available for tracking this issue?

Regards,
Tobias

Hi Tobias,

Thank you for providing this additional information.

In the meantime, would you be able to use the OSAL heap manager while I try to debug this issue some more? I understand with HeapMem and HeapTrack, it is easier to debug heap issues using ROV. However, if you define HEAPMGR_METRICS, you should be able to also see heap metrics when using the OSAL heap manager as detailed in the following link https://dev.ti.com/tirex/explore/content/simplelink_cc13x2_26x2_sdk_4_20_01_04/docs/ble5stack/ble_user_guide/html/ble-stack-5.x-guide/debugging-index.html#debugging-common-heap-issues. I understand if this switch is not possible for now.

I don't believe there is as there is only internal tracking, however, if there are updates to the ticket filed I will update you through this thread.

Best Regards,

Jenny

Hi Jenny,

Using OSAL heap will work for now, but TI reccomends to use HeapMem.
HEAPMGR_METRICS does not show me any information in ROV.

Regards,
Tobias

Hi Tobias,

I also agree with you that debugging the heap with ROV is extremely useful when running into heap issues. Although you are not able to use ROV with HEAPMGR_METRICS predefine symbol/OSAL heap, in the meantime if you need to debug the heap the HEAPMGR_METRICS global variables starting with prefix heapmgr may be useful to you.

Best Regards,

Jenny

Hi Jenny,

do you have any update to the progress of the internal ticket filed for this issue?

Regards,
Tobias

Hi Tobias,

I will try to inquire if there are any updates and get back to you once I have a response.

Best Regards,

Jenny

Hi Jenny,

any updates for that issue?

Best Regards,

Tobias

Hi Tobias,

I checked the ticket and there have not been updates as of yet.

Best Regards,

Jenny

Hi Jenny,

is there still work in progress?

Best Regards

Tobias

Hi Tobias,

I revisited the ticket and there have not been updates as of yet.

Best Regards,

Jenny

Hi Jenny,

still no news?

Best Regards
Tobias

Hi Tobias,

No new updates have been made on the internal ticket. I have requested this to be a fix prioritized for this upcoming SDK.

Best Regards,

Jenny

Hi Jenny,

Any news about the internal ticket?

Best Regards,

Tobias

Any news?

Hi Tobias,

First, I need to apologize as this thread seems to have been misplaced due to some recent changes in the team. I discussed with Jenny and she was able to reproduce this as well and has filed a ticket with the RND team, but the RND team has not been able to get to a place to fix it yet. I can ask internally to have this ticket picked up for the Q2 SDK which will release near the end of Q2. 

What is the current timeline you are working towards?

Hi Evan,

Thank you for reply.
For production we use osal heap manager.
The downside is we cannot use full mem debug capabilities at the same time.
So we will be happy if there will be a solution in  next SDK release.

Best Regards,
Tobias

Any news?

Hi Tobias,

I sincerely apologize for the delay. I am able to reproduce this on the latest SDK, we are looking into the source and I will work closely with R&D to get to the bottom of this. If you have any additional information, then please feel free to provided it as may help us get to resolution more efficiently.

Best Regards,

Jan

Hi Jan,

For that issue the only workaround we found is using the OSAL heap manager. The downside here is that we are not able to debug memory issues in applications using PTM.

Regards,

Tobias

Hi Tobias,

Thank you for the info. I will pass this info along to R&D to try to get to the bottom of this as soon as possible. If there are any updates, then I will let you know. Let me know if you run into any new info, any additional piece of information is always helpful.

Best Regards,

Jan

Hi Jan,

Are there any news?

Best Regards,

Tobias

Hi Tobias,

I have just reached out to R&D to check on the status of this behavior as well as the other behavior you observed in the other thread. I had not heard back from there, so I will check to see if they have any new info to share. I will update you as soon as I have heard back. I truly appreciate your patience.

Best Regards,

Jan

Hello Jan,
any feedback from R&D?

Best Regards,
Tobias

Hi Tobias,

Thank you for your patience and consideration. I have reached out to R&D and I am sorry to say that this ticket has not been addressed yet. Our R&D engineers have a ton of things on their plate and have to prioritize incoming issues by how critical they are and how much they impact development. In this case, they have deprioritized this issue as it does not affect production code. I completely agree with you on the importance of resolving this issue. I will do my best to get this address by R&D as soon as possible, but that is the current status of this.

If you have any additional information that you believe may be helpful, then please share and I will provide it to R&D.

Best Regards,

Jan

Hi Tobias,

I have a small update to share. I have talked to R&D and I have a commitment from them to look into this issue for the Q4 SDK release. I apologize for any inconvenience this may cause. I will follow up with R&D when they are looking into this. It seems that you have a functional workaround for this issue (using OSAL Heap Manager), as a result I will be closing this thread in the meantime. I will follow up with R&D, but as of right now they are planning on looking into this for the Q4 release.

Best Regards,

Jan

Hi Jan,

Thank you for your engagement.
Good to hear that this issue finally has been addressed.
Do I have a possibility to track that issue other than keeping this forum thread alive?

Best Regards,

Tobias

Hi Tobias,

No problem! We are glad to help. I will look into tracking capabilities and get back to you. I truly appreciate your patience in this issue.

Best Regards,

Jan

Hi Jan,

Thank you for opening an external ticket for this issue.

Best regards,
Tobias