Tool/software:
#1. Customer request TI confirm if SDK8.3 R5F DM firmware has 49days crash issue.
#2. If yes, need TI provide SDK8.3 DM firmware source code and fix patch for that.
Hello Tony,
I am reaching out for additional information.
Is this something where you need daily updates, even if I have not made any progress? Or do you just want an update once I am able to have a conversation with the appropriate developers?
Regards,
Nick
Customer request TI solve it ASAP, as it is a know issue for TI, no reason to wait.
Today's update:
I have the software repository. I will not have time to look through the software repository until after we resolve the issues here:
AM623: ti-linux-firmware source code of V10.00.08
AM623: Mbox timeout in resp at continuous operating 4295312 seconds
Please ping this thread if I have not provided another update within a few days.
Regards,
Nick
Now SDK9.2 based 49days crash issue can be closed.
Let's move to check if SDK8.3 is impacted.
Still working on the SDK 8.3 code. Will have another update tomorrow.
Regards,
Nick
Hello Tony,
I am 90% sure that RTOS SDK 8.3 is NOT impacted by the 49 day crash. Please give me one more day to verify.
Note that this is ONLY true if the customer is using the DM R5F firmware from RTOS SDK 8.3. If they are using a later version of DM R5F firmware, that later version of firmware may be susceptible.
Summary of the work the last 2 days
The naming convention for the DM R5F firmware "ipc_echo_testb_mcu1_0_release_strip" actually came from the previous TI-RTOS / PDK version of the IPC project, not the MCU+ SDK ipc_rpmsg_echo_linux version of the IPC project. So I suspect that the example code under ti/drv/ipc/examples is the code that was used to generate the DM R5F firmware.
Assuming that is the code, then SDK 8.3 is not impacted by the 49 day crash. There is no timeout option in the code that waits for Linux to be ready for IPC communication, only infinite loops like this:
examples/common/src/ipc_testsetup_baremetal.c examples/common/src/ipc_testsetup.c while(!Ipc_isRemoteReady(pRemoteProcArray[t])) { TaskP_sleep(10); }
Still waiting on verification that this is the project
My contacts who would be able to verify that this is the project used for the AM62x SDK 8.3 DM R5F firmware have been on vacation the past 2 days. I hope to hear back from them on Thursday.
Regards,
Nick
Hello Tony,
The prebuilt image for SDK 8.3 is not affected by the 49 day crash
I was wrong about the project being used, the prebuilt DM R5F binary for SDK 8.3 was actually
pdk/packages/ti/drv/sciclient/examples/sciserver_testapp
This firmware is not affected by the 49 day crash.
As a reminder, the crash is caused by the DM R5F timing out while waiting for Linux to initialize the RPMsg infrastructure.
There is no RPMsg communication is in this project, so this will never happen.
As discussed in the previous response, even for the other IPC examples in PDK, the "wait for Linux" code does not have a timeout value. So even if there was IPC code, it would just loop forever, instead of timing out after 49 days.
Regards,
Nick