Other Parts Discussed in Thread: SYSCONFIG
Hi,
I've had this problem for a while and I've not been able to find a solution, and it's currently blocking our business. I'm using IPC notify across all 5 cores, configured using Sysconfig.
I have common code to all cores that registers the channels between R5F1_1 and each other core.
R5F0_1 will usually hard fault or fail an assert when the IPC interrupt happens - sometimes it will crash with the first IPC sync all, sometimes it will crash on the first Notify message, other times it will process several messages before crashing. Notably, when it doesn't fault, it takes between 10-20 seconds for the message to be sent.
Here's an example with timestamps. You can see several messages to core 2 are successful, and this is a particularly good run with 32 messages successfully communicated before a hard fault.
Here's another, showing that it can sometimes take a message to core 2 several seconds for success. In this case, 20 seconds, and the next attempt immediately after fails due to a hard fault.
Here's an example from last night, showing the hard fault at a different point of the IPC exchange:
Here's the logs and a screenshot showing the IPC channel registration and Sysconfig for core 1_1:
And here's core 0_1:
In case it's relevant, here's the DM timer config on core 0_1. About 1/10 attempts to debug the project this will cause the core to crash immediately and I have to disconnect (a reset won't do) the LP from power to get it to run again. I don't think it's related, though, as when I remove the timer the IPC message persists.
Given there is nothing to configure in SW for IPC when using Sysconfig, I'm at a loss on how to fix this. I'm not using RP messages, so there should be no shared memory issue. I've tried to switch the clientID, but it's made no difference. Every other IPC channel and other core works perfectly fine, completing IPC exchanges immediately.
Any help is greatly appreciated.
