Bill Lamie

It is good that you have narrowed it down to this callback handler… It would be good to check the thread or ISR stack availability where this callback occurs. Memory...

The trace_buffer_setup processing is non-trivial and, as implemented, could be called from ISR processing. Instead, I suggest using the buffer full callback notification to wake up a thread, which will...

Sounds good. Please let me know how it goes. I’m happy to discuss other issues/improvements at ***@***.******@***.***> From: yf13 ***@***.***> Sent: Wednesday, October 16, 2024 3:30 PM To: eclipse-threadx/threadx ***@***.***>...

In a perfect world, I would have used a "void *" when I created the API some 28 years ago. Alas, the world is not perfect, and changing this would...

On 64-bit architectures, the change could break the port files since the assembly code offsets into the TX_THREAD structure will change. On 32-bit architectures, this could be okay. Of course,...

This is a valid problem. The tx_port.h file must be updated - similar to the Cortex-A9 tx_port.h file - to support floating point save/restore.

Looks like we would need to review at least all the Cortex-R4 and Cortex-R5 ports. The GNU versions for Cortex-R4/R5 are missing the information in tx_port.h. The IAR ports look...

In principle, I agree with this request, and we should have included this in the original APIs. However, now that the API has been public for 28 years, I would...

These both appear to be valid points; we will investigate them further.