ChibiOS Free Embedded RTOS

Hi Bob, chFactoryGetObjectsFIFO() cannot return NULL so it is not necessary to test it. The two chFifoTakeObjectTimeout() should fail because the FIFO is empty. IMO the test is not very useful because the allocator always returns block aligned, at least, to PORT_NATURAL_ALIGN so it would never fail....

I added the missing files.

Giovanni

I committed a tentative fix in trunk, will be backported after some review.

Giovanni

That would be a good idea.

Giovanni

Hi Bob,

The fix is not sufficient, it would align the size of the objects but the base address would not be necessarily aligned too, dyn_create_object_heap() requires an alignment parameter and should use chHeapAllocAligned() internally.

I am looking into it.

Giovanni

Hi,

I applied the latest patches.

Giovanni

Threads are not affected by the cache, all threads go through the cache in the same way, the problem is coherence with other bus masters such as DMAs that don't go through the cache, there is no HW cache coherence mechanism in Cortex-M. ChibiOS critical sections provide implicitly a memory barrier, ...

On F7 DTCM is accessible by DMAs, this is not the case on the H7. F7 is generally much easier to use.

Is the gcc volatile keyword observed correctly for data located in cachable memory?

No, the compiler is not cache-aware.

Giovanni

You can make any combination of the two settings, you could also add extra regions if you want.

Giovanni

It is not an LLD, it is under /os/hal/ports/common/ARMCMx.

Giovanni