Hello,

I've enabled FreeRTOS's support for a MPU (LPC1788). My application, however, needs to _move_ lots of data - log data, screen shots, etc. With a MPU in place, this becomes a difficult or at least - expensive in terms of computation power, as the only way to move data is via kernel provisions (queue...) and signals.
Is there a useful way to share data between tasks? I thought about programming the MPU on the fly so that one of the regions is setup as R/O to allow some access, but what do I do if I need to write (for example, if a task needs to write data to a TX buffer which is located outside of its regions reach...). Having a task per function can cause serious traffic congestion...
Any ideas?

If these issues are not resolved, I am considering leaving the MPU enabled, but only for the kernel.
I would appreciate any input you might have.

You can give more than one task access to the same RAM to share data between tasks. For example, if you set up the MPU of two tasks to have both read and write access to a certain area of RAM. Or, depending on the design, you could set up one task to have read only access and another task to have write only access to a region of RAM so one task can produce data and another consume it (without any other tasks being able to write over the data being consumed).

Does that help your design? Maybe you are doing that already?

Regards.

Thanks for your reply.
I have build a small prototype that allows a task to "R/O - share" its MPU memory (allocated using a LIFO policy but that's off the subject) upon sending a signal to a task that requires access to the memory. That should solve the issue of transferring large blocks of data, but also requires the approval of my colleagues...