Thanks for your reply!
So there is no HW arbitration between two threads accessing the FPGA (PCIe)?
If 2 threads mmap the region of FPGA registers, nothing in HW protects against a single u32 access/read (like the AXI)? It's a purely up to SW to arbitrate?
Thinking about PCIe like a memory, I don't believe I've seen protection around every time I read from RAM... unless there is a difference with-respect-to how the core/MMU views accesses between RAM and PCIe device.
Thanks!