Home /Modules/cpu/CACHE.md

CACHE.md

lib/cpu/cortex-r/CACHE.md · 22 line(s) · markdown

Cortex-R Cache and TCM Model

Principle

Cache and tightly coupled memory (TCM) features differ across Cortex-R implementations. Presence, capacity, line size, associativity, replacement policy, maintenance operations and TCM layout must come from the selected core documentation.

Cache behavior

When a cache is absent or disabled, route accesses according to the configured memory system. When enabled, model cacheability attributes, hit/miss behavior, maintenance operations, and latency at the supported fidelity. Avoid claiming exact timing without target data.

Tightly coupled memory

Represent each TCM region with its address range, access permissions, supported ports, and documented latency. TCM is not interchangeable with cache: it is explicitly mapped memory with implementation-specific integration.

Predictability and real-time analysis

Report whether code/data accesses use TCM, cacheable memory, or external memory. Cache misses, line fills, write buffers, arbitration, and external wait states can affect worst-case execution time. Do not assume all cache traffic is deterministic.

DMA and coherency

DMA/cache interaction is platform-dependent. Include documented coherency or software-maintenance requirements; never assume universal hardware coherency.