This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

GPT and Caching

Hi, ARM communities!

I was wondering whether there is another hardware cache for GPT.

For example, storing (physical address -> GPT entry) so that hardware can quickly check the GPT for specific physical entries. 

Sincerely

Parents
  • The architecture describes a set of behaviours and rules, not a specific implementation.  So for example, the spec says when a GPT entry can/cannot be cached and what steps by software are required to guarantee visibility of a change to the GPTs.  Exactly how that translates into a cache topology is up to the specific processor design

    It is quite possible that different processor designs will make different decisions.  More generally, Arm maintains the division between architecture and implementation (or micro-architecture) specifically so that hardware designers have freedom to innovate while still giving predictable behaviour to software.

    Which is a long winded way of saying: maybe

Reply
  • The architecture describes a set of behaviours and rules, not a specific implementation.  So for example, the spec says when a GPT entry can/cannot be cached and what steps by software are required to guarantee visibility of a change to the GPTs.  Exactly how that translates into a cache topology is up to the specific processor design

    It is quite possible that different processor designs will make different decisions.  More generally, Arm maintains the division between architecture and implementation (or micro-architecture) specifically so that hardware designers have freedom to innovate while still giving predictable behaviour to software.

    Which is a long winded way of saying: maybe

Children
No data