TLB?
TLB?
Has anyone gotten far enough to check to see the PSP's CPU supports a TLB? I had assumed it did, but then a friend pointed out this slide: http://pc.watch.impress.co.jp/docs/2005 ... igai_2.jpg
It would really suck if it didn't.
It would really suck if it didn't.
Re: TLB?
Depends on ones way of thinking. Were I to write an OS for an embedded device, I would try to avoid using a TLB unless I really have to. TLB is only necessary to provide any hardware assist to implementing virtual memory.mrbrown wrote:Has anyone gotten far enough to check to see the PSP's CPU supports a TLB? I had assumed it did, but then a friend pointed out this slide: http://pc.watch.impress.co.jp/docs/2005 ... igai_2.jpg
It would really suck if it didn't.
If I ever port the OS that I want to port to the PS2, you can be sure I won't be using the TLB. :P
-
- Posts: 564
- Joined: Sat Jan 17, 2004 10:22 am
- Location: Sweden
- Contact:
http://www.xiptech.com/uclinuxformips.htmXiptech ported uClinux to MIPS platform. We offer patch of uClinux kernel,uClibc,elf2flt and document. We also provide a software Simulator, which can run MIPS linux binary code, for people who want to learn MIPS but without real hardware. The simulator is a virtual machine, it has a MIPS R3000 CPU and some virtual hardwares, including UART,flash rom,RAM,LCD,touch screen,etc. It provides world leading performance
Well, loser indicated that kernel RAM is protected from user access, so they would have to have at least a basic MMU in place to lock down certain sections of RAM.
Another thing he mentioned was that cop0 was unusable in user programs. So figuring out what type of MMU they have and what it can do is a bit tricky (along with many other kernel-only things).
Another thing he mentioned was that cop0 was unusable in user programs. So figuring out what type of MMU they have and what it can do is a bit tricky (along with many other kernel-only things).
Well, without a TLB then all memory accesses are "unmapped". There can still be user, supervisor, and kernel spaces, which would be sufficient to protect the kernel from userland. The only thing is, they may need to hard code the kernel range, or set it via registers. Just a possibility of course. :)