URL
https://opencores.org/ocsvn/ion/ion/trunk
[/] [ion/] [trunk/] [vhdl/] [mips_cache_stub.vhdl] - Diff between revs 72 and 73
Go to most recent revision |
Show entire file |
Details |
Blame |
View Log
| Rev 72 |
Rev 73 |
| Line 47... |
Line 47... |
--------------------------------------------------------------------------------
|
--------------------------------------------------------------------------------
|
-- KNOWN TROUBLE:
|
-- KNOWN TROUBLE:
|
--
|
--
|
-- Apart from the very rough looks of the code, there's a few known problems:
|
-- Apart from the very rough looks of the code, there's a few known problems:
|
--
|
--
|
-- 1.- Write address setup and hold wrt. WE\ not guaranteed
|
-- 1.- Write cycles too long
|
-- WE\ needs to be asserted later and deasserted earlier. The easy way
|
-- In order to guarantee setup and hold times for WE controlled write
|
-- would be using two extra cycles. Must find some less cosly way.
|
-- cycles, two extra clock cycles are inserted for each SRAM write access.
|
-- So far, in my particular test conditions, this is not giving me trouble
|
-- This is the most reliable way and the easiest but probably not the best.
|
-- so this will have to wait.
|
-- Until I come up with something better, write cycles to SRAM are going
|
|
-- to be very slow.
|
--
|
--
|
-- 2.- Access to unmapped areas will crash the CPU
|
-- 2.- Access to unmapped areas will crash the CPU
|
-- A couple states are missing in the state machine for handling accesses
|
-- A couple states are missing in the state machine for handling accesses
|
-- to unmapped areas. I haven't yet decided how to handle that (return
|
-- to unmapped areas. I haven't yet decided how to handle that (return
|
-- zero, trigger trap, mirror another mapped area...).
|
-- zero, trigger trap, mirror another mapped area...).
|
© copyright 1999-2025
OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.