Subject: Re: more on understanding caches.
To: None <port-mips@netbsd.org, port-m88k@netbsd.org>
From: Toru Nishimura <nisimura@itc.aist-nara.ac.jp>
List: port-m88k
Date: 07/24/2000 11:38:16
>> - Flush
>> Make sure target address range contents synchronized
>> with cache, if any, then discard the copy in cache.
>
> I would prefer to avoid the ambiguous 'flush' term, and have the
> operations be named along these lines.
>
> Invalidate
> Writeback
> Writeback+Invalidate
It was the initial thought of mine, indeed. However, I hesitate to
supercede and conceal 'write-thru' notion with the term of
'write-back' although it's reasonable to construct cache operations
in write-back centric fashion because write-thru is mostly invible
for kernel logics.
Tohru Nishimura