[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-sparse
Subject:    Re: [sparse be] PATCH: more op code
From:       tommy () numba-tu ! com
Date:       2004-04-15 22:44:50
Message-ID: 14709.65.113.40.130.1082069090.squirrel () numba-tu ! com
[Download RAW message or body]

Awesome progress Chris.

> As my understanding, there are two major step on the down stream. One is
> the MIR
> (middle level of representation) which is target independent. Most of the
> compiler optimization stuff is done at this level. e.g. alias analyze,
> common expression eliminate and so on.

MIR is target independent, yes, but many optimization will probably depend
on the target.

Also, C99-style alias assumptions depends on C types, so we'd need to
handle that somehow.

> I don't see a need for adding more OP code for different
> type in MIR. Keeping the MIR instruction set small is good.

Amen!

Something else: I'm a little uncertain about the way parameters to
function calls are represented, eg. the "push".  It seems they aren't
explicitly linked to the call site and the order of them is important. 
While it matches stack based ABI well, it's kind of awkward for register
passing ABI.   It seems like a hard problem to find a model that works
well for both ABIs (and a mixed ABI).

Tommy

-
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[prev in list] [next in list] [prev in thread] [next in thread]