This is the mail archive of the
cgen@sources.redhat.com
mailing list for the CGEN project.
Re: anonymous ifields
- To: fche at redhat dot com (Frank Ch. Eigler)
- Subject: Re: anonymous ifields
- From: Ben Elliston <bje at redhat dot com>
- Date: Fri, 16 Mar 2001 17:50:04 +1100 (EST)
- Cc: cgen at sources dot redhat dot com
- References: <15024.36717.101762.797146.cygnus.local.cgen@scooby.apac.redhat.com><o5g0gfqg9t.fsf@touchme.toronto.redhat.com>
>>>>> "Frank" == Frank Ch Eigler <fche@redhat.com> writes:
Frank> How frequently does each of these anonymous ifields occur in your new
Frank> port? If the ifields in question are used more than once, then a
Frank> pmacro-based declaration plus "f-0" type references would be about as
Frank> compact than an anonymous declaration in situ.
In some cases, they occur dozens of times -- perhaps in 30% of the
entire instruction set.
Frank> If you don't like having to declare the ifields explicitly, we could
Frank> get cgen to automagically declare a bunch of ifields for all
Frank> contiguous bitfields in an instruction word. (O(N**2) of them for N
Frank> bits.)
In some ways, I think that would be *worse*. Things happening
implicitly just gets confusing. Is what I'm proposing really deemed
to be that bad? I can understand it from the perspective of
increasing cgen's syntax.
I'm going to try using pmacros to implement what I'm after and we'll
see how useful it ends up being.
Ben