This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: Your change to final.c
- To: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- Subject: Re: Your change to final.c
- From: Richard Earnshaw <rearnsha at arm dot com>
- Date: Fri, 19 Oct 2001 18:42:41 +0100
- cc: Richard dot Earnshaw at arm dot com, gcc-bugs at gcc dot gnu dot org
- Organization: ARM Ltd.
- Reply-To: Richard dot Earnshaw at arm dot com
> > Your latest change to final.c breaks the ARM bootstrap. There's nothing
> > in the spec for an ASM that says that the number of references to the
> > operands must be less that MAX_RECOG_OPERANDS (which is an internal
> > definition anyway), so we are overflowing this array when we try to output
> > the following ASM statement:
> >
> >
> > I thought there was. Doesn't reload depend on this? What is the proper
> > bound, then?
>
> Well, an asm statement has a limit on the number of distinct operands that
> may be passed to it (10),
Actually, that is no-longer correct. The number of distinct operands can
now be up to 30, since RTH added a named-operand mechanism.
That still doesn't mean that we should restrict the number of references
to 30; it's perfectly legitimate to reference an operand more than once.