Re: [PATCH] checking version of rtl flag access macros

[Janis Johnson <janis187 at us dot ibm dot com>]

On Wed, May 08, 2002 at 01:12:16PM -0400, John David Anglin wrote:
> This patch causes a bootstrap failure under hppa-linux.  There seems to
> be a problem with some macro definitions in the patch.  For example,
> 
> > -/* 1 if insn is a branch that should not unconditionally execute its
> > -   delay slots, i.e., it is an annulled branch.  */
> > -#define INSN_ANNULLED_BRANCH_P(INSN) ((INSN)->unchanging)
> > +/* 1 in an INSN in the delay slot of a branch insn if an annulling branch
> > +   should be used.  */
> > +#define INSN_ANNULLED_BRANCH_P(RTX) (RTL_FLAG_CHECK1((RTX), INSN)->unchanging)
> 
> What's "INSN"?

It's the code for the only kind of rtx that is expected to be used with
the macro INSN_ANNULLED_BRANCH_P.   I changed the name of the macro
argument so it wouldn't conflict with the RTL code name.

Your mail to gcc-bugs says that the macro is used with JUMP_INSN.  If
that's expected, the macro should be changed to add that to the list,
and the documentation in rtl.texi should be updated; otherwise the
hppa-linux port should be fixed to avoid using this macro with such an
rtx.  In the meantime, you can turn off the new rtl flag checking with
--enable-checking=misc,tree,gc (the previous list of defaults).

Janis