This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: RTL_EXPR vs. inlining-on-trees
- To: kenner at vlsi1 dot ultra dot nyu dot edu
- Subject: Re: PATCH: RTL_EXPR vs. inlining-on-trees
- From: Mark Mitchell <mark at codesourcery dot com>
- Date: Sun, 05 Mar 2000 13:06:13 -0800
- Cc: gcc-patches at gcc dot gnu dot org
- Organization: CodeSourcery, LLC
- References: <10003052057.AA06275@vlsi1.ultra.nyu.edu>
>>>>> "Richard" == Richard Kenner <kenner@vlsi1.ultra.nyu.edu> writes:
Richard> It should be. This sort of thing is quite common in Ada
Richard> too. That's what mark_temp_addr_taken is for.
Perhaps that's what I was missing. I'll look at that shortly, and see
if the problem can be fixed that way.
Richard> With those variable *inside* or *outside* of the
Richard> RTL_EXPR?
Inside. There's no fundamental reason that an RTL_EXPR should define
a scope.
Richard> But when that flag is set, what about temporaries
Richard> allocated by the backend which were inside a push/pop?
Hmm? Can you elaborate? With the flag set, all temporaries in the
RTL_EXPR get a strictly longer lifetime than they would have
otherwise, modulo any possible coding bug on my part. I don't see how
that can be dangerous.
--
Mark Mitchell mark@codesourcery.com
CodeSourcery, LLC http://www.codesourcery.com