This is the mail archive of the
mailing list for the GCC project.
Re: RFC: Two possible fixes for g++.dg/torture/pr32950.C failure
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Steve Ellcey <sje at cup dot hp dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 12 Jun 2009 23:22:51 +0000 (UTC)
- Subject: Re: RFC: Two possible fixes for g++.dg/torture/pr32950.C failure
- References: <200906122245.n5CMjO801391@lucas.cup.hp.com>
On Fri, 12 Jun 2009, Steve Ellcey wrote:
> But I am wondering about the optmization done in by compute_record_mode
> which causes this situation, this function changes the mode of a
> structure from BLKmode to the mode of its field when a structure only
> has 1 field (with some exceptions).
> It seems like tree-sra should be handling this now and that it might be
> time to get rid of this ad-hoc optimization done during expansion.
It is or was the case that the modes of structures determined the function
calling ABI on some targets and changing them would break the ABI. If
this is still the case then you can't change them without identifying
every affected target and making it replicate the former rules when
determining how to pass a structure.
Joseph S. Myers