This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix PR86179, stale STMT_VINFO_RELATED_STMT
- From: Richard Sandiford <richard dot sandiford at arm dot com>
- To: Richard Biener <rguenther at suse dot de>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Tue, 19 Jun 2018 09:07:25 +0100
- Subject: Re: [PATCH] Fix PR86179, stale STMT_VINFO_RELATED_STMT
- References: <alpine.LSU.2.20.1806190849130.5043@zhemvz.fhfr.qr>
Richard Biener <rguenther@suse.de> writes:
> Pattern recognition can leave stale STMT_VINFO_RELATED_STMT entries
> around which confuses the correctness asserting I put into
> get_{earlier,later}_stmt. The following makes sure to clear
> those again. To make things more complicated pattern recognition
> sometimes scans forward ... which runs into the issue that a later
> attempt to detect a pattern might overwrite an earlier pattern.
> The patch disallows that as well.
>
> Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
>
> I've just noticed a larger series from Richard S. and hope this
> doesn't interfere.
Not too much, although the unposted parts get rid of the both the
forward replacements and that use of STMT_VINFO_RELATED_STMT.
Thanks,
Richard