Bug 47313 - [4.6 Regression] ICE: PHI argument is not a GIMPLE value
Summary: [4.6 Regression] ICE: PHI argument is not a GIMPLE value
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: middle-end (show other bugs)
Version: 4.6.0
: P3 normal
Target Milestone: 4.6.0
Assignee: Richard Biener
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-01-16 09:33 UTC by Marc Glisse
Modified: 2011-01-17 14:48 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed: 2011-01-16 18:08:47

Attachments
testcase (569 bytes, text/x-c++src)
2011-01-16 09:35 UTC, Marc Glisse 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Marc Glisse 2011-01-16 09:33:27 UTC 
$ g++ -O2 -c a.cc
a.cc: In function 'Periodic_3_Delaunay_triangulation_3< <template-parameter-1-1> >::Vertex_handle Periodic_3_Delaunay_triangulation_3< <template-parameter-1-1> >::nearest_vertex() const [with <template-parameter-1-1> = Triangulation_data_structure_3, Periodic_3_Delaunay_triangulation_3< <template-parameter-1-1> >::Vertex_handle = internal::CC_iterator<Compact_container<int>, false>]':
a.cc:51:1: error: PHI argument is not a GIMPLE value
iftmp.0_2 = PHI <<retval>(2), D.2301_3(D)(3)>

<retval>
a.cc:51:1: internal compiler error: verify_stmts failed
Comment 1 Marc Glisse 2011-01-16 09:35:27 UTC 
Created attachment 22981 [details]
testcase

Try again to attach it, bugzilla apparently failed to upload it in the initial bug report.
Comment 2 H.J. Lu 2011-01-16 18:08:47 UTC 
It is caused by revision 161898:

http://gcc.gnu.org/ml/gcc-cvs/2010-07/msg00252.html
Comment 3 Richard Biener 2011-01-17 12:06:36 UTC 
For some reason IPA-SRA breaks it.  Looking at it.
Comment 4 Richard Biener 2011-01-17 14:48:38 UTC 
Author: rguenth
Date: Mon Jan 17 14:48:35 2011
New Revision: 168907

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168907
Log:
2011-01-17  Richard Guenther  <rguenther@suse.de>

	PR tree-optimization/47313
	* tree-inline.c (tree_function_versioning): Move DECL_RESULT
	handling before copying the body.  Properly deal with
	by-reference result in SSA form.

	* g++.dg/torture/pr47313.C: New testcase.

Added:
    trunk/gcc/testsuite/g++.dg/torture/pr47313.C
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree-inline.c
Comment 5 Richard Biener 2011-01-17 14:48:47 UTC 
Fixed.