Bug 42644 - [4.5 Regression][graphite] wrong code for -fgraphite-identify in 183.equake
Summary: [4.5 Regression][graphite] wrong code for -fgraphite-identify in 183.equake
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: tree-optimization (show other bugs)
Version: 4.5.0
: P1 normal
Target Milestone: 4.5.0
Assignee: Tobias Grosser
URL:
Keywords: wrong-code
Depends on: 42130
Blocks:
  Show dependency tree / graph
 
Reported: 2010-01-07 01:42 UTC by Janis Johnson
Modified: 2010-03-08 17:55 UTC (History)
2 users (show)

See Also:
Host:
Target: powerpc64-linux
Build:
Known to work: 4.4.2
Known to fail:
Last reconfirmed: 2010-01-20 15:50:56

Attachments
minimized executable testcase (600 bytes, text/plain)
2010-01-07 01:43 UTC, Janis Johnson 		Details
Reduced test case for this PR (343 bytes, text/plain)
2010-01-20 15:26 UTC, Ramakrishna Upadrasta 		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 Janis Johnson 2010-01-07 01:42:35 UTC 
Test 183.equake from SPEC CPU2000 gets wrong results on powerpc64-linux (-m32 or -m64) when compiled with either "-O2 -fgraphite-identity" or "-O2 -floop-parallelize-all".  I'll attach an executable testcase that demonstrates the failure.  Test 178.galgel fails with the same options but I haven't investigated it.

The failure starts with the merge from the Graphite branch for r150301.  With options "-O2 -fgraphite-identity" the minimized testcase and the two CPU2000 tests pass with r150248, the last revision before that merge that builds on powerpc64-linux.  They also pass with GCC 4.4.2 with "-O2 -fgraphite-identity".
Comment 1 Janis Johnson 2010-01-07 01:43:35 UTC 
Created attachment 19495 [details]
minimized executable testcase
Comment 2 Sebastian Pop 2010-01-14 08:44:49 UTC 
This might be a duplicate of PR42130.
Comment 3 Ramakrishna Upadrasta 2010-01-20 15:26:08 UTC 
Created attachment 19666 [details]
Reduced test case for this PR

We (Tobias, Li and Ramakrishna) have reduced the test case further. This file shows the problem.
Comment 4 Sebastian Pop 2010-03-08 17:55:45 UTC 
Fixed by http://gcc.gnu.org/viewcvs?view=revision&revision=157286