Bug 17689

Summary: [4.0 Regression] New testsuite failures indicate signal unwinding is broken
Product: gcc Reporter: Andrew Haley <aph>
Component: tree-optimizationAssignee: Andrew Haley <aph>
Status: RESOLVED FIXED    
Severity: critical CC: dnovillo, gcc-bugs, java-prs
Priority: P2 Keywords: wrong-code
Version: 4.0.0   
Target Milestone: 4.0.0   
Host: i686-pc-linux-gnu Target: i686-pc-linux-gnu
Build: i686-pc-linux-gnu Known to work:
Known to fail: Last reconfirmed: 2004-09-27 12:33:20
Bug Depends on:    
Bug Blocks: 17574    

Description Andrew Haley 2004-09-27 10:08:25 UTC
 
Comment 1 Andrew Haley 2004-09-27 10:12:42 UTC
See http://gcc.gnu.org/ml/gcc-regression/2004-09/msg00184.html.

FAIL: Array_3 execution - source compiled test
FAIL: Array_3 -O3 execution - source compiled test
FAIL: Invoke_1 execution - source compiled test
FAIL: Invoke_1 execution - bytecode->native test
FAIL: Invoke_1 -O3 execution - source compiled test
FAIL: Invoke_1 -O3 execution - bytecode->native test

These tests were specially written to test unwinding from SEGV signals.
Comment 2 Andrew Pinski 2004-09-27 12:33:19 UTC
Confirmed via: http://gcc.gnu.org/ml/gcc-testresults/2004-09/msg01195.html
Comment 3 Andrew Pinski 2004-09-27 12:34:06 UTC
It works on powerpc-apple-darwin though: <http://gcc.gnu.org/ml/gcc-testresults/2004-09/
msg01185.html> so this might be a target problem.
Comment 4 Andrew Haley 2004-09-27 12:51:59 UTC
It's i386 specific.  I'm working on a patch now.
Comment 5 Andrew Pinski 2004-09-28 13:49:09 UTC
Fixed correctly by: <http://gcc.gnu.org/ml/gcc-patches/2004-09/msg02867.html>.