Bug 30087 - [4.3 Regression] regressions in the gfortran testsuite with -ftree-vectorize
Summary: [4.3 Regression] regressions in the gfortran testsuite with -ftree-vectorize
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: tree-optimization (show other bugs)
Version: 4.3.0
: P3 normal
Target Milestone: 4.3.0
Assignee: Not yet assigned to anyone
URL:
Keywords: ice-on-valid-code
Depends on:
Blocks:
 
Reported: 2006-12-06 16:43 UTC by Tobias Schlüter
Modified: 2006-12-14 06:48 UTC (History)
3 users (show)

See Also:
Host: i386-darwin
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Tobias Schlüter 2006-12-06 16:43:36 UTC
Running the gfortran testsuite with "-O2 -ftree-vectorize" started giving a number of failures in the span of the last 3-4 days.  Namely, the testcases
 gfortran.dg/actual_array_constructor_3.f90
 gfortran.dg/array_constructor_8.f90
now lead to ICEs in the compiler.

This blocks my effort to make the gfortran testsuite run with -ftree-optimize.
Comment 1 Tobias Burnus 2006-12-08 22:25:11 UTC
Backtrace of the first (second looks similar)

#5  0x000000000077ae99 in tree_class_check_failed (node=0x2b9044b37240, cl=tcc_expression, file=<value optimized out>, line=2578,
    function=0xa8b500 "vect_permute_store_chain") at gcc/tree.c:6409
#6  0x00000000008b7303 in vect_permute_store_chain (dr_chain=0xeb5830, length=2, stmt=0x2b9044b37240, bsi=0x7fff66b46230,
    result_chain=0x7fff66b46158) at gcc/tree-vect-transform.c:2578
#7  0x00000000008bfcc3 in vectorizable_store (stmt=0x2b9044b37240, bsi=dwarf2_read_address: Corrupted DWARF expression.
gcc/tree-vect-transform.c:2835

Comment 2 Ira Rosen 2006-12-10 11:01:29 UTC
I wonder if this is related to vectorizer testcases failures (http://gcc.gnu.org/ml/gcc/2006-12/msg00261.html), since this regression occurred at about the same time.

Ira 
Comment 3 Tobias Schlüter 2006-12-13 20:46:13 UTC
Fixed by Ira in r119730, thanks.