[#6637] gcc test case fails when built with gcc 4.5 while pass built with gcc 4.3

Document created by Aaronwu Employee on Oct 17, 2013
Version 1Show Document
  • View in full screen mode

[#6637] gcc test case fails when built with gcc 4.5 while pass built with gcc 4.3

Submitted By: Mingquan Pan

Open Date

2011-06-13 05:00:04     Close Date

2012-05-15 01:56:55

Priority:

Medium     Assignee:

Mingquan Pan

Stuart Henderson

Board:

N/A     Silicon Revision:

Resolution:

Fixed     Fixed In Release:

N/A

Processor:

ALL     

Host Operating System:

toolchain rev.:

    kernel rev.:

State:

Closed     Found In Release:

2011R1

Is this bug repeatable?:

N/A     

Summary: gcc test case fails when built with gcc 4.5 while pass built with gcc 4.3

Details:

 

gcc test case fails when built with gcc 4.5 while pass built with gcc 4.3 by testing on bf548-ezkit.

 

The diff file compared with last release gcc 4.3 results is like:

 

Tests that now fail, but worked before:

 

bfin-uclinux: gcc.c-torture/execute/pr23047.c execution,  -O0

bfin-uclinux: gcc.c-torture/execute/pr23047.c execution,  -O1

bfin-uclinux: gcc.c-torture/execute/pr23047.c execution,  -O2

bfin-uclinux: gcc.c-torture/execute/pr23047.c execution,  -Os

bfin-uclinux: gcc.dg/fold-abs-2.c execution test

bfin-uclinux: gcc.dg/pr34263.c scan-tree-dump optimized "Cleaned-up latch block of loop with single BB"

bfin-uclinux: gcc.dg/tree-ssa/foldaddr-1.c scan-tree-dump-times original "return 0" 1

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "2 loops found" 1

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "3 loops found" 2

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "4 loops found" 2

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "Found latch edge" 5

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "Merged latch edges" 2

bfin-uclinux: gcc.dg/tree-ssa/ssa-ccp-14.c scan-tree-dump-times optimized "if" 0

bfin-uclinux: gcc.dg/uninit-I.c  (test for warnings, line 6)

bfin-uclinux: gcc.dg/Wstrict-aliasing-float-ptr-int-obj.c  (test for warnings, line 14)

bfin-uclinux: gcc.dg/Wstrict-aliasing-float-ptr-int-obj.c  (test for warnings, line 16)

bfin-uclinux: gcc.target/bfin/builtins/abs_fr1x16-2.c execution, -ffast-math -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/abs_fr1x16-2.c execution, -ffast-math -O2

bfin-uclinux: gcc.target/bfin/builtins/abs_fr1x16-2.c execution, -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/abs_fr1x16-2.c execution, -O1

bfin-uclinux: gcc.target/bfin/builtins/abs_fr1x16-2.c execution, -O2

bfin-uclinux: gcc.target/bfin/builtins/abs_fr1x16-2.c execution, -O3

bfin-uclinux: gcc.target/bfin/builtins/abs_fr1x16-2.c execution, -Os

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation,

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -ffast-math -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -ffast-math -O2

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -O0

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -O1

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -O2

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -O3

bfin-uclinux: gcc.target/bfin/builtins/circptr.c compilation, -Os

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution,

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -ffast-math -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -ffast-math -O2

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -O0

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -O1

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -O2

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -O3

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-1.c execution, -Os

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-2.c execution,

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-2.c execution, -ffast-math -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-2.c execution, -ffast-math -O2

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-2.c execution, -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-2.c execution, -O0

bfin-uclinux: gcc.target/bfin/builtins/cpmlx_msu_fr2x16-2.c execution, -O1

....

 

 

detailed log is like:

 

Executing on host: bfin-uclinux-gcc /home/test/work/cruise/checkouts/toolchain/gcc-4.5/gcc/testsuite/gcc.c-torture/execute/pr23047.c  -w  -O1  -fwrapv -DSTACK_SIZE=75000 -DNO_TRAMPOLINES -Wl,-elf2flt=-s80000  -lm  -mcpu=bf548-0.2  -o /home/test/work/cruise/temp/regtest_build4.5/gcc_build/testsuite/pr23047.x1    (timeout = 300)

spawn bfin-uclinux-gcc /home/test/work/cruise/checkouts/toolchain/gcc-4.5/gcc/testsuite/gcc.c-torture/execute/pr23047.c -w -O1 -fwrapv -DSTACK_SIZE=75000 -DNO_TRAMPOLINES -Wl,-elf2flt=-s80000 -lm -mcpu=bf548-0.2 -o /home/test/work/cruise/temp/regtest_build4.5/gcc_build/testsuite/pr23047.x1^M

PASS: gcc.c-torture/execute/pr23047.c compilation,  -O1

Executing on bfin-uclinux: /tmp/pr23047.x1.3350    (timeout = 300)

Executing on bfin-uclinux: rm -f  /tmp/pr23047.x1.3350    (timeout = 300)

Executed /home/test/work/cruise/temp/regtest_build4.5/gcc_build/testsuite/pr23047.x1, status 1

ABRT

FAIL: gcc.c-torture/execute/pr23047.c execution,  -O1

 

 

kernel is :

 

Testing is running on target with following versions:

kernel:    Linux release 2.6.34.7-ADI-2010R1-svn9508, build #10 Tue May 31 19:56:06 CST 2011

toolchain: bfin-uclinux-gcc release gcc version 4.5.2 (ADI-trunk/svn-5440)

user-dist: release svn-9977, build #4 Tue May 31 19:55:25 CST 2011

 

 

whole list for the failed gcc case and log file are attached.

 

Follow-ups

 

--- Stuart Henderson                                         2011-06-22 09:34:57

reassigning this to me.

 

--- Stuart Henderson                                         2011-06-22 13:29:28

most of these have been fixed on trunk but there's the notable exceptions of:

 

gcc.dg/pr34263.c

gcc.dg/tree-ssa/foldaddr-1.c

gcc.dg/tree-ssa/loop-25.c

gcc.dg/tree-ssa/ssa-ccp-14.c

Which aren't doing optimisations that are expected.

 

gcc.dg/uninit-I.c

not warning of an uninitialised variable in a struct.

 

gcc.target/bfin/builtins/circptr.c

the patch for the circptr builtin needs re-implemented due to changes in

gcc-4.5, so it isn't in yet.

 

--- Stuart Henderson                                         2011-08-10 06:07:52

since everything except the following five tests are now fixed and they are all

XFAILs for 4.5, i'm moving this bug to medium.

gcc.dg/pr34263.c

gcc.dg/tree-ssa/foldaddr-1.c

gcc.dg/tree-ssa/loop-25.c

gcc.dg/tree-ssa/ssa-ccp-14.c

Which aren't doing optimisations that are expected.

 

gcc.dg/uninit-I.c

not warning of an uninitialised variable in a struct.

 

--- Mingquan Pan                                             2011-10-19 04:26:08

Testing with the 2011R1-rc2, the current toolchain 4.5 log compared with 4.3 is

like:

 

#########################################

#

# Comparing gcc-4.5.sum

#

########################################

Tests that now fail, but worked before:

 

bfin-uclinux: gcc.dg/pr34263.c scan-tree-dump optimized "Cleaned-up latch

block of loop with single BB"

bfin-uclinux: gcc.dg/tree-ssa/foldaddr-1.c scan-tree-dump-times original

"return 0" 1

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "2

loops found" 1

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "3

loops found" 2

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile "4

loops found" 2

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile

"Found latch edge" 5

bfin-uclinux: gcc.dg/tree-ssa/loop-25.c scan-tree-dump-times profile

"Merged latch edges" 2

bfin-uclinux: gcc.dg/tree-ssa/ssa-ccp-14.c scan-tree-dump-times optimized

"if" 0

bfin-uclinux: gcc.dg/uninit-I.c  (test for warnings, line 6)

bfin-uclinux: gcc.dg/Wstrict-aliasing-float-ptr-int-obj.c  (test for warnings,

line 14)

bfin-uclinux: gcc.dg/Wstrict-aliasing-float-ptr-int-obj.c  (test for warnings,

line 16)

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation,

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -ffast-math

-mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -ffast-math -O2

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -mfast-fp -O2

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -O0

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -O1

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -O2

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -O3

bfin-uclinux: gcc.target/bfin/builtins/hisilh.c compilation, -Os

bfin-uclinux: tmpdir-gcc.dg-struct-layout-1/t025

c_compat_x_tst.o-c_compat_y_tst.o execute

bfin-uclinux: tmpdir-gcc.dg-struct-layout-1/t027

c_compat_x_tst.o-c_compat_y_tst.o execute

bfin-uclinux: tmpdir-gcc.dg-struct-layout-1/t028

c_compat_x_tst.o-c_compat_y_tst.o execute

 

 

Some of the case failed previously now doesn't seen after fix, while some new

failure comes up.

 

--- Mingquan Pan                                             2011-10-19 04:28:42

when testing the 2011r1-rc2 4.5 toolchain on linux-uclibc, there are more

failure:

 

#########################################

#

# Comparing gcc-4.5.sum

#

########################################

Tests that now fail, but worked before:

 

bfin-linux-uclibc: gcc.c-torture/execute/930608-1.c compilation,  -O3 -g

bfin-linux-uclibc: gcc.c-torture/execute/930702-1.c compilation,  -O3 -g

bfin-linux-uclibc: gcc.c-torture/execute/991112-1.c compilation,  -O3 -g

 

--- Stuart Henderson                                         2012-04-10 10:27:01

bfin-uclinux:

gcc.dg/pr34263.c - Upstream bug, fixed in 4.7

gcc.dg/tree-ssa/foldaddr-1.c - Upstream bug, fixed in 4.7

gcc.dg/tree-ssa/loop-25.c - Upstream bug, fixed in 4.7

gcc.dg/tree-ssa/ssa-ccp-14.c - Upstream bug, fixed in 4.7

gcc.dg/uninit-I.c - Upstream bug, fixed in 4.7

gcc.dg/Wstrict-aliasing-float-ptr-int-obj.c - Upstream bug, fixed in 4.7

gcc.target/bfin/builtins/hisilh.c - test bug, fixed.

bfin-linux-uclibc:

gcc.c-torture/execute/930608-1.c - fixed.

gcc.c-torture/execute/930702-1.c - fixed.

gcc.c-torture/execute/991112-1.c - fixed.

 

Since the upstream bugs are marked as XFAILs and not particularly important, i

suggest we just leave them as such.

 

Leaving...

bfin-uclinux:

tmpdir-gcc.dg-struct-layout-1/t025

tmpdir-gcc.dg-struct-layout-1/t027

tmpdir-gcc.dg-struct-layout-1/t028

which i'm investigating now.

 

--- Stuart Henderson                                         2012-04-18 07:08:08

The final three tests have been fixed up now.

 

--- Mingquan Pan                                             2012-05-15 02:56:57

Looking through the long list of once failing test cases, they are now either

passed or XFAIL as Stuart said in trunk head uclinux 4.5 toolchain testing. So

close it here.

 

The current 4.5 uclinux gcc testing results:

 

                === gcc Summary ===

 

# of expected passes            56603

# of unexpected failures        295

# of expected failures          95

# of unresolved testcases       16

# of unsupported tests          739

 

 

 

    Files

    Changes

    Commits

    Dependencies

    Duplicates

    Associations

    Tags

 

File Name     File Type     File Size     Posted By

gcc_0613.tar.gz    application/gzipped-tar    1405789    Mingquan Pan

Attachments

Outcomes