In my gcc autotester, I check cris-elf using cris-sim (all
sources are in gcc and src trees) on a Fedora 8 amd64.
I'm also using amd64 Debian machines in the gcc build farm, but
not for autotesting. That setup doesn't reveal that many
differences (more later), because I mentioned, Debian tracks
dejagnu closer.
This isn't IMHO a real a difference in test-results, but maybe
it's cause for questions (excerpt of diff between log files):
@@ -20295,9 +20231,9 @@ Schedule of variations:
cris-sim
Running target cris-sim
-Using ~/dejagnuboards/cris-sim.exp as board description file for target.
-Using /usr/share/dejagnu/config/sim.exp as generic interface file for target.
-Using /usr/share/dejagnu/baseboards/basic-sim.exp as board description file
for target.
+Using /tmp/hptstdgp/share/dejagnu/baseboards/cris-sim.exp as board description
file for target.
+Using /tmp/hptstdgp/share/dejagnu/config/sim.exp as generic interface file for
target.
+Using /tmp/hptstdgp/share/dejagnu/baseboards/basic-sim.exp as board
description file for target.
Using /tmp/hpautotest-gcc1/gcc/gcc/testsuite/config/default.exp as
tool-and-target-specific interface file.
Running /tmp/hpautotest-gcc1/gcc/gcc/testsuite/gfortran.dg/dg.exp ...
FAIL: gfortran.dg/PR19872.f -O0 execution test
@@ -20333,53 +20269,29 @@ FAIL: gfortran.dg/backspace_9.f -O3 -fo
FAIL: gfortran.dg/backspace_9.f -O3 -g execution test
FAIL: gfortran.dg/backspace_9.f -Os execution test
FAIL: gfortran.dg/chmod_1.f90 -O0 (test for excess errors)
-WARNING: gfortran.dg/chmod_1.f90 -O0 compilation failed to produce executable
FAIL: gfortran.dg/chmod_1.f90 -O1 (test for excess errors)
-WARNING: gfortran.dg/chmod_1.f90 -O1 compilation failed to produce executable
FAIL: gfortran.dg/chmod_1.f90 -O2 (test for excess errors)
... more similar "loss" of warnings.
and
@@ -20975,6 +20863,7 @@ FAIL: gfortran.fortran-torture/execute/s
# of expected passes 22966
# of unexpected failures 233
# of expected failures 1
+# of unresolved testcases 48
# of unsupported tests 116
/tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/testsuite/gfortran/../../gfortran
version 4.3.0 20080212 (experimental) [trunk revision 132266] (GCC)
The "excess errors" are ld error undefined-reference messages
due to missing functions. (The test should probably be skipped
as unsupported.)
The gfortran.dg/chmod_1.f90 testcase is marked:
! { dg-do run }
! { dg-options "-std=gnu" }
so I think the difference is an improvement: the "run" part is
now properly marked as "unresolved" and doesn't cause spurious
warnings when the test-case doesn't link. All good.
brgds, H-P
_______________________________________________
DejaGnu mailing list
DejaGnu@gnu.org
http://lists.gnu.org/mailman/listinfo/dejagnu
