Current-Users archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
gcc/testsuite/gcc.c-torture/execute
Hi, all.
I'm sorry if this is well known to many people :-)
GCC has testsuites using dejagnu. I noticed some testsuites work
without dejagnu. In gcc/testsuite/gcc.c-torture/execute, there are a lot
of .c programs. All of the .c are compilable.
So, I made a small shell script like:
============= tst.sh =========
#!/bin/sh
for i in $*; do
gcc -O2 $i -o pr
./pr
if [ $? != 0 ]; then
echo $i: FAIL;
else
echo $i: Success;
fi
done
==============================
And:
cd gcc/testsuite/gcc.c-torture/execute
script test.log
sh tst.sh *.c
exit
grep FAIL test.log
For example, the result on evbarm (netbsd-6) shows:
> 980709-1.c: FAIL
> builtin-bitops-1.c: FAIL
> eeprof-1.c: FAIL
> float-floor.c: FAIL
> pr22493-1.c: FAIL
> pr23047.c: FAIL
> pr42544.c: FAIL
> pr44468.c: FAIL
> pr47925.c: FAIL
> pr48197.c: FAIL
> pr48809.c: FAIL
> pr48973-1.c: FAIL
> pr48973-2.c: FAIL
> pr49039.c: FAIL
> pr49123.c: FAIL
> pr49161.c: FAIL
> pr49279.c: FAIL
> pr49644.c: FAIL
on i386:
> 20101011-1.c: FAIL
> 980709-1.c: FAIL
> builtin-bitops-1.c: FAIL
> eeprof-1.c: FAIL
> float-floor.c: FAIL
> pr22493-1.c: FAIL
> pr23047.c: FAIL
> pr42544.c: FAIL
> pr44468.c: FAIL
> pr47237.c: FAIL
> pr47925.c: FAIL
> pr48197.c: FAIL
> pr48809.c: FAIL
> pr48973-1.c: FAIL
> pr48973-2.c: FAIL
> pr49039.c: FAIL
> pr49123.c: FAIL
> pr49161.c: FAIL
> pr49279.c: FAIL
> pr49644.c: FAIL
The difference between -O2 and -O0 on evbarm is:
> % diff -u o1 o2
> --- o1 2012-09-20 11:00:24.000000000 +0900
> +++ o2 2012-09-20 11:00:28.000000000 +0900
> @@ -2,17 +2,6 @@
> builtin-bitops-1.c: FAIL
> eeprof-1.c: FAIL
> float-floor.c: FAIL
> -pr22493-1.c: FAIL
> -pr23047.c: FAIL
> pr42544.c: FAIL
> -pr44468.c: FAIL
> -pr47925.c: FAIL
> pr48197.c: FAIL
> -pr48809.c: FAIL
> -pr48973-1.c: FAIL
> -pr48973-2.c: FAIL
> -pr49039.c: FAIL
> -pr49123.c: FAIL
> -pr49161.c: FAIL
> -pr49279.c: FAIL
> pr49644.c: FAIL
We can use src/external/gpl3/gcc/dist/gcc/testsuite, but the number of
testcase is fewer than gcc 4.7.1's. The above results in this mail is
from gcc 4.7.1's testsuite. The name of many testcases have PR number,
so it's easy to find patches to fix.
I think we are happy if we can integrate this to /usr/tests.
--
-----------------------------------------------
SAITOH Masanobu (msaitoh%execsw.org@localhost
msaitoh%netbsd.org@localhost)
Home |
Main Index |
Thread Index |
Old Index