This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: gas --statistics segfault
On Mon, Feb 06, 2006 at 08:31:06AM +0800, Dasn Clainst wrote:
> On Sun, Feb 05, 2006 at 08:09:35AM -0800, H. J. Lu wrote:
> > On Sun, Feb 05, 2006 at 09:02:19PM +0800, Dasn Clainst wrote:
> > >
> > > Hi, guys.
> > >
> > > In my box, 'as --statistics' sometimes segfault.
> > > -------------------------------------------------------------
> > > $ uname -msr
> > > OpenBSD 3.8 i386
> > > $ echo "main(){}" > p.c
> > > $ cat -n test.sh
> > > 1 #!/bin/sh
> > > 2 count=100;failed=0;i=0
> > > 3 cc -S p.c -o p.s
> > > 4 while [ $i -lt $count ]
> > > 5 do
> > > 6 ./as-new --statistics p.s > /dev/null 2>&1
> > > 7 if [ $? -ne 0 ]; then
> > > 8 let failed=failed+1
> > > 9 fi
> > > 10 let i=i+1
> > > 11 done
> > > 12 echo "$failed/$count failed."
> > > 13 rm -rf as-new.core a.out
> > > $ sh test.sh
> > > 100/100 failed.
> >
> > You were making it harder for people to reproduce it. Please open a bug
> > report at
> >
> > http://www.sourceware.org/bugzilla/
> >
> > with a testcase.
> >
> >
> > H.J.
>
> Usually, --statistics will cause segfault, so no need testcase. The
> test.sh is used to find how often segfaults will produce (usually 100%,
I can't even cut/paste your test.sh since you used "cat -n".
H.J.