From mboxrd@z Thu Jan  1 00:00:00 1970
From: Mike Snitzer <snitzer@redhat.com>
Date: Wed, 17 Mar 2010 10:30:53 -0400
Subject: testsuite: get stacktrace if test drops core
In-Reply-To: <87vdcv8do7.fsf@twilight.int.mornfall.net.>
References: <20100316204454.GA25921@redhat.com>
	<87vdcv8do7.fsf@twilight.int.mornfall.net.>
Message-ID: <20100317143053.GA3836@redhat.com>
List-Id: <lvm-devel.redhat.com>
To: lvm-devel@redhat.com
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

On Wed, Mar 17 2010 at  6:07am -0400,
Petr Rockai <prockai@redhat.com> wrote:

> Hi,
> 
> Mike Snitzer <snitzer@redhat.com> writes:
> > Requires lvm be built with debugging (-g).
> > Also requires ulimit -c be non-zero (to drop core file).
> >
> > diff --git a/test/test-utils.sh b/test/test-utils.sh
> > index 6606c56..06b9636 100644
> > --- a/test/test-utils.sh
> > +++ b/test/test-utils.sh
> > @@ -24,6 +24,21 @@ STACKTRACE() {
> >  		echo "$i ${FUNC}() called from ${BASH_SOURCE[$i]}:${BASH_LINENO[$i]}"
> >  		i=$(($i + 1));
> >  	done
> > +
> > +	# Attempt to get a stacktrace if a core file exists
> > +	# and the lvm binary was built with debugging
> > +	TEST_LVM_BINARY=$(dirname $(which lvm))/../../tools/lvm
> > +	TEST_LVM_CORE=`ls core* | head -1`
> > +	GDB_BINARY=`which gdb`
> > +	READELF_BINARY=`which readelf`
> > +	if [ -n "$TEST_LVM_CORE" -a -n "$GDB_BINARY" -a -n "$READELF_BINARY" ]; then
> > +		if $READELF_BINARY -S $TEST_LVM_BINARY 2>&1 | grep -q .debug_info; then
> > +			echo bt > gdb_commands.txt
> > +			echo l >> gdb_commands.txt
> > +			echo quit >> gdb_commands.txt
> > +			$GDB_BINARY -batch -c $TEST_LVM_CORE -x gdb_commands.txt $TEST_LVM_BINARY
> > +		fi
> > +	fi
> >  }
> >  
> >  init_udev_transaction() {
> Looks good to me. Maybe change "bt" to "bt full"? That would also list
> local variables on the stack, so more useful data.

OK, I'll use "bt full".

> (Maybe even make it "thread apply all bt full", although we don't
> currently use threads here and we don't handle dmeventd coredumps
> either... maybe later.)

Right, we can always change it as needed.

> Otherwise, ACK.

Thanks,
Mike