This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: set print object on should affect MI varobjs (PR mi/13393)
- From: "Ulrich Weigand" <uweigand at de dot ibm dot com>
- To: jan dot kratochvil at redhat dot com (Jan Kratochvil)
- Cc: xgsa at yandex dot ru (xgsa), tromey at redhat dot com (Tom Tromey), gdb-patches at sourceware dot org
- Date: Mon, 23 Jul 2012 19:21:18 +0200 (CEST)
- Subject: Re: set print object on should affect MI varobjs (PR mi/13393)
Jan Kratochvil wrote:
> On Sat, 14 Apr 2012 14:28:26 +0200, xgsa wrote:
> > >Yeah. I'm sorry about the delay. The patch is ok, please check it in.
> >
> > Checked in.
>
> Fedora 17 x86_64:
> +FAIL: gdb.mi/mi-var-rtti.exp: list children of s.public in type_update_when_use_rtti
> +FAIL: gdb.mi/mi-var-rtti.exp: list children of s.ptr (without RTTI) in type_update_when_use_rtti
> +FAIL: gdb.mi/mi-var-rtti.exp: list children of s.ptr.public (without RTTI) in type_update_when_use_rtti
Yes, I'm seeing similar failures on a number of platforms (arm, spu, s390x).
The reason seems to be that the test case is attempting to display local
variables that are not yet initialized, and thus their contents are random.
For C++ objects, this includes a random vtable pointer ...
Now, this particular sub-test appears to do this deliberately; all the other
sub-tests only create the varobj *after* their corresponding objects were
initialized. I'm not quite sure what exactly the test is supposed to be
testing, but I guess it ought to be done differently ...
Bye,
Ulrich
--
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE
Ulrich.Weigand@de.ibm.com