[RFA/RFC] no shared library support warning on powerpc-elf

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

[RFA/RFC] no shared library support warning on powerpc-elf

Joel Brobecker
Hello,

When running a program on the powerpc-elf simulator, I get the following
warning:

        (gdb) run
        Starting program: /[...]/ex/other
 !!! -> warning: no shared library support for this OS / ABI
       
        Breakpoint 1, other () at other.adb:12
        12      end Other;

After having a look, it appears to be a side-effect of a patch
that moved the post-target_create_inferior processing from the
target layer back to the core layer. I think the interesting
part of the ChangeLog entry is the following:

        * infcmd.c (post_create_inferior): Call solib_add,
        solib_create_inferior_hook, and re_enable_breakpoints_in_shlibs.

Before this change was made the remote-sim code was in charge
of doing that, and this code was not calling the solib create_inferior
hook.

In the case of powerpc-elf, GDB links in solib.c and solib-svr4.c.
I don't think the bareboard case is supported by these units, so
I think the sensible thing to do is to remove them from the closure
and not build them.

Opinions?

2007-05-08  Joel Brobecker  <[hidden email]>

        * config/powerpc/ppc-sim.mt (TDEPFILES): Remove solib.o and
        solib-svr4.o.

Is it possible to run the testsuite? The compilers I have use the
powerpc-elf- prefix (eg: powerpc-elf-gcc).

Thanks,
--
Joel

ppc-elf.diff (669 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [RFA/RFC] no shared library support warning on powerpc-elf

Daniel Jacobowitz-2
On Tue, May 08, 2007 at 03:12:53PM -0700, Joel Brobecker wrote:

> Hello,
>
> When running a program on the powerpc-elf simulator, I get the following
> warning:
>
>         (gdb) run
>         Starting program: /[...]/ex/other
>  !!! -> warning: no shared library support for this OS / ABI
>        
>         Breakpoint 1, other () at other.adb:12
>         12      end Other;

> In the case of powerpc-elf, GDB links in solib.c and solib-svr4.c.
> I don't think the bareboard case is supported by these units, so
> I think the sensible thing to do is to remove them from the closure
> and not build them.

Except this interferes with using one GDB for both powerpc-elf and
powerpc-linux.  So it's not a great solution.

Maybe the warning should simply be removed?  I don't think it's very
valuable.  If shared library support is really missing, either the
testsuite will catch it or a user will eventually notice.

> Is it possible to run the testsuite? The compilers I have use the
> powerpc-elf- prefix (eg: powerpc-elf-gcc).

Yes, using the powerpc-eabisim board; but it's easier if you build
everything using --target=powerpc-eabisim.  powerpc-elf-gcc may need
some additional options, like -msim.

--
Daniel Jacobowitz
CodeSourcery
Reply | Threaded
Open this post in threaded view
|

Re: [RFA/RFC] no shared library support warning on powerpc-elf

Joel Brobecker
> > In the case of powerpc-elf, GDB links in solib.c and solib-svr4.c.
> > I don't think the bareboard case is supported by these units, so
> > I think the sensible thing to do is to remove them from the closure
> > and not build them.
>
> Except this interferes with using one GDB for both powerpc-elf and
> powerpc-linux.  So it's not a great solution.

Argh! I didn't notice that powerpc-linux was also affected.
Thanks for catching this...

> Maybe the warning should simply be removed?  I don't think it's very
> valuable.  If shared library support is really missing, either the
> testsuite will catch it or a user will eventually notice.

I agree it's not very useful in this case. It's not something that
is conditional on the environment, but rather a known limitation.
Seeing the debugger emit this warning every time can make the user
think that something's wrong with his setup.

So, I'm all for removing it. Any objection from anyone?

--
Joel
Reply | Threaded
Open this post in threaded view
|

Re: [RFA/RFC] no shared library support warning on powerpc-elf

Daniel Jacobowitz-2
On Fri, May 11, 2007 at 03:50:01PM -0700, Joel Brobecker wrote:

> > Maybe the warning should simply be removed?  I don't think it's very
> > valuable.  If shared library support is really missing, either the
> > testsuite will catch it or a user will eventually notice.
>
> I agree it's not very useful in this case. It's not something that
> is conditional on the environment, but rather a known limitation.
> Seeing the debugger emit this warning every time can make the user
> think that something's wrong with his setup.
>
> So, I'm all for removing it. Any objection from anyone?

There weren't any; could you post a patch (maybe in a new Subject: in
case someone else wants to comment?) removing the warning?  If no one
objects I'll approve it.

--
Daniel Jacobowitz
CodeSourcery