This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: kernel DSO
On Wed, Sep 15, 2004 at 01:37:40PM +0100, James Cownie wrote:
>
> > The debugger can surely get to the DSO mapped in RAM, via the AT entry
> > for example, or glibc could export the entry point... adding more cruft
> > to /proc will probably not happen.
>
> Yes, it _can_ (I've just been working on exactly that code, read
> /proc/<pid>/auxv, use ptrace to extract the DSO), but it's a lot of pain
> and code for something which could be completely transparent to tools
> like debuggers.
>
> I certainly believe that the number of lines of code in the debugger to
> handle this well exceed the number which would be required to provide
> the file as a file either in /proc, or just as a real file somewhere.
>
> The current position is that the debugger can get all the information it
> needs by jumping through rather unpleasant hoops. What I'm proposing
> would give all the same information in a way which the debugger
> necessarily already understands.
FWIW, I agree. We've mostly fixed up GDB to handle this, but anything
that looks at ldd output still gets confused. There shouldn't need to
be this many copies of code to find the vDSO.
--
Daniel Jacobowitz