IA32 bootstrap error

[Richard Kenner]

    Okay, well, then you are introducing a co-dependency here, requiring gcc
    not to emit things gdb's non-standards compliant dwarf2 reader can't
    support in the non-cvs version.
    This is a bad idea.

We do that with assemblers too, or at least we used to.  We don't emit things
that older versions of binutils don't support.

This way, you have dependence that affects the *maintainers*, but not
the users.  That's far better.

    GCC is following the standard here, it's gdb that is at fault.

I don't care who's "at fault" and neither will most users: they will just
complain that GCC produces debugging output that thir GDB can't read.