linker relocation of debug information

Daniel Jacobowitz drow@mvista.com
Wed Jan 28 20:48:00 GMT 2004


On Wed, Jan 28, 2004 at 03:41:06PM -0500, Ian Lance Taylor wrote:
> Jeff Baker <jbaker@qnx.com> writes:
> 
> > Let's say, just for the sake of argument, that a developer was locked into
> > using binutils 2.12.1 and gdb 5.2.1 for an upcoming release.  Let's also say
> > that this developer is unable to step into shared object because the debug
> > sections aren't relocated.  What would you suggest as a solution?
> 
> I don't know, are you allowed to change the tools at all?
> 
> If you can change the tools, I would try changing the binutils to
> relocate debug information.  I don't know if it will help, but it's
> probably worth a try.

That said, how do people feel about changing this behavior back - at
least for relocations which can be resolved without external symbols?

The last time I discussed this, I believe Richard and Geoff were
opposed.  Richard wrote:
> The relocation is the only way to associate the real address with the
> debug info in the case of dynamic linker overrides.

I think this argument is bogus, related to the same problems we have
with debug info for discarded sections: the debug information describes
a particular copy of this symbol.  If the runtime linker chooses a
different copy of the symbol, we will get at best pot luck by using
debug info for this copy.

I don't remember what Geoff's objection was.  Mind reminding me?

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer



More information about the Binutils mailing list