Re: [commit] Improved linker-debugger interface

[Gary Benson <gbenson at redhat dot com>]

Joel Brobecker wrote:
> Also, I am wondering we really want a warning in this case - I think
> this is going to make the average user think that there is something
> wrong and therefore that needs to be fixed. What do people think?

I think this may have been resolved by other messages in this thread,
but for the avoidance of doubt the warning does mean that something is
wrong and needs to be fixed.  If solib-svr4 finds the probes it needs
in the runtime linker it uses them.  If it doesn't find the probes--
either because they aren't there, or because GDB has no probes
support--then solib-svr4 will silently fall back to the original
(non-probes) interface.  The warning is only displayed if solib-svr4
started out using the probes and something unexpected happened, so if
you see the warning it means either the linker or GDB has a bug.

> But also, I think what we will be doing will also depend on how we
> see the future. I get the impression with this warning that we
> eventually expect all GNU/Linux systems to all provide whatever
> feature is needed for the improved method to work. But solib-svr4 is
> used on other platforms too - do we expect it to be the default on
> those as well?

The original (non-probes) interface is still there, and if solib-svr4
can't use the probes interface for whatever reason the original
interface will be used as before.  I don't see this changing any time
soon.  Having said that, nothing in the probes-based linker interface
is SystemTap-specific so it should be possible to implement the
interface on other platforms without touching solib-svr4.c.

Thanks,
Gary