Re: [PATCH 2/4 v2] Implement new features needed for handling SystemTap probes

[Jan Kratochvil <jan dot kratochvil at redhat dot com>]

On Wed, 11 Apr 2012 23:33:18 +0200, Sergio Durigan Junior wrote:
> On Wednesday, April 11 2012, Jan Kratochvil wrote:
> > rtld       lll_futex_wake 0x000000000000ab2e                      /usr/lib/debug/lib64/ld-2.15.so.debug
> > rtld       lll_futex_wake 0x000000000000ab2e                      /usr/lib64/ld-linux-x86-64.so.2
> > [...]
> >  - You do not filter out separate debug info files or rather unify it
> > somehow.
> 
> `info probes stap' accept an argument specifying the provider, name and
> objfile of the probe, so I guess this is the filter you're talking about.

Really not, the main binary and its separate debug info file represent single
object.  You can see the address is the same, it is the same probe.  There is
no reason why to display it twice.  There are already (too) many exceptions in
GDB codebase to ignore this or that if a separate debug info file exists.


> >> +++ b/gdb/probe.c
> >> @@ -0,0 +1,65 @@
> >> +/* Generic SDT probe support for GDB.
> >
> > Not SDT.
> 
> SDT stands for Statically Defined Probe, so I think SDT is correct in
> this case.  Don't you?

"SDT stands for Statically Defined Tracing" and I believe SDT is bound to
DTrace (and therefore also SystemTap) but feel free to write anything there,
I thought it is a typo/thinko.


Thanks,
Jan