[Bug tdep/30028] [gdb/tdep, x86_64] Epilogue unwinder preferred over dwarf unwinder based on gcc 4.4

["vries at gcc dot gnu.org" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=30028

--- Comment #5 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Tom de Vries from comment #4)
> (In reply to Tom de Vries from comment #2)
> > Here ( https://gcc.gnu.org/gcc-4.5/changes.html ) I read:
> > ...
> > GCC now generates unwind info also for epilogues. 
> > ...
> > So, maybe this was fixed in gcc 4.5 ?
> 
> OK, I think this is referring to:
> https://gcc.gnu.org/pipermail/gcc-patches/2009-May/261377.html which was
> committed as cd9c1ca866b6aa5041a352e0ed07ae5f91e235e5 .
> 
> First tag to contain the commit: 4.5.0.

And, we already know this:
...
      if (gcc_4_minor >= 5)
        cust->set_epilogue_unwind_valid (true);
...

So we already scan the producer for
https://sourceware.org/pipermail/gdb-patches/2011-June/083429.html , and
crucially:
...
I find questionable this detection uses DWARF DW_AT_producer.  Therefore on
binaries with .eh_frame but no (DWARF) debug info the detection fails and
epilogue unwinders will get into affect.
...

So, that's what's happening.  We have dwarf apart from .eh_frame, and we don't
use the .eh_frame info because we can't check that the producer is 4.5 or
older.

So, maybe we should just flip the switch on the default assumption.

-- 
You are receiving this mail because:
You are on the CC list for the bug.