This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [patches] Re: x86_64 frame unwind info


[cc'ed to discuss@x86-64.org, because it's certainly an ABI question]

> > 
> > Incorrect.  We do not expect debuggers to be able to interpret .eh_frame,
> > which is of gcc origin.  We expect them to be able to interpret .debug_frame,
> > which is of dwarf2 standards origin.  Thus we emit _both_ .eh_frame and
> > .debug_frame with -g -fexceptions.
> > 
> 
> Aha, I see - so for debugging we actually output both sections....
> Se would that be ok with removed "else"?

So turning the space overhead from 10-20% to 20-40% for C++. Does not look
very attractive. I would strongly advocate to not duplicate them at least
for x86-64.

-Andi


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]