This is the mail archive of the gcc@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]
Other format: [Raw text]

Re: Problems with breakpoints in constructors


On Tue, 6 Jul 2004 18:04:14 +0530, "Amit S. Kale" <amitkale@linsyssoft.com> wrote:

> On Tuesday 06 Jul 2004 12:03 pm, Jason Merrill wrote:

>> That's a bug; since we (correctly) aren't attaching the linkage name to the
>> abstract instance, we need to attach it to the concrete instances.

> Any ideas on how to do that?
> I guess add_name_and_src_coords_attributes should be called for both the 
> concrete instances.

Yes, and it should *only* add DW_AT_MIPS_linkage_name, not the others which
should be inherited from the abstract instance.

> Once debugging information is emitted for an abstrace instance, it's not 
> emitted again. There is a check to that effect. Should we get rid of this 
> check so that debugging information is emitted multiple times for abstract
> instances?

Huh?  There's only one abstract instance, which only exists as debugging
information.  I don't know what it would mean to emit it multiple times.

On Tue, 6 Jul 2004 12:22:58 -0400, Daniel Jacobowitz <drow@mvista.com> wrote:

> On Tue, Jul 06, 2004 at 09:48:32PM +0530, Amit S. Kale wrote:
>> How will GDB search functions correctly without a mangled named? I don't think 
>> we should stop gdb from using DW_AT_MIPS_linkage_name.
>
> By using the explicit scope information, as DWARF-2 is intended to
> work.  The use of mangled names for this is a terrible hack.

I agree.

Jason


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