Re: [RFA] Support for multiple calling conventions, patch 1/3: Add calling convention to struct type

[Corinna Vinschen <vinschen at redhat dot com>]

On Apr 17 12:21, Daniel Jacobowitz wrote:
> On Tue, Apr 15, 2008 at 01:51:38PM +0200, Corinna Vinschen wrote:
> > The result looks like this:
> > 
> >   ftype->calling_convention == 0
> >   
> >     if no debug information is available for the function or if the
> >     debug information is not given as dwarf2.
> > 
> >   ftype->calling_convention == DW_CC_normal
> >   
> >     if dwarf2 debug information is available for the function but no
> >     DW_AT_calling_convention attribute has been given for the function.
> > 
> >   ftype->calling_convention == DW_UNSND(DW_AT_calling_convention)
> > 
> >     if dwarf2 debug information is available for the function and the
> >     DW_AT_calling_convention attribute has been specified.
> 
> This part looks generally fine.  My only complaint is the size
> and placement of the new field.  struct type is a hefty portion
> of GDB's memory usage.
> 
> For a first optimization, it can go in main_type instead.  Then
> there won't be an extra copy for every const or volatile type
> variant.  For a second optimization, it's only for TYPE_CODE_FUNC - so
> I think it will fit in union type_specific.
> 
> Assuming that's right, changed version is OK.

Thanks, my tests show that this works fine.  When the documentation
and NEWS patches are approved, I'll check everything in.


Corinna

-- 
Corinna Vinschen
Cygwin Project Co-Leader
Red Hat