Re: [patch rfc] Per-frame frame-base

[Daniel Jacobowitz <drow@mvista.com>]

On Wed, Mar 26, 2003 at 05:18:32PM -0500, Andrew Cagney wrote:
> >On Wed, Mar 26, 2003 at 04:51:13PM -0500, Andrew Cagney wrote:
> >
> >>>On Wed, Mar 26, 2003 at 03:37:22PM -0500, Andrew Cagney wrote:
> >>>
> >
> >>>>The implementation is very much modeled on the frame-unwind code.  
> >>Debug >>readers are expected to register their own high level frame-base 
> >>handler.
> >
> >>>
> >>>
> >>>So what will the process for getting a dwarf2-debug-info frame
> >>>registered look like?
> >
> >>
> >>The current interface is identical to frame-unwind.
> >>
> >
> >>>How will we figure out that this function has
> >>>dwarf2 debug info?  It's not trivial... we don't have that information
> >>>around any more on a per-PC basis.
> >
> >>
> >>Ask the frame's function's symbol or block.
> >
> >
> >Symbols don't currently have this information, nor do blocks.
> 
> Well, at least the partial symtab does, sort of implicit via 
> read_symtab.  Shame that location_funcs stuff wasn't a symbol wide 
> virtual function table.

I heard a volunteer running away into the woods, where did he go?
It took over a year to get location_funcs in at all.

Obviously using any of these frame cleanups is going to require
cleanups in the symtab.  I don't know who will be interested in doing
them however.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer