Re: [PATCH] On-demand loading of shlib's debuginfo

[Pedro Alves <pedro@codesourcery.com>]

On Thursday 21 July 2011 13:21:04, Jan Kratochvil wrote:
> On Thu, 21 Jul 2011 12:58:44 +0200, Pedro Alves wrote:
> > Is that the reason then that svr4_match_pc_solist goes through
> > the link map to map a PC to a so_list, instead of having a
> > generic implementation that matches the PC to the so_list's
> > loaded (bfd) sections (that is, just call solib_contains_address_p) ?
> > If so, it'd make more sense IMO to remove that from this patch,
> > and add it only along with a change that lazies mapping in the bfd
> > and its sections as well.  I'm not sure how safe that will be.
> 
> I agree the patch as is does not make much sense on its own IMHO, maybe it
> could be pre-approved but checked-in only later with the bfd part of it.

I was under the impression that the debug info loading would be
the major hog, and lazying that alone would stand on its own (as this
patch, but with a solib_contains_address_pc_p call in solib_on_demand_load
instead of the new match_pc_solist stuff).  Do you expect that lazying
opening the bfd and mapping its target sections to save a fair amount of
additional time?

Some benchmarks would indeed by nice...

-- 
Pedro Alves