This is the mail archive of the gdb-patches@sources.redhat.com mailing list for the GDB 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: RFA: Don't use obsavestring in dwarf2read


Daniel Jacobowitz writes:
 > On Thu, Feb 05, 2004 at 03:34:30PM -0500, Elena Zannoni wrote:
 > >  > The obstacks themselves are probably a good idea.  Once upon a time,
 > >  > Peter informed me, there was a plan to free the psymbol obstack when
 > >  > all symbols had been read in; but that doesn't seem like a useful
 > >  > optimization, and I can't think offhand of any use for separate symbol
 > >  > and type obstacks.  I wouldn't object to having a per-objfile obstack
 > >  > instead, and un-seperating them.
 > > 
 > > I think it would be worthwhile to see how much doing that would save us.
 > 
 > Well, it wouldn't save anything by itself - there's immeasurable
 > overhead to the obstacks.  It would let us eliminate this sort of
 > duplication, but they're pretty tricky to identify; it took me a couple
 > of hours to convince myself about this set of 'em.
 > 

I meant in general, yes.  Since the possible 'shorcuts' are difficult
to identify, and they are only for dwarf2 (that I've looked at) I am
bit worried about the cross pointers. I am thinking to kill the triad.

 > >  > > [BTW why are only few obstack properly initialized?]
 > >  > 
 > >  > Which do you mean?
 > >  > 
 > > 
 > > I grepped for obstack_init, and only a few obstacks call that
 > > function. Form the obstack doco, it seems that it needs to be
 > > called. I wonder if the function was introduced later on in libiberty,
 > > as an afterthought.
 > 
 > It looks like obstack_specify_allocation and obstack_init fill the same
 > role.  The objfile's obstacks use the former.
 > 

Which is dumb:

# define obstack_init(h) \
  _obstack_begin ((h), 0, 0, \
		  (void *(*) ()) obstack_chunk_alloc, (void (*) ()) obstack_chunk_free)

# define obstack_specify_allocation(h, size, alignment, chunkfun, freefun) \
  _obstack_begin ((h), (size), (alignment), \
		    (void *(*) ()) (chunkfun), (void (*) ()) (freefun))


all the calls to obstack_specify_allocation use xmalloc and xfree, but
we also have:

/* Unless explicitly specified, GDB obstacks always use xmalloc() and
   xfree().  */
#define obstack_chunk_alloc xmalloc
#define obstack_chunk_free xfree

There is only a call to obstack_specify_allocation that specifies size
and alignment.

I'd prefer them to use just one method. I'll clean up some.

 > > Ah, ok, it's because of the nature of the program you were handling. I
 > > was trying to imagine how the overhead of obstack themselves could be
 > > that large. It seems to me that this is a good argument for an 'on
 > > demand' symbol reading implementaion.  But, yes the various dwarf2
 > > sections are already in the psymbol_obstack. And we are duplicating
 > > that again on the type_obstack. :-( 
 > 
 > Right.  I'm not sure how much of this can be done on demand that isn't
 > already; if I wasn't clear about this, the 100MB was a worst-case
 > number (-readnow).  Without -readnow it's much less.
 > 

ah. Yep, you didn't mention readnow.


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