https://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
Martin Jambor changed:
What|Removed |Added
Resolution|--- |FIXED
Status|NEW
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #17 from Eric Gallager ---
This is the last bug still open blocking bug 45375
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #16 from Jan Hubicka ---
With new polymorphic call code we could handle the testcase from comment 15 by
simple propagation of contextes down during early optimization. This may be
easy enough to be worth to implement.
Still general s
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
Matthijs Kooijman changed:
What|Removed |Added
CC||matthijs at stdin dot nl
--- Comment
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #14 from Martin Jambor 2010-12-15
16:07:31 UTC ---
(In reply to comment #9)
> OK, main() code seems to optimize out that is an imrovement. Is it optimized
> away with your patch pre-IPA too?
Yes. Just before IPA, in fact.
>
> Der
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #13 from Jan Hubicka 2010-12-15
00:12:54 UTC ---
... but obviously the problem still is that we don't devirtualize this early
enough for inlining. The low level code should be able to do so if FRE+CCP was
added as early pass or FRE w
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #12 from Jan Hubicka 2010-12-15
00:10:45 UTC ---
No, this is different, since OBJ_TYPE_REF is sitting here and it imply type
conversion in the way we implement it right now.
There is no type mismatch in between the original address an
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #11 from Martin Jambor 2010-12-14
23:35:29 UTC ---
I believe Richi's comment #14 in PR 46076 applies here as well.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #10 from Jan Hubicka 2010-12-14
23:17:13 UTC ---
Eh,
wanted to paste:
D.2236_10 = (int (*__vtbl_ptr_type) (void)) f;
OBJ_TYPE_REF(D.2236_10;D.2215_3->2) (D.2215_3);
I told ccp should IMO optimize it, but doesn't. I guess it is bec
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #9 from Jan Hubicka 2010-12-14
23:15:16 UTC ---
OK, main() code seems to optimize out that is an imrovement. Is it optimized
away with your patch pre-IPA too?
Derived() is also devirtualizable:
Derived::Derived() (struct Derived *
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
Martin Jambor changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Last reconfirmed|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #7 from Martin Jambor 2010-10-11
18:40:42 UTC ---
And by the way, I'm afraid we really need to somehow address PR 45934
before we attempt to fold more O_T_Rs coming from
constructors/destructors.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
Martin Jambor changed:
What|Removed |Added
CC|mjambor at suse dot cz |jamborm at gcc dot gnu.org
--- Comment #6
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #5 from Jan Hubicka 2010-09-25
22:17:41 UTC ---
Another testcase where we devirtualize via folding is:
// { dg-do assemble }
// { dg-options "-g -O2" }
// Copyright (C) 1999 Free Software Foundation, Inc.
// Contributed by Nathan
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #4 from Jan Hubicka 2010-09-25
22:12:39 UTC ---
Note that the patch attached solves one indirect call in the testcase but has
no effect on mozilla.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
--- Comment #3 from Jan Hubicka 2010-09-25
22:10:15 UTC ---
Hmm,
normally we should see it from COMPONENT_REF:
while (true)
{
if (TREE_CODE (ref) == COMPONENT_REF)
{
tree par_type;
tree binfo, base_binfo;
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45791
Jan Hubicka changed:
What|Removed |Added
Blocks||45375
--- Comment #2 from Jan Hubicka 2010
17 matches
Mail list logo
