http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #54 from mrs at gcc dot gnu.org mrs at gcc dot gnu.org 2013-02-11
22:36:28 UTC ---
Author: mrs
Date: Mon Feb 11 22:36:23 2013
New Revision: 195956
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=195956
Log:
2013-02-11
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #48 from Alexander Potapenko glider at google dot com 2013-02-04
10:11:32 UTC ---
(In reply to comment #40)
if (ctor_recA-position ctor_recB-position)
return -1;
if (ctor_recA-position ctor_recB-position)
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #49 from Alexander Potapenko glider at google dot com 2013-02-04
10:13:49 UTC ---
I agree with Jakub: it's better to return back to the qsort version of the
patch, since it fixes ASan as well, but also provides better support
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #50 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-04
17:24:40 UTC ---
(In reply to comment #49)
I agree with Jakub: it's better to return back to the qsort version of the
patch, since it fixes ASan as well, but
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #51 from mrs at gcc dot gnu.org mrs at gcc dot gnu.org 2013-02-04
20:08:34 UTC ---
Author: mrs
Date: Mon Feb 4 20:08:29 2013
New Revision: 195735
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=195735
Log:
2013-02-04
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #52 from mrs at gcc dot gnu.org mrs at gcc dot gnu.org 2013-02-04
21:07:42 UTC ---
Author: mrs
Date: Mon Feb 4 21:07:35 2013
New Revision: 195737
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=195737
Log:
2013-02-04
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
m...@gcc.gnu.org mrs at gcc dot gnu.org changed:
What|Removed |Added
Status|UNCONFIRMED |RESOLVED
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #47 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-03
15:16:50 UTC ---
posted proposed patch and regression testresults at...
http://gcc.gnu.org/ml/gcc-patches/2013-02/msg00055.html
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #36 from Jakub Jelinek jakub at gcc dot gnu.org 2013-02-02
08:46:55 UTC ---
} ctor_record;
Why? }; should be enough IMHO in C++. Or does GTY still require it?
int ctor_index = -1;
... ctor_index++
What is this for?
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #37 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-02
15:31:37 UTC ---
typedef struct GTY(()) ctor_record {
rtx symbol;
int priority; /* constructor priority */
int position; /* original
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #38 from Jakub Jelinek jakub at gcc dot gnu.org 2013-02-02
15:38:31 UTC ---
Obviously it shouldn't be typedef in that case. Anyway, this part is not a big
deal, just a nit.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #39 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-02
18:16:39 UTC ---
While testing whether the single qsort was sufficient, the origin of the
problem on darwin was clarified. In machopic_asm_out_constructor, after
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
m...@gcc.gnu.org mrs at gcc dot gnu.org changed:
What|Removed |Added
CC||mrs at
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #41 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-02
20:11:07 UTC ---
Created attachment 29338
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=29338
alternative approach of only inserting asan static constructor
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
Jack Howarth howarth at nitro dot med.uc.edu changed:
What|Removed |Added
Attachment #29338|0 |1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #43 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-02
20:19:40 UTC ---
(In reply to comment #40)
Actually I think we should junk sorting entirely and use the alternative
approach of the patch in Comment 42. That
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #44 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-02
20:41:15 UTC ---
(In reply to comment #40)
Doesn't the test case I showed in Comment 28 qualify as working across
translaional units? That test case still
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #45 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-02
22:53:59 UTC ---
(In reply to comment #40)
Also the impact of the proposed patch in Comment 42 could be limited even
further by using...
if (flag_asan
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #46 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-03
00:10:02 UTC ---
(In reply to comment #40)
Also with the patch in Comment 42, the failing test case converted into a
shared library loaded via dlopen works
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #31 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-01
16:46:35 UTC ---
FYI, the proof of concept patch from Comment 27 produces no regressions in the
testsuite on x86_64-apple-darwin12...
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #32 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-01
21:22:06 UTC ---
Created attachment 29332
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=29332
first attempt at va_gc implementation
The attached patch is
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #33 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-01
21:45:42 UTC ---
Replacing the...
ctors-qsort (sort_by_ctor_priority);
with...
qsort(ctors, ctor_index+1, sizeof(ctor_record), sort_by_ctor_priority);
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #34 from Jakub Jelinek jakub at gcc dot gnu.org 2013-02-01
22:19:45 UTC ---
Can you explain why normal qsort wouldn't do the sort in one pass?
You just do
if (ctor_recordA-priority ctor_recordB-priority)
return -1;
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #35 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-02
05:51:31 UTC ---
Created attachment 29334
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=29334
working va_gc implementation
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #23 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-31
22:01:39 UTC ---
Created attachment 29323
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=29323
proposed patch for dynamic allocation
Proposed patch for
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #24 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-31
22:22:50 UTC ---
Suspect we need to use...
ctors[ctor_index].symbol = copy_rtx(symbol);
in machopic_asm_out_constructor although I am unclear on what need
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #25 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-31
22:25:45 UTC ---
(In reply to comment #24)
Suspect we need to use...
ctors[ctor_index].symbol = copy_rtx(symbol);
in machopic_asm_out_constructor
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #26 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-01
02:30:10 UTC ---
Created attachment 29324
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=29324
proposed patch for dynamic allocation and qsort of ctor records
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
Jack Howarth howarth at nitro dot med.uc.edu changed:
What|Removed |Added
Attachment #29324|0 |1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #28 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-01
03:00:37 UTC ---
qsort version of patch works with trivial shared lib test code of...
% cat libcov.C
struct c18 {
virtual void bar() { }
};
c18 ret;
%
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #29 from Jack Howarth howarth at nitro dot med.uc.edu 2013-02-01
05:52:13 UTC ---
The proposed patch with dynamic allocation and qsort of the ctor records by
priority field reduces the number of unexpected failures for...
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #30 from Jakub Jelinek jakub at gcc dot gnu.org 2013-02-01
07:31:24 UTC ---
Don't want to spend too much time on this, so just a few hints:
1) you want to store this in a vector (see vec.h)
2) rtxs are GC allocated, you don't
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #10 from Alexander Potapenko glider at google dot com 2013-01-30
12:29:00 UTC ---
I suppose this isn't important. __mod_term_func are destructors, and they even
aren't called in the crashing program.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #11 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-30
14:23:51 UTC ---
(In reply to comment #10)
I suppose this isn't important. __mod_term_func are destructors, and they even
aren't called in the crashing program.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #12 from Alexander Potapenko glider at google dot com 2013-01-30
14:32:54 UTC ---
The question is why does...
if (builtin_decl_implicit_p (BUILT_IN_ASAN_INIT))
return;
in initialize_sanitizer_builtins() not
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #13 from Jakub Jelinek jakub at gcc dot gnu.org 2013-01-30
14:41:14 UTC ---
(In reply to comment #12)
This one is a necessary one.
asan_finish_file inserts __asan_init into the array of constructors (aka
__mod_init_func
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #14 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-30
15:57:11 UTC ---
(In reply to comment #13)
See in gcc/config/darwin.h...
/* The Apple assembler and linker do not support constructor priorities. */
#undef
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #15 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-30
16:28:17 UTC ---
It also seems that Solaris 2 will suffer from this issue when not using Gold...
#ifndef USE_GLD
/* The Solaris linker doesn't understand
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #16 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-30
16:31:14 UTC ---
This limitation all exists for clang on darwin...
http://llvm.org/bugs/show_bug.cgi?id=12556
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #17 from Jakub Jelinek jakub at gcc dot gnu.org 2013-01-30
16:32:11 UTC ---
Solaris doesn't support Asan in gcc, and perhaps it is time to admit that
Darwin doesn't either.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #18 from Kostya Serebryany kcc at gcc dot gnu.org 2013-01-30
16:36:01 UTC ---
Yea... We don't have interest in supporting gcc-asan-darwin, sorry.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #19 from Jakub Jelinek jakub at gcc dot gnu.org 2013-01-30
16:43:03 UTC ---
Well, if somebody does the work and in a clean way that won't penalize targets
with sane linkers and object formats, I'm not objecting, I just am not
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #20 from Alexander Potapenko glider at google dot com 2013-01-30
17:07:25 UTC ---
(In reply to comment #19)
Well, if somebody does the work and in a clean way that won't penalize targets
with sane linkers and object formats,
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #21 from Alexander Potapenko glider at google dot com 2013-01-30
17:30:18 UTC ---
Created attachment 29309
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=29309
Dummy patch that reverses the order of the constructors
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #22 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-30
23:40:56 UTC ---
(In reply to comment #21)
The proposed patch reduces the number of unexpected failures in the g++
testsuite when using...
make -k check-g++
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
Alexander Potapenko glider at google dot com changed:
What|Removed |Added
CC|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #6 from Alexander Potapenko glider at google dot com 2013-01-29
09:59:09 UTC ---
Looking at the disassembly I see that __asan_init is placed into some
__GLOBAL__sub_I_00099_1_cov.cc function, which isn't being called at runtime
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #7 from Alexander Potapenko glider at google dot com 2013-01-29
11:56:02 UTC ---
Here's the dump of __mod_init_func (the static ctors array):
===
Disassembly of section
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #8 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-29
22:04:28 UTC ---
Created attachment 29302
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=29302
assembly file for reduced testcase from comment 5
generated
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #9 from Jack Howarth howarth at nitro dot med.uc.edu 2013-01-29
22:15:29 UTC ---
Is it significant that in the assembly, the .mod_term_func section section
(which captures the call to __asan_init) is emitted before the
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #1 from Jack Howarth howarth at nitro dot med.uc.edu 2012-12-07
14:48:57 UTC ---
Created attachment 28894
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=28894
assembly file for covariant3.C compiled with -fsanitize=address
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #2 from Jack Howarth howarth at nitro dot med.uc.edu 2012-12-07
14:50:34 UTC ---
Created attachment 28895
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=28895
gdb log of stepi walk from 38th breakpoint of
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #3 from Jack Howarth howarth at nitro dot med.uc.edu 2012-12-07
15:44:16 UTC ---
This might be due to the code...
/* Startup code should go to startup subsection unless it is
unlikely executed (this happens especially
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #4 from Jack Howarth howarth at nitro dot med.uc.edu 2012-12-08
03:14:29 UTC ---
The failing testcase in gdb appears as...
gdb ./covariant3.exe
...
(gdb) br _GLOBAL__sub_I_covariant3.C
Breakpoint 1 at 0x11ce2: file
54 matches
Mail list logo
