[Bug sanitizer/62132] New: [5.0 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[Bug sanitizer/62132] New: [5.0 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

            Bug ID: 62132
           Summary: [5.0 Regression] FAIL:
                    c-c++-common/asan/misalign-[12].c after r213807 on
                    x86_64-apple-darwin13 with -m32
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: sanitizer
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dominiq at lps dot ens.fr
                CC: dodji at gcc dot gnu.org, dvyukov at gcc dot gnu.org,
                    iains at gcc dot gnu.org, jakub at gcc dot gnu.org, kcc at gcc dot gnu.org,
                    ygribov at gcc dot gnu.org
              Host: x86_64-apple-darwin13
            Target: x86_64-apple-darwin13
             Build: x86_64-apple-darwin13

On x86_64-apple-darwin13 with -m32 the following tests fail after r213807:

Running target unix/-m32
FAIL: c-c++-common/asan/misalign-1.c   -O0  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-1.c   -O1  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-1.c   -O2  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-1.c   -O3 -fomit-frame-pointer  output pattern
test, is =================================================================
FAIL: c-c++-common/asan/misalign-1.c   -O3 -g  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-1.c   -Os  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-1.c   -O2 -flto -flto-partition=none  output
pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-1.c   -O2 -flto  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-2.c   -O0  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-2.c   -O1  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-2.c   -O2  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-2.c   -O3 -fomit-frame-pointer  output pattern
test, is =================================================================
FAIL: c-c++-common/asan/misalign-2.c   -O3 -g  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-2.c   -Os  output pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-2.c   -O2 -flto -flto-partition=none  output
pattern test, is
=================================================================
FAIL: c-c++-common/asan/misalign-2.c   -O2 -flto  output pattern test, is
=================================================================

In 64 bit mode the output is

=================================================================
==48043==ERROR: AddressSanitizer: unknown-crash on address 0x7fff5ba3f3bf at pc
0x1041c0bf2 bp 0x7fff5ba3f2f0 sp 0x7fff5ba3f2e8
READ of size 4 at 0x7fff5ba3f3bf thread T0
    #0 0x1041c0bf1
(/Users/dominiq/Documents/Fortran/g95bench/win/f90/bug/a.out+0x100000bf1)
    #1 0x1041c0e8f
(/Users/dominiq/Documents/Fortran/g95bench/win/f90/bug/a.out+0x100000e8f)

Address 0x7fff5ba3f3bf is located in stack of thread T0 at offset 175 in frame
    #0 0x1041c0caf
(/Users/dominiq/Documents/Fortran/g95bench/win/f90/bug/a.out+0x100000caf)
...

while in 32 bit mode, it is

=================================================================
==48035==ERROR: AddressSanitizer: unknown-crash on address 0xbffda4cf at pc
0x26b19 bp 0xbffda3d8 sp 0xbffda3cc
READ of size 4 at 0xbffda4cf thread T0
    #0 0x26b18
(/Users/dominiq/Documents/Fortran/g95bench/win/f90/bug/a.out+0x1b18)

Address 0xbffda4cf is located in stack of thread T0 at offset 175 in frame
    #0 0x26bef
(/Users/dominiq/Documents/Fortran/g95bench/win/f90/bug/a.out+0x1bef)
...

i.e., the #1 line is missing. Revision r213806 is OK.

[Bug sanitizer/62132] [5.0 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |5.0

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[y.gribov at samsung dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

Yury Gribov <y.gribov at samsung dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |y.gribov at samsung dot com

--- Comment #1 from Yury Gribov <y.gribov at samsung dot com> ---
I can confirm this but I'm not sure that this has anything to do with r213806
(besides it triggering already existing problem). Libasan uses frame pointers
unconditionally to unwind stack on Darwin. By default 32-bit Darwins seem to
use -fomit-frame-pointer so unwinding can (and in our case does) fail. Adding
-fno-omit-frame-pointer to compile options fixed the problem for me. Perhaps we
should add this flag by default when running Asan tests on Darwin? Or maybe
libsanitizer could be updated to unwind with fp?

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[y.gribov at samsung dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #2 from Yury Gribov <y.gribov at samsung dot com> ---
s/unwind with fp/unwind without fp/

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-08-20
     Ever confirmed|0                           |1

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
The following patch fixes the failures:

--- ../_clean/gcc/testsuite/c-c++-common/asan/misalign-1.c    2014-05-31
12:40:03.000000000 +0200
+++ gcc/testsuite/c-c++-common/asan/misalign-1.c    2014-08-20
11:30:44.000000000 +0200
@@ -1,5 +1,6 @@
 /* { dg-do run { target { ilp32 || lp64 } } } */
 /* { dg-options "-O2" } */
+/* { dg-additional-options "-fno-omit-frame-pointer" { target *-*-darwin* } }
*/
 /* { dg-shouldfail "asan" } */

 struct S { int i; } __attribute__ ((packed));
--- ../_clean/gcc/testsuite/c-c++-common/asan/misalign-2.c    2014-05-31
12:40:03.000000000 +0200
+++ gcc/testsuite/c-c++-common/asan/misalign-2.c    2014-08-20
11:30:56.000000000 +0200
@@ -1,5 +1,6 @@
 /* { dg-do run { target { ilp32 || lp64 } } } */
 /* { dg-options "-O2" } */
+/* { dg-additional-options "-fno-omit-frame-pointer" { target *-*-darwin* } }
*/
 /* { dg-shouldfail "asan" } */

 struct S { int i; } __attribute__ ((packed));

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[y.gribov at samsung dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #4 from Yury Gribov <y.gribov at samsung dot com> ---
Note that same problem may pop up in other tests as well so IMHO better fix
this in general.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #6 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Yury Gribov from comment #2)
> s/unwind with fp/unwind without fp/

it should be possible to do this, if m32 code is compiled with
"-funwind-tables".

for m64, that's already the default and, also for m32 when -fexceptions is
given.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[y.gribov at samsung dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #7 from Yury Gribov <y.gribov at samsung dot com> ---
(In reply to Iain Sandoe from comment #6)
> (In reply to Yury Gribov from comment #2)
> > s/unwind with fp/unwind without fp/
> 
> it should be possible to do this, if m32 code is compiled with
> "-funwind-tables".

This was my impression as well but it looks like current libsanitizer does not
support anything except fp on Macs. Could you report this on address-sanitizer
tracker (https://code.google.com/p/address-sanitizer/issues/list)?

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #8 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> This was my impression as well but it looks like current libsanitizer does
> not support anything except fp on Macs. Could you report this on
> address-sanitizer tracker
> (https://code.google.com/p/address-sanitizer/issues/list)?

I have filled issue 340,
https://code.google.com/p/address-sanitizer/issues/detail?id=340&thanks=340&ts=1410715229.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #9 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Per comments 6 and 7 I have tried

--- ../_clean/gcc/testsuite/c-c++-common/asan/misalign-1.c    2014-05-31
12:40:03.000000000 +0200
+++ gcc/testsuite/c-c++-common/asan/misalign-1.c    2014-11-05
14:35:09.000000000 +0100
@@ -1,6 +1,7 @@
 /* { dg-do run { target { ilp32 || lp64 } } } */
 /* { dg-options "-O2" } */
 /* { dg-shouldfail "asan" } */
+/* { dg-additional-options "-funwind-tables" { target *-*-darwin* } } */

 struct S { int i; } __attribute__ ((packed));

but it does not fix the failures. What am I misunderstanding?

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[y.gribov at samsung dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #10 from Yury Gribov <y.gribov at samsung dot com> ---
(In reply to Dominique d'Humieres from comment #9)
> Per comments 6 and 7 I have tried
> ...
> but it does not fix the failures. What am I misunderstanding?

That wouldn't help because libsanitizer does not use -funwind-tables even if
they are available.  I think the problem is in StackTrace::WillUseFastUnwind in
libsanitizer/sanitizer_common/sanitizer_stacktrace.h.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #11 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Yury Gribov from comment #10)
> (In reply to Dominique d'Humieres from comment #9)
> > Per comments 6 and 7 I have tried
> > ...
> > but it does not fix the failures. What am I misunderstanding?
> 
> That wouldn't help because libsanitizer does not use -funwind-tables even if
> they are available.  I think the problem is in StackTrace::WillUseFastUnwind
> in libsanitizer/sanitizer_common/sanitizer_stacktrace.h.

Dominique,
in summary: as things stand (with the current asan) your first fix LGTM. 
However, you should post it for approval by an appropriate maintainer.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #12 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
I now see the failures on 4.9 due to r216459.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #13 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
The #c3 change is preapproved, please post the patch to gcc-patches though.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING

--- Comment #14 from Richard Biener <rguenth at gcc dot gnu.org> ---
Status?

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fxcoudert at gcc dot gnu.org

--- Comment #15 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
*** Bug 63964 has been marked as a duplicate of this bug. ***

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #17 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
Author: fxcoudert
Date: Wed Nov 19 14:29:43 2014
New Revision: 217777

URL: https://gcc.gnu.org/viewcvs?rev=217777&root=gcc&view=rev
Log:
    PR sanitizer/62132
    * c-c++-common/asan/misalign-1.c: Pass -fno-omit-frame-pointer on
    darwin, adjust dg-output.
    * c-c++-common/asan/misalign-2.c: Likewise.

Modified:
    trunk/gcc/testsuite/c-c++-common/asan/misalign-1.c
    trunk/gcc/testsuite/c-c++-common/asan/misalign-2.c

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #18 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
Author: fxcoudert
Date: Wed Nov 19 14:32:09 2014
New Revision: 217779

URL: https://gcc.gnu.org/viewcvs?rev=217779&root=gcc&view=rev
Log:
Fixing the mess I did with the two previous commits. Sorry!

    PR sanitizer/62132
    * c-c++-common/asan/misalign-1.c: Pass -fno-omit-frame-pointer on
    darwin, adjust dg-output.
    * c-c++-common/asan/misalign-2.c: Likewise.

Modified:
    trunk/gcc/testsuite/c-c++-common/asan/misalign-1.c
    trunk/gcc/testsuite/c-c++-common/asan/misalign-2.c

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #19 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
I have committed a patch adding -fno-omit-frame-pointer to the testcases, which
makes them pass on x86_64-apple-darwin14.

I'm not closing this PR, since I don't know if the maintainers consider it
fixing the bug, or simply working around it :)

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

--- Comment #20 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Francois-Xavier Coudert from comment #19)
> I have committed a patch adding -fno-omit-frame-pointer to the testcases,
> which makes them pass on x86_64-apple-darwin14.
> 
> I'm not closing this PR, since I don't know if the maintainers consider it
> fixing the bug, or simply working around it :)

IMO, this is fixed here - i.e. the issue of whether a different unwind scheme
should be available is an issue for "upstream" (see c#8)

If/when that changes, we might wish to alter the flag (or more likely test that
both methods work) here.

- however, let's give other folks 24h to chime in.

[Bug sanitizer/62132] [5 Regression] FAIL: c-c++-common/asan/misalign-[12].c after r213807 on x86_64-apple-darwin13 with -m32

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62132

Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|---                         |FIXED

--- Comment #21 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
Thus closing.