[Bug target/38736] New: [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[Bug target/38736] New: [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[pinskia at gcc dot gnu dot org]

using __attribute__((aligned)) can change the alignment dependent on if you use
-mavx or not.
This is because BIGGEST_ALIGNMENT is defined using -mavx.  This is not correct
as -mavx should not change the ABI unless you use the avx sized vectors.


-- 
           Summary: [4.4 Regression] -mavx can change the ABI via
                    BIGGEST_ALIGNMENT
           Product: gcc
           Version: 4.4.0
            Status: UNCONFIRMED
          Keywords: wrong-code, ABI
          Severity: blocker
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pinskia at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 GCC target triplet|                            |i?86-*-* x86_64-*-*
   Target Milestone|---                         |4.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2009-01-05 23:05 -------
*** Bug 38737 has been marked as a duplicate of this bug. ***


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |schwab at suse dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2009-01-05 23:14 -------
Most likely ADJUST_FIELD_ALIGN needs to be defined for AVX vectors. 
ADJUST_FIELD_ALIGN overrides both BIGGEST_FIELD_ALIGNMENT for sure.

x86_field_alignment seems wrong anyways.  It does not even take into account
vector types ...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[schwab at suse dot de]

------- Comment #3 from schwab at suse dot de  2009-01-05 23:22 -------
ADJUST_FIELD_ALIGN does not override user-specified alignment.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #4 from hjl dot tools at gmail dot com  2009-01-05 23:30 -------
There is another place:

libstdc++-v3/include/std/type_traits:   struct __attribute__((__aligned__)) { }
__align; 


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2009-01-05 23:30:01
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[Joey dot ye at intel dot com]

------- Comment #5 from Joey dot ye at intel dot com  2009-01-07 02:45 -------
More places with BIGGEST_ALIGN:
$ grep -r "(aligned)" .|grep attribute|grep -v testsuite|grep -v texi
./libstdc++-v3/libsupc++/eh_alloc.cc:typedef char
one_buffer[EMERGENCY_OBJ_SIZE] __attribute__((aligned));
./libjava/exception.cc:  char end[0] __attribute__((aligned));
./libjava/exception.cc:    __attribute__((aligned));
./gcc/unwind-sjlj.c:  jmp_buf jbuf __attribute__((aligned));


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #6 from hjl dot tools at gmail dot com  2009-01-07 02:48 -------
(In reply to comment #4)
> There is another place:
> 
> libstdc++-v3/include/std/type_traits:   struct __attribute__((__aligned__)) { }
> __align; 
> 

It may not be a problem since it is used to get the default alignment,
which is encoded in the class template:

  template<std::size_t _Len>
    struct __aligned_storage_msa
    {
      union __type
      {
        unsigned char __data[_Len];
        struct __attribute__((__aligned__)) { } __align;
      };
    };
...
  template<std::size_t _Len, std::size_t _Align =
           __alignof__(typename __aligned_storage_msa<_Len>::__type)>
    struct aligned_storage
    {
      union type
      {
        unsigned char __data[_Len];
        struct __attribute__((__aligned__((_Align)))) { } __align;
      };
    };


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2009-01-05 23:30:01         |2009-01-07 02:48:15
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #7 from hjl dot tools at gmail dot com  2009-01-07 02:50 -------
(In reply to comment #5)
> More places with BIGGEST_ALIGN:
> $ grep -r "(aligned)" .|grep attribute|grep -v testsuite|grep -v texi
> ./libstdc++-v3/libsupc++/eh_alloc.cc:typedef char
> one_buffer[EMERGENCY_OBJ_SIZE] __attribute__((aligned));
> ./libjava/exception.cc:  char end[0] __attribute__((aligned));
> ./libjava/exception.cc:    __attribute__((aligned));
> ./gcc/unwind-sjlj.c:  jmp_buf jbuf __attribute__((aligned));
> 

Will they cause problems if __attribute__((aligned)) is changed by -mavx?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #8 from hjl dot tools at gmail dot com  2009-01-07 02:58 -------
(In reply to comment #5)
> More places with BIGGEST_ALIGN:
> ./gcc/unwind-sjlj.c:  jmp_buf jbuf __attribute__((aligned));
> 

It may be handled the same way as struct _Unwind_Exception.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #9 from hjl dot tools at gmail dot com  2009-01-07 18:17 -------
A patch is posted at

http://gcc.gnu.org/ml/gcc-patches/2009-01/msg00350.html


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2009-
                   |                            |01/msg00350.html


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #10 from hjl dot tools at gmail dot com  2009-01-07 18:23 -------
The updated patch is at

http://gcc.gnu.org/ml/gcc-patches/2009-01/msg00352.html


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|http://gcc.gnu.org/ml/gcc-  |http://gcc.gnu.org/ml/gcc-
                   |patches/2009-               |patches/2009-
                   |01/msg00350.html            |01/msg00352.html


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #11 from hjl dot tools at gmail dot com  2009-01-07 19:21 -------
The updated patch is posted at

http://gcc.gnu.org/ml/gcc-patches/2009-01/msg00358.html


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|http://gcc.gnu.org/ml/gcc-  |http://gcc.gnu.org/ml/gcc-
                   |patches/2009-               |patches/2009-
                   |01/msg00352.html            |01/msg00358.html


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #12 from hjl dot tools at gmail dot com  2009-01-14 01:52 -------
The current patch is at

http://gcc.gnu.org/ml/gcc-patches/2009-01/msg00675.html


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|http://gcc.gnu.org/ml/gcc-  |http://gcc.gnu.org/ml/gcc-
                   |patches/2009-               |patches/2009-
                   |01/msg00358.html            |01/msg00675.html


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl at gcc dot gnu dot org]

------- Comment #13 from hjl at gcc dot gnu dot org  2009-01-18 16:26 -------
Subject: Bug 38736

Author: hjl
Date: Sun Jan 18 16:26:26 2009
New Revision: 143486

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=143486
Log:
gcc/

2009-01-18  H.J. Lu  <hongjiu.lu@intel.com>

        PR target/38736
        * c-common.c (handle_aligned_attribute): Use
        ATTRIBUTE_ALIGNED_VALUE instead of BIGGEST_ALIGNMENT for
        default alignment value.

        * c-cppbuiltin.c (c_cpp_builtins): Define __BIGGEST_ALIGNMENT__.

        * defaults.h (ATTRIBUTE_ALIGNED_VALUE): New.
        * config/i386/i386.h (ATTRIBUTE_ALIGNED_VALUE): Likewise.

        * doc/extend.texi: Update __attribute__ ((aligned)).  Document
        __BIGGEST_ALIGNMENT__.

        * doc/tm.texi: Document ATTRIBUTE_ALIGNED_VALUE.

gcc/testsuite/

2009-01-18  H.J. Lu  <hongjiu.lu@intel.com>

        PR target/38736
        * g++.dg/compat/abi/pr38736_main.C: New.
        * g++.dg/compat/abi/pr38736_x.C: Likewise.
        * g++.dg/compat/abi/pr38736_y.C: Likewise.
        * g++.dg/other/macro-1.C: Likewise.
        * gcc.dg/macro-1.c: Likewise.
        * gcc.dg/compat/pr38736_main.c: Likewise.
        * gcc.dg/compat/pr38736_x.c: Likewise.
        * gcc.dg/compat/pr38736_y.c: Likewise.

Added:
    trunk/gcc/testsuite/g++.dg/compat/abi/pr38736_main.C
    trunk/gcc/testsuite/g++.dg/compat/abi/pr38736_x.C
    trunk/gcc/testsuite/g++.dg/compat/abi/pr38736_y.C
    trunk/gcc/testsuite/g++.dg/other/macro-1.C
    trunk/gcc/testsuite/gcc.dg/compat/pr38736_main.c
    trunk/gcc/testsuite/gcc.dg/compat/pr38736_x.c
    trunk/gcc/testsuite/gcc.dg/compat/pr38736_y.c
    trunk/gcc/testsuite/gcc.dg/macro-1.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/c-common.c
    trunk/gcc/c-cppbuiltin.c
    trunk/gcc/config/i386/i386.h
    trunk/gcc/defaults.h
    trunk/gcc/doc/extend.texi
    trunk/gcc/doc/tm.texi
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736

[Bug target/38736] [4.4 Regression] -mavx can change the ABI via BIGGEST_ALIGNMENT

[hjl dot tools at gmail dot com]

------- Comment #14 from hjl dot tools at gmail dot com  2009-01-18 22:12 -------
Fixed.


-- 

hjl dot tools at gmail dot com changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38736