r245459 - According to i686 ABI, long double size on x86 is 12 bytes not 16 bytes.

Wed Aug 19 11:42:10 PDT 2015

Yes, it looks like a legacy issue. Documentation says so:

*https://gcc.gnu.org/onlinedocs/gcc-4.8.2/gcc/i386-and-x86-64-Options.html
<https://gcc.gnu.org/onlinedocs/gcc-4.8.2/gcc/i386-and-x86-64-Options.html>*

-m96bit-long-double-m128bit-long-doubleThese switches control the size of long
double type. The i386 application binary interface specifies the size to be
96 bits, so -m96bit-long-double is the default in 32-bit mode.

Modern architectures (Pentium and newer) prefer long double to be aligned
to an 8- or 16-byte boundary. In arrays or structures conforming to the
ABI, this is not possible. So specifying -m128bit-long-double aligns long
double to a 16-byte boundary by padding the long double with an additional
32-bit zero.

In the x86-64 compiler, -m128bit-long-double is the default choice as its
ABI specifies that long double is aligned on 16-byte boundary.

Notice that neither of these options enable any extra precision over the
x87 standard of 80 bits for a long double.

*Warning:* if you override the default value for your target ABI, this
changes the size of structures and arrays containing long double variables,
as well as modifying the function calling convention for functions taking long
double. Hence they are not binary-compatible with code compiled without
that switch.

And practical testing agrees:

sh-4.3$ cat < a.cpp
#include <iostream>
int main() {
  long double a;
  std::cout<<sizeof(a)<<std::endl;
}
sh-4.3$ g++ -v
Using built-in specs.
COLLECT_GCC=C:\mingw32\bin\g++.exe
COLLECT_LTO_WRAPPER=C:/mingw32/bin/../libexec/gcc/i686-w64-mingw32/5.1.0/lto-wrapper.exe
Target: i686-w64-mingw32
Configured with: ../../../src/gcc-5.1.0/configure --host=i686-w64-mingw32
--build=i686-w64-mingw32 --target=i686-w64-mingw32 --prefix=/mingw32
--with-sysroot=/c/mingw510/i686-510-posix-dwarf-rt_v4-rev0/mingw32
--with-gxx-include-dir=/mingw32/i686-w64-mingw32/include/c++
--enable-shared --enable-static --disable-multilib
--enable-languages=c,c++,fortran,objc,obj-c++,lto
--enable-libstdcxx-time=yes --enable-threads=posix --enable-libgomp
--enable-libatomic --enable-lto --enable-graphite --enable-checking=release
--enable-fully-dynamic-string --enable-version-specific-runtime-libs
--disable-sjlj-exceptions --with-dwarf2 --disable-isl-version-check
--disable-libstdcxx-pch --disable-libstdcxx-debug --enable-bootstrap
--disable-rpath --disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-gnu-as --with-gnu-ld --with-arch=i686
--with-tune=generic --with-libiconv --with-system-zlib
--with-gmp=/c/mingw510/prerequisites/i686-w64-mingw32-static
--with-mpfr=/c/mingw510/prerequisites/i686-w64-mingw32-static
--with-mpc=/c/mingw510/prerequisites/i686-w64-mingw32-static
--with-isl=/c/mingw510/prerequisites/i686-w64-mingw32-static
--with-pkgversion='i686-posix-dwarf-rev0, Built by MinGW-W64 project'
--with-bugurl=http://sourceforge.net/projects/mingw-w64 CFLAGS='-O2 -pipe
-I/c/mingw510/i686-510-posix-dwarf-rt_v4-rev0/mingw32/opt/include
-I/c/mingw510/prerequisites/i686-zlib-static/include
-I/c/mingw510/prerequisites/i686-w64-mingw32-static/include' CXXFLAGS='-O2
-pipe -I/c/mingw510/i686-510-posix-dwarf-rt_v4-rev0/mingw32/opt/include
-I/c/mingw510/prerequisites/i686-zlib-static/include
-I/c/mingw510/prerequisites/i686-w64-mingw32-static/include' CPPFLAGS=
LDFLAGS='-pipe
-L/c/mingw510/i686-510-posix-dwarf-rt_v4-rev0/mingw32/opt/lib
-L/c/mingw510/prerequisites/i686-zlib-static/lib
-L/c/mingw510/prerequisites/i686-w64-mingw32-static/lib
-Wl,--large-address-aware'
Thread model: posix
gcc version 5.1.0 (i686-posix-dwarf-rev0, Built by MinGW-W64 project)

sh-4.3$ g++ a.cpp
sh-4.3$ ./a.exe
12

Without the patch clang outputs 16 and seg faults on a boost::math example.

2015-08-19 21:29 GMT+03:00 Richard Smith <richard at metafoo.co.uk>:

> On Wed, Aug 19, 2015 at 10:02 AM, Yaron Keren via cfe-commits <
> cfe-commits at lists.llvm.org> wrote:
>
>> Author: yrnkrn
>> Date: Wed Aug 19 12:02:32 2015
>> New Revision: 245459
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=245459&view=rev
>> Log:
>> According to i686 ABI, long double size on x86 is 12 bytes not 16 bytes.
>> See
>>  https://gcc.gnu.org/onlinedocs/gcc-3.2/gcc/i386-and-x86-64-Options.html
>>
>>
>> Added:
>>     cfe/trunk/test/CodeGen/mingw-long-double-size.c
>> Modified:
>>     cfe/trunk/lib/Basic/Targets.cpp
>>
>> Modified: cfe/trunk/lib/Basic/Targets.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Basic/Targets.cpp?rev=245459&r1=245458&r2=245459&view=diff
>>
>> ==============================================================================
>> --- cfe/trunk/lib/Basic/Targets.cpp (original)
>> +++ cfe/trunk/lib/Basic/Targets.cpp Wed Aug 19 12:02:32 2015
>> @@ -3785,7 +3785,8 @@ class MinGWX86_32TargetInfo : public Win
>>  public:
>>    MinGWX86_32TargetInfo(const llvm::Triple &Triple)
>>        : WindowsX86_32TargetInfo(Triple) {
>> -    LongDoubleWidth = LongDoubleAlign = 128;
>> +    LongDoubleWidth = 96;
>> +    LongDoubleAlign = 128;
>>
>
> Is this really correct? It's deeply suspicious that the size is not a
> multiple of the alignment.
>
>
>>      LongDoubleFormat = &llvm::APFloat::x87DoubleExtended;
>>    }
>>    void getTargetDefines(const LangOptions &Opts,
>>
>> Added: cfe/trunk/test/CodeGen/mingw-long-double-size.c
>> URL:
>> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/mingw-long-double-size.c?rev=245459&view=auto
>>
>> ==============================================================================
>> --- cfe/trunk/test/CodeGen/mingw-long-double-size.c (added)
>> +++ cfe/trunk/test/CodeGen/mingw-long-double-size.c Wed Aug 19 12:02:32
>> 2015
>> @@ -0,0 +1,5 @@
>> +// RUN: %clang_cc1 -triple i686-pc-windows-gnu -S %s  -o - | FileCheck
>> %s -check-prefix=CHECK_I686
>> +// CHECK_I686: lda,12
>> +// RUN: %clang_cc1 -triple x86_64-pc-windows-gnu -S %s  -o - | FileCheck
>> %s -check-prefix=CHECK_X86_64
>> +// CHECK_X86_64: lda,16
>> +long double lda;
>>
>>
>> _______________________________________________
>> cfe-commits mailing list
>> cfe-commits at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20150819/20f010f4/attachment.html>