[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #20 from github-bugzi...@puremagic.com ---
Commits pushed to newCTFE at https://github.com/dlang/dmd

https://github.com/dlang/dmd/commit/3b34132a9dd6fd3ac6bd32c572dc6a9c2a75b8f8
fix Issue 16536 - OSX mangling mismatch in dmd's C++/D ABI

https://github.com/dlang/dmd/commit/131668b53dea8833e1d922f180800793e30703f2
Merge pull request #6180 from MartinNowak/fix16536

https://github.com/dlang/dmd/commit/01a60612755c505350ebd26e701f46ec48255eda
fix OSX64 ABI mismatch with divcoeff

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #19 from github-bugzi...@puremagic.com ---
Commits pushed to master at https://github.com/dlang/dmd

https://github.com/dlang/dmd/commit/3b34132a9dd6fd3ac6bd32c572dc6a9c2a75b8f8
fix Issue 16536 - OSX mangling mismatch in dmd's C++/D ABI

https://github.com/dlang/dmd/commit/131668b53dea8833e1d922f180800793e30703f2
Merge pull request #6180 from MartinNowak/fix16536

https://github.com/dlang/dmd/commit/01a60612755c505350ebd26e701f46ec48255eda
fix OSX64 ABI mismatch with divcoeff

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #18 from github-bugzi...@puremagic.com ---
Commit pushed to stable at https://github.com/dlang/dmd

https://github.com/dlang/dmd/commit/01a60612755c505350ebd26e701f46ec48255eda
fix OSX64 ABI mismatch with divcoeff

- also apply fix from #6180 (Issue 16536) to divcoeff.c
- add comment that declarations must be kept in sync

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #17 from Steven Schveighoffer  ---
And now:

Undefined symbols for architecture x86_64:
  "choose_multiplier(int, unsigned long, int, unsigned long*, int*)",
referenced from:
  cdmul(elem*, unsigned int*) in backend.a(cod2.o)
  "udiv_coefficients(int, unsigned long, int*, unsigned long*, int*)",
referenced from:
  cdmul(elem*, unsigned int*) in backend.a(cod2.o)
ld: symbol(s) not found for architecture x86_64

Looks like the prototype in cod2.c does not match the type (one is targ_ullong,
one is ullong). I tried the obvious fix of changing the prototype in cod2.c,
but it complains of not knowing what ullong is there.

I'll let the more experienced take this :)

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #16 from github-bugzi...@puremagic.com ---
Commits pushed to stable at https://github.com/dlang/dmd

https://github.com/dlang/dmd/commit/3b34132a9dd6fd3ac6bd32c572dc6a9c2a75b8f8
fix Issue 16536 - OSX mangling mismatch in dmd's C++/D ABI

- just fix the specific dmd problem of picking up UINT64_TYPE (unsigned
  long long) on newer XCode/clang versions
- the D mangling of ulong should eventually be fixed to always match uint64_t

https://github.com/dlang/dmd/commit/131668b53dea8833e1d922f180800793e30703f2
Merge pull request #6180 from MartinNowak/fix16536

fix Issue 16536 - OSX mangling mismatch in dmd's C++/D ABI

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

github-bugzi...@puremagic.com changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |FIXED

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

Martin Nowak  changed:

   What|Removed |Added

   Keywords||pull

--- Comment #15 from Martin Nowak  ---
https://github.com/dlang/dmd/pull/6180

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #14 from Martin Nowak  ---
(In reply to Martin Nowak from comment #13)
> Let's please answer the question of comment 1, which typedef is active.
> Looks like you have a __UINT64_TYPE__ long long unsigned int now.

Indeed the __UINT64_TYPE__ isn't present on my version of XCode/clang.
It seems the better fix would be to rely on  and always use uint64_t.
That would be unsigned long long, so we'd need to change D's ulong to match
that.

I'll go with a workaround for newer XCode versions for now, but this needs to
be sorted out clearly.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #13 from Martin Nowak  ---
Let's please answer the question of comment 1, which typedef is active.
Looks like you have a __UINT64_TYPE__ long long unsigned int now.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #12 from Jacob Carlborg  ---
I have the same error with Clang 7.0.2 as well. I also have the same output.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #9 from Martin Nowak  ---
Looks like the difference between Xcode 7.3.x and older versions (tried 5.1.1)
is the lack of __UINTMAX_TYPE__.
Strangely David reports __UINTMAX_TYPE__ to be "long unsigned int" (compatible
w/ dmd's ulong).

Could someone who can reproduce the problem please run.

touch test.c
clang -dM -E test.c | grep UINTMAX

prints
#define __UINTMAX_TYPE__ long unsigned int

for me.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #11 from Steven Schveighoffer  ---
(In reply to David Nadlinger from comment #10)

> Just for the sake of completeness, I still get the same in both C and C++
> mode
> 

I get the same.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #10 from David Nadlinger  ---
(In reply to Martin Nowak from comment #9)
> Looks like the difference between Xcode 7.3.x and older versions (tried
> 5.1.1) is the lack of __UINTMAX_TYPE__.
> Strangely David reports __UINTMAX_TYPE__ to be "long unsigned int"
> (compatible w/ dmd's ulong).
> 
> Could someone who can reproduce the problem […]

Just for the sake of completeness, I still get the same in both C and C++ mode

$ clang -dM -E test.c | grep UINTMAX
#define __UINTMAX_C_SUFFIX__ UL
#define __UINTMAX_FMTX__ "lX"
#define __UINTMAX_FMTo__ "lo"
#define __UINTMAX_FMTu__ "lu"
#define __UINTMAX_FMTx__ "lx"
#define __UINTMAX_MAX__ 18446744073709551615UL
#define __UINTMAX_TYPE__ long unsigned int
#define __UINTMAX_WIDTH__ 64

$ clang++ -dM -E test.cpp | grep UINTMAX
#define __UINTMAX_C_SUFFIX__ UL
#define __UINTMAX_FMTX__ "lX"
#define __UINTMAX_FMTo__ "lo"
#define __UINTMAX_FMTu__ "lu"
#define __UINTMAX_FMTx__ "lx"
#define __UINTMAX_MAX__ 18446744073709551615UL
#define __UINTMAX_TYPE__ long unsigned int
#define __UINTMAX_WIDTH__ 64

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

Martin Nowak  changed:

   What|Removed |Added

 CC||timothee.co...@gmail.com

--- Comment #8 from Martin Nowak  ---
*** Issue 16596 has been marked as a duplicate of this issue. ***

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

Martin Nowak  changed:

   What|Removed |Added

 CC||c...@dawg.eu

--- Comment #7 from Martin Nowak  ---
Well, from the error message, it seems to me we're using `unsigned long long`
on the C side (typedef unsigned long long targ_ullong), but `unsigned long` on
the D side (alias targ_ullong = ulong).

(In reply to Jacob Carlborg from comment #6)

> I manually patch the compiler by replacing targ_size_t with size_t/d_size_t
> where the linker complains. I could create a PR but I don't know if it's the
> correct solution.

Well we want to always use unsigned long long on both sides (and for 32 and
64-bit dmds), b/c it must be able to hold any target's size_t (hence the name
targ_size_t).
It's a slightly different problem than the  fix for Issue 16000
https://github.com/dlang/dmd/pull/5788/files#diff-5910c4bad5eadf90caa32a46ab678aa0R12,
where we wanted size_t (4/8 bytes depending on 32/64-bit dmd) on both sides.

Not too sure, but I guess if you `typedef unsigned long targ_ullong` on OSX/64
it should work, but that stuff starts to become really messy.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #6 from Jacob Carlborg  ---
I manually patch the compiler by replacing targ_size_t with size_t/d_size_t
where the linker complains. I could create a PR but I don't know if it's the
correct solution.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

John Colvin  changed:

   What|Removed |Added

 CC||john.loughran.colvin@gmail.
   ||com

--- Comment #5 from John Colvin  ---
This is preventing testing 2.072.0-b1 on os x.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

Steven Schveighoffer  changed:

   What|Removed |Added

 CC||schvei...@yahoo.com

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #4 from Jacob Carlborg  ---
Should it be using d_size_t instead of targ_ullong?

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

Jacob Carlborg  changed:

   What|Removed |Added

 CC||d...@me.com

--- Comment #3 from Jacob Carlborg  ---
Same problem with Xcode/Clang 8. I looks like "size_t" is defined as
"__SIZE_TYPE__" which is defined as "long unsigned int". Replacing defining
"targ_ullong" as "size_t" fixes the problem for 64bit, but not for 32bit.

--

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

--- Comment #2 from David Nadlinger  ---
I'm using the stock Apple Clang shipped with Xcode 7.3.1. Is this the unsigned
long vs. unsigned long long mangling problem coming back to haunt us again?

---
$ clang -dM -E test.cpp
#define OBJC_NEW_PROPERTIES 1
#define _LP64 1
#define __APPLE_CC__ 6000
#define __APPLE__ 1
#define __ATOMIC_ACQUIRE 2
#define __ATOMIC_ACQ_REL 4
#define __ATOMIC_CONSUME 1
#define __ATOMIC_RELAXED 0
#define __ATOMIC_RELEASE 3
#define __ATOMIC_SEQ_CST 5
#define __BIGGEST_ALIGNMENT__ 16
#define __BLOCKS__ 1
#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
#define __CHAR16_TYPE__ unsigned short
#define __CHAR32_TYPE__ unsigned int
#define __CHAR_BIT__ 8
#define __CONSTANT_CFSTRINGS__ 1
#define __DBL_DECIMAL_DIG__ 17
#define __DBL_DENORM_MIN__ 4.9406564584124654e-324
#define __DBL_DIG__ 15
#define __DBL_EPSILON__ 2.2204460492503131e-16
#define __DBL_HAS_DENORM__ 1
#define __DBL_HAS_INFINITY__ 1
#define __DBL_HAS_QUIET_NAN__ 1
#define __DBL_MANT_DIG__ 53
#define __DBL_MAX_10_EXP__ 308
#define __DBL_MAX_EXP__ 1024
#define __DBL_MAX__ 1.7976931348623157e+308
#define __DBL_MIN_10_EXP__ (-307)
#define __DBL_MIN_EXP__ (-1021)
#define __DBL_MIN__ 2.2250738585072014e-308
#define __DECIMAL_DIG__ __LDBL_DECIMAL_DIG__
#define __DEPRECATED 1
#define __DYNAMIC__ 1
#define __ENVIRONMENT_MAC_OS_X_VERSION_MIN_REQUIRED__ 101100
#define __EXCEPTIONS 1
#define __FINITE_MATH_ONLY__ 0
#define __FLT_DECIMAL_DIG__ 9
#define __FLT_DENORM_MIN__ 1.40129846e-45F
#define __FLT_DIG__ 6
#define __FLT_EPSILON__ 1.19209290e-7F
#define __FLT_EVAL_METHOD__ 0
#define __FLT_HAS_DENORM__ 1
#define __FLT_HAS_INFINITY__ 1
#define __FLT_HAS_QUIET_NAN__ 1
#define __FLT_MANT_DIG__ 24
#define __FLT_MAX_10_EXP__ 38
#define __FLT_MAX_EXP__ 128
#define __FLT_MAX__ 3.40282347e+38F
#define __FLT_MIN_10_EXP__ (-37)
#define __FLT_MIN_EXP__ (-125)
#define __FLT_MIN__ 1.17549435e-38F
#define __FLT_RADIX__ 2
#define __FXSR__ 1
#define __GCC_ATOMIC_BOOL_LOCK_FREE 2
#define __GCC_ATOMIC_CHAR16_T_LOCK_FREE 2
#define __GCC_ATOMIC_CHAR32_T_LOCK_FREE 2
#define __GCC_ATOMIC_CHAR_LOCK_FREE 2
#define __GCC_ATOMIC_INT_LOCK_FREE 2
#define __GCC_ATOMIC_LLONG_LOCK_FREE 2
#define __GCC_ATOMIC_LONG_LOCK_FREE 2
#define __GCC_ATOMIC_POINTER_LOCK_FREE 2
#define __GCC_ATOMIC_SHORT_LOCK_FREE 2
#define __GCC_ATOMIC_TEST_AND_SET_TRUEVAL 1
#define __GCC_ATOMIC_WCHAR_T_LOCK_FREE 2
#define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1 1
#define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16 1
#define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2 1
#define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 1
#define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 1
#define __GNUC_GNU_INLINE__ 1
#define __GNUC_MINOR__ 2
#define __GNUC_PATCHLEVEL__ 1
#define __GNUC__ 4
#define __GNUG__ 4
#define __GXX_ABI_VERSION 1002
#define __GXX_RTTI 1
#define __GXX_WEAK__ 1
#define __INT16_C_SUFFIX__
#define __INT16_FMTd__ "hd"
#define __INT16_FMTi__ "hi"
#define __INT16_MAX__ 32767
#define __INT16_TYPE__ short
#define __INT32_C_SUFFIX__
#define __INT32_FMTd__ "d"
#define __INT32_FMTi__ "i"
#define __INT32_MAX__ 2147483647
#define __INT32_TYPE__ int
#define __INT64_C_SUFFIX__ LL
#define __INT64_FMTd__ "lld"
#define __INT64_FMTi__ "lli"
#define __INT64_MAX__ 9223372036854775807LL
#define __INT64_TYPE__ long long int
#define __INT8_C_SUFFIX__
#define __INT8_FMTd__ "hhd"
#define __INT8_FMTi__ "hhi"
#define __INT8_MAX__ 127
#define __INT8_TYPE__ signed char
#define __INTMAX_C_SUFFIX__ L
#define __INTMAX_FMTd__ "ld"
#define __INTMAX_FMTi__ "li"
#define __INTMAX_MAX__ 9223372036854775807L
#define __INTMAX_TYPE__ long int
#define __INTMAX_WIDTH__ 64
#define __INTPTR_FMTd__ "ld"
#define __INTPTR_FMTi__ "li"
#define __INTPTR_MAX__ 9223372036854775807L
#define __INTPTR_TYPE__ long int
#define __INTPTR_WIDTH__ 64
#define __INT_FAST16_FMTd__ "hd"
#define __INT_FAST16_FMTi__ "hi"
#define __INT_FAST16_MAX__ 32767
#define __INT_FAST16_TYPE__ short
#define __INT_FAST32_FMTd__ "d"
#define __INT_FAST32_FMTi__ "i"
#define __INT_FAST32_MAX__ 2147483647
#define __INT_FAST32_TYPE__ int
#define __INT_FAST64_FMTd__ "ld"
#define __INT_FAST64_FMTi__ "li"
#define __INT_FAST64_MAX__ 9223372036854775807L
#define __INT_FAST64_TYPE__ long int
#define __INT_FAST8_FMTd__ "hhd"
#define __INT_FAST8_FMTi__ "hhi"
#define __INT_FAST8_MAX__ 127
#define __INT_FAST8_TYPE__ signed char
#define __INT_LEAST16_FMTd__ "hd"
#define __INT_LEAST16_FMTi__ "hi"
#define __INT_LEAST16_MAX__ 32767
#define __INT_LEAST16_TYPE__ short
#define __INT_LEAST32_FMTd__ "d"
#define __INT_LEAST32_FMTi__ "i"
#define __INT_LEAST32_MAX__ 2147483647
#define __INT_LEAST32_TYPE__ int
#define __INT_LEAST64_FMTd__ "ld"
#define __INT_LEAST64_FMTi__ "li"
#define __INT_LEAST64_MAX__ 9223372036854775807L
#define __INT_LEAST64_TYPE__ long int
#define __INT_LEAST8_FMTd__ "hhd"
#define __INT_LEAST8_FMTi__ "hhi"
#define __INT_LEAST8_MAX__ 127
#define __INT_LEAST8_TYPE__ signed char
#define __INT_MAX__ 2147483647
#define __LDBL_DECIMAL_DIG__ 21
#define 

[Issue 16536] DMD master does not build on OS X 10.11.6/Xcode 7.3.1

[via Digitalmars-d-bugs]

https://issues.dlang.org/show_bug.cgi?id=16536

Walter Bright  changed:

   What|Removed |Added

 CC||bugzi...@digitalmars.com

--- Comment #1 from Walter Bright  ---
Note the following from cdef.h:

#if defined(__UINT64_TYPE__)
typedef __INT64_TYPE__ targ_llong;
typedef __UINT64_TYPE__targ_ullong;
#elif defined(__UINTMAX_TYPE__)
typedef __INTMAX_TYPE__targ_llong;
typedef __UINTMAX_TYPE__   targ_ullong;
#else
typedef long long  targ_llong;
typedef unsigned long long targ_ullong;
#endif

Which of these is the C++ compiler on your machine doing? Is there another
macro it should be doing? Getting this right is necessary for the name mangling
to match what D generates for 'long' and 'ulong'.

The way to determine the pertinent macro is to have your C++ compiler dump all
its predefined macros, once for -m32 and once for -m64. Diff the two, and find
the relevant macro.

--