numbers test crashes under wine

[Keno]

If my debugger stack is to be trusted (and I'm not 100% sure it is), then this originates as a divide by zero in LLVM:

=> 0x000000006acc1581 <_ZN12_GLOBAL__N_111DAGCombiner19visitVECTOR_SHUFFLEEPN4llvm6SDNodeE+15969>:	idivl  (%rcx)
   0x000000006acc1583 <_ZN12_GLOBAL__N_111DAGCombiner19visitVECTOR_SHUFFLEEPN4llvm6SDNodeE+15971>:	mov    0x83536(%rip),%rdx        # 0x6ad44ac0 <_ZN4llvm16DAGTypeLegalizer25SoftenFloatRes_FP16_TO_FPEPNS_6SDNodeE+80>
   0x000000006acc158a <_ZN12_GLOBAL__N_111DAGCombiner19visitVECTOR_SHUFFLEEPN4llvm6SDNodeE+15978>:	mov    $0x2,%ecx
   0x000000006acc158f <_ZN12_GLOBAL__N_111DAGCombiner19visitVECTOR_SHUFFLEEPN4llvm6SDNodeE+15983>:	mov    %eax,(%rdx)
   0x000000006acc1591 <_ZN12_GLOBAL__N_111DAGCombiner19visitVECTOR_SHUFFLEEPN4llvm6SDNodeE+15985>:	call   0x6acc1540 <_ZN12_GLOBAL__N_111DAGCombiner19visitVECTOR_SHUFFLEEPN4llvm6SDNodeE+15904>

[Keno]

          0x6acc0000         0x6acc1000     0x1000        0x0 /home/keno/julia-mywin/usr/bin/libgmp.dll
          0x6acc1000         0x6ad36000    0x75000        0x0
          0x6ad36000         0x6ad37000     0x1000        0x0

Actually, might be in gmp and gdb is confused. Will look into it.

[Keno]

Alright, small GDB tweak gives the much more plausible:

   0x000000006acc1570 <+0>:	sub    $0x28,%rsp
   0x000000006acc1574 <+4>:	mov    0x83435(%rip),%rcx        # 0x6ad449b0 <.refptr.__gmp_0>
   0x000000006acc157b <+11>:	mov    $0xa,%eax
   0x000000006acc1580 <+16>:	cltd
=> 0x000000006acc1581 <+17>:	idivl  (%rcx)
   0x000000006acc1583 <+19>:	mov    0x83536(%rip),%rdx        # 0x6ad44ac0 <.refptr.__gmp_junk>
   0x000000006acc158a <+26>:	mov    $0x2,%ecx
   0x000000006acc158f <+31>:	mov    %eax,(%rdx)
   0x000000006acc1591 <+33>:	call   0x6acc1540 <__gmp_exception>

So the SIGFPE here is expected. Let's see why that segfaults.