Runtime crash with unreachable reached

[AaronGhost]

Hi, thanks for developing Julia!

I encountered a runtime crash on 1.10.4 while including the script below. The runtime crash still exists in 1.11.0rc1 and nightly. I tried to reduce it to the best of my ability:

• A StructB contains a StructA and an additional array
• Both StructA and StructB have a "parameter" version (AParams and BParams)
• a StructA/B can be with buildA/B and A/BParams
• Calling buildB triggers a crash, only when using a boolean array (which does not depend on T).

abstract type AbstractA{T <: AbstractFloat} end
const UnionT{T, N} = Union{Array{T, N}, Array{Bool, N}}

struct StructA{T} <: AbstractA{T}
    x::T
end

struct AParams
    x
end

function buildA(params::AParams)
    return StructA(params.x)
end

struct StructB{T, DIM, S <: AbstractA{T}, SMT<: UnionT{T, DIM}}
    StructB(::S, arr::SMT) where {T, DIM, S <: AbstractA{T}, SMT <: UnionT{T, DIM}} = new{T, DIM, S, SMT}()
end

struct BParams{T}
    arr::UnionT{T}
    op::AParams
end

function buildB(objparams::BParams)
    op = buildA(objparams.op)
    return StructB(op, objparams.arr)
end

opparams = AParams(1.0f0)
objparams = BParams(ones(Bool, (10,10)), opparams)
obj = buildB(objparams)

Please find below the crash log

Unreachable reached at 0000023570eb17ff

Please submit a bug report with steps to reproduce this fault, and any error messages that follow (in their entirety). Thanks.
Exception: EXCEPTION_ILLEGAL_INSTRUCTION at 0x23570eb17ff -- buildB at D:\Documents\Julia\reproduceissue.jl:27
in expression starting at D:\Documents\Julia\reproduceissue.jl:32
buildB at D:\Documents\Julia\reproduceissue.jl:27
unknown function (ip: 0000023570eb183b)
jl_apply at C:/workdir/src\julia.h:1982 [inlined]
do_call at C:/workdir/src\interpreter.c:126
eval_value at C:/workdir/src\interpreter.c:223
eval_stmt_value at C:/workdir/src\interpreter.c:174 [inlined]
eval_body at C:/workdir/src\interpreter.c:635
jl_interpret_toplevel_thunk at C:/workdir/src\interpreter.c:775
jl_toplevel_eval_flex at C:/workdir/src\toplevel.c:934
jl_toplevel_eval_flex at C:/workdir/src\toplevel.c:877
ijl_toplevel_eval at C:/workdir/src\toplevel.c:943 [inlined]
ijl_toplevel_eval_in at C:/workdir/src\toplevel.c:985
eval at .\boot.jl:385 [inlined]
include_string at .\loading.jl:2076
_include at .\loading.jl:2136
include at .\client.jl:489
unknown function (ip: 0000023570eb052b)
jl_apply at C:/workdir/src\julia.h:1982 [inlined]
do_call at C:/workdir/src\interpreter.c:126
eval_value at C:/workdir/src\interpreter.c:223
eval_stmt_value at C:/workdir/src\interpreter.c:174 [inlined]
eval_body at C:/workdir/src\interpreter.c:635
jl_interpret_toplevel_thunk at C:/workdir/src\interpreter.c:775
jl_toplevel_eval_flex at C:/workdir/src\toplevel.c:934
jl_toplevel_eval_flex at C:/workdir/src\toplevel.c:877
ijl_toplevel_eval at C:/workdir/src\toplevel.c:943 [inlined]
ijl_toplevel_eval_in at C:/workdir/src\toplevel.c:985
eval at .\boot.jl:385 [inlined]
eval_user_input at C:\workdir\usr\share\julia\stdlib\v1.10\REPL\src\REPL.jl:150
repl_backend_loop at C:\workdir\usr\share\julia\stdlib\v1.10\REPL\src\REPL.jl:246
#start_repl_backend#46 at C:\workdir\usr\share\julia\stdlib\v1.10\REPL\src\REPL.jl:231
#run_repl#59 at C:\workdir\usr\share\julia\stdlib\v1.10\REPL\src\REPL.jl:389
jfptr_run_repl_95791.1 at C:\Users\anon\AppData\Local\Programs\Julia-1.10.4\lib\julia\sys.dll (unknown line)
#1013 at .\client.jl:432
jfptr_YY.1013_86566.1 at C:\Users\anon\AppData\Local\Programs\Julia-1.10.4\lib\julia\sys.dll (unknown line)
jl_apply at C:/workdir/src\julia.h:1982 [inlined]
jl_f__call_latest at C:/workdir/src\builtins.c:812
#invokelatest#2 at .\essentials.jl:892 [inlined]
invokelatest at .\essentials.jl:889 [inlined]
run_main_repl at .\client.jl:416
exec_options at .\client.jl:333
_start at .\client.jl:552
jfptr__start_86591.1 at C:\Users\anon\AppData\Local\Programs\Julia-1.10.4\lib\julia\sys.dll (unknown line)
jl_apply at C:/workdir/src\julia.h:1982 [inlined]
true_main at C:/workdir/src\jlapi.c:582
jl_repl_entrypoint at C:/workdir/src\jlapi.c:731
mainCRTStartup at C:/workdir/cli\loader_exe.c:58
BaseThreadInitThunk at C:\Windows\System32\KERNEL32.DLL (unknown line)
RtlUserThreadStart at C:\Windows\SYSTEM32\ntdll.dll (unknown line)
Allocations: 2908 (Pool: 2900; Big: 8); GC: 0

Output from versioninfo:

Julia Version 1.10.4
Commit 48d4fd4843 (2024-06-04 10:41 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: 16 × 11th Gen Intel(R) Core(TM) i9-11900K @ 3.50GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-15.0.7 (ORCJIT, rocketlake)
Threads: 1 default, 0 interactive, 1 GC (on 16 virtual cores)

[vtjnash]

Looks like this is assumed to be un-constructible

│   %6 = Core.apply_type(%1, %2, %3, %4, %5)::Type{StructB{_A, _B, S, SMT}} where {_A, _B, S<:StructA, SMT<:(Array{Bool})}
│        %new(%6)::Union{}

due to this incorrect type intersection result

julia> typeintersect(StructB, StructB{_A, _B, S, SMT} where {_A, _B, S<:StructA, SMT<:(Array{Bool})})
Union{}

This appears to occur due to an incorrect assignment of T = Bool terminating the algorithm, which is not instead correctly propagating the constraint from AbstractA against the Union to instead derive the correct assignment of T = _A

[JeffBezanson]

@AaronGhost Thanks for the useful report. A good workaround in a case like this is to remove or simplify the constraints on struct type parameters.

bisect points to 51e3bc3, but I guess that only exposed this intersection problem. I will try to separately bisect the intersection bug.

[JeffBezanson]

Bisect points to 748149e @N5N3

It looks like this commit was a precision improvement and not a correctness fix, so it should be ok to revert it?

[N5N3]

Ah, yes. That commit brings about more correctness issues. Reverting seems the best choice here.