Intermittent segfault instead of StackOverflowError when promote_rule is defined in a certain way

[ikirill]

I don't know how to phrase it better because I don't understand what it's doing. The minimal example is reduced from the ApproxFun example below:

julia> struct A <: Number end

julia> Base.promote_rule(::Type{A}, ::Type{N}) where {N<:Number} = N

julia> promote(A(), false)
Segmentation fault: 11

julia> versioninfo()
Julia Version 1.1.0-DEV.352
Commit 4851fab9b4 (2018-10-01 13:53 UTC)
Platform Info:
  OS: macOS (x86_64-apple-darwin14.5.0)
  CPU: Intel(R) Core(TM) i7-4770HQ CPU @ 2.20GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.1 (ORCJIT, haswell)

In ApproxFun:

(v1.1) pkg> activate nightly

(nightly) pkg> st
    Status `~/Sandboxes/Julia-Misc/nightly/Project.toml`
  [28f2ccd6] ApproxFun v0.9.0+ #522d95f (https://github.com/JuliaApproximation/ApproxFun.jl.git)

julia> using ApproxFun
[ Info: Recompiling stale cache file /Users/kirill/.julia/compiled/v1.1/ApproxFun/jGqLz.ji for ApproxFun [28f2ccd6-bb30-5033-b560-165f7b14dc2f]

julia> Fun(true)
true on ApproxFun.AnyDomain()

julia> Multiplication(Fun(true))
ERROR: StackOverflowError:
Stacktrace:
 [1] promote_type at ./promotion.jl:221 [inlined]
 [2] promote_result(::Type, ::Type, ::Type{Bool}, ::Type{ApproxFun.UnsetNumber}) at ./promotion.jl:239
 ... (the last 2 lines are repeated 79998 more times)
 [159999] promote_type at ./promotion.jl:221 [inlined]

julia> Multiplication(Fun(true))
Segmentation fault: 11

[fredrikekre]

Looks a lot like #29639.

[dlfivefifty]

Hmm, I always assumed that Segfaults can happen whenever you have stack overflows, since this seems to happen often whenever there's an accidental recursive self-call.

[dlfivefifty]

Ah, there's already an issue about recursive calls causing Segfaults: #17109

I think this issue can be closed as redundant.

[ikirill]

@dlfivefifty Usually, the correct behaviour when executing an incorrect/invalid program is to generate the specific type of error that is the correct error. Generating an incorrect error or segfaulting should definitely be a bug, even though in both cases your program doesn't work right. And a correct runtime should never segfault anyway so long as the program doesn't do anything unsafe.