InexactError when running on 32-bit platform

I am reliably hitting an InexactError when using @benchmark on a 32-bit ARM platform (a Zynq7020 with a Cortex A9).

Here is an example of when it does and doesn't happen:

julia> using BenchmarkTools

julia> @benchmark sleep(1)
BenchmarkTools.Trial: 
  memory estimate:  64 bytes
  allocs estimate:  5
  --------------
  minimum time:     1.001 s (0.00% GC)
  median time:      1.002 s (0.00% GC)
  mean time:        1.002 s (0.00% GC)
  maximum time:     1.002 s (0.00% GC)
  --------------
  samples:          5
  evals/sample:     1

julia> @benchmark sleep(5)
ERROR: InexactError: trunc(Int32, 5.005859015e9)
Stacktrace:
 [1] trunc at ./float.jl:682 [inlined]
 [2] ceil(::Type{Int32}, ::Float64) at ./float.jl:365
 [3] tune!(::BenchmarkTools.Benchmark{Symbol("##benchmark#258")}, ::BenchmarkTools.Parameters; progressid::Nothing, nleaves::Float64, ndone::Float64, verbose::Bool, pad::String, kwargs::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at /root/.julia/packages/BenchmarkTools/eCEpo/src/execution.jl:209
 [4] tune! at /root/.julia/packages/BenchmarkTools/eCEpo/src/execution.jl:208 [inlined] (repeats 2 times)
 [5] top-level scope at /root/.julia/packages/BenchmarkTools/eCEpo/src/execution.jl:288

Start at src/execution.jl around line 209, where the stack trace shows tune! failing, and reproduce the issue with @benchmark sleep(5) on a 32-bit ARM platform. Compare it with the working sleep(1) case; done means the longer benchmark completes without InexactError.