Unhandled exception when executing programs that use the 'parallel_for' primitive.

[H-N41K]

open Domainslib

let num_domains = try int_of_string Sys.argv.(1) with _ -> 4
let n = try int_of_string Sys.argv.(2) with _ -> 100
let a = Array.create_float n

let _ =
let pool = Task.setup_pool ~num_additional_domains:(num_domains - 1) () in
try Task.parallel_for pool ~start:0
~finish:(n - 1) ~body:(fun i -> Array.set a i (Random.float 1000.)) with e -> (Printexc.default_uncaught_exception_handler e (Printexc.get_raw_backtrace ()));
Task.teardown_pool pool

I am getting the following error when I try to run the above code with num_domains > 1.

hemen@hemen-VirtualBox:~/Desktop/multicore_ocaml_practice$ dune exec ./float_init_par.exe
Fatal error: exception Unhandled

After tracing the stack I got the following:

hemen@hemen-VirtualBox:~/Desktop/multicore_ocaml_practice$ dune build ./float_init_par.ml

hemen@hemen-VirtualBox:~/Desktop/multicore_ocaml_practice$ OCAMLRUNPARAM=b ./_build/default/float_init_par.exe
Fatal error: exception Unhandled
Raised by primitive operation at Domainslib__Task.parallel_for.work in file "lib/task.ml", line 198, characters 6-26
Called from Domainslib__Task.parallel_for.work in file "lib/task.ml", line 198, characters 6-26
Called from Domainslib__Task.parallel_for.work in file "lib/task.ml", line 198, characters 6-26
Called from Dune__exe__Float_init_par in file "float_init_par.ml", line 9, characters 6-107

I am unable to resolve what caused the exception to be raised at Domainslib__Task.parallel_for.work
I am getting this issue when I set num_domains to be greater than 1 in every program that uses the parallel_for primitive.
I used ocamldebug to debug this , however the debugger crashes with a segmentation fault as shown below:

hemen@hemen-VirtualBox:~/Desktop/multicore_ocaml_practice$ dune build ./float_init_par.bc

hemen@hemen-VirtualBox:~/Desktop/multicore_ocaml_practice$ ocamldebug ./_build/default/float_init_par.bc
	OCaml Debugger version 4.12.0+domains+effects

(ocd) r
Loading program... done.
Lost connection with process 14884 (active process)
at time 0
Trying to recover...
Segmentation fault (core dumped)

Kindly suggest how to resolve this.

[Sudha247]

Thanks for reporting! This repo was a bit behind domainslib.0.4.0, I've fixed it in #12. The programs should now work with #12, please let me know otherwise.

Also, I'm afraid ocamldebug is currently not supported with Multicore. We recommend using gdb or rr for debugging Multicore OCaml programs.

[Sudha247]

Fixed by #12, I'm closing this. Feel free to make new issues for any other questions.

[H-N41K]

Thanks for resolving the issue. The programs work fine now. 👍