[Pkg-opencl-devel] Bug#892426: pocl: FTBFS on most non-x86 architectures: 12-16% of tests pass

Fri Mar 9 02:51:30 UTC 2018

Source: pocl
Version: 1.1~rc2-1
Severity: normal
Tags: upstream
User: debian-mips at lists.debian.org
Usertags: mips mips64el mipsel

Builds of pocl for most non-x86 architectures, including several
release architectures, failed with heavy test suite errors: between
12% and 16% of tests passed, depending on the architecture [1].  The
only exceptions are arm64, which succeeded, and armhf, on which 43% of
tests passed [2] -- still not great, but better.

More specifically, the failures in question occurred on armel, mips,
mips64el, mipsel, ppc64el, s390x, and the non-release architectures
powerpc, ppc64, and sparc64.  94 tests failed identically on all nine
of these architectures, as detailed below.  In addition, some tests
failed on only a subset of architectures:

  25 - kernel/test_shuffle_half (Failed): armel
  46/47 - regression/infinite_loop_REPL (Failed): sparc64
  60/61 - regression/infinite_loop_LOOPS (Failed): sparc64
  67/68 - regression/passing_a_constant_array_as_an_arg (Failed): sparc64
  79/80 - runtime/test_event_cycle (Failed): powerpc
  83/84 - runtime/clCreateKernel (Failed): sparc64
  84/85 - runtime/clGetKernelArgInfo (Failed): sparc64
  113/114 - examples/example1_dot_product (Child aborted): all but sparc64
  114/115 - examples/example2_matrix_transpose (Child aborted): all but sparc64
  115/116 - examples/example2_matrix_transpose_alocals (Child aborted): all but sparc64
  116/117 - poclcc (Child aborted): all but sparc64
  118/119 - examples/trig (Child aborted): all but sparc64
  119/120 - EinsteinToolkit (Child aborted): all but mips*
  120/121 - EinsteinToolkit_SubDev (Child aborted): all but mips*

NB: Test numbers vary somewhat because kernel/test_shuffle_half is #25
on armel and does not exist on the other eight affected architectures.

Could you please take a look?

Thanks!

[1] https://buildd.debian.org/status/logs.php?pkg=pocl&ver=1.1%7Erc2-1
[2] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=888063

Here's the list of tests that failed on all nine of these architectures:

2 - kernel/test_as_type (Failed)
3 - kernel/test_convert_type_1 (Failed)
4 - kernel/test_convert_type_2 (Failed)
5 - kernel/test_convert_type_4 (Failed)
6 - kernel/test_convert_type_8 (Failed)
7 - kernel/test_convert_type_16 (Failed)
8 - kernel/test_bitselect (Failed)
9 - kernel/test_hadd_loopvec (Failed)
10 - kernel/test_hadd_loops (Failed)
11 - kernel/test_min_max (Failed)
12 - kernel/test_length_distance (Failed)
13 - kernel/test_fmin_fmax_fma (Failed)
14 - kernel/test_convert_sat_regression (Failed)
15 - kernel/test_rotate (Failed)
16 - kernel/test_fabs (Failed)
17 - kernel/test_short16 (Failed)
18 - kernel/test_frexp_modf (Failed)
19 - kernel/test_local_struct_array (Failed)
20 - kernel/test_sampler_address_clamp (Failed)
21 - kernel/test_image_query_funcs (Failed)
22 - kernel/test_shuffle_char (Failed)
23 - kernel/test_shuffle_short (Failed)
24 - kernel/test_shuffle_ushort (Failed)
26/27 - kernel/test_shuffle_uint (Failed)
27/28 - kernel/test_shuffle_float (Failed)
28/29 - kernel/test_shuffle_long (Failed)
29/30 - kernel/test_shuffle_ulong (Failed)
30/31 - kernel/test_shuffle_double (Failed)
31/32 - kernel/test_printf (Failed)
32/33 - kernel/test_sizeof_uint (Failed)
33/34 - regression/test_issue_231 (Failed)
34/35 - regression/test_issue_445 (Failed)
35/36 - regression/test_issue_553 (Failed)
37/38 - regression/phi_nodes_not_replicated_REPL (Failed)
38/39 - regression/issues_with_local_pointers_REPL (Failed)
39/40 - regression/barrier_between_two_for_loops_REPL (Failed)
40/41 - regression/simple_for-loop_with_a_barrier_inside_REPL (Failed)
41/42 - regression/for-loop_with_computation_after_the_brexit_REPL (Failed)
42/43 - regression/for-loop_with_a_variable_iteration_count_REPL (Failed)
43/44 - regression/early_return_before_a_barrier_region_REPL (Failed)
44/45 - regression/id-dependent_computation_before_kernel_exit_REPL (Failed)
45/46 - regression/barrier_just_before_return_REPL (Failed)
47/48 - regression/undominated_variable_from_conditional_barrier_handling_REPL (Failed)
48/49 - regression/assigning_a_loop_iterator_variable_to_a_private_makes_it_local_REPL (Failed)
49/50 - regression/assigning_a_loop_iterator_variable_to_a_private_makes_it_local_2_REPL (Failed)
50/51 - regression/test_program_from_binary_with_local_1_1_1_REPL (Failed)
51/52 - regression/phi_nodes_not_replicated_LOOPS (Failed)
52/53 - regression/issues_with_local_pointers_LOOPS (Failed)
53/54 - regression/barrier_between_two_for_loops_LOOPS (Failed)
54/55 - regression/simple_for-loop_with_a_barrier_inside_LOOPS (Failed)
55/56 - regression/for-loop_with_computation_after_the_brexit_LOOPS (Failed)
56/57 - regression/for-loop_with_a_variable_iteration_count_LOOPS (Failed)
57/58 - regression/early_return_before_a_barrier_region_LOOPS (Failed)
58/59 - regression/id-dependent_computation_before_kernel_exit_LOOPS (Failed)
59/60 - regression/barrier_just_before_return_LOOPS (Failed)
61/62 - regression/undominated_variable_from_conditional_barrier_handling_LOOPS (Failed)
62/63 - regression/assigning_a_loop_iterator_variable_to_a_private_makes_it_local_LOOPS (Failed)
63/64 - regression/assigning_a_loop_iterator_variable_to_a_private_makes_it_local_2_LOOPS (Failed)
64/65 - regression/test_program_from_binary_with_local_1_1_1_LOOPS (Failed)
65/66 - regression/setting_a_buffer_argument_to_NULL_causes_a_segfault (Failed)
66/67 - regression/clSetKernelArg_overwriting_the_previous_kernel's_args (Failed)
68/69 - regression/case_with_multiple_variable_length_loops_and_a_barrier_in_one (Failed)
69/70 - regression/autolocals_in_constexprs (Failed)
71/72 - regression/vector_kernel_arguments (Failed)
75/76 - runtime/clCreateProgramWithBinary (Failed)
76/77 - runtime/clBuildProgram (Failed)
77/78 - runtime/test_kernel_cache_includes (Failed)
78/79 - runtime/clFinish (Failed)
85/86 - runtime/clSetEventCallback (Failed)
87/88 - runtime/clCreateKernelsInProgram (Failed)
88/89 - runtime/clCreateSubDevices (Failed)
90/91 - runtime/test_enqueue_kernel_from_binary (Failed)
93/94 - workgroup/unconditional_barriers_REPL (Failed)
94/95 - workgroup/unbarriered_for_loops_REPL (Failed)
95/96 - workgroup/barriered_for_loops_REPL (Failed)
96/97 - workgroup/conditional_barrier_REPL (Failed)
97/98 - workgroup/b_loop_with_none_of_the_WIs_reaching_the_barrier_REPL (Failed)
98/99 - workgroup/forcing_horizontal_parallelization_to_some_outer_loops_REPL (Failed)
99/100 - workgroup/loop_with_two_paths_to_the_latch_REPL (Failed)
100/101 - workgroup/b_loop_with_two_latches_REPL (Failed)
101/102 - workgroup/workgroup_sizes_work_items_get_wrong_ids_REPL (Failed)
102/103 - workgroup/different_implicit_barrier_injection_scenarios_LOOPS (Failed)
103/104 - workgroup/unconditional_barriers_LOOPS (Failed)
104/105 - workgroup/unbarriered_for_loops_LOOPS (Failed)
105/106 - workgroup/barriered_for_loops_LOOPS (Failed)
106/107 - workgroup/conditional_barrier_LOOPS (Failed)
107/108 - workgroup/b_loop_with_none_of_the_WIs_reaching_the_barrier_LOOPS (Failed)
108/109 - workgroup/forcing_horizontal_parallelization_to_some_outer_loops_LOOPS (Failed)
109/110 - workgroup/loop_with_two_paths_to_the_latch_LOOPS (Failed)
110/111 - workgroup/b_loop_with_two_latches_LOOPS (Failed)
111/112 - workgroup/workgroup_sizes_work_items_get_wrong_ids_LOOPS (Failed)
112/113 - workgroup/issue_548_convergent_propagation_LOOPS (Failed)
117/118 - examples/scalarwave (Child aborted)

-- 
Aaron M. Ucko, KB1CJC (amu at alum.mit.edu, ucko at debian.org)
http://www.mit.edu/~amu/ | http://stuff.mit.edu/cgi/finger/?amu@monk.mit.edu