"Fatal Python error: Aborted" in test_dparray_T

[PokhodenkoSA]

Test:
dpctl/tests/test_dparray.py::Test_dparray::test_dparray_T

Environment:

runner@fv-az213-658:~/work/dpctl/dpctl$ clinfo
Number of platforms                               0

gdb:

Thread 1 "python" received signal SIGABRT, Aborted.
__GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:50
50      ../sysdeps/unix/sysv/linux/raise.c: No such file or directory.
(gdb) bt
#0  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:50
#1  0x00007ffff7a81859 in __GI_abort () at abort.c:79
#2  0x00007ffff3a3d951 in ?? () from /lib/x86_64-linux-gnu/libstdc++.so.6
#3  0x00007ffff3a4947c in ?? () from /lib/x86_64-linux-gnu/libstdc++.so.6
#4  0x00007ffff3a494e7 in std::terminate() () from /lib/x86_64-linux-gnu/libstdc++.so.6
#5  0x00007ffff3a49799 in __cxa_throw () from /lib/x86_64-linux-gnu/libstdc++.so.6
#6  0x00007ffff5c0265f in cl::sycl::device_selector::select_device() const () from /opt/intel/oneapi/compiler/2021.1.1/linux/lib/libsycl.so.5
#7  0x00007ffff5c8581e in (anonymous namespace)::QueueManager::getQueueStack() () from /home/runner/work/dpctl/dpctl/dpctl/libDPCTLSyclInterface.so
#8  0x00007ffff5c85d28 in DPCTLQueueMgr_GetCurrentQueue () from /home/runner/work/dpctl/dpctl/dpctl/libDPCTLSyclInterface.so
#9  0x00007ffff2f72110 in __pyx_f_5dpctl_19_sycl_queue_manager_17_SyclQueueManager_get_current_queue (__pyx_v_self=<optimized out>, __pyx_skip_dispatch=__pyx_skip_dispatch@entry=1) at dpctl/_sycl_queue_manager.cpp:3206
#10 0x00007ffff2f72492 in __pyx_pf_5dpctl_19_sycl_queue_manager_17_SyclQueueManager_8get_current_queue (__pyx_v_self=<optimized out>) at dpctl/_sycl_queue_manager.cpp:3257
#11 __pyx_pw_5dpctl_19_sycl_queue_manager_17_SyclQueueManager_9get_current_queue (__pyx_v_self=<optimized out>, unused=<optimized out>) at dpctl/_sycl_queue_manager.cpp:3241
#12 0x00007ffff2f725c1 in __Pyx_PyObject_CallMethO (arg=0x0, func=0x7fffeb680f50) at dpctl/_sycl_queue_manager.cpp:20470
#13 __Pyx_PyObject_CallNoArg (func=0x7fffeb680f50) at dpctl/_sycl_queue_manager.cpp:20576
#14 __pyx_f_5dpctl_19_sycl_queue_manager_get_current_queue (__pyx_skip_dispatch=<optimized out>) at dpctl/_sycl_queue_manager.cpp:3893
#15 0x00007ffff2fc0108 in __pyx_f_5dpctl_6memory_7_memory_7_Memory__cinit_alloc (__pyx_v_self=0x7fffeb0a4d70, __pyx_v_alignment=0, __pyx_v_nbytes=8192, __pyx_v_ptr_type=0x7ffff5ca2d20, __pyx_v_queue=0x7ffff7f59710 <_Py_NoneStruct>)
    at dpctl/memory/_memory.cpp:4530
#16 0x00007ffff2fb9505 in __pyx_pf_5dpctl_6memory_7_memory_15MemoryUSMShared___cinit__ (__pyx_v_copy=<optimized out>, __pyx_v_queue=<optimized out>, __pyx_v_alignment=0, __pyx_v_other=<optimized out>, __pyx_v_self=0x7fffeb0a4d70)
    at dpctl/memory/_memory.cpp:8262
#17 __pyx_pw_5dpctl_6memory_7_memory_15MemoryUSMShared_1__cinit__ (__pyx_kwds=<optimized out>, __pyx_args=<optimized out>, __pyx_v_self=0x7fffeb0a4d70) at dpctl/memory/_memory.cpp:8215
#18 __pyx_tp_new_5dpctl_6memory_7_memory_MemoryUSMShared (t=<optimized out>, a=<optimized out>, k=<optimized out>) at dpctl/memory/_memory.cpp:22933
#19 0x00007ffff7d91a8a in type_call (type=type@entry=0x7ffff2fd1e80 <__pyx_type_5dpctl_6memory_7_memory_MemoryUSMShared>, args=args@entry=0x7fffeb0ac190, kwds=kwds@entry=0x0) at Objects/typeobject.c:951

[oleksandr-pavlyk]

Simpler:

(idp) [17:35:00 ansatnuc04 dpctl]$ SYCL_DEVICE_FILTER=cuda python -c "import dpctl; dpctl.get_current_queue()"
terminate called after throwing an instance of 'cl::sycl::runtime_error'
  what():  No device of requested type available. Please check https://software.intel.com/content/www/us/en/develop/articles/intel-oneapi-dpcpp-system-requirements.html -1 (CL_DEVICE_NOT_FOUND)
Aborted (core dumped)

[oleksandr-pavlyk]

Evidently DS.select_device() throws an exception different that is being caught...

Indeed using

]$ cat default_select_device.cpp
#include <iostream>
#include <CL/sycl.hpp>

int main() {
    auto DS = sycl::default_selector();
    auto p = new sycl::device(DS.select_device());

    delete p;
    return 0;
}

and compiling it using dpcpp default_select_device.cpp -o crash.out using

Intel(R) oneAPI DPC++ Compiler 2021.2.0 (2021.2.0.20210317)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /localdisk/intel/oneapi/compiler/2021.2.0/linux/bin

I also get the crash:

]$ SYCL_DEVICE_FILTER=cuda ./crash.out
terminate called after throwing an instance of 'cl::sycl::runtime_error'
  what():  No device of requested type available. Please check https://software.intel.com/content/www/us/en/develop/articles/intel-oneapi-dpcpp-system-requirements.html -1 (CL_DEVICE_NOT_FOUND)
Aborted (core dumped)

so the code in dpct-capi/source/dpctl_sycl_queue_manager.cpp also needs to handle sycl::runtime_error.

[diptorupd]

so the code in dpct-capi/source/dpctl_sycl_queue_manager.cpp also needs to handle sycl::runtime_error.

Yes, and the same thing is there in platform causing the other crash.

[diptorupd]

Yes, and the same thing is there in platform causing the other crash.

This is what happens when you keep hacking when your attention-levels drop. Bad misses.