SIGSEGV happens when setting a break

(cuda-gdb) b nest::RingBuffer::get_value(long)

cuda-gdb has received a SIGSEGV and will attempt to get its own backtrace.

       cuda-gdb| segv_handler() +0x4a| ?????
       cuda-gdb| iterate_over_all_matching_symtabs() +0x110
       cuda-gdb| add_matching_symbols_to_info() +0x288
       cuda-gdb| find_linespec_symbols() +0xab7
       cuda-gdb| parse_linespec() +0x96c
       cuda-gdb| event_location_to_sals() +0x30d
       cuda-gdb| decode_line_full() +0xc9
       cuda-gdb| parse_breakpoint_sals() +0xdc
       cuda-gdb| create_breakpoint() +0xc8
       cuda-gdb| break_command_1() +0xb3
       cuda-gdb| cmd_func() +0x29
       cuda-gdb| execute_command() +0x42b
       cuda-gdb| command_handler() +0x6c
       cuda-gdb| command_line_handler() +0x38
       cuda-gdb| gdb_rl_callback_handler() +0x3c
       cuda-gdb| rl_callback_read_char() +0x113
       cuda-gdb| gdb_rl_callback_read_char_wrapper_noexcept() +0x4e
       cuda-gdb| gdb_rl_callback_read_char_wrapper() +0x9
       cuda-gdb| stdin_event_handler() +0x60
       cuda-gdb| gdb_wait_for_event() +0x4e5
       cuda-gdb| gdb_do_one_event() +0x77
       cuda-gdb| start_event_loop() +0x1e
       cuda-gdb| captured_command_loop() +0x28
       cuda-gdb| gdb_main() +0xaed
       cuda-gdb| main() +0x25
...64/| __libc_start_main() +0xf5
       cuda-gdb| _start() +0x29

I’m using the latest CUDA Toolkit on JURECA at JSC.

Hello! Thank you very much for the reporting this issue.

Could you provide some additional information:

  • cuda-gdb version you are using (output of cuda-gdb --version command).
  • nvidia-smi command output
  • deviceQuery sample output

Also if you are not using the latest CUDA-GDB version, could you check whether this issue is still reproducible with the latest CUDA-GDB: CUDA-GDB | NVIDIA Developer