Examining the Symbol Table

See GDB* documentation.

info functions

At least following syntax is supported:

info functions
info functions [regexp]

Note

Running the info functions command without arguments may produce a lot of output as the list of all functions in all loaded shared libraries is typically very long.

Example

Source file numba_dpex/examples/debug/simple_sum.py:

import dpnp as np

import numba_dpex as ndpx


@ndpx.kernel(debug=True)
def data_parallel_sum(item, a, b, c):
    i = item.get_id(0)
    c[i] = a[i] + b[i]  # Condition breakpoint location


global_size = 10
N = global_size

a = np.array(np.random.random(N), dtype=np.float32)
b = np.array(np.random.random(N), dtype=np.float32)
c = np.ones_like(a)

ndpx.call_kernel(data_parallel_sum, ndpx.Range(global_size), a, b, c)

print("Done...")

Output of the debug session:

$ NUMBA_OPT=0 gdb-oneapi -q python
(gdb) set breakpoint pending on
(gdb) break simple_sum.py:22
(gdb) run simple_sum.py
(gdb) info functions data_parallel_sum
...
All functions matching regular expression "data_parallel_sum":
File simple_sum.py:
20:         void __main__::data_parallel_sum(Array<float, 1, C, mutable, aligned>, Array<float, 1, C, mutable, aligned>, Array<float, 1, C, mutable, aligned>);
(gdb) continue
(gdb) info functions __main__
...
All functions matching regular expression "__main__":
20:         void __main__::data_parallel_sum(Array<float, 1, C, mutable, aligned>, Array<float, 1, C, mutable, aligned>, Array<float, 1, C, mutable, aligned>);
(gdb) continue
...
Done...

whatis [arg] and ptype [arg]

To print the type of a variable, run the ptype <variable> or whatis <variable> commands:

(gdb) whatis a
type = byte [56]
(gdb) ptype l1
type = double
(gdb) whatis l1
type = double
(gdb) continue
...
Done...

Example - Complex Data Types

Source code numba_dpex/examples/debug/side-by-side-2.py:

def common_loop_body(i, a, b):
    param_a = a[i]
    param_b = b[i]
    param_c = param_a + 10  # Set breakpoint here
    param_d = param_b * 0.5
    result = param_c + param_d
    return result

Debug session:

$ gdb-oneapi -q python
...
(gdb) set environment NUMBA_OPT 0
(gdb) set environment NUMBA_EXTEND_VARIABLE_LIFETIMES 1
(gdb) break side-by-side-2.py:29 if param_a == 5
...
(gdb) run numba_dpex/examples/debug/side-by-side-2.py --api=numba-dpex-kernel
...
Thread 2.1 hit Breakpoint 1, with SIMD lane 5, __main__::common_loop_body (i=5, a=..., b=...) at side-by-side-2.py:29
29          result = param_c + param_d
(gdb) ptype a
type = struct array(float32, 1d, C) ({float addrspace(1)*, float addrspace(1)*, i64, i64, float addrspace(1)*, [1 x i64], [1 x i64]}) {
    float *meminfo;
    float *parent;
    int64 nitems;
    int64 itemsize;
    float *data;
    i64 shape[1];
    i64 strides[1];
}
(gdb) whatis a
type = array(float32, 1d, C) ({float addrspace(1)*, float addrspace(1)*, i64, i64, float addrspace(1)*, [1 x i64], [1 x i64]})