failing to debug thrust vector type content in vscode

[amit-ruf]

I'm trying to use latest vscode (v1.97.1) with latest Nsight Visual Code Studio (v 2024.1.34572442) to debug a simple .cu script using thrust::device_vectors and thrust::host_vectors. My system includes Quadro 3000 GPU.
cuda-gdb shows the contents of the variables, but vscode just fails to show them (showing empty values). When I try to access these variable from cuda-gpu commandline vscode clears its RUN AND DEBUG completely, not showing even the names of the local variables.

The simple script I'm trying to debug is:

#include <random>
#include <stdexcept>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <iostream>
#include <algorithm>
#include <cassert>
#include <cstdlib>
int main(int argc, char *argv[])
{
    cudaDeviceReset();
    thrust::host_vector<int> visible_vec(30);
    thrust::device_vector<float> test(30);    
}

I'm using the following .cuda-gdbinit to configure pretty-printing for thrust classes ( I tried many variations , including without .cuda-gdbinit and pretty-printing):

python
import gdb
import sys

# Ensure pretty printing is enabled and auto-load Python scripts are on.
gdb.execute("set print pretty on")
gdb.execute("set auto-load python-scripts on")

# Python2 compatibility for iterators.
if sys.version_info[0] > 2:
    Iterator = object
else:
    class Iterator:
        def next(self):
            return self.__next__()

#---------------------------------------------------------------------
# Pretty Printer for thrust::*_vector
# Attempts a legacy layout first; if that fails, falls back to an
# alternative layout (adjust field names as needed).
#---------------------------------------------------------------------
class ThrustVectorPrinter(object):
    "Pretty printer for thrust::*_vector"

    def __init__(self, val):
        self.val = val
        # Try legacy layout first.
        try:
            self.pointer = val["m_storage"]["m_begin"]["m_iterator"]
            self.size = int(val["m_size"])
            self.capacity = int(val["m_storage"]["m_size"])
        except Exception:
            # Fallback layout.
            try:
                self.pointer = val["data"]
                self.size = int(val["size"])
                self.capacity = self.size  # Update if you have a proper capacity field
            except Exception:
                # Neither layout worked; mark as empty.
                self.pointer = None
                self.size = 0
                self.capacity = 0
        # Check for device pointer.
        self.is_device = False
        if self.pointer is not None and str(self.pointer.type).startswith("thrust::device_ptr"):
            self.pointer = self.pointer["m_iterator"]
            self.is_device = True
        self._children_cache = None

    class _host_accessible_iterator(Iterator):
        def __init__(self, start, size):
            self.item = start
            self.size = size
            self.count = 0

        def __iter__(self):
            return self

        def __next__(self):
            if self.count >= self.size:
                raise StopIteration
            elt = self.item.dereference()
            result = ("[%d]" % self.count, elt)
            self.item = self.item + 1
            self.count += 1
            return result
        next = __next__  # For Python2

    class _device_iterator(Iterator):
        def __init__(self, start, size):
            self.item = start
            self.size = size
            self.count = 0
            self.sizeof = self.item.dereference().type.sizeof
            # Limit buffer to at most 1MB (or at least one element)
            self.buffer_size = min(size, max(1, 2**20 // self.sizeof))
            self.buffer = gdb.parse_and_eval("(void*)malloc(%d)" % (self.buffer_size * self.sizeof))
            self.buffer.fetch_lazy()
            self.buffer_count = self.buffer_size
            self.update_buffer()

        def update_buffer(self):
            if self.buffer_count >= self.buffer_size:
                self.buffer_item = gdb.parse_and_eval(hex(self.buffer)).cast(self.item.type)
                self.buffer_count = 0
                buffer_addr = hex(self.buffer)
                device_addr = hex(self.item.dereference().address)
                num_bytes = min(self.buffer_size, self.size - self.count) * self.sizeof
                status = gdb.parse_and_eval(
                    "(cudaError)cudaMemcpy(%s, %s, %d, cudaMemcpyDeviceToHost)" % (buffer_addr, device_addr, num_bytes)
                )
                if status != 0:
                    raise gdb.MemoryError("memcpy from device failed: %s" % status)

        def __iter__(self):
            return self

        def __next__(self):
            if self.count >= self.size:
                raise StopIteration
            self.update_buffer()
            elt = self.buffer_item.dereference()
            self.buffer_item = self.buffer_item + 1
            self.buffer_count += 1
            result = ("[%d]" % self.count, elt)
            self.item = self.item + 1
            self.count += 1
            return result
        next = __next__

        def __del__(self):
            try:
                gdb.parse_and_eval("(void)free(%s)" % hex(self.buffer)).fetch_lazy()
            except Exception:
                pass

    def _get_children(self):
        if self.pointer is None:
            return []
        if self.is_device:
            return list(ThrustVectorPrinter._device_iterator(self.pointer, self.size))
        else:
            return list(ThrustVectorPrinter._host_accessible_iterator(self.pointer, self.size))

    def children(self):
        if self._children_cache is None:
            self._children_cache = self._get_children()
        return self._children_cache

    def num_children(self):
        return len(self.children())

    def child(self, index):
        return self.children()[index]

    def to_string(self):
        return "thrust vector of length %d, capacity %d" % (self.size, self.capacity)

    def display_hint(self):
        return "array"

#---------------------------------------------------------------------
# Pretty Printer for thrust::device_reference
#---------------------------------------------------------------------
class ThrustReferencePrinter(object):
    "Pretty printer for thrust::device_reference"

    def __init__(self, val):
        self.val = val
        self.pointer = val["ptr"]["m_iterator"]
        self.type = self.pointer.dereference().type
        self.sizeof = self.type.sizeof
        self.buffer = gdb.parse_and_eval("(void*)malloc(%d)" % self.sizeof)
        device_addr = hex(self.pointer)
        buffer_addr = hex(self.buffer)
        status = gdb.parse_and_eval(
            "(cudaError)cudaMemcpy(%s, %s, %d, cudaMemcpyDeviceToHost)" % (buffer_addr, device_addr, self.sizeof)
        )
        if status != 0:
            raise gdb.MemoryError("memcpy from device failed: %s" % status)
        self.buffer_val = gdb.parse_and_eval(hex(self.buffer)).cast(self.pointer.type).dereference()

    def __del__(self):
        try:
            gdb.parse_and_eval("(void)free(%s)" % hex(self.buffer)).fetch_lazy()
        except Exception:
            pass

    def children(self):
        return []

    def to_string(self):
        return "(%s) @%s: %s" % (self.val.type, self.pointer, self.buffer_val)

    def display_hint(self):
        return None

#---------------------------------------------------------------------
# Pretty Printer for a user-defined type: SingleFibleType.
# Adjust the member names as needed.
#---------------------------------------------------------------------
class SingleFibleTypePrinter(object):
    "Pretty printer for SingleFibleType"

    def __init__(self, val):
        self.val = val

    def to_string(self):
        return ("x: %s, y: %s, z: %s, r: %s, pressure: %s, n_collisions: %s, visible: %s, "
                "deflated: %s, splitable: %s" %
                (self.val["x"], self.val["y"], self.val["z"],
                 self.val["r"], self.val["pressure"], self.val["n_collisions"],
                 self.val["visible"], self.val["deflated"], self.val["splitable"]))

#---------------------------------------------------------------------
# Pretty Printer for std::unordered_map<size_t, SingleFibleType>
#---------------------------------------------------------------------
class UnorderedMapPrinter(object):
    "Pretty printer for std::unordered_map<size_t, SingleFibleType>"

    def __init__(self, val):
        self.val = val

    def to_string(self):
        size = self.val["_M_h"]["_M_element_count"]
        return "std::unordered_map with %d elements" % size

    def children(self):
        elements = []
        buckets = self.val["_M_h"]["_M_buckets"]
        for bucket in buckets:
            node = bucket
            while node != 0:
                node = node.dereference()
                pair = node["_M_v"]
                key = pair["first"]
                value = pair["second"]
                elements.append(("Key %s" % key, key))
                elements.append(("Value %s" % key, SingleFibleTypePrinter(value).to_string()))
                node = node["_M_next"]
        return elements

#---------------------------------------------------------------------
# Build a RegexpCollection of pretty printers for our custom types.
#---------------------------------------------------------------------
def build_pretty_printers():
    pp = gdb.printing.RegexpCollectionPrettyPrinter("custom")
    pp.add_printer("thrust_vector", "^thrust::(host_vector|device_vector)<", ThrustVectorPrinter)
    pp.add_printer("thrust_reference", "^thrust::device_reference<", ThrustReferencePrinter)
    pp.add_printer("unordered_map", "^std::unordered_map<.*SingleFibleType", UnorderedMapPrinter)
    return pp

#---------------------------------------------------------------------
# Register our pretty printers globally.
#---------------------------------------------------------------------
gdb.printing.register_pretty_printer(None, build_pretty_printers())

#---------------------------------------------------------------------
# Also register libstdc++ printers if available.
#---------------------------------------------------------------------
sys.path.insert(0, "/usr/share/gcc/python")
try:
    from libstdcxx.v6.printers import register_libstdcxx_printers
    register_libstdcxx_printers(None)
except ImportError:
    pass

end

Below is a detailed log of vscode DEBUG CONSOLE trying to show visible_vec:

-exec p visible_vec
From client: evaluate({"expression":"-exec p visible_vec","frameId":1000,"context":"repl"})
GDB command: 43 p visible_vec
To client: {"seq":0,"type":"event","event":"output","body":{"category":"log","output":"p visible_vec\n"}}
p visible_vec
To client: {"seq":0,"type":"event","event":"output","body":{"category":"stdout","output":"$2 = thrust vector of length 30, capacity 30 = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}\n"}}
$2 = thrust vector of length 30, capacity 30 = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
GDB result: 43 done
To client: {"seq":0,"type":"response","request_seq":20,"command":"evaluate","success":true}
To client: {"seq":0,"type":"event","event":"stopped","body":{"reason":"Refreshing State","allThreadsStopped":false,"threadId":99999}}
To client: {"seq":0,"type":"event","event":"changedCudaFocus","body":{}}
From client: scopes({"frameId":1000})
To client: {"seq":0,"type":"response","request_seq":21,"command":"scopes","success":false,"message":"TypeError: Cannot read properties of undefined (reading 'threadId')\n\tat _.scopesRequest (/home/myuser/.vscode-server/extensions/nvidia.nsight-vscode-edition-2024.1.34572442/dist/extension.js:2:170520)\n\tat _.dispatchRequest (/home/myuser/.vscode-server/extensions/nvidia.nsight-vscode-edition-2024.1.34572442/dist/extension.js:2:6469)\n\tat _.dispatchRequest (/home/myuser/.vscode-server/extensions/nvidia.nsight-vscode-edition-2024.1.34572442/dist/extension.js:2:17965)\n\tat _.dispatchRequest (/home/myuser/.vscode-server/extensions/nvidia.nsight-vscode-edition-2024.1.34572442/dist/extension.js:2:159643)\n\tat _.handleMessage (/home/myuser/.vscode-server/extensions/nvidia.nsight-vscode-edition-2024.1.34572442/dist/extension.js:2:22600)\n\tat ej.sendMessage (file:///vscode/vscode-server/bin/linux-x64/e249dada235c2083c83813bd65b7f4707fb97b76/out/vs/workbench/api/node/extensionHostProcess.js:120:697)\n\tat Eb.$sendDAMessage (file:///vscode/vscode-server/bin/linux-x64/e249dada235c2083c83813bd65b7f4707fb97b76/out/vs/workbench/api/node/extensionHostProcess.js:119:97915)\n\tat Uy.S (file:///vscode/vscode-server/bin/linux-x64/e249dada235c2083c83813bd65b7f4707fb97b76/out/vs/workbench/api/node/extensionHostProcess.js:29:158274)\n\tat Uy.Q (file:///vscode/vscode-server/bin/linux-x64/e249dada235c2083c83813bd65b7f4707fb97b76/out/vs/workbench/api/node/extensionHostProcess.js:29:158054)\n\tat[...]
From client: threads(undefined)
GDB command: 44 -thread-info
GDB result: 44 done,threads=[{id="1",target-id="Thread 0x7ffff7a09000 (LWP 34557)",name="test_fist",frame={level="0",addr="0x00005555555618ab",func="test_FiblesSystem__random_generator",args=[{name="n",value="30"}],file="/fiSt/source/core/cuda/test_fist.cu",fullname="/fiSt/source/core/cuda/test_fist.cu",line="96",arch="i386:x86-64"},state="stopped",core="10"},{id="2",target-id="Thread 0x7ffff3200000 (LWP 34561)",name="cuda00001400006",frame={level="0",addr="0x00007ffff7b26bcf",func="poll",args=[],from="/usr/lib/x86_64-linux-gnu/libc.so.6",arch="i386:x86-64"},state="stopped",core="3"},{id="3",target-id="Thread 0x7ffff1c00000 (LWP 34562)",name="test_fist",frame={level="0",addr="0x00007ffff7b33e2e",func="epoll_wait",args=[],from="/usr/lib/x86_64-linux-gnu/libc.so.6",arch="i386:x86-64"},state="stopped",core="6"},{id="4",target-id="Thread 0x7ffff0e00000 (LWP 34572)",name="test_fist",frame={level="0",addr="0x00007ffff7a9f117",func="??",args=[],from="/usr/lib/x86_64-linux-gnu/libc.so.6",arch="i386:x86-64"},
GDB -cont-: 44 state="stopped",core="9"},{id="5",target-id="Thread 0x7fffeaa00000 (LWP 34573)",name="cuda-EvtHandlr",frame={level="0",addr="0x00007ffff7b26bcf",func="poll",args=[],from="/usr/lib/x86_64-linux-gnu/libc.so.6",arch="i386:x86-64"},state="stopped",core="2"}],current-thread-id="1"
To client: {"seq":0,"type":"response","request_seq":22,"command":"threads","success":true,"body":{"threads":[{"id":1,"name":"test_fist","running":false},{"id":2,"name":"cuda00001400006","running":false},{"id":3,"name":"test_fist","running":false},{"id":4,"name":"test_fist","running":false},{"id":5,"name":"cuda-EvtHandlr","running":false},{"id":99999,"name":"(CUDA)"}]}}
From client: stackTrace({"startFrame":0,"levels":20})
GDB command: 45 -stack-info-depth 100
GDB result: 45 done,depth="2"
GDB command: 46 -stack-list-frames 0 1
GDB result: 46 done,stack=[frame={level="0",addr="0x00005555555618ab",func="test_FiblesSystem__random_generator",file="/fiSt/source/core/cuda/test_fist.cu",fullname="/fiSt/source/core/cuda/test_fist.cu",line="96",arch="i386:x86-64"},frame={level="1",addr="0x000055555556308a",func="main",file="/fiSt/source/core/cuda/test_fist.cu",fullname="/fiSt/source/core/cuda/test_fist.cu",line="279",arch="i386:x86-64"}]
To client: {"seq":0,"type":"response","request_seq":23,"command":"stackTrace","success":true,"body":{"stackFrames":[{"id":1000,"source":{"name":"test_fist.cu","path":"/fiSt/source/core/cuda/test_fist.cu","sourceReference":0},"line":96,"column":0,"name":"test_FiblesSystem__random_generator","instructionPointerReference":"0x00005555555618ab"},{"id":1001,"source":{"name":"test_fist.cu","path":"/fiSt/source/core/cuda/test_fist.cu","sourceReference":0},"line":279,"column":0,"name":"main","instructionPointerReference":"0x000055555556308a"}],"totalFrames":2}}
From client: scopes({"frameId":1000})
To client: {"seq":0,"type":"response","request_seq":24,"command":"scopes","success":true,"body":{"scopes":[{"name":"Local","variablesReference":1000,"expensive":false},{"name":"Registers","variablesReference":1001,"expensive":false,"presentationHint":"registers"}]}}
From client: variables({"variablesReference":1000})
To client: {"seq":0,"type":"response","request_seq":25,"command":"variables","success":true,"body":{"variables":[]}}

I assume its either a vscode bug or some setup issue on my end.
Your help is highly appreciated