Modules/_remote_debugging/object_reading.c

/******************************************************************************
 * Remote Debugging Module - Object Reading Functions
 *
 * This file contains functions for reading Python objects from remote
 * process memory, including strings, bytes, and integers.
 ******************************************************************************/

#include "_remote_debugging.h"
#include <limits.h>

/* ============================================================================
 * MEMORY READING FUNCTIONS
 * ============================================================================ */

#define DEFINE_MEMORY_READER(type_name, c_type, error_msg) \
int \
read_##type_name(RemoteUnwinderObject *unwinder, uintptr_t address, c_type *result) \
{ \
    int res = _Py_RemoteDebug_PagedReadRemoteMemory(&unwinder->handle, address, sizeof(c_type), result); \
    if (res < 0) { \
        set_exception_cause(unwinder, PyExc_RuntimeError, error_msg); \
        return -1; \
    } \
    return 0; \
}

DEFINE_MEMORY_READER(ptr, uintptr_t, "Failed to read pointer from remote memory")
DEFINE_MEMORY_READER(char, char, "Failed to read char from remote memory")

int
read_py_ptr(RemoteUnwinderObject *unwinder, uintptr_t address, uintptr_t *ptr_addr)
{
    if (read_ptr(unwinder, address, ptr_addr)) {
        set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to read Python pointer");
        return -1;
    }
    *ptr_addr &= ~Py_TAG_BITS;
    return 0;
}

/* ============================================================================
 * PYTHON OBJECT READING FUNCTIONS
 * ============================================================================ */

PyObject *
read_py_str(
    RemoteUnwinderObject *unwinder,
    uintptr_t address,
    Py_ssize_t max_len
) {
    // Read the entire PyUnicodeObject at once; for short strings the data
    // is inline right after the header and we'll already have (some of) it.
    char unicode_obj[SIZEOF_UNICODE_OBJ];
    int res = _Py_RemoteDebug_PagedReadRemoteMemory(
        &unwinder->handle,
        address,
        SIZEOF_UNICODE_OBJ,
        unicode_obj
    );
    if (res < 0) {
        set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to read PyUnicodeObject");
        return NULL;
    }

    Py_ssize_t len = GET_MEMBER(Py_ssize_t, unicode_obj, unwinder->debug_offsets.unicode_object.length);
    if (len < 0) {
        PyErr_Format(PyExc_RuntimeError,
                     "Invalid string length (%zd) at 0x%lx", len, address);
        set_exception_cause(unwinder, PyExc_RuntimeError, "Invalid string length in remote Unicode object");
        return NULL;
    }
    if (len > max_len) {
        // gh-157788: a long name must not fail the whole read
        len = max_len;
    }

    // Inspect state to pick the right data offset and character width.
    // We rely on the remote process sharing this Python version's
    // PyASCIIObject layout, the same assumption already used for `length`.
    struct _PyUnicodeObject_state state = GET_MEMBER(
        struct _PyUnicodeObject_state,
        unicode_obj,
        unwinder->debug_offsets.unicode_object.state);

    if (!state.compact) {
        PyErr_Format(PyExc_RuntimeError,
                     "Cannot read non-compact Unicode object at 0x%lx", address);
        set_exception_cause(unwinder, PyExc_RuntimeError,
                            "Legacy (non-compact) Unicode objects are not supported");
        return NULL;
    }

    int kind = (int)state.kind;
    Py_UCS4 max_char;
    switch (kind) {
        case PyUnicode_1BYTE_KIND:
            max_char = state.ascii ? 0x7F : 0xFF;
            break;
        case PyUnicode_2BYTE_KIND:
            max_char = 0xFFFF;
            break;
        case PyUnicode_4BYTE_KIND:
            max_char = 0x10FFFF;
            break;
        default:
            PyErr_Format(PyExc_RuntimeError,
                         "Invalid Unicode kind %d at 0x%lx", kind, address);
            set_exception_cause(unwinder, PyExc_RuntimeError,
                                "Invalid kind in remote Unicode object");
            return NULL;
    }

    size_t header_size = state.ascii
        ? (size_t)unwinder->debug_offsets.unicode_object.asciiobject_size
        : (size_t)unwinder->debug_offsets.unicode_object.compactunicodeobject_size;

    // len * kind is bounded by max_len * 4 (kind <= 4, len <= max_len), so
    // the multiplication can't overflow for any caller-sane max_len, but the
    // explicit cap here keeps a corrupted remote `length` from later turning
    // into a giant allocation.
    size_t nbytes = (size_t)len * (size_t)kind;
    if ((size_t)len > (SIZE_MAX / 4) || nbytes > (size_t)max_len * 4) {
        PyErr_Format(PyExc_RuntimeError,
                     "Implausible Unicode byte size %zu at 0x%lx", nbytes, address);
        set_exception_cause(unwinder, PyExc_RuntimeError,
                            "Garbage byte size in remote Unicode object");
        return NULL;
    }

    PyObject *result = PyUnicode_New(len, max_char);
    if (result == NULL) {
        set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to allocate PyUnicode for remote string");
        return NULL;
    }
    if (nbytes == 0) {
        return result;
    }

    void *data = PyUnicode_DATA(result);

    // Reuse data already present in the header read; only round-trip for
    // whatever spills past it.
    size_t inline_avail = (header_size < SIZEOF_UNICODE_OBJ)
        ? SIZEOF_UNICODE_OBJ - header_size
        : 0;
    size_t inline_bytes = nbytes < inline_avail ? nbytes : inline_avail;
    if (inline_bytes > 0) {
        memcpy(data, unicode_obj + header_size, inline_bytes);
    }

    if (nbytes > inline_bytes) {
        res = _Py_RemoteDebug_PagedReadRemoteMemory(
            &unwinder->handle,
            address + header_size + inline_bytes,
            nbytes - inline_bytes,
            (char *)data + inline_bytes);
        if (res < 0) {
            Py_DECREF(result);
            set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to read string data from remote memory");
            return NULL;
        }
    }

    return result;
}

PyObject *
read_py_bytes(
    RemoteUnwinderObject *unwinder,
    uintptr_t address,
    Py_ssize_t max_len
) {
    PyObject *result = NULL;
    char *buf = NULL;

    // Read the entire PyBytesObject at once
    char bytes_obj[SIZEOF_BYTES_OBJ];
    int res = _Py_RemoteDebug_PagedReadRemoteMemory(
        &unwinder->handle,
        address,
        SIZEOF_BYTES_OBJ,
        bytes_obj
    );
    if (res < 0) {
        set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to read PyBytesObject");
        goto err;
    }

    Py_ssize_t len = GET_MEMBER(Py_ssize_t, bytes_obj, unwinder->debug_offsets.bytes_object.ob_size);
    if (len < 0 || len > max_len) {
        PyErr_Format(PyExc_RuntimeError,
                     "Invalid bytes length (%zd) at 0x%lx", len, address);
        set_exception_cause(unwinder, PyExc_RuntimeError, "Invalid bytes length in remote bytes object");
        return NULL;
    }

    buf = (char *)PyMem_RawMalloc(len+1);
    if (buf == NULL) {
        PyErr_NoMemory();
        set_exception_cause(unwinder, PyExc_MemoryError, "Failed to allocate buffer for bytes reading");
        return NULL;
    }

    size_t offset = (size_t)unwinder->debug_offsets.bytes_object.ob_sval;
    res = _Py_RemoteDebug_PagedReadRemoteMemory(&unwinder->handle, address + offset, len, buf);
    if (res < 0) {
        set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to read bytes data from remote memory");
        goto err;
    }
    buf[len] = '\0';

    result = PyBytes_FromStringAndSize(buf, len);
    if (result == NULL) {
        set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to create PyBytes from remote bytes data");
        goto err;
    }

    PyMem_RawFree(buf);
    assert(result != NULL);
    return result;

err:
    if (buf != NULL) {
        PyMem_RawFree(buf);
    }
    return NULL;
}

long
read_py_long(
    RemoteUnwinderObject *unwinder,
    uintptr_t address
)
{
    unsigned int shift = PYLONG_BITS_IN_DIGIT;

    // Read the entire PyLongObject at once
    char long_obj[SIZEOF_LONG_OBJ];
    int bytes_read = _Py_RemoteDebug_PagedReadRemoteMemory(
        &unwinder->handle,
        address,
        (size_t)unwinder->debug_offsets.long_object.size,
        long_obj);
    if (bytes_read < 0) {
        set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to read PyLongObject");
        return -1;
    }

    uintptr_t lv_tag = GET_MEMBER(uintptr_t, long_obj, unwinder->debug_offsets.long_object.lv_tag);
    int negative = (lv_tag & 3) == 2;
    Py_ssize_t size = lv_tag >> 3;

    if (size == 0) {
        return 0;
    }

    // Validate size: reject garbage (negative or unreasonably large)
    if (size < 0 || size > MAX_LONG_DIGITS) {
        PyErr_Format(PyExc_RuntimeError,
            "Invalid PyLong digit count: %zd (expected 0-%d)", size, MAX_LONG_DIGITS);
        set_exception_cause(unwinder, PyExc_RuntimeError,
            "Invalid PyLong size (corrupted remote memory)");
        return -1;
    }

    // Calculate how many digits fit inline in our local buffer
    Py_ssize_t ob_digit_offset = unwinder->debug_offsets.long_object.ob_digit;
    Py_ssize_t inline_digits_space = SIZEOF_LONG_OBJ - ob_digit_offset;
    Py_ssize_t max_inline_digits = inline_digits_space / (Py_ssize_t)sizeof(digit);

    digit *digits = (digit *)PyMem_RawMalloc(size * sizeof(digit));
    if (!digits) {
        PyErr_NoMemory();
        set_exception_cause(unwinder, PyExc_MemoryError, "Failed to allocate digits for PyLong");
        return -1;
    }

    if (size <= max_inline_digits && size <= _PY_NSMALLNEGINTS + _PY_NSMALLPOSINTS) {
        memcpy(digits, long_obj + ob_digit_offset, size * sizeof(digit));
    } else {
        bytes_read = _Py_RemoteDebug_PagedReadRemoteMemory(
            &unwinder->handle,
            address + (uintptr_t)unwinder->debug_offsets.long_object.ob_digit,
            sizeof(digit) * size,
            digits
        );
        if (bytes_read < 0) {
            set_exception_cause(unwinder, PyExc_RuntimeError, "Failed to read PyLong digits from remote memory");
            goto error;
        }
    }

    unsigned long limit = negative
        ? (unsigned long)LONG_MAX + 1UL
        : (unsigned long)LONG_MAX;
    unsigned long value = 0;

    for (Py_ssize_t i = size; i-- > 0;) {
        if (digits[i] >= PyLong_BASE) {
            PyErr_Format(PyExc_RuntimeError,
                "Invalid PyLong digit: %u (base %u)", digits[i], PyLong_BASE);
            set_exception_cause(unwinder, PyExc_RuntimeError,
                "Invalid PyLong digit (corrupted remote memory)");
            goto error;
        }
        if (value > ((limit - (unsigned long)digits[i]) >> shift)) {
            PyErr_SetString(PyExc_OverflowError,
                "Remote PyLong value does not fit in C long");
            set_exception_cause(unwinder, PyExc_OverflowError,
                "Remote PyLong value is too large");
            goto error;
        }
        value = (value << shift) | (unsigned long)digits[i];
    }
    PyMem_RawFree(digits);
    if (negative) {
        if (value == (unsigned long)LONG_MAX + 1UL) {
            return LONG_MIN;
        }
        return -(long)value;
    }
    return (long)value;
error:
    PyMem_RawFree(digits);
    return -1;
}