bpo-38256: Fix binascii.crc32() when inputs are 4+GiB (GH-32000)

When compiled with `USE_ZLIB_CRC32` defined (`configure` sets this on POSIX systems), `binascii.crc32(...)` failed to compute the correct value when the input data was >= 4GiB. Because the zlib crc32 API is limited to a 32-bit length.

This lines it up with the `zlib.crc32(...)` implementation that doesn't have that flaw.

**Performance:** This also adopts the same GIL releasing for larger inputs logic that `zlib.crc32` has, and causes the Windows build to always use zlib's crc32 instead of our slow C code as zlib is a required build dependency on Windows.

Modules/binascii.c

@@ -737,6 +737,21 @@ static const unsigned int crc_32_tab[256] = {
 0x5d681b02U, 0x2a6f2b94U, 0xb40bbe37U, 0xc30c8ea1U, 0x5a05df1bU,
 0x2d02ef8dU
 };
+
+static unsigned int
+internal_crc32(const unsigned char *bin_data, Py_ssize_t len, unsigned int crc)
+{ /* By Jim Ahlstrom; All rights transferred to CNRI */
+    unsigned int result;
+
+    crc = ~ crc;
+    while (len-- > 0) {
+        crc = crc_32_tab[(crc ^ *bin_data++) & 0xff] ^ (crc >> 8);
+        /* Note:  (crc >> 8) MUST zero fill on left */
+    }
+
+    result = (crc ^ 0xFFFFFFFF);
+    return result & 0xffffffff;
+}
 #endif  /* USE_ZLIB_CRC32 */
 
 /*[clinic input]
@@ -754,34 +769,46 @@ binascii_crc32_impl(PyObject *module, Py_buffer *data, unsigned int crc)
 /*[clinic end generated code: output=52cf59056a78593b input=bbe340bc99d25aa8]*/
 
 #ifdef USE_ZLIB_CRC32
-/* This was taken from zlibmodule.c PyZlib_crc32 (but is PY_SSIZE_T_CLEAN) */
+/* This is the same as zlibmodule.c zlib_crc32_impl. It exists in two
+ * modules for historical reasons. */
 {
-    const Byte *buf;
-    Py_ssize_t len;
-    int signed_val;
+    /* Releasing the GIL for very small buffers is inefficient
+       and may lower performance */
+    if (data->len > 1024*5) {
+        unsigned char *buf = data->buf;
+        Py_ssize_t len = data->len;
 
-    buf = (Byte*)data->buf;
-    len = data->len;
-    signed_val = crc32(crc, buf, len);
-    return (unsigned int)signed_val & 0xffffffffU;
+        Py_BEGIN_ALLOW_THREADS
+        /* Avoid truncation of length for very large buffers. crc32() takes
+           length as an unsigned int, which may be narrower than Py_ssize_t. */
+        while ((size_t)len > UINT_MAX) {
+            crc = crc32(crc, buf, UINT_MAX);
+            buf += (size_t) UINT_MAX;
+            len -= (size_t) UINT_MAX;
+        }
+        crc = crc32(crc, buf, (unsigned int)len);
+        Py_END_ALLOW_THREADS
+    } else {
+        crc = crc32(crc, data->buf, (unsigned int)data->len);
+    }
+    return crc & 0xffffffff;
 }
 #else  /* USE_ZLIB_CRC32 */
-{ /* By Jim Ahlstrom; All rights transferred to CNRI */
-    const unsigned char *bin_data;
-    Py_ssize_t len;
-    unsigned int result;
+{
+    const unsigned char *bin_data = data->buf;
+    Py_ssize_t len = data->len;
 
-    bin_data = data->buf;
-    len = data->len;
-
-    crc = ~ crc;
-    while (len-- > 0) {
-        crc = crc_32_tab[(crc ^ *bin_data++) & 0xff] ^ (crc >> 8);
-        /* Note:  (crc >> 8) MUST zero fill on left */
+    /* Releasing the GIL for very small buffers is inefficient
+       and may lower performance */
+    if (len > 1024*5) {
+        unsigned int result;
+        Py_BEGIN_ALLOW_THREADS
+        result = internal_crc32(bin_data, len, crc);
+        Py_END_ALLOW_THREADS
+        return result;
+    } else {
+        return internal_crc32(bin_data, len, crc);
     }
-
-    result = (crc ^ 0xFFFFFFFF);
-    return result & 0xffffffff;
 }
 #endif  /* USE_ZLIB_CRC32 */
 

Modules/clinic/zlibmodule.c.h

@@ -753,7 +753,7 @@ PyDoc_STRVAR(zlib_crc32__doc__,
 #define ZLIB_CRC32_METHODDEF    \
     {"crc32", (PyCFunction)(void(*)(void))zlib_crc32, METH_FASTCALL, zlib_crc32__doc__},
 
-static PyObject *
+static unsigned int
 zlib_crc32_impl(PyObject *module, Py_buffer *data, unsigned int value);
 
 static PyObject *
@@ -762,6 +762,7 @@ zlib_crc32(PyObject *module, PyObject *const *args, Py_ssize_t nargs)
     PyObject *return_value = NULL;
     Py_buffer data = {NULL, NULL};
     unsigned int value = 0;
+    unsigned int _return_value;
 
     if (!_PyArg_CheckPositional("crc32", nargs, 1, 2)) {
         goto exit;
@@ -781,7 +782,11 @@ zlib_crc32(PyObject *module, PyObject *const *args, Py_ssize_t nargs)
         goto exit;
     }
 skip_optional:
-    return_value = zlib_crc32_impl(module, &data, value);
+    _return_value = zlib_crc32_impl(module, &data, value);
+    if ((_return_value == (unsigned int)-1) && PyErr_Occurred()) {
+        goto exit;
+    }
+    return_value = PyLong_FromUnsignedLong((unsigned long)_return_value);
 
 exit:
     /* Cleanup for data */
@@ -815,4 +820,4 @@ exit:
 #ifndef ZLIB_DECOMPRESS___DEEPCOPY___METHODDEF
     #define ZLIB_DECOMPRESS___DEEPCOPY___METHODDEF
 #endif /* !defined(ZLIB_DECOMPRESS___DEEPCOPY___METHODDEF) */
-/*[clinic end generated code: output=e3e8a6142ea045a7 input=a9049054013a1b77]*/
+/*[clinic end generated code: output=f009d194565416d1 input=a9049054013a1b77]*/

Modules/zlibmodule.c

@@ -1420,7 +1420,7 @@ zlib_adler32_impl(PyObject *module, Py_buffer *data, unsigned int value)
 }
 
 /*[clinic input]
-zlib.crc32
+zlib.crc32 -> unsigned_int
 
     data: Py_buffer
     value: unsigned_int(bitwise=True) = 0
@@ -1432,9 +1432,9 @@ Compute a CRC-32 checksum of data.
 The returned checksum is an integer.
 [clinic start generated code]*/
 
-static PyObject *
+static unsigned int
 zlib_crc32_impl(PyObject *module, Py_buffer *data, unsigned int value)
-/*[clinic end generated code: output=63499fa20af7ea25 input=26c3ed430fa00b4c]*/
+/*[clinic end generated code: output=b217562e4fe6d6a6 input=1229cb2fb5ea948a]*/
 {
     /* Releasing the GIL for very small buffers is inefficient
        and may lower performance */
@@ -1455,7 +1455,7 @@ zlib_crc32_impl(PyObject *module, Py_buffer *data, unsigned int value)
     } else {
         value = crc32(value, data->buf, (unsigned int)data->len);
     }
-    return PyLong_FromUnsignedLong(value & 0xffffffffU);
+    return value;
 }