[3.10] bpo-38256: Fix binascii.crc32 large input. (GH-32000) (GH-32013)

Inputs >= 4GiB to `binascii.crc32(...)` when compiled to use the zlib
crc32 implementation (the norm on POSIX) no longer return the wrong
result.

1 files changed, 13 insertions, 9 deletions

diff --git a/Modules/binascii.c b/Modules/binascii.c
index 1f3248b..3777580 100644
--- a/Modules/binascii.c
+++ b/Modules/binascii.c
@@ -1120,16 +1120,20 @@ 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) */
+/* The same core as zlibmodule.c zlib_crc32_impl. */
 {
-    const Byte *buf;
-    Py_ssize_t len;
-    int signed_val;
-
-    buf = (Byte*)data->buf;
-    len = data->len;
-    signed_val = crc32(crc, buf, len);
-    return (unsigned int)signed_val & 0xffffffffU;
+    unsigned char *buf = data->buf;
+    Py_ssize_t len = data->len;
+
+    /* 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);
+    return crc & 0xffffffff;
 }
 #else  /* USE_ZLIB_CRC32 */
 { /* By Jim Ahlstrom; All rights transferred to CNRI */