Replace KB unit with KiB (#4293)

kB (*kilo* byte) unit means 1000 bytes, whereas KiB ("kibibyte")
means 1024 bytes. KB was misused: replace kB or KB with KiB when
appropriate.

Same change for MB and GB which become MiB and GiB.

Change the output of Tools/iobench/iobench.py.

Round also the size of the documentation from 5.5 MB to 5 MiB.

Doc/c-api/memory.rst

@@ -450,7 +450,7 @@ The pymalloc allocator
 
 Python has a *pymalloc* allocator optimized for small objects (smaller or equal
 to 512 bytes) with a short lifetime. It uses memory mappings called "arenas"
-with a fixed size of 256 KB. It falls back to :c:func:`PyMem_RawMalloc` and
+with a fixed size of 256 KiB. It falls back to :c:func:`PyMem_RawMalloc` and
 :c:func:`PyMem_RawRealloc` for allocations larger than 512 bytes.
 
 *pymalloc* is the default allocator of the :c:data:`PYMEM_DOMAIN_MEM` (ex:

Doc/library/hashlib.rst

@@ -267,7 +267,7 @@ include a `salt <https://en.wikipedia.org/wiki/Salt_%28cryptography%29>`_.
    should be about 16 or more bytes from a proper source, e.g. :func:`os.urandom`.
 
    *n* is the CPU/Memory cost factor, *r* the block size, *p* parallelization
-   factor and *maxmem* limits memory (OpenSSL 1.1.0 defaults to 32 MB).
+   factor and *maxmem* limits memory (OpenSSL 1.1.0 defaults to 32 MiB).
    *dklen* is the length of the derived key.
 
    Availability: OpenSSL 1.1+

Doc/library/locale.rst

@@ -373,7 +373,7 @@ The :mod:`locale` module defines the following exception and functions:
 
    Please note that this function works like :meth:`format_string` but will
    only work for exactly one ``%char`` specifier.  For example, ``'%f'`` and
-   ``'%.0f'`` are both valid specifiers, but ``'%f kB'`` is not.
+   ``'%.0f'`` are both valid specifiers, but ``'%f KiB'`` is not.
 
    For whole format strings, use :func:`format_string`.
 

Doc/library/multiprocessing.rst

@@ -1034,7 +1034,7 @@ Connection objects are usually created using :func:`Pipe` -- see also
       Send an object to the other end of the connection which should be read
       using :meth:`recv`.
 
-      The object must be picklable.  Very large pickles (approximately 32 MB+,
+      The object must be picklable.  Very large pickles (approximately 32 MiB+,
       though it depends on the OS) may raise a :exc:`ValueError` exception.
 
    .. method:: recv()
@@ -1071,7 +1071,7 @@ Connection objects are usually created using :func:`Pipe` -- see also
 
       If *offset* is given then data is read from that position in *buffer*.  If
       *size* is given then that many bytes will be read from buffer.  Very large
-      buffers (approximately 32 MB+, though it depends on the OS) may raise a
+      buffers (approximately 32 MiB+, though it depends on the OS) may raise a
       :exc:`ValueError` exception
 
    .. method:: recv_bytes([maxlength])

Doc/tools/templates/download.html

@@ -18,23 +18,23 @@ in the table are the size of the download files in megabytes.</p>
 <table class="docutils">
   <tr><th>Format</th><th>Packed as .zip</th><th>Packed as .tar.bz2</th></tr>
   <tr><td>PDF (US-Letter paper size)</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-letter.zip">Download</a> (ca. 13 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-letter.zip">Download</a> (ca. 13 MiB)</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-letter.tar.bz2">Download</a> (ca. 13 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-letter.tar.bz2">Download</a> (ca. 13 MiB)</td>
   </tr>
   <tr><td>PDF (A4 paper size)</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-a4.zip">Download</a> (ca. 13 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-a4.zip">Download</a> (ca. 13 MiB)</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-a4.tar.bz2">Download</a> (ca. 13 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-pdf-a4.tar.bz2">Download</a> (ca. 13 MiB)</td>
   </tr>
   <tr><td>HTML</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-html.zip">Download</a> (ca. 9 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-html.zip">Download</a> (ca. 9 MiB)</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-html.tar.bz2">Download</a> (ca. 6 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-html.tar.bz2">Download</a> (ca. 6 MiB)</td>
   </tr>
   <tr><td>Plain Text</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-text.zip">Download</a> (ca. 3 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-text.zip">Download</a> (ca. 3 MiB)</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs-text.tar.bz2">Download</a> (ca. 2 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs-text.tar.bz2">Download</a> (ca. 2 MiB)</td>
   </tr>
  <tr><td>EPUB</td>
-    <td><a href="{{ dlbase }}/python-{{ release }}-docs.epub">Download</a> (ca. 5.5 MB)</td>
+    <td><a href="{{ dlbase }}/python-{{ release }}-docs.epub">Download</a> (ca. 5 MiB)</td>
     <td></td>
   </tr>
 </table>

Include/internal/pymalloc.h

@@ -160,7 +160,7 @@
  */
 #ifdef WITH_MEMORY_LIMITS
 #ifndef SMALL_MEMORY_LIMIT
-#define SMALL_MEMORY_LIMIT      (64 * 1024 * 1024)      /* 64 MB -- more? */
+#define SMALL_MEMORY_LIMIT      (64 * 1024 * 1024)      /* 64 MiB -- more? */
 #endif
 #endif
 
@@ -177,7 +177,7 @@
  * Arenas are allocated with mmap() on systems supporting anonymous memory
  * mappings to reduce heap fragmentation.
  */
-#define ARENA_SIZE              (256 << 10)     /* 256KB */
+#define ARENA_SIZE              (256 << 10)     /* 256 KiB */
 
 #ifdef WITH_MEMORY_LIMITS
 #define MAX_ARENAS              (SMALL_MEMORY_LIMIT / ARENA_SIZE)
@@ -435,7 +435,7 @@ currently in use isn't on either list.
 */
 
 /* How many arena_objects do we initially allocate?
- * 16 = can allocate 16 arenas = 16 * ARENA_SIZE = 4MB before growing the
+ * 16 = can allocate 16 arenas = 16 * ARENA_SIZE = 4 MiB before growing the
  * `arenas` vector.
  */
 #define INITIAL_ARENA_OBJECTS 16

Lib/distutils/cygwinccompiler.py

@@ -234,8 +234,8 @@ class CygwinCCompiler(UnixCCompiler):
         # who wants symbols and a many times larger output file
         # should explicitly switch the debug mode on
         # otherwise we let dllwrap/ld strip the output file
-        # (On my machine: 10KB < stripped_file < ??100KB
+        # (On my machine: 10KiB < stripped_file < ??100KiB
-        #   unstripped_file = stripped_file + XXX KB
+        #   unstripped_file = stripped_file + XXX KiB
         #  ( XXX=254 for a typical python extension))
         if not debug:
             extra_preargs.append("-s")

Lib/gzip.py

@@ -308,7 +308,7 @@ class GzipFile(_compression.BaseStream):
             if self.mode == WRITE:
                 fileobj.write(self.compress.flush())
                 write32u(fileobj, self.crc)
-                # self.size may exceed 2GB, or even 4GB
+                # self.size may exceed 2 GiB, or even 4 GiB
                 write32u(fileobj, self.size & 0xffffffff)
             elif self.mode == READ:
                 self._buffer.close()

Lib/test/_test_multiprocessing.py

@@ -4221,7 +4221,7 @@ class TestIgnoreEINTR(unittest.TestCase):
         conn.send('ready')
         x = conn.recv()
         conn.send(x)
-        conn.send_bytes(b'x'*(1024*1024))   # sending 1 MB should block
+        conn.send_bytes(b'x' * (1024 * 1024))   # sending 1 MiB should block
 
     @unittest.skipUnless(hasattr(signal, 'SIGUSR1'), 'requires SIGUSR1')
     def test_ignore(self):

Lib/test/libregrtest/cmdline.py

@@ -96,7 +96,7 @@ resources to test.  Currently only the following are defined:
 
     largefile - It is okay to run some test that may create huge
                 files.  These tests can take a long time and may
-                consume >2GB of disk space temporarily.
+                consume >2 GiB of disk space temporarily.
 
     network -   It is okay to run tests that use external network
                 resource, e.g. testing SSL support for sockets.

Lib/test/pickletester.py

@@ -2276,7 +2276,7 @@ class AbstractPickleTests(unittest.TestCase):
 
 class BigmemPickleTests(unittest.TestCase):
 
-    # Binary protocols can serialize longs of up to 2GB-1
+    # Binary protocols can serialize longs of up to 2 GiB-1
 
     @bigmemtest(size=_2G, memuse=3.6, dry_run=False)
     def test_huge_long_32b(self, size):
@@ -2291,7 +2291,7 @@ class BigmemPickleTests(unittest.TestCase):
         finally:
             data = None
 
-    # Protocol 3 can serialize up to 4GB-1 as a bytes object
+    # Protocol 3 can serialize up to 4 GiB-1 as a bytes object
     # (older protocols don't have a dedicated opcode for bytes and are
     # too inefficient)
 

Lib/test/test_bigaddrspace.py

@@ -3,7 +3,7 @@ These tests are meant to exercise that requests to create objects bigger
 than what the address space allows are properly met with an OverflowError
 (rather than crash weirdly).
 
-Primarily, this means 32-bit builds with at least 2 GB of available memory.
+Primarily, this means 32-bit builds with at least 2 GiB of available memory.
 You need to pass the -M option to regrtest (e.g. "-M 2.1G") for tests to
 be enabled.
 """

Lib/test/test_bz2.py

@@ -62,7 +62,7 @@ class BaseTest(unittest.TestCase):
     BAD_DATA = b'this is not a valid bzip2 file'
 
     # Some tests need more than one block of uncompressed data. Since one block
-    # is at least 100 kB, we gather some data dynamically and compress it.
+    # is at least 100,000 bytes, we gather some data dynamically and compress it.
     # Note that this assumes that compression works correctly, so we cannot
     # simply use the bigger test data for all tests.
     test_size = 0

Lib/test/test_io.py

@@ -564,7 +564,7 @@ class IOTest(unittest.TestCase):
 
     def test_large_file_ops(self):
         # On Windows and Mac OSX this test comsumes large resources; It takes
-        # a long time to build the >2GB file and takes >2GB of disk space
+        # a long time to build the >2 GiB file and takes >2 GiB of disk space
         # therefore the resource must be enabled to run this test.
         if sys.platform[:3] == 'win' or sys.platform == 'darwin':
             support.requires(
@@ -736,7 +736,7 @@ class IOTest(unittest.TestCase):
         if sys.maxsize > 0x7FFFFFFF:
             self.skipTest("test can only run in a 32-bit address space")
         if support.real_max_memuse < support._2G:
-            self.skipTest("test requires at least 2GB of memory")
+            self.skipTest("test requires at least 2 GiB of memory")
         with self.open(zero, "rb", buffering=0) as f:
             self.assertRaises(OverflowError, f.read)
         with self.open(zero, "rb") as f:
@@ -1421,7 +1421,7 @@ class CBufferedReaderTest(BufferedReaderTest, SizeofTest):
     def test_constructor(self):
         BufferedReaderTest.test_constructor(self)
         # The allocation can succeed on 32-bit builds, e.g. with more
-        # than 2GB RAM and a 64-bit kernel.
+        # than 2 GiB RAM and a 64-bit kernel.
         if sys.maxsize > 0x7FFFFFFF:
             rawio = self.MockRawIO()
             bufio = self.tp(rawio)
@@ -1733,7 +1733,7 @@ class CBufferedWriterTest(BufferedWriterTest, SizeofTest):
     def test_constructor(self):
         BufferedWriterTest.test_constructor(self)
         # The allocation can succeed on 32-bit builds, e.g. with more
-        # than 2GB RAM and a 64-bit kernel.
+        # than 2 GiB RAM and a 64-bit kernel.
         if sys.maxsize > 0x7FFFFFFF:
             rawio = self.MockRawIO()
             bufio = self.tp(rawio)
@@ -2206,7 +2206,7 @@ class CBufferedRandomTest(BufferedRandomTest, SizeofTest):
     def test_constructor(self):
         BufferedRandomTest.test_constructor(self)
         # The allocation can succeed on 32-bit builds, e.g. with more
-        # than 2GB RAM and a 64-bit kernel.
+        # than 2 GiB RAM and a 64-bit kernel.
         if sys.maxsize > 0x7FFFFFFF:
             rawio = self.MockRawIO()
             bufio = self.tp(rawio)

Lib/test/test_largefile.py

@@ -9,12 +9,12 @@ from test.support import TESTFN, requires, unlink
 import io  # C implementation of io
 import _pyio as pyio # Python implementation of io
 
-# size of file to create (>2GB; 2GB == 2147483648 bytes)
+# size of file to create (>2 GiB; 2 GiB == 2,147,483,648 bytes)
 size = 2500000000
 
 class LargeFileTest:
     """Test that each file function works as expected for large
-    (i.e. > 2GB) files.
+    (i.e. > 2 GiB) files.
     """
 
     def setUp(self):
@@ -142,7 +142,7 @@ def setUpModule():
         pass
 
     # On Windows and Mac OSX this test comsumes large resources; It
-    # takes a long time to build the >2GB file and takes >2GB of disk
+    # takes a long time to build the >2 GiB file and takes >2 GiB of disk
     # space therefore the resource must be enabled to run this test.
     # If not, nothing after this line stanza will be executed.
     if sys.platform[:3] == 'win' or sys.platform == 'darwin':

Lib/test/test_mmap.py

@@ -777,7 +777,7 @@ class LargeMmapTests(unittest.TestCase):
             with mmap.mmap(f.fileno(), 0x10000, access=mmap.ACCESS_READ) as m:
                 self.assertEqual(m.size(), 0x180000000)
 
-    # Issue 11277: mmap() with large (~4GB) sparse files crashes on OS X.
+    # Issue 11277: mmap() with large (~4 GiB) sparse files crashes on OS X.
 
     def _test_around_boundary(self, boundary):
         tail = b'  DEARdear  '

Lib/test/test_os.py

@@ -171,7 +171,7 @@ class FileTests(unittest.TestCase):
         with open(support.TESTFN, "rb") as fp:
             data = os.read(fp.fileno(), size)
 
-        # The test does not try to read more than 2 GB at once because the
+        # The test does not try to read more than 2 GiB at once because the
         # operating system is free to return less bytes than requested.
         self.assertEqual(data, b'test')
 
@@ -2573,7 +2573,7 @@ class SendfileTestServer(asyncore.dispatcher, threading.Thread):
 @unittest.skipUnless(hasattr(os, 'sendfile'), "test needs os.sendfile()")
 class TestSendfile(unittest.TestCase):
 
-    DATA = b"12345abcde" * 16 * 1024  # 160 KB
+    DATA = b"12345abcde" * 16 * 1024  # 160 KiB
     SUPPORT_HEADERS_TRAILERS = not sys.platform.startswith("linux") and \
                                not sys.platform.startswith("solaris") and \
                                not sys.platform.startswith("sunos")

Lib/test/test_socket.py

@@ -5299,7 +5299,7 @@ class SendfileUsingSendTest(ThreadedTCPSocketTest):
     Test the send() implementation of socket.sendfile().
     """
 
-    FILESIZE = (10 * 1024 * 1024)  # 10MB
+    FILESIZE = (10 * 1024 * 1024)  # 10 MiB
     BUFSIZE = 8192
     FILEDATA = b""
     TIMEOUT = 2

Lib/test/test_tarfile.py

@@ -779,12 +779,12 @@ class Bz2DetectReadTest(Bz2Test, DetectReadTest):
     def test_detect_stream_bz2(self):
         # Originally, tarfile's stream detection looked for the string
         # "BZh91" at the start of the file. This is incorrect because
-        # the '9' represents the blocksize (900kB). If the file was
+        # the '9' represents the blocksize (900,000 bytes). If the file was
         # compressed using another blocksize autodetection fails.
         with open(tarname, "rb") as fobj:
             data = fobj.read()
 
-        # Compress with blocksize 100kB, the file starts with "BZh11".
+        # Compress with blocksize 100,000 bytes, the file starts with "BZh11".
         with bz2.BZ2File(tmpname, "wb", compresslevel=1) as fobj:
             fobj.write(data)
 

Lib/test/test_threading.py

@@ -132,10 +132,10 @@ class ThreadTests(BaseTestCase):
         # Kill the "immortal" _DummyThread
         del threading._active[ident[0]]
 
-    # run with a small(ish) thread stack size (256kB)
+    # run with a small(ish) thread stack size (256 KiB)
     def test_various_ops_small_stack(self):
         if verbose:
-            print('with 256kB thread stack size...')
+            print('with 256 KiB thread stack size...')
         try:
             threading.stack_size(262144)
         except _thread.error:
@@ -144,10 +144,10 @@ class ThreadTests(BaseTestCase):
         self.test_various_ops()
         threading.stack_size(0)
 
-    # run with a large thread stack size (1MB)
+    # run with a large thread stack size (1 MiB)
     def test_various_ops_large_stack(self):
         if verbose:
-            print('with 1MB thread stack size...')
+            print('with 1 MiB thread stack size...')
         try:
             threading.stack_size(0x100000)
         except _thread.error:

Lib/test/test_zipfile64.py

@@ -39,7 +39,7 @@ class TestsWithSourceFile(unittest.TestCase):
         # Create the ZIP archive.
         zipfp = zipfile.ZipFile(f, "w", compression)
 
-        # It will contain enough copies of self.data to reach about 6GB of
+        # It will contain enough copies of self.data to reach about 6 GiB of
         # raw data to store.
         filecount = 6*1024**3 // len(self.data)
 

Lib/test/test_zlib.py

@@ -72,7 +72,7 @@ class ChecksumTestCase(unittest.TestCase):
         self.assertEqual(binascii.crc32(b'spam'), zlib.crc32(b'spam'))
 
 
-# Issue #10276 - check that inputs >=4GB are handled correctly.
+# Issue #10276 - check that inputs >=4 GiB are handled correctly.
 class ChecksumBigBufferTestCase(unittest.TestCase):
 
     @bigmemtest(size=_4G + 4, memuse=1, dry_run=False)
@@ -130,7 +130,7 @@ class ExceptionTestCase(unittest.TestCase):
 class BaseCompressTestCase(object):
     def check_big_compress_buffer(self, size, compress_func):
         _1M = 1024 * 1024
-        # Generate 10MB worth of random, and expand it by repeating it.
+        # Generate 10 MiB worth of random, and expand it by repeating it.
         # The assumption is that zlib's memory is not big enough to exploit
         # such spread out redundancy.
         data = b''.join([random.getrandbits(8 * _1M).to_bytes(_1M, 'little')

Lib/xmlrpc/client.py

@@ -1046,7 +1046,7 @@ def gzip_encode(data):
 # in the HTTP header, as described in RFC 1952
 #
 # @param data The encoded data
-# @keyparam max_decode Maximum bytes to decode (20MB default), use negative
+# @keyparam max_decode Maximum bytes to decode (20 MiB default), use negative
 #    values for unlimited decoding
 # @return the unencoded data
 # @raises ValueError if data is not correctly coded.

Misc/NEWS.d/3.5.0a1.rst

@@ -3035,7 +3035,7 @@ by Phil Elson.
 .. section: Library
 
 os.read() now uses a :c:func:`Py_ssize_t` type instead of :c:type:`int` for
-the size to support reading more than 2 GB at once. On Windows, the size is
+the size to support reading more than 2 GiB at once. On Windows, the size is
 truncted to INT_MAX. As any call to os.read(), the OS may read less bytes
 than the number of requested bytes.
 
@@ -3144,7 +3144,7 @@ by Pablo Torres Navarrete and SilentGhost.
 .. nonce: u_oiv9
 .. section: Library
 
-ssl.RAND_add() now supports strings longer than 2 GB.
+ssl.RAND_add() now supports strings longer than 2 GiB.
 
 ..
 

Misc/README.AIX

@@ -51,24 +51,24 @@ Note: this section may not apply when compiling Python as a 64 bit
 application.
 
 By default on AIX each program gets one segment register for its data
-segment. As each segment register covers 256 MB, a Python program that
+segment. As each segment register covers 256 MiB, a Python program that
-would use more than 256MB will raise a MemoryError.  The standard
+would use more than 256 MiB will raise a MemoryError.  The standard
 Python test suite is one such application.
 
 To allocate more segment registers to Python, you must use the linker
 option -bmaxdata or the ldedit tool to specify the number of bytes you
 need in the data segment.
 
-For example, if you want to allow 512MB of memory for Python (this is
+For example, if you want to allow 512 MiB of memory for Python (this is
 enough for the test suite to run without MemoryErrors), you should run
 the following command at the end of compilation:
 
 ldedit -b maxdata:0x20000000 ./python
 
-You can allow up to 2GB of memory for Python by using the value
+You can allow up to 2 GiB of memory for Python by using the value
 0x80000000 for maxdata.
 
-It is also possible to go beyond 2GB of memory by activating Large
+It is also possible to go beyond 2 GiB of memory by activating Large
 Page Use. You should consult the IBM documentation if you need to use
 this option. You can also follow the discussion of this problem
 in issue 11212 at bugs.python.org.

Modules/_hashopenssl.c

@@ -801,7 +801,7 @@ _hashlib_scrypt_impl(PyObject *module, Py_buffer *password, Py_buffer *salt,
     }
 
     if (maxmem < 0 || maxmem > INT_MAX) {
-        /* OpenSSL 1.1.0 restricts maxmem to 32MB. It may change in the
+        /* OpenSSL 1.1.0 restricts maxmem to 32 MiB. It may change in the
            future. The maxmem constant is private to OpenSSL. */
         PyErr_Format(PyExc_ValueError,
                      "maxmem must be positive and smaller than %d",

Modules/_io/fileio.c

@@ -41,7 +41,7 @@
 #if BUFSIZ < (8*1024)
 #define SMALLCHUNK (8*1024)
 #elif (BUFSIZ >= (2 << 25))
-#error "unreasonable BUFSIZ > 64MB defined"
+#error "unreasonable BUFSIZ > 64 MiB defined"
 #else
 #define SMALLCHUNK BUFSIZ
 #endif

Modules/_io/winconsoleio.c

@@ -31,7 +31,7 @@
 #if BUFSIZ < (16*1024)
 #define SMALLCHUNK (2*1024)
 #elif (BUFSIZ >= (2 << 25))
-#error "unreasonable BUFSIZ > 64MB defined"
+#error "unreasonable BUFSIZ > 64 MiB defined"
 #else
 #define SMALLCHUNK BUFSIZ
 #endif

Modules/_threadmodule.c

@@ -1276,10 +1276,10 @@ exception is raised, and the stack size is unmodified.  32k bytes\n\
 sufficient stack space for the interpreter itself.\n\
 \n\
 Note that some platforms may have particular restrictions on values for\n\
-the stack size, such as requiring a minimum stack size larger than 32kB or\n\
+the stack size, such as requiring a minimum stack size larger than 32 KiB or\n\
 requiring allocation in multiples of the system memory page size\n\
 - platform documentation should be referred to for more information\n\
-(4kB pages are common; using multiples of 4096 for the stack size is\n\
+(4 KiB pages are common; using multiples of 4096 for the stack size is\n\
 the suggested approach in the absence of more specific information).");
 
 static PyMethodDef thread_methods[] = {

Modules/faulthandler.c

@@ -14,8 +14,8 @@
 #  include <sys/resource.h>
 #endif
 
-/* Allocate at maximum 100 MB of the stack to raise the stack overflow */
+/* Allocate at maximum 100 MiB of the stack to raise the stack overflow */
-#define STACK_OVERFLOW_MAX_SIZE (100*1024*1024)
+#define STACK_OVERFLOW_MAX_SIZE (100 * 1024 * 1024)
 
 #define FAULTHANDLER_LATER
 

Modules/hashtable.c

@@ -243,7 +243,7 @@ _Py_hashtable_print_stats(_Py_hashtable_t *ht)
            ht, ht->entries, ht->num_buckets, load * 100.0);
     if (nchains)
         printf("avg_chain_len=%.1f, ", (double)total_chain_len / nchains);
-    printf("max_chain_len=%" PY_FORMAT_SIZE_T "u, %" PY_FORMAT_SIZE_T "u kB\n",
+    printf("max_chain_len=%" PY_FORMAT_SIZE_T "u, %" PY_FORMAT_SIZE_T "u KiB\n",
            max_chain_len, size / 1024);
 }
 #endif

Objects/accu.c

@@ -64,7 +64,7 @@ _PyAccu_Accumulate(_PyAccu *acc, PyObject *unicode)
      * builds) of:
      *   - 8 bytes for the list slot
      *   - 56 bytes for the header of the unicode object
-     * that is, 64 bytes.  100000 such objects waste more than 6MB
+     * that is, 64 bytes.  100000 such objects waste more than 6 MiB
      * compared to a single concatenated string.
      */
     if (nsmall < 100000)

Objects/typeobject.c

@@ -206,7 +206,7 @@ PyType_ClearCache(void)
             method_cache_misses, (int) (100.0 * method_cache_misses / total));
     fprintf(stderr, "-- Method cache collisions  = %zd (%d%%)\n",
             method_cache_collisions, (int) (100.0 * method_cache_collisions / total));
-    fprintf(stderr, "-- Method cache size        = %zd KB\n",
+    fprintf(stderr, "-- Method cache size        = %zd KiB\n",
             sizeof(method_cache) / 1024);
 #endif
 

Python/fileutils.c

@@ -599,8 +599,8 @@ _Py_attribute_data_to_stat(BY_HANDLE_FILE_INFORMATION *info, ULONG reparse_tag,
    On POSIX, use fstat().
 
    On Windows, use GetFileType() and GetFileInformationByHandle() which support
-   files larger than 2 GB.  fstat() may fail with EOVERFLOW on files larger
+   files larger than 2 GiB.  fstat() may fail with EOVERFLOW on files larger
-   than 2 GB because the file size type is a signed 32-bit integer: see issue
+   than 2 GiB because the file size type is a signed 32-bit integer: see issue
    #23152.
 
    On Windows, set the last Windows error and return nonzero on error. On
@@ -665,8 +665,8 @@ _Py_fstat_noraise(int fd, struct _Py_stat_struct *status)
    On POSIX, use fstat().
 
    On Windows, use GetFileType() and GetFileInformationByHandle() which support
-   files larger than 2 GB.  fstat() may fail with EOVERFLOW on files larger
+   files larger than 2 GiB.  fstat() may fail with EOVERFLOW on files larger
-   than 2 GB because the file size type is a signed 32-bit integer: see issue
+   than 2 GiB because the file size type is a signed 32-bit integer: see issue
    #23152.
 
    Raise an exception and return -1 on error. On Windows, set the last Windows

Python/thread_nt.h

@@ -323,8 +323,8 @@ PyThread_release_lock(PyThread_type_lock aLock)
 }
 
 /* minimum/maximum thread stack sizes supported */
-#define THREAD_MIN_STACKSIZE    0x8000          /* 32kB */
+#define THREAD_MIN_STACKSIZE    0x8000          /* 32 KiB */
-#define THREAD_MAX_STACKSIZE    0x10000000      /* 256MB */
+#define THREAD_MAX_STACKSIZE    0x10000000      /* 256 MiB */
 
 /* set the thread stack size.
  * Return 0 if size is valid, -1 otherwise.

Python/thread_pthread.h

@@ -35,7 +35,7 @@
 #define THREAD_STACK_SIZE       0x400000
 #endif
 /* for safety, ensure a viable minimum stacksize */
-#define THREAD_STACK_MIN        0x8000  /* 32kB */
+#define THREAD_STACK_MIN        0x8000  /* 32 KiB */
 #else  /* !_POSIX_THREAD_ATTR_STACKSIZE */
 #ifdef THREAD_STACK_SIZE
 #error "THREAD_STACK_SIZE defined but _POSIX_THREAD_ATTR_STACKSIZE undefined"

Tools/iobench/iobench.py

@@ -29,9 +29,9 @@ def text_open(fn, mode, encoding=None):
         return open(fn, mode)
 
 def get_file_sizes():
-    for s in ['20 KB', '400 KB', '10 MB']:
+    for s in ['20 KiB', '400 KiB', '10 MiB']:
         size, unit = s.split()
-        size = int(size) * {'KB': 1024, 'MB': 1024 ** 2}[unit]
+        size = int(size) * {'KiB': 1024, 'MiB': 1024 ** 2}[unit]
         yield s.replace(' ', ''), size
 
 def get_binary_files():
@@ -273,7 +273,7 @@ def run_all_tests(options):
 
     def print_results(size, n, real, cpu):
         bw = n * float(size) / 1024 ** 2 / real
-        bw = ("%4d MB/s" if bw > 100 else "%.3g MB/s") % bw
+        bw = ("%4d MiB/s" if bw > 100 else "%.3g MiB/s") % bw
         out.write(bw.rjust(12) + "\n")
         if cpu < 0.90 * real:
             out.write("   warning: test above used only %d%% CPU, "

setup.py

@@ -907,7 +907,7 @@ class PyBuildExt(build_ext):
                 missing.append('_hashlib')
 
         # We always compile these even when OpenSSL is available (issue #14693).
-        # It's harmless and the object code is tiny (40-50 KB per module,
+        # It's harmless and the object code is tiny (40-50 KiB per module,
         # only loaded when actually used).
         exts.append( Extension('_sha256', ['sha256module.c'],
                                depends=['hashlib.h']) )