cosmopolitan/third_party/python/Python/pytime.c

886 lines
23 KiB
C
Raw Normal View History

/*-*- mode:c;indent-tabs-mode:nil;c-basic-offset:4;tab-width:8;coding:utf-8 -*-│
vi: set net ft=c ts=4 sts=4 sw=4 fenc=utf-8 :vi
Python 3
https://docs.python.org/3/license.html │
*/
Undiamond Python headers This change gets the Python codebase into a state where it conforms to the conventions of this codebase. It's now possible to include headers from Python, without worrying about ordering. Python has traditionally solved that problem by "diamonding" everything in Python.h, but that's problematic since it means any change to any Python header invalidates all the build artifacts. Lastly it makes tooling not work. Since it is hard to explain to Emacs when I press C-c C-h to add an import line it shouldn't add the header that actually defines the symbol, and instead do follow the nonstandard Python convention. Progress has been made on letting Python load source code from the zip executable structure via the standard C library APIs. System calss now recognizes zip!FILENAME alternative URIs as equivalent to zip:FILENAME since Python uses colon as its delimiter. Some progress has been made on embedding the notice license terms into the Python object code. This is easier said than done since Python has an extremely complicated ownership story. - Some termios APIs have been added - Implement rewinddir() dirstream API - GetCpuCount() API added to Cosmopolitan Libc - More bugs in Cosmopolitan Libc have been fixed - zipobj.com now has flags for mangling the path - Fixed bug a priori with sendfile() on certain BSDs - Polyfill F_DUPFD and F_DUPFD_CLOEXEC across platforms - FIOCLEX / FIONCLEX now polyfilled for fast O_CLOEXEC changes - APE now supports a hybrid solution to no-self-modify for builds - Many BSD-only magnums added, e.g. O_SEARCH, O_SHLOCK, SF_NODISKIO
2021-08-12 07:42:14 +00:00
#include "libc/calls/weirdtypes.h"
#include "libc/math.h"
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
#include "libc/nt/synchronization.h"
Undiamond Python headers This change gets the Python codebase into a state where it conforms to the conventions of this codebase. It's now possible to include headers from Python, without worrying about ordering. Python has traditionally solved that problem by "diamonding" everything in Python.h, but that's problematic since it means any change to any Python header invalidates all the build artifacts. Lastly it makes tooling not work. Since it is hard to explain to Emacs when I press C-c C-h to add an import line it shouldn't add the header that actually defines the symbol, and instead do follow the nonstandard Python convention. Progress has been made on letting Python load source code from the zip executable structure via the standard C library APIs. System calss now recognizes zip!FILENAME alternative URIs as equivalent to zip:FILENAME since Python uses colon as its delimiter. Some progress has been made on embedding the notice license terms into the Python object code. This is easier said than done since Python has an extremely complicated ownership story. - Some termios APIs have been added - Implement rewinddir() dirstream API - GetCpuCount() API added to Cosmopolitan Libc - More bugs in Cosmopolitan Libc have been fixed - zipobj.com now has flags for mangling the path - Fixed bug a priori with sendfile() on certain BSDs - Polyfill F_DUPFD and F_DUPFD_CLOEXEC across platforms - FIOCLEX / FIONCLEX now polyfilled for fast O_CLOEXEC changes - APE now supports a hybrid solution to no-self-modify for builds - Many BSD-only magnums added, e.g. O_SEARCH, O_SHLOCK, SF_NODISKIO
2021-08-12 07:42:14 +00:00
#include "libc/sysv/consts/clock.h"
#include "libc/time/time.h"
#include "third_party/python/Include/floatobject.h"
#include "third_party/python/Include/longobject.h"
#include "third_party/python/Include/object.h"
#include "third_party/python/Include/pyerrors.h"
#include "third_party/python/Include/pymacro.h"
#include "third_party/python/Include/pymath.h"
#include "third_party/python/Include/pytime.h"
/* clang-format off */
#define _PyTime_check_mul_overflow(a, b) \
(assert(b > 0), \
(_PyTime_t)(a) < _PyTime_MIN / (_PyTime_t)(b) \
|| _PyTime_MAX / (_PyTime_t)(b) < (_PyTime_t)(a))
/* To millisecond (10^-3) */
#define SEC_TO_MS 1000
/* To microseconds (10^-6) */
#define MS_TO_US 1000
#define SEC_TO_US (SEC_TO_MS * MS_TO_US)
/* To nanoseconds (10^-9) */
#define US_TO_NS 1000
#define MS_TO_NS (MS_TO_US * US_TO_NS)
#define SEC_TO_NS (SEC_TO_MS * MS_TO_NS)
/* Conversion from nanoseconds */
#define NS_TO_MS (1000 * 1000)
#define NS_TO_US (1000)
typedef int clockid_t;
static void
error_time_t_overflow(void)
{
PyErr_SetString(PyExc_OverflowError,
"timestamp out of range for platform time_t");
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
_PyTime_t
_PyTime_MulDiv(_PyTime_t ticks, _PyTime_t mul, _PyTime_t div)
{
_PyTime_t intpart, remaining;
/* Compute (ticks * mul / div) in two parts to prevent integer overflow:
compute integer part, and then the remaining part.
(ticks * mul) / div == (ticks / div) * mul + (ticks % div) * mul / div
The caller must ensure that "(div - 1) * mul" cannot overflow. */
intpart = ticks / div;
ticks %= div;
remaining = ticks * mul;
remaining /= div;
return intpart * mul + remaining;
}
time_t
_PyLong_AsTime_t(PyObject *obj)
{
#if SIZEOF_TIME_T == SIZEOF_LONG_LONG
long long val;
val = PyLong_AsLongLong(obj);
#else
long val;
Py_BUILD_ASSERT(sizeof(time_t) <= sizeof(long));
val = PyLong_AsLong(obj);
#endif
if (val == -1 && PyErr_Occurred()) {
if (PyErr_ExceptionMatches(PyExc_OverflowError))
error_time_t_overflow();
return -1;
}
return (time_t)val;
}
PyObject *
_PyLong_FromTime_t(time_t t)
{
#if SIZEOF_TIME_T == SIZEOF_LONG_LONG
return PyLong_FromLongLong((long long)t);
#else
Py_BUILD_ASSERT(sizeof(time_t) <= sizeof(long));
return PyLong_FromLong((long)t);
#endif
}
/* Round to nearest with ties going to nearest even integer
(_PyTime_ROUND_HALF_EVEN) */
static double
_PyTime_RoundHalfEven(double x)
{
double rounded = round(x);
if (fabs(x-rounded) == 0.5)
/* halfway case: round to even */
rounded = 2.0*round(x/2.0);
return rounded;
}
static double
_PyTime_Round(double x, _PyTime_round_t round)
{
/* volatile avoids optimization changing how numbers are rounded */
volatile double d;
d = x;
if (round == _PyTime_ROUND_HALF_EVEN){
d = _PyTime_RoundHalfEven(d);
}
else if (round == _PyTime_ROUND_CEILING){
d = ceil(d);
}
else if (round == _PyTime_ROUND_FLOOR) {
d = floor(d);
}
else {
assert(round == _PyTime_ROUND_UP);
d = (d >= 0.0) ? ceil(d) : floor(d);
}
return d;
}
static int
_PyTime_DoubleToDenominator(double d, time_t *sec, long *numerator,
double denominator, _PyTime_round_t round)
{
double intpart;
/* volatile avoids optimization changing how numbers are rounded */
volatile double floatpart;
floatpart = modf(d, &intpart);
floatpart *= denominator;
floatpart = _PyTime_Round(floatpart, round);
if (floatpart >= denominator) {
floatpart -= denominator;
intpart += 1.0;
}
else if (floatpart < 0) {
floatpart += denominator;
intpart -= 1.0;
}
assert(0.0 <= floatpart && floatpart < denominator);
if (!_Py_InIntegralTypeRange(time_t, intpart)) {
error_time_t_overflow();
return -1;
}
*sec = (time_t)intpart;
*numerator = (long)floatpart;
return 0;
}
static int
_PyTime_ObjectToDenominator(PyObject *obj, time_t *sec, long *numerator,
double denominator, _PyTime_round_t round)
{
assert(denominator <= (double)LONG_MAX);
if (PyFloat_Check(obj)) {
double d = PyFloat_AsDouble(obj);
if (Py_IS_NAN(d)) {
*numerator = 0;
PyErr_SetString(PyExc_ValueError, "Invalid value NaN (not a number)");
return -1;
}
return _PyTime_DoubleToDenominator(d, sec, numerator,
denominator, round);
}
else {
*sec = _PyLong_AsTime_t(obj);
*numerator = 0;
if (*sec == (time_t)-1 && PyErr_Occurred())
return -1;
return 0;
}
}
int
_PyTime_ObjectToTime_t(PyObject *obj, time_t *sec, _PyTime_round_t round)
{
if (PyFloat_Check(obj)) {
double intpart;
/* volatile avoids optimization changing how numbers are rounded */
volatile double d;
d = PyFloat_AsDouble(obj);
if (Py_IS_NAN(d)) {
PyErr_SetString(PyExc_ValueError, "Invalid value NaN (not a number)");
return -1;
}
d = _PyTime_Round(d, round);
(void)modf(d, &intpart);
if (!_Py_InIntegralTypeRange(time_t, intpart)) {
error_time_t_overflow();
return -1;
}
*sec = (time_t)intpart;
return 0;
}
else {
*sec = _PyLong_AsTime_t(obj);
if (*sec == (time_t)-1 && PyErr_Occurred())
return -1;
return 0;
}
}
int
_PyTime_ObjectToTimespec(PyObject *obj, time_t *sec, long *nsec,
_PyTime_round_t round)
{
int res;
res = _PyTime_ObjectToDenominator(obj, sec, nsec, 1e9, round);
if (res == 0) {
assert(0 <= *nsec && *nsec < SEC_TO_NS);
}
return res;
}
int
_PyTime_ObjectToTimeval(PyObject *obj, time_t *sec, long *usec,
_PyTime_round_t round)
{
int res;
res = _PyTime_ObjectToDenominator(obj, sec, usec, 1e6, round);
if (res == 0) {
assert(0 <= *usec && *usec < SEC_TO_US);
}
return res;
}
static void
_PyTime_overflow(void)
{
PyErr_SetString(PyExc_OverflowError,
"timestamp too large to convert to C _PyTime_t");
}
_PyTime_t
_PyTime_FromSeconds(int seconds)
{
_PyTime_t t;
t = (_PyTime_t)seconds;
/* ensure that integer overflow cannot happen, int type should have 32
bits, whereas _PyTime_t type has at least 64 bits (SEC_TO_MS takes 30
bits). */
Py_BUILD_ASSERT(INT_MAX <= _PyTime_MAX / SEC_TO_NS);
Py_BUILD_ASSERT(INT_MIN >= _PyTime_MIN / SEC_TO_NS);
assert((t >= 0 && t <= _PyTime_MAX / SEC_TO_NS)
|| (t < 0 && t >= _PyTime_MIN / SEC_TO_NS));
t *= SEC_TO_NS;
return t;
}
_PyTime_t
_PyTime_FromNanoseconds(long long ns)
{
_PyTime_t t;
Py_BUILD_ASSERT(sizeof(long long) <= sizeof(_PyTime_t));
t = Py_SAFE_DOWNCAST(ns, long long, _PyTime_t);
return t;
}
static int
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
pytime_fromtimespec(_PyTime_t *tp, struct timespec *ts, int raise)
{
int res = 0;
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
_PyTime_t t;
Py_BUILD_ASSERT(sizeof(ts->tv_sec) <= sizeof(_PyTime_t));
t = (_PyTime_t)ts->tv_sec;
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
if (__builtin_mul_overflow(t, SEC_TO_NS, &t)) {
if (raise)
_PyTime_overflow();
res = -1;
}
t += ts->tv_nsec;
*tp = t;
return res;
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
int
_PyTime_FromTimespec(_PyTime_t *tp, struct timespec *ts)
{
return pytime_fromtimespec(tp, ts, 1);
}
int
pytime_fromtimeval(_PyTime_t *tp, struct timeval *tv, int raise)
{
_PyTime_t t;
int res = 0;
Py_BUILD_ASSERT(sizeof(tv->tv_sec) <= sizeof(_PyTime_t));
t = (_PyTime_t)tv->tv_sec;
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
if (__builtin_mul_overflow(t, SEC_TO_NS, &t)) {
if (raise)
_PyTime_overflow();
res = -1;
}
t += (_PyTime_t)tv->tv_usec * US_TO_NS;
*tp = t;
return res;
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
int
_PyTime_FromTimeval(_PyTime_t *tp, struct timeval *tv)
{
return pytime_fromtimeval(tp, tv, 1);
}
static int
_PyTime_FromFloatObject(_PyTime_t *t, double value, _PyTime_round_t round,
long unit_to_ns)
{
/* volatile avoids optimization changing how numbers are rounded */
volatile double d;
/* convert to a number of nanoseconds */
d = value;
d *= (double)unit_to_ns;
d = _PyTime_Round(d, round);
if (!_Py_InIntegralTypeRange(_PyTime_t, d)) {
_PyTime_overflow();
return -1;
}
*t = (_PyTime_t)d;
return 0;
}
static int
_PyTime_FromObject(_PyTime_t *t, PyObject *obj, _PyTime_round_t round,
long unit_to_ns)
{
if (PyFloat_Check(obj)) {
double d;
d = PyFloat_AsDouble(obj);
if (Py_IS_NAN(d)) {
PyErr_SetString(PyExc_ValueError, "Invalid value NaN (not a number)");
return -1;
}
return _PyTime_FromFloatObject(t, d, round, unit_to_ns);
}
else {
long long sec;
Py_BUILD_ASSERT(sizeof(long long) <= sizeof(_PyTime_t));
sec = PyLong_AsLongLong(obj);
if (sec == -1 && PyErr_Occurred()) {
if (PyErr_ExceptionMatches(PyExc_OverflowError))
_PyTime_overflow();
return -1;
}
if (_PyTime_check_mul_overflow(sec, unit_to_ns)) {
_PyTime_overflow();
return -1;
}
*t = sec * unit_to_ns;
return 0;
}
}
int
_PyTime_FromSecondsObject(_PyTime_t *t, PyObject *obj, _PyTime_round_t round)
{
return _PyTime_FromObject(t, obj, round, SEC_TO_NS);
}
int
_PyTime_FromMillisecondsObject(_PyTime_t *t, PyObject *obj, _PyTime_round_t round)
{
return _PyTime_FromObject(t, obj, round, MS_TO_NS);
}
double
_PyTime_AsSecondsDouble(_PyTime_t t)
{
/* volatile avoids optimization changing how numbers are rounded */
volatile double d;
if (t % SEC_TO_NS == 0) {
_PyTime_t secs;
/* Divide using integers to avoid rounding issues on the integer part.
1e-9 cannot be stored exactly in IEEE 64-bit. */
secs = t / SEC_TO_NS;
d = (double)secs;
}
else {
d = (double)t;
d /= 1e9;
}
return d;
}
PyObject *
_PyTime_AsNanosecondsObject(_PyTime_t t)
{
Py_BUILD_ASSERT(sizeof(long long) >= sizeof(_PyTime_t));
return PyLong_FromLongLong((long long)t);
}
static _PyTime_t
_PyTime_Divide(const _PyTime_t t, const _PyTime_t k,
const _PyTime_round_t round)
{
assert(k > 1);
if (round == _PyTime_ROUND_HALF_EVEN) {
_PyTime_t x, r, abs_r;
x = t / k;
r = t % k;
abs_r = Py_ABS(r);
if (abs_r > k / 2 || (abs_r == k / 2 && (Py_ABS(x) & 1))) {
if (t >= 0)
x++;
else
x--;
}
return x;
}
else if (round == _PyTime_ROUND_CEILING) {
if (t >= 0){
return (t + k - 1) / k;
}
else{
return t / k;
}
}
else if (round == _PyTime_ROUND_FLOOR){
if (t >= 0) {
return t / k;
}
else{
return (t - (k - 1)) / k;
}
}
else {
assert(round == _PyTime_ROUND_UP);
if (t >= 0) {
return (t + k - 1) / k;
}
else {
return (t - (k - 1)) / k;
}
}
}
_PyTime_t
_PyTime_AsMilliseconds(_PyTime_t t, _PyTime_round_t round)
{
return _PyTime_Divide(t, NS_TO_MS, round);
}
_PyTime_t
_PyTime_AsMicroseconds(_PyTime_t t, _PyTime_round_t round)
{
return _PyTime_Divide(t, NS_TO_US, round);
}
static int
_PyTime_AsTimeval_impl(_PyTime_t t, _PyTime_t *p_secs, int *p_us,
_PyTime_round_t round)
{
_PyTime_t secs, ns;
int usec;
int res = 0;
secs = t / SEC_TO_NS;
ns = t % SEC_TO_NS;
usec = (int)_PyTime_Divide(ns, US_TO_NS, round);
if (usec < 0) {
usec += SEC_TO_US;
if (secs != _PyTime_MIN)
secs -= 1;
else
res = -1;
}
else if (usec >= SEC_TO_US) {
usec -= SEC_TO_US;
if (secs != _PyTime_MAX)
secs += 1;
else
res = -1;
}
assert(0 <= usec && usec < SEC_TO_US);
*p_secs = secs;
*p_us = usec;
return res;
}
static int
_PyTime_AsTimevalStruct_impl(_PyTime_t t, struct timeval *tv,
_PyTime_round_t round, int raise)
{
_PyTime_t secs, secs2;
int us;
int res;
res = _PyTime_AsTimeval_impl(t, &secs, &us, round);
#ifdef MS_WINDOWS
tv->tv_sec = (long)secs;
#else
tv->tv_sec = secs;
#endif
tv->tv_usec = us;
secs2 = (_PyTime_t)tv->tv_sec;
if (res < 0 || secs2 != secs) {
if (raise)
error_time_t_overflow();
return -1;
}
return 0;
}
int
_PyTime_AsTimeval(_PyTime_t t, struct timeval *tv, _PyTime_round_t round)
{
return _PyTime_AsTimevalStruct_impl(t, tv, round, 1);
}
int
_PyTime_AsTimeval_noraise(_PyTime_t t, struct timeval *tv, _PyTime_round_t round)
{
return _PyTime_AsTimevalStruct_impl(t, tv, round, 0);
}
int
_PyTime_AsTimevalTime_t(_PyTime_t t, time_t *p_secs, int *us,
_PyTime_round_t round)
{
_PyTime_t secs;
int res;
res = _PyTime_AsTimeval_impl(t, &secs, us, round);
*p_secs = secs;
if (res < 0 || (_PyTime_t)*p_secs != secs) {
error_time_t_overflow();
return -1;
}
return 0;
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
static int
win_perf_counter_frequency(int64_t *pfrequency, int raise)
{
int64_t frequency;
if (!QueryPerformanceFrequency(&frequency)) {
if (raise) {
PyErr_SetFromWindowsErr(0);
}
return -1;
}
/* Sanity check: should never occur in practice */
if (frequency < 1) {
if (raise) {
PyErr_SetString(PyExc_RuntimeError,
"invalid QueryPerformanceFrequency");
}
return -1;
}
/* Check that frequency can be casted to _PyTime_t.
*
* Make also sure that (ticks * SEC_TO_NS) cannot overflow in
* _PyTime_MulDiv(), with ticks < frequency.
*
* Known QueryPerformanceFrequency() values:
*
* - 10,000,000 (10 MHz): 100 ns resolution
* - 3,579,545 Hz (3.6 MHz): 279 ns resolution
*
* None of these frequencies can overflow with 64-bit _PyTime_t, but
* check for overflow, just in case.
*/
if (frequency > _PyTime_MAX
|| frequency > (int64_t)_PyTime_MAX / (int64_t)SEC_TO_NS)
{
if (raise) {
PyErr_SetString(PyExc_OverflowError,
"QueryPerformanceFrequency is too large");
}
return -1;
}
*pfrequency = frequency;
return 0;
}
static int
py_get_win_perf_counter(_PyTime_t *tp, _Py_clock_info_t *info, int raise)
{
static int64_t frequency;
if (!frequency) {
if (win_perf_counter_frequency(&frequency, raise) < 0) {
return -1;
}
}
if (info) {
info->implementation = "QueryPerformanceCounter()";
info->resolution = 1 / (double)frequency;
info->monotonic = 1;
info->adjustable = 0;
}
int64_t ticksll;
QueryPerformanceCounter(&ticksll);
/* Make sure that casting int64_t to _PyTime_t cannot overflow,
both types are signed */
_PyTime_t ticks;
Py_BUILD_ASSERT(sizeof(ticksll) <= sizeof(ticks));
ticks = (_PyTime_t)ticksll;
*tp = _PyTime_MulDiv(ticks, SEC_TO_NS, (_PyTime_t)frequency);
return 0;
}
int
_PyTime_GetPerfCounterWithInfo(_PyTime_t *t, _Py_clock_info_t *info)
{
if (!IsWindows()) {
return _PyTime_GetMonotonicClockWithInfo(t, info);
} else {
return py_get_win_perf_counter(t, info, 1);
}
}
int
_PyTime_AsTimespec(_PyTime_t t, struct timespec *ts)
{
_PyTime_t secs, nsec;
secs = t / SEC_TO_NS;
nsec = t % SEC_TO_NS;
if (nsec < 0) {
nsec += SEC_TO_NS;
secs -= 1;
}
ts->tv_sec = (time_t)secs;
assert(0 <= nsec && nsec < SEC_TO_NS);
ts->tv_nsec = nsec;
if ((_PyTime_t)ts->tv_sec != secs) {
error_time_t_overflow();
return -1;
}
return 0;
}
static int
pygettimeofday(_PyTime_t *tp, _Py_clock_info_t *info, int raise)
{
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
if (IsWindows()) {
uint64_t large;
struct NtFileTime system_time;
assert(info == NULL || raise);
GetSystemTimeAsFileTime(&system_time);
large = system_time.dwHighDateTime;
large <<= 32;
large |= system_time.dwLowDateTime;
/* 11,644,473,600,000,000,000: number of nanoseconds between
the 1st january 1601 and the 1st january 1970 (369 years + 89 leap
days). */
*tp = large * 100 - 11644473600000000000ull;
if (info) {
bool32 isTimeAdjustmentDisabled;
uint32_t timeAdjustment, timeIncrement;
info->implementation = "GetSystemTimeAsFileTime()";
info->monotonic = 0;
if (!GetSystemTimeAdjustment(&timeAdjustment, &timeIncrement,
&isTimeAdjustmentDisabled)) {
PyErr_SetFromWindowsErr(0);
return -1;
}
info->resolution = timeIncrement * 1e-7;
info->adjustable = 1;
}
} else {
int err;
struct timespec ts;
assert(info == NULL || raise);
err = clock_gettime(CLOCK_REALTIME, &ts);
if (err) {
if (raise)
PyErr_SetFromErrno(PyExc_OSError);
return -1;
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
if (pytime_fromtimespec(tp, &ts, raise) < 0)
return -1;
if (info) {
struct timespec res;
info->implementation = "clock_gettime(CLOCK_REALTIME)";
info->monotonic = 0;
info->adjustable = 1;
if (clock_getres(CLOCK_REALTIME, &res) == 0)
info->resolution = res.tv_sec + res.tv_nsec * 1e-9;
else
info->resolution = 1e-9;
}
}
return 0;
}
_PyTime_t
_PyTime_GetSystemClock(void)
{
_PyTime_t t;
if (pygettimeofday(&t, NULL, 0) < 0) {
/* should not happen, _PyTime_Init() checked the clock at startup */
assert(0);
/* use a fixed value instead of a random value from the stack */
t = 0;
}
return t;
}
int
_PyTime_GetSystemClockWithInfo(_PyTime_t *t, _Py_clock_info_t *info)
{
return pygettimeofday(t, info, 1);
}
static int
pymonotonic(_PyTime_t *tp, _Py_clock_info_t *info, int raise)
{
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
if (IsWindows()) {
uint64_t ticks;
_PyTime_t t;
assert(info == NULL || raise);
ticks = GetTickCount64();
Py_BUILD_ASSERT(sizeof(ticks) <= sizeof(_PyTime_t));
t = (_PyTime_t)ticks;
if (__builtin_mul_overflow(t, MS_TO_NS, &t)) {
if (raise) {
_PyTime_overflow();
return -1;
}
/* Hello, time traveler! */
assert(0);
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
*tp = t * MS_TO_NS;
if (info) {
uint32_t timeAdjustment, timeIncrement;
bool32 isTimeAdjustmentDisabled, ok;
info->implementation = "GetTickCount64()";
info->monotonic = 1;
ok = GetSystemTimeAdjustment(&timeAdjustment, &timeIncrement,
&isTimeAdjustmentDisabled);
if (!ok) {
PyErr_SetFromWindowsErr(0);
return -1;
}
info->resolution = timeIncrement * 1e-7;
info->adjustable = 0;
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
return 0;
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
#ifdef __APPLE__
static mach_timebase_info_data_t timebase;
uint64_t time;
if (timebase.denom == 0) {
/* According to the Technical Q&A QA1398, mach_timebase_info() cannot
fail: https://developer.apple.com/library/mac/#qa/qa1398/ */
(void)mach_timebase_info(&timebase);
}
time = mach_absolute_time();
/* apply timebase factor */
time *= timebase.numer;
time /= timebase.denom;
*tp = time;
if (info) {
info->implementation = "mach_absolute_time()";
info->resolution = (double)timebase.numer / timebase.denom * 1e-9;
info->monotonic = 1;
info->adjustable = 0;
}
#else
struct timespec ts;
#ifdef CLOCK_HIGHRES
const clockid_t clk_id = CLOCK_HIGHRES;
const char *implementation = "clock_gettime(CLOCK_HIGHRES)";
#else
const clockid_t clk_id = CLOCK_MONOTONIC;
const char *implementation = "clock_gettime(CLOCK_MONOTONIC)";
#endif
assert(info == NULL || raise);
if (clock_gettime(clk_id, &ts) != 0) {
if (raise) {
PyErr_SetFromErrno(PyExc_OSError);
return -1;
}
return -1;
}
if (info) {
struct timespec res;
info->monotonic = 1;
info->implementation = implementation;
info->adjustable = 0;
if (clock_getres(clk_id, &res) != 0) {
PyErr_SetFromErrno(PyExc_OSError);
return -1;
}
info->resolution = res.tv_sec + res.tv_nsec * 1e-9;
}
Productionize new APE loader and more The APE_NO_MODIFY_SELF loader payload has been moved out of the examples folder and improved so that it works on BSD systems, and permits general elf program headers. This brings its quality up enough that it should be acceptable to use by default for many programs, e.g. Python, Lua, SQLite and Python. It's the responsibility of the user to define an appropriate TMPDIR if /tmp is considered an adversarial environment. Mac OS shall be supported by APE_NO_MODIFY_SELF soon. Fixes and improvements have been made to program_executable_name as it's now the one true way to get the absolute path of the executing image. This change fixes a memory leak in linenoise history loading, introduced by performance optimizations in 51904e2687c04d7ae20410cd94c2148972d6bae6 This change fixes a longstanding regression with Mach system calls, that 23ae9dfceb6f657862f00e674a8e4dc357a9d24d back in February which impacted our sched_yield() implementation, which is why no one noticed until now. The Blinkenlights PC emulator has been improved. We now fix rendering on XNU and BSD by not making the assumption that the kernel terminal driver understands UTF8 since that seems to break its internal modeling of \r\n which is now being addressed by using \e[𝑦H instead. The paneling is now more compact in real mode so you won't need to make your font as tiny if you're only emulating an 8086 program. The CLMUL ISA is now emulated too This change also makes improvement to time. CLOCK_MONOTONIC now does the right thing on Windows NT. The nanosecond time module functions added in Python 3.7 have been backported. This change doubles the performance of Argon2 password stretching simply by not using its copy_block and xor_block helper functions, as they were trivial to inline thus resulting in us needing to iterate over each 1024 byte block four fewer times. This change makes code size improvements. _PyUnicode_ToNumeric() was 64k in size and now it's 10k. The CJK codec lookup tables now use lazy delta zigzag deflate (δzd) encoding which reduces their size from 600k to 200k plus the code bloat caused by macro abuse in _decimal.c is now addressed so our fully-loaded statically-linked hermetically-sealed Python virtual interpreter container is now 9.4 megs in the default build mode and 5.5m in MODE=tiny which leaves plenty of room for chibicc. The pydoc web server now accommodates the use case of people who work by SSH'ing into a different machine w/ python.com -m pydoc -p8080 -h0.0.0.0 Finally Python Capsulae delenda est and won't be supported in the future
2021-10-02 15:17:04 +00:00
if (pytime_fromtimespec(tp, &ts, raise) < 0)
return -1;
#endif
return 0;
}
_PyTime_t
_PyTime_GetMonotonicClock(void)
{
_PyTime_t t;
if (pymonotonic(&t, NULL, 0) < 0) {
/* should not happen, _PyTime_Init() checked that monotonic clock at
startup */
assert(0);
/* use a fixed value instead of a random value from the stack */
t = 0;
}
return t;
}
int
_PyTime_GetMonotonicClockWithInfo(_PyTime_t *tp, _Py_clock_info_t *info)
{
return pymonotonic(tp, info, 1);
}
int
_PyTime_Init(void)
{
_PyTime_t t;
/* ensure that the system clock works */
if (_PyTime_GetSystemClockWithInfo(&t, NULL) < 0)
return -1;
/* ensure that the operating system provides a monotonic clock */
if (_PyTime_GetMonotonicClockWithInfo(&t, NULL) < 0)
return -1;
return 0;
}
int
_PyTime_localtime(time_t t, struct tm *tm)
{
#ifdef MS_WINDOWS
int error;
error = localtime_s(tm, &t);
if (error != 0) {
errno = error;
PyErr_SetFromErrno(PyExc_OSError);
return -1;
}
return 0;
#else /* !MS_WINDOWS */
if (localtime_r(&t, tm) == NULL) {
#ifdef EINVAL
if (errno == 0)
errno = EINVAL;
#endif
PyErr_SetFromErrno(PyExc_OSError);
return -1;
}
return 0;
#endif /* MS_WINDOWS */
}
int
_PyTime_gmtime(time_t t, struct tm *tm)
{
#ifdef MS_WINDOWS
int error;
error = gmtime_s(tm, &t);
if (error != 0) {
errno = error;
PyErr_SetFromErrno(PyExc_OSError);
return -1;
}
return 0;
#else /* !MS_WINDOWS */
if (gmtime_r(&t, tm) == NULL) {
#ifdef EINVAL
if (errno == 0)
errno = EINVAL;
#endif
PyErr_SetFromErrno(PyExc_OSError);
return -1;
}
return 0;
#endif /* MS_WINDOWS */
}