LADI
/
cpython
1
Fork 0
Code

ARM OABI 

https://bugs.gentoo.org/show_bug.cgi?id=266703
https://bugs.python.org/issue1762561
This commit is contained in:
Mike Gilbert 2018-06-17 11:14:43 -04:00 committed by Nedko Arnaudov
parent 98c48f7370
commit 2cc385ad81
1 changed files with 76 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

103
Objects/floatobject.c
View File

@ -1858,9 +1858,18 @@ float_getnewargs(PyFloatObject *v)
/* this is for the benefit of the pack/unpack routines below */
typedef enum {
unknown_format, ieee_big_endian_format, ieee_little_endian_format
unknown_format,
ieee_big_endian_format,
ieee_little_endian_format,
ieee_arm_mixed_endian_format
} float_format_type;
/* byte order of a C double for each of the recognised IEEE formats */
static const unsigned char BIG_ENDIAN_BYTEORDER[8] = {7,6,5,4,3,2,1,0};
static const unsigned char LITTLE_ENDIAN_BYTEORDER[8] = {0,1,2,3,4,5,6,7};
static const unsigned char ARM_MIXED_ENDIAN_BYTEORDER[8] = {4,5,6,7,0,1,2,3};
static float_format_type double_format, float_format;
static float_format_type detected_double_format, detected_float_format;
@ -1897,6 +1906,8 @@ float_getformat(PyTypeObject *v, PyObject* arg)
return PyString_FromString("IEEE, little-endian");
case ieee_big_endian_format:
return PyString_FromString("IEEE, big-endian");
case ieee_arm_mixed_endian_format:
return PyString_FromString("IEEE, ARM mixed-endian");
default:
Py_FatalError("insane float_format or double_format");
return NULL;
@ -1910,8 +1921,9 @@ PyDoc_STRVAR(float_getformat_doc,
"used in Python's test suite.\n"
"\n"
"typestr must be 'double' or 'float'. This function returns whichever of\n"
"'unknown', 'IEEE, big-endian' or 'IEEE, little-endian' best describes the\n"
"format of floating point numbers used by the C type named by typestr.");
"'unknown', 'IEEE, big-endian', 'IEEE, little-endian' or\n"
"'IEEE, ARM mixed-endian' best describes the format of floating-point\n"
"numbers used by the C type named by typestr.");
static PyObject *
float_setformat(PyTypeObject *v, PyObject* args)
@ -1949,11 +1961,15 @@ float_setformat(PyTypeObject *v, PyObject* args)
else if (strcmp(format, "IEEE, big-endian") == 0) {
f = ieee_big_endian_format;
}
else if (strcmp(format, "IEEE, ARM mixed-endian") == 0 &&
p == &double_format) {
f = ieee_arm_mixed_endian_format;
}
else {
PyErr_SetString(PyExc_ValueError,
"__setformat__() argument 2 must be "
"'unknown', 'IEEE, little-endian' or "
"'IEEE, big-endian'");
"'unknown', 'IEEE, little-endian', "
"'IEEE, big-endian' or 'IEEE, ARM mixed-endian'");
return NULL;
}
@ -1976,8 +1992,10 @@ PyDoc_STRVAR(float_setformat_doc,
"used in Python's test suite.\n"
"\n"
"typestr must be 'double' or 'float'. fmt must be one of 'unknown',\n"
"'IEEE, big-endian' or 'IEEE, little-endian', and in addition can only be\n"
"one of the latter two if it appears to match the underlying C reality.\n"
"'IEEE, big-endian', 'IEEE, little-endian' or 'IEEE, ARM mixed-endian'\n"
"and in addition can only be one of the last three if it appears to\n"
"match the underlying C reality. Note that the ARM mixed-endian\n"
"format can only be set for the 'double' type, not for 'float'.\n"
"\n"
"Override the automatic determination of C-level floating point type.\n"
"This affects how floats are converted to and from binary strings.");
@ -2172,7 +2190,11 @@ _PyFloat_Init(void)
Note that if we're on some whacked-out platform which uses
IEEE formats but isn't strictly little-endian or big-
endian, we will fall back to the portable shifts & masks
method. */
method.
Addendum: We also attempt to detect the mixed-endian IEEE format
used by the ARM old ABI (OABI) and also used by the FPA
floating-point unit on some older ARM processors. */
#if SIZEOF_DOUBLE == 8
{
@ -2181,6 +2203,8 @@ _PyFloat_Init(void)
detected_double_format = ieee_big_endian_format;
else if (memcmp(&x, "\x05\x04\x03\x02\x01\xff\x3f\x43", 8) == 0)
detected_double_format = ieee_little_endian_format;
else if (memcmp(&x, "\x01\xff\x3f\x43\x05\x04\x03\x02", 8) == 0)
detected_double_format = ieee_arm_mixed_endian_format;
else
detected_double_format = unknown_format;
}
@ -2526,17 +2550,31 @@ _PyFloat_Pack8(double x, unsigned char *p, int le)
}
else {
const char *s = (char*)&x;
int i, incr = 1;
int i;
const unsigned char *byteorder;
if ((double_format == ieee_little_endian_format && !le)
|| (double_format == ieee_big_endian_format && le)) {
p += 7;
incr = -1;
switch (double_format) {
case ieee_little_endian_format:
byteorder = LITTLE_ENDIAN_BYTEORDER;
break;
case ieee_big_endian_format:
byteorder = BIG_ENDIAN_BYTEORDER;
break;
case ieee_arm_mixed_endian_format:
byteorder = ARM_MIXED_ENDIAN_BYTEORDER;
break;
default:
Py_FatalError("insane float_format or double_format");
return -1;
}
for (i = 0; i < 8; i++) {
*p = *s++;
p += incr;
if (le) {
for (i = 0; i < 8; i++)
p[byteorder[i]] = *s++;
}
else {
for (i = 0; i < 8; i++)
p[7-byteorder[i]] = *s++;
}
return 0;
}
@ -2695,22 +2733,33 @@ _PyFloat_Unpack8(const unsigned char *p, int le)
}
else {
double x;
char *s = (char*)&x;
const unsigned char *byteorder;
int i;
if ((double_format == ieee_little_endian_format && !le)
|| (double_format == ieee_big_endian_format && le)) {
char buf[8];
char *d = &buf[7];
int i;
switch (double_format) {
case ieee_little_endian_format:
byteorder = LITTLE_ENDIAN_BYTEORDER;
break;
case ieee_big_endian_format:
byteorder = BIG_ENDIAN_BYTEORDER;
break;
case ieee_arm_mixed_endian_format:
byteorder = ARM_MIXED_ENDIAN_BYTEORDER;
break;
default:
Py_FatalError("insane float_format or double_format");
return -1.0;
}
for (i = 0; i < 8; i++) {
*d-- = *p++;
}
memcpy(&x, buf, 8);
if (le) {
for (i=0; i<8; i++)
*s++ = p[byteorder[i]];
}
else {
memcpy(&x, p, 8);
for (i=0; i<8; i++)
*s++ = p[7-byteorder[i]];
}
return x;
}
}