qemu-e2k/tests/tcg/loongarch64/test_fclass.c

131 lines
5.1 KiB
C
Raw Normal View History

#include <stdio.h>
/* float class */
#define FLOAT_CLASS_SIGNALING_NAN 0x001
#define FLOAT_CLASS_QUIET_NAN 0x002
#define FLOAT_CLASS_NEGATIVE_INFINITY 0x004
#define FLOAT_CLASS_NEGATIVE_NORMAL 0x008
#define FLOAT_CLASS_NEGATIVE_SUBNORMAL 0x010
#define FLOAT_CLASS_NEGATIVE_ZERO 0x020
#define FLOAT_CLASS_POSITIVE_INFINITY 0x040
#define FLOAT_CLASS_POSITIVE_NORMAL 0x080
#define FLOAT_CLASS_POSITIVE_SUBNORMAL 0x100
#define FLOAT_CLASS_POSITIVE_ZERO 0x200
#define TEST_FCLASS(N) \
void test_fclass_##N(long s) \
{ \
double fd; \
long rd; \
\
asm volatile("fclass."#N" %0, %2\n\t" \
"movfr2gr."#N" %1, %2\n\t" \
: "=f"(fd), "=r"(rd) \
: "f"(s) \
: ); \
switch (rd) { \
case FLOAT_CLASS_SIGNALING_NAN: \
case FLOAT_CLASS_QUIET_NAN: \
case FLOAT_CLASS_NEGATIVE_INFINITY: \
case FLOAT_CLASS_NEGATIVE_NORMAL: \
case FLOAT_CLASS_NEGATIVE_SUBNORMAL: \
case FLOAT_CLASS_NEGATIVE_ZERO: \
case FLOAT_CLASS_POSITIVE_INFINITY: \
case FLOAT_CLASS_POSITIVE_NORMAL: \
case FLOAT_CLASS_POSITIVE_SUBNORMAL: \
case FLOAT_CLASS_POSITIVE_ZERO: \
break; \
default: \
printf("fclass."#N" test failed.\n"); \
break; \
} \
}
/*
* float format
* type | S | Exponent | Fraction | example value
* 31 | 30 --23 | 22 | 21 --0 |
* | bit |
* SNAN 0/1 | 0xFF | 0 | !=0 | 0x7FBFFFFF
* QNAN 0/1 | 0xFF | 1 | | 0x7FCFFFFF
* -infinity 1 | 0xFF | 0 | 0xFF800000
* -normal 1 | [1, 0xFE] | [0, 0x7FFFFF]| 0xFF7FFFFF
* -subnormal 1 | 0 | !=0 | 0x807FFFFF
* -0 1 | 0 | 0 | 0x80000000
* +infinity 0 | 0xFF | 0 | 0x7F800000
* +normal 0 | [1, 0xFE] | [0, 0x7FFFFF]| 0x7F7FFFFF
* +subnormal 0 | 0 | !=0 | 0x007FFFFF
* +0 0 | 0 | 0 | 0x00000000
*/
long float_snan = 0x7FBFFFFF;
long float_qnan = 0x7FCFFFFF;
long float_neg_infinity = 0xFF800000;
long float_neg_normal = 0xFF7FFFFF;
long float_neg_subnormal = 0x807FFFFF;
long float_neg_zero = 0x80000000;
long float_post_infinity = 0x7F800000;
long float_post_normal = 0x7F7FFFFF;
long float_post_subnormal = 0x007FFFFF;
long float_post_zero = 0x00000000;
/*
* double format
* type | S | Exponent | Fraction | example value
* 63 | 62 -- 52 | 51 | 50 -- 0 |
* | bit |
* SNAN 0/1 | 0x7FF | 0 | !=0 | 0x7FF7FFFFFFFFFFFF
* QNAN 0/1 | 0x7FF | 1 | | 0x7FFFFFFFFFFFFFFF
* -infinity 1 | 0x7FF | 0 | 0xFFF0000000000000
* -normal 1 |[1, 0x7FE] | | 0xFFEFFFFFFFFFFFFF
* -subnormal 1 | 0 | !=0 | 0x8007FFFFFFFFFFFF
* -0 1 | 0 | 0 | 0x8000000000000000
* +infinity 0 | 0x7FF | 0 | 0x7FF0000000000000
* +normal 0 |[1, 0x7FE] | | 0x7FEFFFFFFFFFFFFF
* +subnormal 0 | 0 | !=0 | 0x000FFFFFFFFFFFFF
* +0 0 | 0 | 0 | 0x0000000000000000
*/
long double_snan = 0x7FF7FFFFFFFFFFFF;
long double_qnan = 0x7FFFFFFFFFFFFFFF;
long double_neg_infinity = 0xFFF0000000000000;
long double_neg_normal = 0xFFEFFFFFFFFFFFFF;
long double_neg_subnormal = 0x8007FFFFFFFFFFFF;
long double_neg_zero = 0x8000000000000000;
long double_post_infinity = 0x7FF0000000000000;
long double_post_normal = 0x7FEFFFFFFFFFFFFF;
long double_post_subnormal = 0x000FFFFFFFFFFFFF;
long double_post_zero = 0x0000000000000000;
TEST_FCLASS(s)
TEST_FCLASS(d)
int main()
{
/* fclass.s */
test_fclass_s(float_snan);
test_fclass_s(float_qnan);
test_fclass_s(float_neg_infinity);
test_fclass_s(float_neg_normal);
test_fclass_s(float_neg_subnormal);
test_fclass_s(float_neg_zero);
test_fclass_s(float_post_infinity);
test_fclass_s(float_post_normal);
test_fclass_s(float_post_subnormal);
test_fclass_s(float_post_zero);
/* fclass.d */
test_fclass_d(double_snan);
test_fclass_d(double_qnan);
test_fclass_d(double_neg_infinity);
test_fclass_d(double_neg_normal);
test_fclass_d(double_neg_subnormal);
test_fclass_d(double_neg_zero);
test_fclass_d(double_post_infinity);
test_fclass_d(double_post_normal);
test_fclass_d(double_post_subnormal);
test_fclass_d(double_post_zero);
return 0;
}