As stated in the riscv reference here: https://docs.riscv.org/reference/isa/unpriv/f-st-ext.html#21-1-3-nan-generation-and-propagation
Except when otherwise stated, if the result of a floating-point operation is NaN, it is the canonical NaN. The canonical NaN has a positive sign and all significand bits clear except the MSB, a.k.a. the quiet bit.
With relevance to awk, since awk uses double-precision floats for numbers, canonical NaN translates to +nan = 7ff8000000000000.
In
|
-inf -inform inform -nan -nancy nancy -123 0 123 +123 nancy +nancy +nan inform +inform +inf |
, line 4 asserts that
-nan + 0 = -nan, but this is NOT true on riscv64.
Here's a minimal repro that shows the issue on both gcc and clang on riscv64:
#include <inttypes.h>
#include <math.h>
#include <stdio.h>
#include <stdint.h>
#include <string.h>
#define myprint(d) do { \
double _d = (d); \
uint64_t _u; \
memcpy(&_u, &_d, sizeof(_d)); \
printf("%s = %f %" PRIx64 "\n", #d, _d, _u); \
} while (0)
int main(void)
{
double f = -NAN;
double g = 0;
myprint(f);
myprint(g);
myprint(-NAN + 0);
myprint(f + g);
}Thus, that testcase is not portable. Downstream, we can see the test failure on FreeBSD in CI here: https://ci.freebsd.org/job/FreeBSD-main-riscv64-test/16606/testReport/usr.bin.awk.bugs-fixed/bug_fix_test/inf_nan_torture/
As stated in the riscv reference here: https://docs.riscv.org/reference/isa/unpriv/f-st-ext.html#21-1-3-nan-generation-and-propagation
With relevance to awk, since awk uses
double-precision floats for numbers, canonical NaN translates to+nan = 7ff8000000000000.In
awk/bugs-fixed/inf-nan-torture.in
Line 1 in 5739fd7
-nan + 0 = -nan, but this is NOT true on riscv64.Here's a minimal repro that shows the issue on both gcc and clang on riscv64:
Thus, that testcase is not portable. Downstream, we can see the test failure on FreeBSD in CI here: https://ci.freebsd.org/job/FreeBSD-main-riscv64-test/16606/testReport/usr.bin.awk.bugs-fixed/bug_fix_test/inf_nan_torture/