--- Additional Comments From asl at math dot spbu dot ru 2009-12-20 17:42
---
Testcase:
$ cat vcvt.s
.text
.globl test
.align 2
test:
vcvt.s32.f64s0, d0
$ arm-eabi-as vcvt.s -mfpu=vfp -o test.o && arm-eabi-objdump -d test.o &&
cat test.o
:
According to ARM docs instructions for conversion between 32-bit integers and
floats have the following format:
vcvt{r}.S32.F64 Sd, Dm (same for other directions)
here "r" specifies rounding mode, if it's used then operation uses rounding
mode from FPSCR, if "r" is omitted - then rounding towar
