The o32 abi programs could only use even-numbered floating-point registers by appointment. The GCC generate assembly source code using odd-numbered floating-point register when I -march=mips64 and -mabi=32. Every double floating-point instructions will use 32-bit even/odd-numbered paired 64-bit fpr when 0 = bit FR in Status register. However kernel set bit FR to a thread at loading ELF for N32 or 64 only. In other words, o32 abi process can't use odd-numbered fpr. How ought I to act? Thanks! On Sun, Nov 27, 2011 at 1:20 PM, Ian Lance Taylor <iant@xxxxxxxxxx> wrote: > Heiher <admin@xxxxxxxxxxx> writes: > >> How to use even-numbered floating-point registers only in march=mips64 >> and mabi=o32 ? > > As far as I know there is no gcc option for that. Why do you want to do > that? Does -msingle-float do what you want? > > I think it would be fairly easy to modify the source code to use only > the even-numbered FP registers. > > Ian -- Best regards! Heiher
