"-msoft-float" makes no sense for ARC because there's no such thing as "-mhard-float" on ARC. Instead we use our own "-mfpu=XXX" option when a particular floating-point related HW feature is enabled in the ARC core.
We used to live with that phony option for quite some time but with migration to newer GCC following warning now appears: -------------------->8----------------- arc-linux-gcc: warning: ‘-msoft-float’ is deprecated cc1: warning: ‘-msoft-float’ is deprecated [enabled by default] -------------------->8-----------------
And that warning gets printed for each invocation of gcc with the option in question, which makes compilation output barely readable.
So we disable that phony option for ARC now.
Signed-off-by: Alexey Brodkin abrodkin@synopsys.com Cc: Waldemar Brodkorb wbx@uclibc-ng.org Cc: Vineet Gupta vgupta@synopsys.com Cc: Anton Kolesov akolesov@synopsys.com --- Rules.mak | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/Rules.mak b/Rules.mak index 207c4ec..0517f32 100644 --- a/Rules.mak +++ b/Rules.mak @@ -303,6 +303,7 @@ ifneq ($(TARGET_ARCH),nios2) ifneq ($(TARGET_ARCH),sh) ifneq ($(TARGET_ARCH),c6x) ifneq ($(TARGET_ARCH),h8300) +ifneq ($(TARGET_ARCH),aarch) CPU_CFLAGS-y += -msoft-float endif endif @@ -311,6 +312,7 @@ endif endif endif endif +endif
$(eval $(call check-gcc-var,-std=gnu99)) CPU_CFLAGS-y += $(CFLAG_-std=gnu99)