replace armhf math asm source files with inline asm

this makes it possible to inline them with LTO, and is the simplest
approach to eliminating the use of .sub files.

this also makes VFP sqrt available for use with the standard EABI
(plain arm rather than armhf subarch) when libc is built with
-mfloat-abi=softfp. the same could have been done for fabs, but when
the argument and return value are in integer registers, moving to VFP
registers and back is almost certainly more costly than a simple
integer operation.

src/math/arm/fabs.c

+#include <math.h>
+
+#if __ARM_PCS_VFP
+
+double fabs(double x)
+{
+	__asm__ ("vabs.f64 %P0, %P1" : "=w"(x) : "w"(x));
+	return x;
+}
+
+#else
+
+#include "../fabs.c"
+
+#endif

src/math/arm/fabsf.c

+#include <math.h>
+
+#if __ARM_PCS_VFP
+
+float fabsf(float x)
+{
+	__asm__ ("vabs.f32 %0, %1" : "=t"(x) : "t"(x));
+	return x;
+}
+
+#else
+
+#include "../fabsf.c"
+
+#endif

src/math/arm/sqrt.c

+#include <math.h>
+
+#if __VFP_FP__ && !__SOFTFP__
+
+double sqrt(double x)
+{
+	__asm__ ("vsqrt.f64 %P0, %P1" : "=w"(x) : "w"(x));
+	return x;
+}
+
+#else
+
+#include "../sqrt.c"
+
+#endif

src/math/arm/sqrtf.c

+#include <math.h>
+
+#if __VFP_FP__ && !__SOFTFP__
+
+float sqrtf(float x)
+{
+	__asm__ ("vsqrt.f32 %0, %1" : "=t"(x) : "t"(x));
+	return x;
+}
+
+#else
+
+#include "../sqrtf.c"
+
+#endif

src/math/armebhf/fabs.sub

-../armhf/fabs.s

src/math/armebhf/fabsf.sub

-../armhf/fabsf.s

src/math/armebhf/sqrt.sub

-../armhf/sqrt.s

src/math/armebhf/sqrtf.sub

-../armhf/sqrtf.s

src/math/armhf/fabs.s

-.syntax unified
-.fpu vfp
-.text
-.global fabs
-.type   fabs,%function
-fabs:
-	vabs.f64 d0, d0
-	bx lr

src/math/armhf/fabs.sub

-fabs.s

src/math/armhf/fabsf.s

-.syntax unified
-.fpu vfp
-.text
-.global fabsf
-.type   fabsf,%function
-fabsf:
-	vabs.f32 s0, s0
-	bx lr

src/math/armhf/fabsf.sub

-fabsf.s

src/math/armhf/sqrt.s

-.syntax unified
-.fpu vfp
-.text
-.global sqrt
-.type   sqrt,%function
-sqrt:
-	vsqrt.f64 d0, d0
-	bx lr

src/math/armhf/sqrt.sub

-sqrt.s

src/math/armhf/sqrtf.s

-.syntax unified
-.fpu vfp
-.text
-.global sqrtf
-.type   sqrtf,%function
-sqrtf:
-	vsqrt.f32 s0, s0
-	bx lr

src/math/armhf/sqrtf.sub

-sqrtf.s