diff --git a/newlib/libc/include/machine/ieeefp.h b/newlib/libc/include/machine/ieeefp.h
index 29c08fa43..b1b44665e 100644
--- a/newlib/libc/include/machine/ieeefp.h
+++ b/newlib/libc/include/machine/ieeefp.h
@@ -48,6 +48,23 @@
         This represents what type a float arg is passed as.  It is used when the type is
         not promoted to double.
 	
+
+   __OBSOLETE_MATH_DEFAULT
+
+	Default value for __OBSOLETE_MATH if that's not set by the user.
+	It should be set here based on predefined feature macros.
+
+   __OBSOLETE_MATH
+
+	If set to 1 then some new math code will be disabled and older libm
+	code will be used instead.  This is necessary because the new math
+	code does not support all targets, it assumes that the toolchain has
+	ISO C99 support (hexfloat literals, standard fenv semantics), the
+	target has IEEE-754 conforming binary32 float and binary64 double
+	(not mixed endian) representation, standard SNaN representation,
+	double and single precision arithmetics has similar latency and it
+	has no legacy SVID matherr support, only POSIX errno and fenv
+	exception based error handling.
 */
 
 #if (defined(__arm__) || defined(__thumb__)) && !defined(__MAVERICK__)
@@ -61,6 +78,7 @@
 # else
 #  define __IEEE_BIG_ENDIAN
 # endif
+# define __OBSOLETE_MATH_DEFAULT 0
 #else
 # define __IEEE_BIG_ENDIAN
 # ifdef __ARMEL__
@@ -75,6 +93,7 @@
 #else
 #define __IEEE_BIG_ENDIAN
 #endif
+#define __OBSOLETE_MATH_DEFAULT 0
 #endif
 
 #ifdef __epiphany__
@@ -427,6 +446,14 @@
 #define __IEEE_BIG_ENDIAN
 #endif
 
+#ifndef __OBSOLETE_MATH_DEFAULT
+/* Use old math code by default.  */
+#define __OBSOLETE_MATH_DEFAULT 1
+#endif
+#ifndef __OBSOLETE_MATH
+#define __OBSOLETE_MATH __OBSOLETE_MATH_DEFAULT
+#endif
+
 #ifndef __IEEE_BIG_ENDIAN
 #ifndef __IEEE_LITTLE_ENDIAN
 #error Endianess not declared!!
diff --git a/newlib/libm/common/Makefile.am b/newlib/libm/common/Makefile.am
index 1268e17db..3e2620c7e 100644
--- a/newlib/libm/common/Makefile.am
+++ b/newlib/libm/common/Makefile.am
@@ -18,11 +18,14 @@ fsrc =	sf_finite.c sf_copysign.c sf_modf.c sf_scalbn.c \
 	sf_cbrt.c sf_exp10.c sf_expm1.c sf_ilogb.c \
 	sf_infinity.c sf_isinf.c sf_isinff.c sf_isnan.c sf_isnanf.c \
 	sf_log1p.c sf_nan.c sf_nextafter.c sf_pow10.c \
-	sf_rint.c sf_logb.c sf_log2.c \
+	sf_rint.c sf_logb.c \
 	sf_fdim.c sf_fma.c sf_fmax.c sf_fmin.c sf_fpclassify.c \
 	sf_lrint.c sf_llrint.c \
 	sf_lround.c sf_llround.c sf_nearbyint.c sf_remquo.c sf_round.c \
-	sf_scalbln.c sf_trunc.c
+	sf_scalbln.c sf_trunc.c \
+	sf_exp.c sf_exp2.c sf_exp2_data.c sf_log.c sf_log_data.c \
+	sf_log2.c sf_log2_data.c sf_pow_log2_data.c sf_pow.c \
+	math_errf.c
 
 lsrc =	atanl.c cosl.c sinl.c tanl.c tanhl.c frexpl.c modfl.c ceill.c fabsl.c \
 	floorl.c log1pl.c expm1l.c acosl.c asinl.c atan2l.c coshl.c sinhl.c \
diff --git a/newlib/libm/common/Makefile.in b/newlib/libm/common/Makefile.in
index 9b367653d..c603be361 100644
--- a/newlib/libm/common/Makefile.in
+++ b/newlib/libm/common/Makefile.in
@@ -78,21 +78,21 @@ am__objects_1 = lib_a-s_finite.$(OBJEXT) lib_a-s_copysign.$(OBJEXT) \
 	lib_a-s_modf.$(OBJEXT) lib_a-s_scalbn.$(OBJEXT) \
 	lib_a-s_cbrt.$(OBJEXT) lib_a-s_exp10.$(OBJEXT) \
 	lib_a-s_expm1.$(OBJEXT) lib_a-s_ilogb.$(OBJEXT) \
-	lib_a-s_infinity.$(OBJEXT) \
-	lib_a-s_isinf.$(OBJEXT) lib_a-s_isinfd.$(OBJEXT) \
-	lib_a-s_isnan.$(OBJEXT) lib_a-s_isnand.$(OBJEXT) \
-	lib_a-s_log1p.$(OBJEXT) lib_a-s_nan.$(OBJEXT) \
-	lib_a-s_nextafter.$(OBJEXT) lib_a-s_pow10.$(OBJEXT) \
-	lib_a-s_rint.$(OBJEXT) lib_a-s_logb.$(OBJEXT) \
-	lib_a-s_log2.$(OBJEXT) lib_a-s_matherr.$(OBJEXT) \
-	lib_a-s_lib_ver.$(OBJEXT) lib_a-s_fdim.$(OBJEXT) \
-	lib_a-s_fma.$(OBJEXT) lib_a-s_fmax.$(OBJEXT) \
-	lib_a-s_fmin.$(OBJEXT) lib_a-s_fpclassify.$(OBJEXT) \
-	lib_a-s_lrint.$(OBJEXT) lib_a-s_llrint.$(OBJEXT) \
-	lib_a-s_lround.$(OBJEXT) lib_a-s_llround.$(OBJEXT) \
-	lib_a-s_nearbyint.$(OBJEXT) lib_a-s_remquo.$(OBJEXT) \
-	lib_a-s_round.$(OBJEXT) lib_a-s_scalbln.$(OBJEXT) \
-	lib_a-s_signbit.$(OBJEXT) lib_a-s_trunc.$(OBJEXT)
+	lib_a-s_infinity.$(OBJEXT) lib_a-s_isinf.$(OBJEXT) \
+	lib_a-s_isinfd.$(OBJEXT) lib_a-s_isnan.$(OBJEXT) \
+	lib_a-s_isnand.$(OBJEXT) lib_a-s_log1p.$(OBJEXT) \
+	lib_a-s_nan.$(OBJEXT) lib_a-s_nextafter.$(OBJEXT) \
+	lib_a-s_pow10.$(OBJEXT) lib_a-s_rint.$(OBJEXT) \
+	lib_a-s_logb.$(OBJEXT) lib_a-s_log2.$(OBJEXT) \
+	lib_a-s_matherr.$(OBJEXT) lib_a-s_lib_ver.$(OBJEXT) \
+	lib_a-s_fdim.$(OBJEXT) lib_a-s_fma.$(OBJEXT) \
+	lib_a-s_fmax.$(OBJEXT) lib_a-s_fmin.$(OBJEXT) \
+	lib_a-s_fpclassify.$(OBJEXT) lib_a-s_lrint.$(OBJEXT) \
+	lib_a-s_llrint.$(OBJEXT) lib_a-s_lround.$(OBJEXT) \
+	lib_a-s_llround.$(OBJEXT) lib_a-s_nearbyint.$(OBJEXT) \
+	lib_a-s_remquo.$(OBJEXT) lib_a-s_round.$(OBJEXT) \
+	lib_a-s_scalbln.$(OBJEXT) lib_a-s_signbit.$(OBJEXT) \
+	lib_a-s_trunc.$(OBJEXT)
 am__objects_2 = lib_a-sf_finite.$(OBJEXT) lib_a-sf_copysign.$(OBJEXT) \
 	lib_a-sf_modf.$(OBJEXT) lib_a-sf_scalbn.$(OBJEXT) \
 	lib_a-sf_cbrt.$(OBJEXT) lib_a-sf_exp10.$(OBJEXT) \
@@ -102,14 +102,19 @@ am__objects_2 = lib_a-sf_finite.$(OBJEXT) lib_a-sf_copysign.$(OBJEXT) \
 	lib_a-sf_isnanf.$(OBJEXT) lib_a-sf_log1p.$(OBJEXT) \
 	lib_a-sf_nan.$(OBJEXT) lib_a-sf_nextafter.$(OBJEXT) \
 	lib_a-sf_pow10.$(OBJEXT) lib_a-sf_rint.$(OBJEXT) \
-	lib_a-sf_logb.$(OBJEXT) lib_a-sf_log2.$(OBJEXT) \
-	lib_a-sf_fdim.$(OBJEXT) lib_a-sf_fma.$(OBJEXT) \
-	lib_a-sf_fmax.$(OBJEXT) lib_a-sf_fmin.$(OBJEXT) \
-	lib_a-sf_fpclassify.$(OBJEXT) lib_a-sf_lrint.$(OBJEXT) \
-	lib_a-sf_llrint.$(OBJEXT) lib_a-sf_lround.$(OBJEXT) \
-	lib_a-sf_llround.$(OBJEXT) lib_a-sf_nearbyint.$(OBJEXT) \
-	lib_a-sf_remquo.$(OBJEXT) lib_a-sf_round.$(OBJEXT) \
-	lib_a-sf_scalbln.$(OBJEXT) lib_a-sf_trunc.$(OBJEXT)
+	lib_a-sf_logb.$(OBJEXT) lib_a-sf_fdim.$(OBJEXT) \
+	lib_a-sf_fma.$(OBJEXT) lib_a-sf_fmax.$(OBJEXT) \
+	lib_a-sf_fmin.$(OBJEXT) lib_a-sf_fpclassify.$(OBJEXT) \
+	lib_a-sf_lrint.$(OBJEXT) lib_a-sf_llrint.$(OBJEXT) \
+	lib_a-sf_lround.$(OBJEXT) lib_a-sf_llround.$(OBJEXT) \
+	lib_a-sf_nearbyint.$(OBJEXT) lib_a-sf_remquo.$(OBJEXT) \
+	lib_a-sf_round.$(OBJEXT) lib_a-sf_scalbln.$(OBJEXT) \
+	lib_a-sf_trunc.$(OBJEXT) lib_a-sf_exp.$(OBJEXT) \
+	lib_a-sf_exp2.$(OBJEXT) lib_a-sf_exp2_data.$(OBJEXT) \
+	lib_a-sf_log.$(OBJEXT) lib_a-sf_log_data.$(OBJEXT) \
+	lib_a-sf_log2.$(OBJEXT) lib_a-sf_log2_data.$(OBJEXT) \
+	lib_a-sf_pow_log2_data.$(OBJEXT) lib_a-sf_pow.$(OBJEXT) \
+	lib_a-math_errf.$(OBJEXT)
 am__objects_3 = lib_a-atanl.$(OBJEXT) lib_a-cosl.$(OBJEXT) \
 	lib_a-sinl.$(OBJEXT) lib_a-tanl.$(OBJEXT) \
 	lib_a-tanhl.$(OBJEXT) lib_a-frexpl.$(OBJEXT) \
@@ -148,21 +153,23 @@ lib_a_OBJECTS = $(am_lib_a_OBJECTS)
 LTLIBRARIES = $(noinst_LTLIBRARIES)
 libcommon_la_LIBADD =
 am__objects_5 = s_finite.lo s_copysign.lo s_modf.lo s_scalbn.lo \
-	s_cbrt.lo s_exp10.lo s_expm1.lo s_ilogb.lo \
-	s_infinity.lo s_isinf.lo s_isinfd.lo s_isnan.lo s_isnand.lo \
-	s_log1p.lo s_nan.lo s_nextafter.lo s_pow10.lo s_rint.lo \
-	s_logb.lo s_log2.lo s_matherr.lo s_lib_ver.lo s_fdim.lo \
-	s_fma.lo s_fmax.lo s_fmin.lo s_fpclassify.lo s_lrint.lo \
-	s_llrint.lo s_lround.lo s_llround.lo s_nearbyint.lo \
-	s_remquo.lo s_round.lo s_scalbln.lo s_signbit.lo s_trunc.lo
+	s_cbrt.lo s_exp10.lo s_expm1.lo s_ilogb.lo s_infinity.lo \
+	s_isinf.lo s_isinfd.lo s_isnan.lo s_isnand.lo s_log1p.lo \
+	s_nan.lo s_nextafter.lo s_pow10.lo s_rint.lo s_logb.lo \
+	s_log2.lo s_matherr.lo s_lib_ver.lo s_fdim.lo s_fma.lo \
+	s_fmax.lo s_fmin.lo s_fpclassify.lo s_lrint.lo s_llrint.lo \
+	s_lround.lo s_llround.lo s_nearbyint.lo s_remquo.lo s_round.lo \
+	s_scalbln.lo s_signbit.lo s_trunc.lo
 am__objects_6 = sf_finite.lo sf_copysign.lo sf_modf.lo sf_scalbn.lo \
 	sf_cbrt.lo sf_exp10.lo sf_expm1.lo sf_ilogb.lo sf_infinity.lo \
 	sf_isinf.lo sf_isinff.lo sf_isnan.lo sf_isnanf.lo sf_log1p.lo \
 	sf_nan.lo sf_nextafter.lo sf_pow10.lo sf_rint.lo sf_logb.lo \
-	sf_log2.lo sf_fdim.lo sf_fma.lo sf_fmax.lo sf_fmin.lo \
-	sf_fpclassify.lo sf_lrint.lo sf_llrint.lo sf_lround.lo \
-	sf_llround.lo sf_nearbyint.lo sf_remquo.lo sf_round.lo \
-	sf_scalbln.lo sf_trunc.lo
+	sf_fdim.lo sf_fma.lo sf_fmax.lo sf_fmin.lo sf_fpclassify.lo \
+	sf_lrint.lo sf_llrint.lo sf_lround.lo sf_llround.lo \
+	sf_nearbyint.lo sf_remquo.lo sf_round.lo sf_scalbln.lo \
+	sf_trunc.lo sf_exp.lo sf_exp2.lo sf_exp2_data.lo sf_log.lo \
+	sf_log_data.lo sf_log2.lo sf_log2_data.lo sf_pow_log2_data.lo \
+	sf_pow.lo math_errf.lo
 am__objects_7 = atanl.lo cosl.lo sinl.lo tanl.lo tanhl.lo frexpl.lo \
 	modfl.lo ceill.lo fabsl.lo floorl.lo log1pl.lo expm1l.lo \
 	acosl.lo asinl.lo atan2l.lo coshl.lo sinhl.lo expl.lo \
@@ -347,11 +354,14 @@ fsrc = sf_finite.c sf_copysign.c sf_modf.c sf_scalbn.c \
 	sf_cbrt.c sf_exp10.c sf_expm1.c sf_ilogb.c \
 	sf_infinity.c sf_isinf.c sf_isinff.c sf_isnan.c sf_isnanf.c \
 	sf_log1p.c sf_nan.c sf_nextafter.c sf_pow10.c \
-	sf_rint.c sf_logb.c sf_log2.c \
+	sf_rint.c sf_logb.c \
 	sf_fdim.c sf_fma.c sf_fmax.c sf_fmin.c sf_fpclassify.c \
 	sf_lrint.c sf_llrint.c \
 	sf_lround.c sf_llround.c sf_nearbyint.c sf_remquo.c sf_round.c \
-	sf_scalbln.c sf_trunc.c
+	sf_scalbln.c sf_trunc.c \
+	sf_exp.c sf_exp2.c sf_exp2_data.c sf_log.c sf_log_data.c \
+	sf_log2.c sf_log2_data.c sf_pow_log2_data.c sf_pow.c \
+	math_errf.c
 
 lsrc = atanl.c cosl.c sinl.c tanl.c tanhl.c frexpl.c modfl.c ceill.c fabsl.c \
 	floorl.c log1pl.c expm1l.c acosl.c asinl.c atan2l.c coshl.c sinhl.c \
@@ -796,12 +806,6 @@ lib_a-sf_logb.o: sf_logb.c
 lib_a-sf_logb.obj: sf_logb.c
 	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_logb.obj `if test -f 'sf_logb.c'; then $(CYGPATH_W) 'sf_logb.c'; else $(CYGPATH_W) '$(srcdir)/sf_logb.c'; fi`
 
-lib_a-sf_log2.o: sf_log2.c
-	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log2.o `test -f 'sf_log2.c' || echo '$(srcdir)/'`sf_log2.c
-
-lib_a-sf_log2.obj: sf_log2.c
-	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log2.obj `if test -f 'sf_log2.c'; then $(CYGPATH_W) 'sf_log2.c'; else $(CYGPATH_W) '$(srcdir)/sf_log2.c'; fi`
-
 lib_a-sf_fdim.o: sf_fdim.c
 	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_fdim.o `test -f 'sf_fdim.c' || echo '$(srcdir)/'`sf_fdim.c
 
@@ -886,6 +890,66 @@ lib_a-sf_trunc.o: sf_trunc.c
 lib_a-sf_trunc.obj: sf_trunc.c
 	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_trunc.obj `if test -f 'sf_trunc.c'; then $(CYGPATH_W) 'sf_trunc.c'; else $(CYGPATH_W) '$(srcdir)/sf_trunc.c'; fi`
 
+lib_a-sf_exp.o: sf_exp.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_exp.o `test -f 'sf_exp.c' || echo '$(srcdir)/'`sf_exp.c
+
+lib_a-sf_exp.obj: sf_exp.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_exp.obj `if test -f 'sf_exp.c'; then $(CYGPATH_W) 'sf_exp.c'; else $(CYGPATH_W) '$(srcdir)/sf_exp.c'; fi`
+
+lib_a-sf_exp2.o: sf_exp2.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_exp2.o `test -f 'sf_exp2.c' || echo '$(srcdir)/'`sf_exp2.c
+
+lib_a-sf_exp2.obj: sf_exp2.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_exp2.obj `if test -f 'sf_exp2.c'; then $(CYGPATH_W) 'sf_exp2.c'; else $(CYGPATH_W) '$(srcdir)/sf_exp2.c'; fi`
+
+lib_a-sf_exp2_data.o: sf_exp2_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_exp2_data.o `test -f 'sf_exp2_data.c' || echo '$(srcdir)/'`sf_exp2_data.c
+
+lib_a-sf_exp2_data.obj: sf_exp2_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_exp2_data.obj `if test -f 'sf_exp2_data.c'; then $(CYGPATH_W) 'sf_exp2_data.c'; else $(CYGPATH_W) '$(srcdir)/sf_exp2_data.c'; fi`
+
+lib_a-sf_log.o: sf_log.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log.o `test -f 'sf_log.c' || echo '$(srcdir)/'`sf_log.c
+
+lib_a-sf_log.obj: sf_log.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log.obj `if test -f 'sf_log.c'; then $(CYGPATH_W) 'sf_log.c'; else $(CYGPATH_W) '$(srcdir)/sf_log.c'; fi`
+
+lib_a-sf_log_data.o: sf_log_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log_data.o `test -f 'sf_log_data.c' || echo '$(srcdir)/'`sf_log_data.c
+
+lib_a-sf_log_data.obj: sf_log_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log_data.obj `if test -f 'sf_log_data.c'; then $(CYGPATH_W) 'sf_log_data.c'; else $(CYGPATH_W) '$(srcdir)/sf_log_data.c'; fi`
+
+lib_a-sf_log2.o: sf_log2.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log2.o `test -f 'sf_log2.c' || echo '$(srcdir)/'`sf_log2.c
+
+lib_a-sf_log2.obj: sf_log2.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log2.obj `if test -f 'sf_log2.c'; then $(CYGPATH_W) 'sf_log2.c'; else $(CYGPATH_W) '$(srcdir)/sf_log2.c'; fi`
+
+lib_a-sf_log2_data.o: sf_log2_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log2_data.o `test -f 'sf_log2_data.c' || echo '$(srcdir)/'`sf_log2_data.c
+
+lib_a-sf_log2_data.obj: sf_log2_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_log2_data.obj `if test -f 'sf_log2_data.c'; then $(CYGPATH_W) 'sf_log2_data.c'; else $(CYGPATH_W) '$(srcdir)/sf_log2_data.c'; fi`
+
+lib_a-sf_pow_log2_data.o: sf_pow_log2_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_pow_log2_data.o `test -f 'sf_pow_log2_data.c' || echo '$(srcdir)/'`sf_pow_log2_data.c
+
+lib_a-sf_pow_log2_data.obj: sf_pow_log2_data.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_pow_log2_data.obj `if test -f 'sf_pow_log2_data.c'; then $(CYGPATH_W) 'sf_pow_log2_data.c'; else $(CYGPATH_W) '$(srcdir)/sf_pow_log2_data.c'; fi`
+
+lib_a-sf_pow.o: sf_pow.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_pow.o `test -f 'sf_pow.c' || echo '$(srcdir)/'`sf_pow.c
+
+lib_a-sf_pow.obj: sf_pow.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-sf_pow.obj `if test -f 'sf_pow.c'; then $(CYGPATH_W) 'sf_pow.c'; else $(CYGPATH_W) '$(srcdir)/sf_pow.c'; fi`
+
+lib_a-math_errf.o: math_errf.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-math_errf.o `test -f 'math_errf.c' || echo '$(srcdir)/'`math_errf.c
+
+lib_a-math_errf.obj: math_errf.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-math_errf.obj `if test -f 'math_errf.c'; then $(CYGPATH_W) 'math_errf.c'; else $(CYGPATH_W) '$(srcdir)/math_errf.c'; fi`
+
 lib_a-atanl.o: atanl.c
 	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-atanl.o `test -f 'atanl.c' || echo '$(srcdir)/'`atanl.c
 
diff --git a/newlib/libm/common/math_config.h b/newlib/libm/common/math_config.h
new file mode 100644
index 000000000..c5667844a
--- /dev/null
+++ b/newlib/libm/common/math_config.h
@@ -0,0 +1,179 @@
+/* Configuration for math routines.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#ifndef _MATH_CONFIG_H
+#define _MATH_CONFIG_H
+
+#include <math.h>
+#include <stdint.h>
+
+#ifndef WANT_ROUNDING
+/* Correct special case results in non-nearest rounding modes.  */
+# define WANT_ROUNDING 1
+#endif
+#ifndef WANT_ERRNO
+/* Set errno according to ISO C with (math_errhandling & MATH_ERRNO) != 0.  */
+# define WANT_ERRNO 1
+#endif
+#ifndef WANT_ERRNO_UFLOW
+/* Set errno to ERANGE if result underflows to 0 (in all rounding modes).  */
+# define WANT_ERRNO_UFLOW (WANT_ROUNDING && WANT_ERRNO)
+#endif
+
+#ifndef TOINT_INTRINSICS
+# define TOINT_INTRINSICS 0
+#endif
+#ifndef TOINT_RINT
+# define TOINT_RINT 0
+#endif
+#ifndef TOINT_SHIFT
+# define TOINT_SHIFT 1
+#endif
+
+static inline uint32_t
+asuint (float f)
+{
+  union
+  {
+    float f;
+    uint32_t i;
+  } u = {f};
+  return u.i;
+}
+
+static inline float
+asfloat (uint32_t i)
+{
+  union
+  {
+    uint32_t i;
+    float f;
+  } u = {i};
+  return u.f;
+}
+
+static inline uint64_t
+asuint64 (double f)
+{
+  union
+  {
+    double f;
+    uint64_t i;
+  } u = {f};
+  return u.i;
+}
+
+static inline double
+asdouble (uint64_t i)
+{
+  union
+  {
+    uint64_t i;
+    double f;
+  } u = {i};
+  return u.f;
+}
+
+#ifndef IEEE_754_2008_SNAN
+# define IEEE_754_2008_SNAN 1
+#endif
+static inline int
+issignalingf_inline (float x)
+{
+  uint32_t ix = asuint (x);
+  if (!IEEE_754_2008_SNAN)
+    return (ix & 0x7fc00000) == 0x7fc00000;
+  return 2 * (ix ^ 0x00400000) > 2u * 0x7fc00000;
+}
+
+#ifdef __GNUC__
+# define HIDDEN __attribute__ ((__visibility__ ("hidden")))
+# define NOINLINE __attribute__ ((noinline))
+#else
+# define HIDDEN
+# define NOINLINE
+#endif
+
+HIDDEN float __math_oflowf (unsigned long);
+HIDDEN float __math_uflowf (unsigned long);
+HIDDEN float __math_may_uflowf (unsigned long);
+HIDDEN float __math_divzerof (unsigned long);
+HIDDEN float __math_invalidf (float);
+
+/* Shared between expf, exp2f and powf.  */
+#define EXP2F_TABLE_BITS 5
+#define EXP2F_POLY_ORDER 3
+extern const struct exp2f_data
+{
+  uint64_t tab[1 << EXP2F_TABLE_BITS];
+  double shift_scaled;
+  double poly[EXP2F_POLY_ORDER];
+  double shift;
+  double invln2_scaled;
+  double poly_scaled[EXP2F_POLY_ORDER];
+} __exp2f_data HIDDEN;
+
+#define LOGF_TABLE_BITS 4
+#define LOGF_POLY_ORDER 4
+extern const struct logf_data
+{
+  struct
+  {
+    double invc, logc;
+  } tab[1 << LOGF_TABLE_BITS];
+  double ln2;
+  double poly[LOGF_POLY_ORDER - 1]; /* First order coefficient is 1.  */
+} __logf_data HIDDEN;
+
+#define LOG2F_TABLE_BITS 4
+#define LOG2F_POLY_ORDER 4
+extern const struct log2f_data
+{
+  struct
+  {
+    double invc, logc;
+  } tab[1 << LOG2F_TABLE_BITS];
+  double poly[LOG2F_POLY_ORDER];
+} __log2f_data HIDDEN;
+
+#define POWF_LOG2_TABLE_BITS 4
+#define POWF_LOG2_POLY_ORDER 5
+#if TOINT_INTRINSICS
+# define POWF_SCALE_BITS EXP2F_TABLE_BITS
+#else
+# define POWF_SCALE_BITS 0
+#endif
+#define POWF_SCALE ((double) (1 << POWF_SCALE_BITS))
+extern const struct powf_log2_data
+{
+  struct
+  {
+    double invc, logc;
+  } tab[1 << POWF_LOG2_TABLE_BITS];
+  double poly[POWF_LOG2_POLY_ORDER];
+} __powf_log2_data HIDDEN;
+
+#endif
diff --git a/newlib/libm/common/math_errf.c b/newlib/libm/common/math_errf.c
new file mode 100644
index 000000000..60b1bb3f2
--- /dev/null
+++ b/newlib/libm/common/math_errf.c
@@ -0,0 +1,89 @@
+/* Single-precision math error handling.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include "math_config.h"
+
+#if WANT_ERRNO
+#include <errno.h>
+/* NOINLINE reduces code size and avoids making math functions non-leaf
+   when the error handling is inlined.  */
+NOINLINE static float
+with_errnof (float y, int e)
+{
+  errno = e;
+  return y;
+}
+#else
+#define with_errnof(x, e) (x)
+#endif
+
+/* NOINLINE prevents fenv semantics breaking optimizations.  */
+NOINLINE static float
+xflowf (unsigned long sign, float y)
+{
+  y = (sign ? -y : y) * y;
+  return with_errnof (y, ERANGE);
+}
+
+HIDDEN float
+__math_uflowf (unsigned long sign)
+{
+  return xflowf (sign, 0x1p-95f);
+}
+
+#if WANT_ERRNO_UFLOW
+/* Underflows to zero in some non-nearest rounding mode, setting errno
+   is valid even if the result is non-zero, but in the subnormal range.  */
+HIDDEN float
+__math_may_uflowf (unsigned long sign)
+{
+  return xflowf (sign, 0x1.4p-75f);
+}
+#endif
+
+HIDDEN float
+__math_oflowf (unsigned long sign)
+{
+  return xflowf (sign, 0x1p97f);
+}
+
+HIDDEN float
+__math_divzerof (unsigned long sign)
+{
+  float y = 0;
+  return with_errnof ((sign ? -1 : 1) / y, ERANGE);
+}
+
+HIDDEN float
+__math_invalidf (float x)
+{
+  float y = (x - x) / (x - x);
+  return isnan (x) ? y : with_errnof (y, EDOM);
+}
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_exp.c b/newlib/libm/common/sf_exp.c
new file mode 100644
index 000000000..79ec62bf5
--- /dev/null
+++ b/newlib/libm/common/sf_exp.c
@@ -0,0 +1,113 @@
+/* Single-precision e^x function.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include <math.h>
+#include <stdint.h>
+#include "math_config.h"
+
+/*
+EXP2F_TABLE_BITS = 5
+EXP2F_POLY_ORDER = 3
+
+ULP error: 0.502 (nearest rounding.)
+Relative error: 1.69 * 2^-34 in [-ln2/64, ln2/64] (before rounding.)
+Wrong count: 170635 (all nearest rounding wrong results with fma.)
+Non-nearest ULP error: 1 (rounded ULP error)
+*/
+
+#define N (1 << EXP2F_TABLE_BITS)
+#define InvLn2N __exp2f_data.invln2_scaled
+#define T __exp2f_data.tab
+#define C __exp2f_data.poly_scaled
+
+static inline uint32_t
+top12 (float x)
+{
+  return asuint (x) >> 20;
+}
+
+float
+expf (float x)
+{
+  uint32_t abstop;
+  uint64_t ki, t;
+  /* double_t for better performance on targets with FLT_EVAL_METHOD==2.  */
+  double_t kd, xd, z, r, r2, y, s;
+
+  xd = (double_t) x;
+  abstop = top12 (x) & 0x7ff;
+  if (__builtin_expect (abstop >= top12 (88.0f), 0))
+    {
+      /* |x| >= 88 or x is nan.  */
+      if (asuint (x) == asuint (-INFINITY))
+	return 0.0f;
+      if (abstop >= top12 (INFINITY))
+	return x + x;
+      if (x > 0x1.62e42ep6f) /* x > log(0x1p128) ~= 88.72 */
+	return __math_oflowf (0);
+      if (x < -0x1.9fe368p6f) /* x < log(0x1p-150) ~= -103.97 */
+	return __math_uflowf (0);
+#if WANT_ERRNO_UFLOW
+      if (x < -0x1.9d1d9ep6f) /* x < log(0x1p-149) ~= -103.28 */
+	return __math_may_uflowf (0);
+#endif
+    }
+
+  /* x*N/Ln2 = k + r with r in [-1/2, 1/2] and int k.  */
+  z = InvLn2N * xd;
+
+  /* Round and convert z to int, the result is in [-150*N, 128*N] and
+     ideally ties-to-even rule is used, otherwise the magnitude of r
+     can be bigger which gives larger approximation error.  */
+#if TOINT_INTRINSICS
+  kd = roundtoint (z);
+  ki = converttoint (z);
+#elif TOINT_RINT
+  kd = rint (z);
+  ki = (long) kd;
+#elif TOINT_SHIFT
+# define SHIFT __exp2f_data.shift
+  kd = (double) (z + SHIFT); /* Rounding to double precision is required.  */
+  ki = asuint64 (kd);
+  kd -= SHIFT;
+#endif
+  r = z - kd;
+
+  /* exp(x) = 2^(k/N) * 2^(r/N) ~= s * (C0*r^3 + C1*r^2 + C2*r + 1) */
+  t = T[ki % N];
+  t += ki << (52 - EXP2F_TABLE_BITS);
+  s = asdouble (t);
+  z = C[0] * r + C[1];
+  r2 = r * r;
+  y = C[2] * r + 1;
+  y = z * r2 + y;
+  y = y * s;
+  return (float) y;
+}
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_exp2.c b/newlib/libm/common/sf_exp2.c
new file mode 100644
index 000000000..75f2551c1
--- /dev/null
+++ b/newlib/libm/common/sf_exp2.c
@@ -0,0 +1,99 @@
+/* Single-precision 2^x function.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include <math.h>
+#include <stdint.h>
+#include "math_config.h"
+
+/*
+EXP2F_TABLE_BITS = 5
+EXP2F_POLY_ORDER = 3
+
+ULP error: 0.502 (nearest rounding.)
+Relative error: 1.69 * 2^-34 in [-1/64, 1/64] (before rounding.)
+Wrong count: 168353 (all nearest rounding wrong results with fma.)
+Non-nearest ULP error: 1 (rounded ULP error)
+*/
+
+#define N (1 << EXP2F_TABLE_BITS)
+#define T __exp2f_data.tab
+#define C __exp2f_data.poly
+#define SHIFT __exp2f_data.shift_scaled
+
+static inline uint32_t
+top12 (float x)
+{
+  return asuint (x) >> 20;
+}
+
+float
+exp2f (float x)
+{
+  uint32_t abstop;
+  uint64_t ki, t;
+  /* double_t for better performance on targets with FLT_EVAL_METHOD==2.  */
+  double_t kd, xd, z, r, r2, y, s;
+
+  xd = (double_t) x;
+  abstop = top12 (x) & 0x7ff;
+  if (__builtin_expect (abstop >= top12 (128.0f), 0))
+    {
+      /* |x| >= 128 or x is nan.  */
+      if (asuint (x) == asuint (-INFINITY))
+	return 0.0f;
+      if (abstop >= top12 (INFINITY))
+	return x + x;
+      if (x > 0.0f)
+	return __math_oflowf (0);
+      if (x <= -150.0f)
+	return __math_uflowf (0);
+#if WANT_ERRNO_UFLOW
+      if (x < -149.0f)
+	return __math_may_uflowf (0);
+#endif
+    }
+
+  /* x = k/N + r with r in [-1/(2N), 1/(2N)] and int k.  */
+  kd = (double) (xd + SHIFT); /* Rounding to double precision is required.  */
+  ki = asuint64 (kd);
+  kd -= SHIFT; /* k/N for int k.  */
+  r = xd - kd;
+
+  /* exp2(x) = 2^(k/N) * 2^r ~= s * (C0*r^3 + C1*r^2 + C2*r + 1) */
+  t = T[ki % N];
+  t += ki << (52 - EXP2F_TABLE_BITS);
+  s = asdouble (t);
+  z = C[0] * r + C[1];
+  r2 = r * r;
+  y = C[2] * r + 1;
+  y = z * r2 + y;
+  y = y * s;
+  return (float) y;
+}
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_exp2_data.c b/newlib/libm/common/sf_exp2_data.c
new file mode 100644
index 000000000..63b6f391d
--- /dev/null
+++ b/newlib/libm/common/sf_exp2_data.c
@@ -0,0 +1,56 @@
+/* Shared data between expf, exp2f and powf.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include "math_config.h"
+
+#define N (1 << EXP2F_TABLE_BITS)
+
+const struct exp2f_data __exp2f_data = {
+  /* tab[i] = uint(2^(i/N)) - (i << 52-BITS)
+     used for computing 2^(k/N) for an int |k| < 150 N as
+     double(tab[k%N] + (k << 52-BITS)) */
+  .tab = {
+0x3ff0000000000000, 0x3fefd9b0d3158574, 0x3fefb5586cf9890f, 0x3fef9301d0125b51,
+0x3fef72b83c7d517b, 0x3fef54873168b9aa, 0x3fef387a6e756238, 0x3fef1e9df51fdee1,
+0x3fef06fe0a31b715, 0x3feef1a7373aa9cb, 0x3feedea64c123422, 0x3feece086061892d,
+0x3feebfdad5362a27, 0x3feeb42b569d4f82, 0x3feeab07dd485429, 0x3feea47eb03a5585,
+0x3feea09e667f3bcd, 0x3fee9f75e8ec5f74, 0x3feea11473eb0187, 0x3feea589994cce13,
+0x3feeace5422aa0db, 0x3feeb737b0cdc5e5, 0x3feec49182a3f090, 0x3feed503b23e255d,
+0x3feee89f995ad3ad, 0x3feeff76f2fb5e47, 0x3fef199bdd85529c, 0x3fef3720dcef9069,
+0x3fef5818dcfba487, 0x3fef7c97337b9b5f, 0x3fefa4afa2a490da, 0x3fefd0765b6e4540,
+  },
+  .shift_scaled = 0x1.8p+52 / N,
+  .poly = { 0x1.c6af84b912394p-5, 0x1.ebfce50fac4f3p-3, 0x1.62e42ff0c52d6p-1 },
+  .shift = 0x1.8p+52,
+  .invln2_scaled = 0x1.71547652b82fep+0 * N,
+  .poly_scaled = {
+0x1.c6af84b912394p-5/N/N/N, 0x1.ebfce50fac4f3p-3/N/N, 0x1.62e42ff0c52d6p-1/N
+  },
+};
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_log.c b/newlib/libm/common/sf_log.c
new file mode 100644
index 000000000..ba5558a54
--- /dev/null
+++ b/newlib/libm/common/sf_log.c
@@ -0,0 +1,98 @@
+/* Single-precision log function.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include <math.h>
+#include <stdint.h>
+#include "math_config.h"
+
+/*
+LOGF_TABLE_BITS = 4
+LOGF_POLY_ORDER = 4
+
+ULP error: 0.818 (nearest rounding.)
+Relative error: 1.957 * 2^-26 (before rounding.)
+*/
+
+#define T __logf_data.tab
+#define A __logf_data.poly
+#define Ln2 __logf_data.ln2
+#define N (1 << LOGF_TABLE_BITS)
+#define OFF 0x3f330000
+
+float
+logf (float x)
+{
+  /* double_t for better performance on targets with FLT_EVAL_METHOD==2.  */
+  double_t z, r, r2, y, y0, invc, logc;
+  uint32_t ix, iz, tmp;
+  int k, i;
+
+  ix = asuint (x);
+#if WANT_ROUNDING
+  /* Fix sign of zero with downward rounding when x==1.  */
+  if (__builtin_expect (ix == 0x3f800000, 0))
+    return 0;
+#endif
+  if (__builtin_expect (ix - 0x00800000 >= 0x7f800000 - 0x00800000, 0))
+    {
+      /* x < 0x1p-126 or inf or nan.  */
+      if (ix * 2 == 0)
+	return __math_divzerof (1);
+      if (ix == 0x7f800000) /* log(inf) == inf.  */
+	return x;
+      if ((ix & 0x80000000) || ix * 2 >= 0xff000000)
+	return __math_invalidf (x);
+      /* x is subnormal, normalize it.  */
+      ix = asuint (x * 0x1p23f);
+      ix -= 23 << 23;
+    }
+
+  /* x = 2^k z; where z is in range [OFF,2*OFF] and exact.
+     The range is split into N subintervals.
+     The ith subinterval contains z and c is near its center.  */
+  tmp = ix - OFF;
+  i = (tmp >> (23 - LOGF_TABLE_BITS)) % N;
+  k = (int32_t) tmp >> 23; /* arithmetic shift */
+  iz = ix - (tmp & 0x1ff << 23);
+  invc = T[i].invc;
+  logc = T[i].logc;
+  z = (double_t) asfloat (iz);
+
+  /* log(x) = log1p(z/c-1) + log(c) + k*Ln2 */
+  r = z * invc - 1;
+  y0 = logc + (double_t) k * Ln2;
+
+  /* Pipelined polynomial evaluation to approximate log1p(r).  */
+  r2 = r * r;
+  y = A[1] * r + A[2];
+  y = A[0] * r2 + y;
+  y = y * r2 + (y0 + r);
+  return (float) y;
+}
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_log2.c b/newlib/libm/common/sf_log2.c
index 9fbaaca72..f52cd385d 100644
--- a/newlib/libm/common/sf_log2.c
+++ b/newlib/libm/common/sf_log2.c
@@ -1,48 +1,99 @@
-/* sf_log2.c -- float version of s_log2.c.
- * Modification of sf_exp10.c by Yaakov Selkowitz 2009.
- */
+/* Single-precision log2 function.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
 
-/*
- * ====================================================
- * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
- *
- * Developed at SunPro, a Sun Microsystems, Inc. business.
- * Permission to use, copy, modify, and distribute this
- * software is freely granted, provided that this notice 
- * is preserved.
- * ====================================================
- */
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
 
-/* 
- * wrapper log2f(x)
- */
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
 
 #include "fdlibm.h"
-#include <errno.h>
+#if !__OBSOLETE_MATH
+
 #include <math.h>
-#undef log2
-#undef log2f
+#include <stdint.h>
+#include "math_config.h"
 
-#ifdef __STDC__
-	float log2f(float x)		/* wrapper log2f */
-#else
-	float log2f(x)			/* wrapper log2f */
-	float x;
-#endif
+/*
+LOG2F_TABLE_BITS = 4
+LOG2F_POLY_ORDER = 4
+
+ULP error: 0.752 (nearest rounding.)
+Relative error: 1.9 * 2^-26 (before rounding.)
+*/
+
+#define N (1 << LOG2F_TABLE_BITS)
+#define T __log2f_data.tab
+#define A __log2f_data.poly
+#define OFF 0x3f330000
+
+float
+log2f (float x)
 {
-  return (logf(x) / (float_t) M_LN2);
-}
+  /* double_t for better performance on targets with FLT_EVAL_METHOD==2.  */
+  double_t z, r, r2, p, y, y0, invc, logc;
+  uint32_t ix, iz, top, tmp;
+  int k, i;
 
-#ifdef _DOUBLE_IS_32BITS
-
-#ifdef __STDC__
-	double log2(double x)
-#else
-	double log2(x)
-	double x;
+  ix = asuint (x);
+#if WANT_ROUNDING
+  /* Fix sign of zero with downward rounding when x==1.  */
+  if (__builtin_expect (ix == 0x3f800000, 0))
+    return 0;
 #endif
-{
-	return (double) log2f((float) x);
-}
+  if (__builtin_expect (ix - 0x00800000 >= 0x7f800000 - 0x00800000, 0))
+    {
+      /* x < 0x1p-126 or inf or nan.  */
+      if (ix * 2 == 0)
+	return __math_divzerof (1);
+      if (ix == 0x7f800000) /* log2(inf) == inf.  */
+	return x;
+      if ((ix & 0x80000000) || ix * 2 >= 0xff000000)
+	return __math_invalidf (x);
+      /* x is subnormal, normalize it.  */
+      ix = asuint (x * 0x1p23f);
+      ix -= 23 << 23;
+    }
 
-#endif /* defined(_DOUBLE_IS_32BITS) */
+  /* x = 2^k z; where z is in range [OFF,2*OFF] and exact.
+     The range is split into N subintervals.
+     The ith subinterval contains z and c is near its center.  */
+  tmp = ix - OFF;
+  i = (tmp >> (23 - LOG2F_TABLE_BITS)) % N;
+  top = tmp & 0xff800000;
+  iz = ix - top;
+  k = (int32_t) tmp >> 23; /* arithmetic shift */
+  invc = T[i].invc;
+  logc = T[i].logc;
+  z = (double_t) asfloat (iz);
+
+  /* log2(x) = log1p(z/c-1)/ln2 + log2(c) + k */
+  r = z * invc - 1;
+  y0 = logc + (double_t) k;
+
+  /* Pipelined polynomial evaluation to approximate log1p(r)/ln2.  */
+  r2 = r * r;
+  y = A[1] * r + A[2];
+  y = A[0] * r2 + y;
+  p = A[3] * r + y0;
+  y = y * r2 + p;
+  return (float) y;
+}
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_log2_data.c b/newlib/libm/common/sf_log2_data.c
new file mode 100644
index 000000000..f0e80947c
--- /dev/null
+++ b/newlib/libm/common/sf_log2_data.c
@@ -0,0 +1,56 @@
+/* Data definition for log2f.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include "math_config.h"
+
+const struct log2f_data __log2f_data = {
+  .tab = {
+  { 0x1.661ec79f8f3bep+0, -0x1.efec65b963019p-2 },
+  { 0x1.571ed4aaf883dp+0, -0x1.b0b6832d4fca4p-2 },
+  { 0x1.49539f0f010bp+0, -0x1.7418b0a1fb77bp-2 },
+  { 0x1.3c995b0b80385p+0, -0x1.39de91a6dcf7bp-2 },
+  { 0x1.30d190c8864a5p+0, -0x1.01d9bf3f2b631p-2 },
+  { 0x1.25e227b0b8eap+0, -0x1.97c1d1b3b7afp-3 },
+  { 0x1.1bb4a4a1a343fp+0, -0x1.2f9e393af3c9fp-3 },
+  { 0x1.12358f08ae5bap+0, -0x1.960cbbf788d5cp-4 },
+  { 0x1.0953f419900a7p+0, -0x1.a6f9db6475fcep-5 },
+  { 0x1p+0, 0x0p+0 },
+  { 0x1.e608cfd9a47acp-1, 0x1.338ca9f24f53dp-4 },
+  { 0x1.ca4b31f026aap-1, 0x1.476a9543891bap-3 },
+  { 0x1.b2036576afce6p-1, 0x1.e840b4ac4e4d2p-3 },
+  { 0x1.9c2d163a1aa2dp-1, 0x1.40645f0c6651cp-2 },
+  { 0x1.886e6037841edp-1, 0x1.88e9c2c1b9ff8p-2 },
+  { 0x1.767dcf5534862p-1, 0x1.ce0a44eb17bccp-2 },
+  },
+  .poly = {
+  -0x1.712b6f70a7e4dp-2, 0x1.ecabf496832ep-2, -0x1.715479ffae3dep-1,
+  0x1.715475f35c8b8p0,
+  }
+};
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_log_data.c b/newlib/libm/common/sf_log_data.c
new file mode 100644
index 000000000..28231b4dc
--- /dev/null
+++ b/newlib/libm/common/sf_log_data.c
@@ -0,0 +1,56 @@
+/* Data definition for logf.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include "math_config.h"
+
+const struct logf_data __logf_data = {
+  .tab = {
+  { 0x1.661ec79f8f3bep+0, -0x1.57bf7808caadep-2 },
+  { 0x1.571ed4aaf883dp+0, -0x1.2bef0a7c06ddbp-2 },
+  { 0x1.49539f0f010bp+0, -0x1.01eae7f513a67p-2 },
+  { 0x1.3c995b0b80385p+0, -0x1.b31d8a68224e9p-3 },
+  { 0x1.30d190c8864a5p+0, -0x1.6574f0ac07758p-3 },
+  { 0x1.25e227b0b8eap+0, -0x1.1aa2bc79c81p-3 },
+  { 0x1.1bb4a4a1a343fp+0, -0x1.a4e76ce8c0e5ep-4 },
+  { 0x1.12358f08ae5bap+0, -0x1.1973c5a611cccp-4 },
+  { 0x1.0953f419900a7p+0, -0x1.252f438e10c1ep-5 },
+  { 0x1p+0, 0x0p+0 },
+  { 0x1.e608cfd9a47acp-1, 0x1.aa5aa5df25984p-5 },
+  { 0x1.ca4b31f026aap-1, 0x1.c5e53aa362eb4p-4 },
+  { 0x1.b2036576afce6p-1, 0x1.526e57720db08p-3 },
+  { 0x1.9c2d163a1aa2dp-1, 0x1.bc2860d22477p-3 },
+  { 0x1.886e6037841edp-1, 0x1.1058bc8a07ee1p-2 },
+  { 0x1.767dcf5534862p-1, 0x1.4043057b6ee09p-2 },
+  },
+  .ln2 = 0x1.62e42fefa39efp-1,
+  .poly = {
+  -0x1.00ea348b88334p-2, 0x1.5575b0be00b6ap-2, -0x1.ffffef20a4123p-2,
+  }
+};
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_pow.c b/newlib/libm/common/sf_pow.c
new file mode 100644
index 000000000..6741a349b
--- /dev/null
+++ b/newlib/libm/common/sf_pow.c
@@ -0,0 +1,230 @@
+/* Single-precision pow function.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include <math.h>
+#include <stdint.h>
+#include "math_config.h"
+
+/*
+POWF_LOG2_POLY_ORDER = 5
+EXP2F_TABLE_BITS = 5
+
+ULP error: 0.82 (~ 0.5 + relerr*2^24)
+relerr: 1.27 * 2^-26 (Relative error ~= 128*Ln2*relerr_log2 + relerr_exp2)
+relerr_log2: 1.83 * 2^-33 (Relative error of logx.)
+relerr_exp2: 1.69 * 2^-34 (Relative error of exp2(ylogx).)
+*/
+
+#define N (1 << POWF_LOG2_TABLE_BITS)
+#define T __powf_log2_data.tab
+#define A __powf_log2_data.poly
+#define OFF 0x3f330000
+
+/* Subnormal input is normalized so ix has negative biased exponent.
+   Output is multiplied by N (POWF_SCALE) if TOINT_INTRINICS is set.  */
+static inline double_t
+log2_inline (uint32_t ix)
+{
+  /* double_t for better performance on targets with FLT_EVAL_METHOD==2.  */
+  double_t z, r, r2, r4, p, q, y, y0, invc, logc;
+  uint32_t iz, top, tmp;
+  int k, i;
+
+  /* x = 2^k z; where z is in range [OFF,2*OFF] and exact.
+     The range is split into N subintervals.
+     The ith subinterval contains z and c is near its center.  */
+  tmp = ix - OFF;
+  i = (tmp >> (23 - POWF_LOG2_TABLE_BITS)) % N;
+  top = tmp & 0xff800000;
+  iz = ix - top;
+  k = (int32_t) top >> (23 - POWF_SCALE_BITS); /* arithmetic shift */
+  invc = T[i].invc;
+  logc = T[i].logc;
+  z = (double_t) asfloat (iz);
+
+  /* log2(x) = log1p(z/c-1)/ln2 + log2(c) + k */
+  r = z * invc - 1;
+  y0 = logc + (double_t) k;
+
+  /* Pipelined polynomial evaluation to approximate log1p(r)/ln2.  */
+  r2 = r * r;
+  y = A[0] * r + A[1];
+  p = A[2] * r + A[3];
+  r4 = r2 * r2;
+  q = A[4] * r + y0;
+  q = p * r2 + q;
+  y = y * r4 + q;
+  return y;
+}
+
+#undef N
+#undef T
+#define N (1 << EXP2F_TABLE_BITS)
+#define T __exp2f_data.tab
+#define SIGN_BIAS (1 << (EXP2F_TABLE_BITS + 11))
+
+/* The output of log2 and thus the input of exp2 is either scaled by N
+   (in case of fast toint intrinsics) or not.  The unscaled xd must be
+   in [-1021,1023], sign_bias sets the sign of the result.  */
+static inline double_t
+exp2_inline (double_t xd, unsigned long sign_bias)
+{
+  uint64_t ki, ski, t;
+  /* double_t for better performance on targets with FLT_EVAL_METHOD==2.  */
+  double_t kd, z, r, r2, y, s;
+
+#if TOINT_INTRINSICS
+# define C __exp2f_data.poly_scaled
+  /* N*x = k + r with r in [-1/2, 1/2] */
+  kd = roundtoint (xd); /* k */
+  ki = converttoint (xd);
+#else
+# define C __exp2f_data.poly
+# define SHIFT __exp2f_data.shift_scaled
+  /* x = k/N + r with r in [-1/(2N), 1/(2N)] */
+  kd = (double) (xd + SHIFT); /* Rounding to double precision is required.  */
+  ki = asuint64 (kd);
+  kd -= SHIFT; /* k/N */
+#endif
+  r = xd - kd;
+
+  /* exp2(x) = 2^(k/N) * 2^r ~= s * (C0*r^3 + C1*r^2 + C2*r + 1) */
+  t = T[ki % N];
+  ski = ki + sign_bias;
+  t += ski << (52 - EXP2F_TABLE_BITS);
+  s = asdouble (t);
+  z = C[0] * r + C[1];
+  r2 = r * r;
+  y = C[2] * r + 1;
+  y = z * r2 + y;
+  y = y * s;
+  return y;
+}
+
+/* Returns 0 if not int, 1 if odd int, 2 if even int.  */
+static inline int
+checkint (uint32_t iy)
+{
+  int e = iy >> 23 & 0xff;
+  if (e < 0x7f)
+    return 0;
+  if (e > 0x7f + 23)
+    return 2;
+  if (iy & ((1 << (0x7f + 23 - e)) - 1))
+    return 0;
+  if (iy & (1 << (0x7f + 23 - e)))
+    return 1;
+  return 2;
+}
+
+static inline int
+zeroinfnan (uint32_t ix)
+{
+  return 2 * ix - 1 >= 2u * 0x7f800000 - 1;
+}
+
+float
+powf (float x, float y)
+{
+  unsigned long sign_bias = 0;
+  uint32_t ix, iy;
+
+  ix = asuint (x);
+  iy = asuint (y);
+  if (__builtin_expect (ix - 0x00800000 >= 0x7f800000 - 0x00800000
+			  || zeroinfnan (iy),
+			0))
+    {
+      /* Either (x < 0x1p-126 or inf or nan) or (y is 0 or inf or nan).  */
+      if (__builtin_expect (zeroinfnan (iy), 0))
+	{
+	  if (2 * iy == 0)
+	    return issignalingf_inline (x) ? x + y : 1.0f;
+	  if (ix == 0x3f800000)
+	    return issignalingf_inline (y) ? x + y : 1.0f;
+	  if (2 * ix > 2u * 0x7f800000 || 2 * iy > 2u * 0x7f800000)
+	    return x + y;
+	  if (2 * ix == 2 * 0x3f800000)
+	    return 1.0f;
+	  if ((2 * ix < 2 * 0x3f800000) == !(iy & 0x80000000))
+	    return 0.0f; /* |x|<1 && y==inf or |x|>1 && y==-inf.  */
+	  return y * y;
+	}
+      if (__builtin_expect (zeroinfnan (ix), 0))
+	{
+	  float_t x2 = x * x;
+	  if (ix & 0x80000000 && checkint (iy) == 1)
+	    {
+	      x2 = -x2;
+	      sign_bias = 1;
+	    }
+#if WANT_ERRNO
+	  if (2 * ix == 0 && iy & 0x80000000)
+	    return __math_divzerof (sign_bias);
+#endif
+	  return iy & 0x80000000 ? 1 / x2 : x2;
+	}
+      /* x and y are non-zero finite.  */
+      if (ix & 0x80000000)
+	{
+	  /* Finite x < 0.  */
+	  int yint = checkint (iy);
+	  if (yint == 0)
+	    return __math_invalidf (x);
+	  if (yint == 1)
+	    sign_bias = SIGN_BIAS;
+	  ix &= 0x7fffffff;
+	}
+      if (ix < 0x00800000)
+	{
+	  /* Normalize subnormal x so exponent becomes negative.  */
+	  ix = asuint (x * 0x1p23f);
+	  ix &= 0x7fffffff;
+	  ix -= 23 << 23;
+	}
+    }
+  double_t logx = log2_inline (ix);
+  double_t ylogx = y * logx; /* Note: cannot overflow, y is single prec.  */
+  if (__builtin_expect ((asuint64 (ylogx) >> 47 & 0xffff)
+			  >= asuint64 (126.0 * POWF_SCALE) >> 47,
+			0))
+    {
+      /* |y*log(x)| >= 126.  */
+      if (ylogx > 0x1.fffffffd1d571p+6 * POWF_SCALE)
+	return __math_oflowf (sign_bias);
+      if (ylogx <= -150.0 * POWF_SCALE)
+	return __math_uflowf (sign_bias);
+#if WANT_ERRNO_UFLOW
+      if (ylogx < -149.0 * POWF_SCALE)
+	return __math_may_uflowf (sign_bias);
+#endif
+    }
+  return (float) exp2_inline (ylogx, sign_bias);
+}
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/common/sf_pow_log2_data.c b/newlib/libm/common/sf_pow_log2_data.c
new file mode 100644
index 000000000..24f638c19
--- /dev/null
+++ b/newlib/libm/common/sf_pow_log2_data.c
@@ -0,0 +1,57 @@
+/* Data definition for powf.
+   Copyright (c) 2017 ARM Ltd.  All rights reserved.
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+   1. Redistributions of source code must retain the above copyright
+      notice, this list of conditions and the following disclaimer.
+   2. Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+   3. The name of the company may not be used to endorse or promote
+      products derived from this software without specific prior written
+      permission.
+
+   THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS AND ANY EXPRESS OR IMPLIED
+   WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+   MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+   IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
+   TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
+
+#include "fdlibm.h"
+#if !__OBSOLETE_MATH
+
+#include "math_config.h"
+
+const struct powf_log2_data __powf_log2_data = {
+  .tab = {
+  { 0x1.661ec79f8f3bep+0, -0x1.efec65b963019p-2 * POWF_SCALE },
+  { 0x1.571ed4aaf883dp+0, -0x1.b0b6832d4fca4p-2 * POWF_SCALE },
+  { 0x1.49539f0f010bp+0, -0x1.7418b0a1fb77bp-2 * POWF_SCALE },
+  { 0x1.3c995b0b80385p+0, -0x1.39de91a6dcf7bp-2 * POWF_SCALE },
+  { 0x1.30d190c8864a5p+0, -0x1.01d9bf3f2b631p-2 * POWF_SCALE },
+  { 0x1.25e227b0b8eap+0, -0x1.97c1d1b3b7afp-3 * POWF_SCALE },
+  { 0x1.1bb4a4a1a343fp+0, -0x1.2f9e393af3c9fp-3 * POWF_SCALE },
+  { 0x1.12358f08ae5bap+0, -0x1.960cbbf788d5cp-4 * POWF_SCALE },
+  { 0x1.0953f419900a7p+0, -0x1.a6f9db6475fcep-5 * POWF_SCALE },
+  { 0x1p+0, 0x0p+0 * POWF_SCALE },
+  { 0x1.e608cfd9a47acp-1, 0x1.338ca9f24f53dp-4 * POWF_SCALE },
+  { 0x1.ca4b31f026aap-1, 0x1.476a9543891bap-3 * POWF_SCALE },
+  { 0x1.b2036576afce6p-1, 0x1.e840b4ac4e4d2p-3 * POWF_SCALE },
+  { 0x1.9c2d163a1aa2dp-1, 0x1.40645f0c6651cp-2 * POWF_SCALE },
+  { 0x1.886e6037841edp-1, 0x1.88e9c2c1b9ff8p-2 * POWF_SCALE },
+  { 0x1.767dcf5534862p-1, 0x1.ce0a44eb17bccp-2 * POWF_SCALE },
+  },
+  .poly = {
+  0x1.27616c9496e0bp-2 * POWF_SCALE, -0x1.71969a075c67ap-2 * POWF_SCALE,
+  0x1.ec70a6ca7baddp-2 * POWF_SCALE, -0x1.7154748bef6c8p-1 * POWF_SCALE,
+  0x1.71547652ab82bp0 * POWF_SCALE,
+  }
+};
+#endif /* !__OBSOLETE_MATH */
diff --git a/newlib/libm/math/Makefile.am b/newlib/libm/math/Makefile.am
index 9e8e92840..e745159ae 100644
--- a/newlib/libm/math/Makefile.am
+++ b/newlib/libm/math/Makefile.am
@@ -48,7 +48,8 @@ fsrc =	kf_rem_pio2.c \
 	sf_frexp.c sf_ldexp.c \
 	sf_signif.c sf_sin.c \
 	sf_tan.c sf_tanh.c \
-	wf_exp2.c wf_tgamma.c
+	wf_exp2.c wf_tgamma.c \
+	wf_log2.c
 
 lsrc = el_hypot.c
 
diff --git a/newlib/libm/math/Makefile.in b/newlib/libm/math/Makefile.in
index 8f76aad4e..1f5a69406 100644
--- a/newlib/libm/math/Makefile.in
+++ b/newlib/libm/math/Makefile.in
@@ -137,7 +137,8 @@ am__objects_2 = lib_a-kf_rem_pio2.$(OBJEXT) lib_a-kf_cos.$(OBJEXT) \
 	lib_a-sf_frexp.$(OBJEXT) lib_a-sf_ldexp.$(OBJEXT) \
 	lib_a-sf_signif.$(OBJEXT) lib_a-sf_sin.$(OBJEXT) \
 	lib_a-sf_tan.$(OBJEXT) lib_a-sf_tanh.$(OBJEXT) \
-	lib_a-wf_exp2.$(OBJEXT) lib_a-wf_tgamma.$(OBJEXT)
+	lib_a-wf_exp2.$(OBJEXT) lib_a-wf_tgamma.$(OBJEXT) \
+	lib_a-wf_log2.$(OBJEXT)
 am__objects_3 = lib_a-el_hypot.$(OBJEXT)
 @USE_LIBTOOL_FALSE@am_lib_a_OBJECTS = $(am__objects_1) \
 @USE_LIBTOOL_FALSE@	$(am__objects_2) $(am__objects_3)
@@ -169,7 +170,7 @@ am__objects_5 = kf_rem_pio2.lo kf_cos.lo kf_sin.lo kf_tan.lo \
 	wf_sincos.lo wf_drem.lo sf_asinh.lo sf_atan.lo sf_ceil.lo \
 	sf_cos.lo sf_erf.lo sf_fabs.lo sf_floor.lo sf_frexp.lo \
 	sf_ldexp.lo sf_signif.lo sf_sin.lo sf_tan.lo sf_tanh.lo \
-	wf_exp2.lo wf_tgamma.lo
+	wf_exp2.lo wf_tgamma.lo wf_log2.lo
 am__objects_6 = el_hypot.lo
 @USE_LIBTOOL_TRUE@am_libmath_la_OBJECTS = $(am__objects_4) \
 @USE_LIBTOOL_TRUE@	$(am__objects_5) $(am__objects_6)
@@ -372,7 +373,8 @@ fsrc = kf_rem_pio2.c \
 	sf_frexp.c sf_ldexp.c \
 	sf_signif.c sf_sin.c \
 	sf_tan.c sf_tanh.c \
-	wf_exp2.c wf_tgamma.c
+	wf_exp2.c wf_tgamma.c \
+	wf_log2.c
 
 lsrc = el_hypot.c
 libmath_la_LDFLAGS = -Xcompiler -nostdlib
@@ -1270,6 +1272,12 @@ lib_a-wf_tgamma.o: wf_tgamma.c
 lib_a-wf_tgamma.obj: wf_tgamma.c
 	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-wf_tgamma.obj `if test -f 'wf_tgamma.c'; then $(CYGPATH_W) 'wf_tgamma.c'; else $(CYGPATH_W) '$(srcdir)/wf_tgamma.c'; fi`
 
+lib_a-wf_log2.o: wf_log2.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-wf_log2.o `test -f 'wf_log2.c' || echo '$(srcdir)/'`wf_log2.c
+
+lib_a-wf_log2.obj: wf_log2.c
+	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-wf_log2.obj `if test -f 'wf_log2.c'; then $(CYGPATH_W) 'wf_log2.c'; else $(CYGPATH_W) '$(srcdir)/wf_log2.c'; fi`
+
 lib_a-el_hypot.o: el_hypot.c
 	$(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(lib_a_CFLAGS) $(CFLAGS) -c -o lib_a-el_hypot.o `test -f 'el_hypot.c' || echo '$(srcdir)/'`el_hypot.c
 
diff --git a/newlib/libm/math/ef_exp.c b/newlib/libm/math/ef_exp.c
index 6824f991a..e817370ac 100644
--- a/newlib/libm/math/ef_exp.c
+++ b/newlib/libm/math/ef_exp.c
@@ -15,6 +15,7 @@
 
 #include "fdlibm.h"
 
+#if __OBSOLETE_MATH
 #ifdef __v810__
 #define const
 #endif
@@ -97,3 +98,4 @@ P5   =  4.1381369442e-08; /* 0x3331bb4c */
 	    return y*twom100;
 	}
 }
+#endif /* __OBSOLETE_MATH */
diff --git a/newlib/libm/math/ef_log.c b/newlib/libm/math/ef_log.c
index 619fe9090..8dc902478 100644
--- a/newlib/libm/math/ef_log.c
+++ b/newlib/libm/math/ef_log.c
@@ -15,6 +15,7 @@
 
 #include "fdlibm.h"
 
+#if __OBSOLETE_MATH
 #ifdef __STDC__
 static const float
 #else
@@ -90,3 +91,4 @@ static float zero   =  0.0;
 		     return dk*ln2_hi-((s*(f-R)-dk*ln2_lo)-f);
 	}
 }
+#endif /* __OBSOLETE_MATH */
diff --git a/newlib/libm/math/ef_pow.c b/newlib/libm/math/ef_pow.c
index b3041dbdc..524e3f9b0 100644
--- a/newlib/libm/math/ef_pow.c
+++ b/newlib/libm/math/ef_pow.c
@@ -15,6 +15,7 @@
 
 #include "fdlibm.h"
 
+#if __OBSOLETE_MATH
 #ifdef __v810__
 #define const 
 #endif
@@ -253,3 +254,4 @@ ivln2_l  =  7.0526075433e-06; /* 0x36eca570 =1/ln2 tail*/
 	else SET_FLOAT_WORD(z,j);
 	return s*z;
 }
+#endif /* __OBSOLETE_MATH */
diff --git a/newlib/libm/math/wf_exp.c b/newlib/libm/math/wf_exp.c
index 70f4459b4..f16af1d1b 100644
--- a/newlib/libm/math/wf_exp.c
+++ b/newlib/libm/math/wf_exp.c
@@ -18,6 +18,7 @@
  */
 
 #include "fdlibm.h"
+#if __OBSOLETE_MATH
 #include <errno.h>
 
 #ifdef __STDC__
@@ -101,3 +102,4 @@ u_threshold= -1.0397208405e+02;  /* 0xc2cff1b5 */
 }
 
 #endif /* defined(_DOUBLE_IS_32BITS) */
+#endif /* __OBSOLETE_MATH */
diff --git a/newlib/libm/math/wf_exp2.c b/newlib/libm/math/wf_exp2.c
index 944031405..5dce9c832 100644
--- a/newlib/libm/math/wf_exp2.c
+++ b/newlib/libm/math/wf_exp2.c
@@ -18,6 +18,7 @@
  */
 
 #include "fdlibm.h"
+#if __OBSOLETE_MATH
 #include <errno.h>
 #include <math.h>
 
@@ -44,3 +45,4 @@
 }
 
 #endif /* defined(_DOUBLE_IS_32BITS) */
+#endif /* __OBSOLETE_MATH */
diff --git a/newlib/libm/math/wf_log.c b/newlib/libm/math/wf_log.c
index 4518b863b..07be8d63c 100644
--- a/newlib/libm/math/wf_log.c
+++ b/newlib/libm/math/wf_log.c
@@ -18,6 +18,7 @@
  */
 
 #include "fdlibm.h"
+#if __OBSOLETE_MATH
 #include <errno.h>
 
 #ifdef __STDC__
@@ -84,3 +85,4 @@
 }
 
 #endif /* defined(_DOUBLE_IS_32BITS) */
+#endif /* __OBSOLETE_MATH */
diff --git a/newlib/libm/math/wf_log2.c b/newlib/libm/math/wf_log2.c
new file mode 100644
index 000000000..e32153747
--- /dev/null
+++ b/newlib/libm/math/wf_log2.c
@@ -0,0 +1,50 @@
+/* wf_log2.c -- float version of s_log2.c.
+ * Modification of sf_exp10.c by Yaakov Selkowitz 2009.
+ */
+
+/*
+ * ====================================================
+ * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
+ *
+ * Developed at SunPro, a Sun Microsystems, Inc. business.
+ * Permission to use, copy, modify, and distribute this
+ * software is freely granted, provided that this notice
+ * is preserved.
+ * ====================================================
+ */
+
+/*
+ * wrapper log2f(x)
+ */
+
+#include "fdlibm.h"
+#if __OBSOLETE_MATH
+#include <errno.h>
+#include <math.h>
+#undef log2
+#undef log2f
+
+#ifdef __STDC__
+	float log2f(float x)		/* wrapper log2f */
+#else
+	float log2f(x)			/* wrapper log2f */
+	float x;
+#endif
+{
+  return (logf(x) / (float_t) M_LN2);
+}
+
+#ifdef _DOUBLE_IS_32BITS
+
+#ifdef __STDC__
+	double log2(double x)
+#else
+	double log2(x)
+	double x;
+#endif
+{
+	return (double) log2f((float) x);
+}
+
+#endif /* defined(_DOUBLE_IS_32BITS) */
+#endif /* __OBSOLETE_MATH */
diff --git a/newlib/libm/math/wf_pow.c b/newlib/libm/math/wf_pow.c
index a30f8808e..be453558b 100644
--- a/newlib/libm/math/wf_pow.c
+++ b/newlib/libm/math/wf_pow.c
@@ -18,6 +18,7 @@
  */
 
 #include "fdlibm.h"
+#if __OBSOLETE_MATH
 #include <errno.h>
 
 #ifdef __STDC__
@@ -177,3 +178,4 @@
 }
 
 #endif /* defined(_DOUBLE_IS_32BITS) */
+#endif /* __OBSOLETE_MATH */