Get --ftrace working on aarch64

This change implements a new approach to function call logging, that's
based on the GCC flag: -fpatchable-function-entry. Read the commentary
in build/config.mk to learn how it works.

dsp/bmp/bmp.mk

@@ -41,11 +41,11 @@ $(DSP_BMP_A).pkg:				\
 
 o/$(MODE)/dsp/bmp/float2short.o			\
 o/$(MODE)/dsp/bmp/scalevolume.o: private	\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			$(MATHEMATICAL)
 
 o/tiny/dsp/bmp/scalevolume.o: private		\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			-Os
 
 DSP_BMP_LIBS = $(foreach x,$(DSP_BMP_ARTIFACTS),$($(x)))

dsp/core/core.mk

@@ -40,15 +40,15 @@ o/$(MODE)/dsp/core/magikarp.o			\
 o/$(MODE)/dsp/core/c93654369.o			\
 o/$(MODE)/dsp/core/float2short.o		\
 o/$(MODE)/dsp/core/scalevolume.o: private	\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			$(MATHEMATICAL)
 
 o/tiny/dsp/core/scalevolume.o: private		\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			-Os
 
 o/$(MODE)/dsp/core/det3.o: private		\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			-ffast-math
 
 DSP_CORE_LIBS = $(foreach x,$(DSP_CORE_ARTIFACTS),$($(x)))

dsp/mpeg/mpeg.mk

@@ -52,7 +52,7 @@ $(DSP_MPEG_A).pkg:				\
 		$(foreach x,$(DSP_MPEG_A_DIRECTDEPS),$($(x)_A).pkg)
 
 o/$(MODE)/dsp/mpeg/clamp4int256-k8.o: private	\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			-Os
 
 DSP_MPEG_LIBS = $(foreach x,$(DSP_MPEG_ARTIFACTS),$($(x)))

dsp/scale/scale.mk

@@ -52,7 +52,7 @@ o/$(MODE)/dsp/scale/cdecimate2xuint8x8.o	\
 o/$(MODE)/dsp/scale/gyarados.o			\
 o/$(MODE)/dsp/scale/magikarp.o			\
 o/$(MODE)/dsp/scale/scale.o: private		\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			$(MATHEMATICAL)
 
 DSP_SCALE_LIBS = $(foreach x,$(DSP_SCALE_ARTIFACTS),$($(x)))

dsp/tty/tty.mk

@@ -54,12 +54,12 @@ $(DSP_TTY_A).pkg:				\
 		$(foreach x,$(DSP_TTY_A_DIRECTDEPS),$($(x)_A).pkg)
 
 o/$(MODE)/dsp/tty/ttyraster.o: private		\
-		OVERRIDE_CFLAGS +=		\
+		CFLAGS +=			\
 			$(MATHEMATICAL)
 
 ifeq ($(ARCH), aarch64)
 # takes 14 seconds to compile with aarch64 gcc
-o/$(MODE)/dsp/tty/ttyraster.o: private OVERRIDE_CFLAGS += -O1
+o/$(MODE)/dsp/tty/ttyraster.o: private CFLAGS += -O1
 o/$(MODE)/dsp/tty/ttyraster.o: private QUOTA += -C128
 endif
 

dsp/tty/windex-avx2.S

@@ -23,10 +23,11 @@
 //	@param	rdi points to nonempty array
 //	@param	rsi is item count divisible by 16
 //	@note	needs avx2 (haswell+)
+	.ftrace1
 windex_avx2:
+	.ftrace2
 	push	%rbp
 	mov	%rsp,%rbp
-	.profilable
 	and	$-32,%rsp
 	sub	$32,%rsp
 	vmovdqa	(%rdi),%ymm1

dsp/tty/windex-sse4.S

@@ -23,10 +23,11 @@
 //	@param	rdi points to nonempty array
 //	@param	esi is 16-byte aligned 8+ / 8 multiple array item count
 //	@note	needs sse4 (nehalem+)
+	.ftrace1
 windex_sse4:
+	.ftrace2
 	push	%rbp
 	mov	%rsp,%rbp
-	.profilable
 	mov	$8,%eax
 	sub	$32,%rsp
 	movdqa	(%rdi),%xmm2