libm: arm: Add arm specific floor() optimization

Add arm specific floor() implementation which avoids VMSR and VMRS
instructions.

Change-Id: Ibd4cd7147aa2f98c9b5bbaf74948843ea619dba4

2 files changed, 164 insertions, 1 deletions

diff --git a/libm/Android.mk b/libm/Android.mk
index dc6c704..ebc3c9f 100644
--- a/libm/Android.mk
+++ b/libm/Android.mk
@@ -130,7 +130,6 @@ LOCAL_SRC_FILES := \
     upstream-freebsd/lib/msun/src/s_fdim.c \
     upstream-freebsd/lib/msun/src/s_finite.c \
     upstream-freebsd/lib/msun/src/s_finitef.c \
-    upstream-freebsd/lib/msun/src/s_floor.c \
     upstream-freebsd/lib/msun/src/s_floorf.c \
     upstream-freebsd/lib/msun/src/s_fma.c \
     upstream-freebsd/lib/msun/src/s_fmaf.c \
@@ -264,20 +263,39 @@ LOCAL_SRC_FILES += \
 LOCAL_SRC_FILES_arm += \
     arm/fenv.c \
 
+# s_floor.S requires neon instructions.
+ifdef TARGET_2ND_ARCH
+arch_variant := $(TARGET_2ND_ARCH_VARIANT)
+else
+arch_variant := $(TARGET_ARCH_VARIANT)
+endif
+
+# Use the C version on armv7-a since it doesn't support neon instructions.
+ifeq ($(arch_variant),armv7-a)
+LOCAL_SRC_FILES_arm += upstream-freebsd/lib/msun/src/s_floor.c
+else
+LOCAL_SRC_FILES_arm += arm/s_floor.S
+endif
+
 LOCAL_SRC_FILES_arm64 += \
     arm64/fenv.c \
+    upstream-freebsd/lib/msun/src/s_floor.c \
 
 LOCAL_SRC_FILES_mips += \
     mips/fenv.c \
+    upstream-freebsd/lib/msun/src/s_floor.c \
 
 LOCAL_SRC_FILES_mips64 += \
     mips/fenv.c \
+    upstream-freebsd/lib/msun/src/s_floor.c \
 
 LOCAL_SRC_FILES_x86 += \
     i387/fenv.c \
+    upstream-freebsd/lib/msun/src/s_floor.c \
 
 LOCAL_SRC_FILES_x86_64 += \
     amd64/fenv.c \
+    upstream-freebsd/lib/msun/src/s_floor.c \
 
 LOCAL_C_INCLUDES_x86 += $(LOCAL_PATH)/i387
 
@@ -297,6 +315,9 @@ LOCAL_CFLAGS := \
     -Wno-unknown-pragmas \
     -fvisibility=hidden \
 
+LOCAL_ASFLAGS := \
+    -Ibionic/libc \
+
 # Workaround the GCC "(long)fn -> lfn" optimization bug which will result in
 # self recursions for lrint, lrintf, and lrintl.
 # BUG: 14225968
diff --git a/libm/arm/s_floor.S b/libm/arm/s_floor.S
new file mode 100644
index 0000000..4405358
--- /dev/null
+++ b/libm/arm/s_floor.S
@@ -0,0 +1,142 @@
+/*
+ * Copyright (c) 2013-2014, NVIDIA Corporation.  All rights reserved.
+ * Johnny Qiu <joqiu@nvidia.com>
+ * Shu Zhang <chazhang@nvidia.com>
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *     * 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.
+ *     * Neither the name of The Linux Foundation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED "AS IS" AND ANY EXPRESS OR IMPLIED
+ * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS
+ * 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 <float.h>
+#include <private/bionic_asm.h>
+
+ENTRY(floor)    /* x in r0, r1 */
+
+        and             r3, r1, #0x80000000     /* sign(x) */
+        bic             r1, r1, #0x80000000     /* x = abs(x) */
+
+        /* extract exp of x */
+        lsr             r2, r1, #20
+        sub             r2, r2, #0x3fc
+        subs            r2, r2, #0x3            /* r2 <- exp */
+
+        /* |x| < 1.0? */
+        blt             .Lx_lt_one
+
+        /* x < 0? */
+        cmp             r3, #0
+        bne             .Lclr_frac_neg
+
+        /* |x| <= 2^20? */
+        cmp             r2, #20
+        ble             .Lclr_frac_r1
+
+        /* |x| < 2^52? */
+        cmp             r2, #52
+        blt             .Lclr_frac_r0
+
+        /* return x */
+        bx              lr
+
+.Lclr_frac_r1:
+        rsb             r2, r2, #20
+        lsr             r1, r1, r2
+        lsl             r1, r1, r2
+        mov             r0, #0
+        bx              lr
+
+.Lclr_frac_r0:
+        rsb             r2, r2, #52
+        lsr             r0, r0, r2
+        lsl             r0, r0, r2
+        bx              lr
+
+.Lclr_frac_neg:
+        /* |x| <= 2^20? */
+        cmp             r2, #20
+        ble             .Lclr_frac_r1_neg
+
+        /* |x| < 2^52? */
+        cmp             r2, #52
+        blt             .Lclr_frac_r0_neg
+
+        /* return x */
+        orr             r1, r1, #0x80000000
+        bx              lr
+
+.Lclr_frac_r1_neg:
+        rsb             r2, r2, #20
+        mov             r3, #1
+        lsl             r3, r3, r2
+        sub             r3, r3, #1
+        and             r3, r1, r3
+        orr             r3, r3, r0
+        lsr             r1, r1, r2
+        lsl             r1, r1, r2
+        mov             r0, #0
+        b               .Lreturn_x_neg
+
+.Lclr_frac_r0_neg:
+        rsb             r2, r2, #52
+        mov             r3, #1
+        lsl             r3, r3, r2
+        sub             r3, r3, #1
+        and             r3, r0, r3
+        lsr             r0, r0, r2
+        lsl             r0, r0, r2
+        b               .Lreturn_x_neg
+
+.Lx_lt_one:
+        /* x == +-0? */
+        cmp             r0, #0
+        cmpeq           r1, #0
+        orreq           r1, r1, r3
+        bxeq            lr
+
+        /* (x > 0) ? 0 : -1 */
+        mov             r1, #0x00100000
+        mov             r0, #0
+        cmp             r3, #0
+        movne           r1, #0xc0000000
+        sub             r1, r1, #0x00100000
+        bx              lr
+
+.Lreturn_x_neg:
+        cmp             r3, #0
+        orr             r1, r1, #0x80000000
+        bxeq            lr
+
+        vmov            d16, r0, r1
+        vmov.f64        d18, #1.0
+        vsub.f64        d16, d16, d18
+        vmov            r0, r1, d16
+        bx              lr
+
+END(floor)
+
+#if LDBL_MANT_DIG == 53
+        .weak           floorl
+        .equ            floorl,floor
+#endif