[lk][arch] Add spin_cycles

Add an architecture specific function which spins for a specific
number of CPU cycles.  Currently implemented for ARM-M only.

Change-Id: Idbf2a83186cf5ffa239d644dc732fe3d419431c1
Signed-off-by: John Grossman <johngro@google.com>

arch/arm/arm-m/spin_cycles.c

@@ -0,0 +1,62 @@
+/* vim: set expandtab ts=4 sw=4 tw=100: */
+/*
+ * Copyright (c) 2013 Google Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files
+ * (the "Software"), to deal in the Software without restriction,
+ * including without limitation the rights to use, copy, modify, merge,
+ * publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so,
+ * subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+ * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#include <compiler.h>
+#include <kernel/debug.h>
+
+__ALIGNED(8) __NAKED
+void spin_cycles(uint32_t cycles)
+{
+    asm (
+        /* 4 cycles per loop, subtract out 8 cycles for the overhead of the next
+         * 4 instructions, plus the call into and return from the function.
+         * Then, add 3 then >> 2 to round up to the number of loop iterations.
+         */
+        "subs r1, %[cycles], #5\n"
+        "asrs r1, r1, #2\n"
+        "ble .Ldone\n"
+
+        /* Padding to stay aligned on an 8 byte boundary, also has the added
+         * advantage of normalizing the overhead (1+1+2 cycles if the branch is
+         * take, or 1+1+1+1 cycles if the branch is skipped and the nop is
+         * executed)
+         */
+        "nop\n"
+
+        /* Main delay loop.
+         * sub is 1 cycle
+         * nop is 1 cycle
+         * branch is 2 cycles
+         */
+        ".Lloop:\n"
+        "subs r1, r1, #1\n"
+        "nop\n"
+        "bne .Lloop\n"
+
+        ".Ldone:\n"
+        "bx lr\n"
+        :                       /* no output */
+        : [cycles] "r" (cycles) /* input is cycles */
+        : "r1"                  /* r1 gets clobbered */
+    );
+}

arch/arm/rules.mk

@@ -182,6 +182,7 @@ MODULE_SRCS += \
 	$(LOCAL_DIR)/arm-m/arch.c \
 	$(LOCAL_DIR)/arm-m/vectab.c \
 	$(LOCAL_DIR)/arm-m/start.c \
+	$(LOCAL_DIR)/arm-m/spin_cycles.c \
 	$(LOCAL_DIR)/arm-m/exceptions.c \
 	$(LOCAL_DIR)/arm-m/thread.c
 

include/debug.h

@@ -80,6 +80,9 @@ void _panic(void *caller, const char *fmt, ...) __PRINTFLIKE(2, 3) __NO_RETURN;
 /* spin the cpu for a period of (short) time */
 void spin(uint32_t usecs);
 
+/* spin the cpu for a certain number of cpu cycles */
+void spin_cycles(uint32_t usecs);
+
 __END_CDECLS
 
 #endif