MIPS: Outline udelay and fix a few issues.

Outlining fixes the issue were on certain CPUs such as the R10000 family
the delay loop would need an extra cycle if it overlaps a cacheline
boundary.

The rewrite also fixes build errors with GCC 4.4 which was changed in
way incompatible with the kernel's inline assembly.

Relying on pure C for computation of the delay value removes the need for
explicit.  The price we pay is a slight slowdown of the computation - to
be fixed on another day.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

5 files changed, 66 insertions, 92 deletions

diff --git a/arch/mips/include/asm/cpu-info.h b/arch/mips/include/asm/cpu-info.h
index 744cd8fb107..126044308de 100644
--- a/arch/mips/include/asm/cpu-info.h
+++ b/arch/mips/include/asm/cpu-info.h
@@ -39,8 +39,8 @@ struct cache_desc {
 #define MIPS_CACHE_PINDEX	0x00000020	/* Physically indexed cache */
 
 struct cpuinfo_mips {
-	unsigned long		udelay_val;
-	unsigned long		asid_cache;
+	unsigned int		udelay_val;
+	unsigned int		asid_cache;
 
 	/*
 	 * Capability and feature descriptor structure for MIPS CPU
diff --git a/arch/mips/include/asm/delay.h b/arch/mips/include/asm/delay.h
index b0bccd2c4ed..a07e51b2be1 100644
--- a/arch/mips/include/asm/delay.h
+++ b/arch/mips/include/asm/delay.h
@@ -11,94 +11,12 @@
 #ifndef _ASM_DELAY_H
 #define _ASM_DELAY_H
 
-#include <linux/param.h>
-#include <linux/smp.h>
+extern void __delay(unsigned int loops);
+extern void __ndelay(unsigned int ns);
+extern void __udelay(unsigned int us);
 
-#include <asm/compiler.h>
-#include <asm/war.h>
-
-static inline void __delay(unsigned long loops)
-{
-	if (sizeof(long) == 4)
-		__asm__ __volatile__ (
-		"	.set	noreorder				\n"
-		"	.align	3					\n"
-		"1:	bnez	%0, 1b					\n"
-		"	subu	%0, 1					\n"
-		"	.set	reorder					\n"
-		: "=r" (loops)
-		: "0" (loops));
-	else if (sizeof(long) == 8 && !DADDI_WAR)
-		__asm__ __volatile__ (
-		"	.set	noreorder				\n"
-		"	.align	3					\n"
-		"1:	bnez	%0, 1b					\n"
-		"	dsubu	%0, 1					\n"
-		"	.set	reorder					\n"
-		: "=r" (loops)
-		: "0" (loops));
-	else if (sizeof(long) == 8 && DADDI_WAR)
-		__asm__ __volatile__ (
-		"	.set	noreorder				\n"
-		"	.align	3					\n"
-		"1:	bnez	%0, 1b					\n"
-		"	dsubu	%0, %2					\n"
-		"	.set	reorder					\n"
-		: "=r" (loops)
-		: "0" (loops), "r" (1));
-}
-
-
-/*
- * Division by multiplication: you don't have to worry about
- * loss of precision.
- *
- * Use only for very small delays ( < 1 msec).  Should probably use a
- * lookup table, really, as the multiplications take much too long with
- * short delays.  This is a "reasonable" implementation, though (and the
- * first constant multiplications gets optimized away if the delay is
- * a constant)
- */
-
-static inline void __udelay(unsigned long usecs, unsigned long lpj)
-{
-	unsigned long hi, lo;
-
-	/*
-	 * The rates of 128 is rounded wrongly by the catchall case
-	 * for 64-bit.  Excessive precission?  Probably ...
-	 */
-#if defined(CONFIG_64BIT) && (HZ == 128)
-	usecs *= 0x0008637bd05af6c7UL;		/* 2**64 / (1000000 / HZ) */
-#elif defined(CONFIG_64BIT)
-	usecs *= (0x8000000000000000UL / (500000 / HZ));
-#else /* 32-bit junk follows here */
-	usecs *= (unsigned long) (((0x8000000000000000ULL / (500000 / HZ)) +
-	                           0x80000000ULL) >> 32);
-#endif
-
-	if (sizeof(long) == 4)
-		__asm__("multu\t%2, %3"
-		: "=h" (usecs), "=l" (lo)
-		: "r" (usecs), "r" (lpj)
-		: GCC_REG_ACCUM);
-	else if (sizeof(long) == 8 && !R4000_WAR)
-		__asm__("dmultu\t%2, %3"
-		: "=h" (usecs), "=l" (lo)
-		: "r" (usecs), "r" (lpj)
-		: GCC_REG_ACCUM);
-	else if (sizeof(long) == 8 && R4000_WAR)
-		__asm__("dmultu\t%3, %4\n\tmfhi\t%0"
-		: "=r" (usecs), "=h" (hi), "=l" (lo)
-		: "r" (usecs), "r" (lpj)
-		: GCC_REG_ACCUM);
-
-	__delay(usecs);
-}
-
-#define __udelay_val cpu_data[raw_smp_processor_id()].udelay_val
-
-#define udelay(usecs) __udelay((usecs), __udelay_val)
+#define ndelay(ns) __udelay(ns)
+#define udelay(us) __udelay(us)
 
 /* make sure "usecs *= ..." in udelay do not overflow. */
 #if HZ >= 1000
diff --git a/arch/mips/kernel/proc.c b/arch/mips/kernel/proc.c
index 26760cad8b6..e0a4ac18fa0 100644
--- a/arch/mips/kernel/proc.c
+++ b/arch/mips/kernel/proc.c
@@ -42,7 +42,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
 	seq_printf(m, fmt, __cpu_name[n],
 	                           (version >> 4) & 0x0f, version & 0x0f,
 	                           (fp_vers >> 4) & 0x0f, fp_vers & 0x0f);
-	seq_printf(m, "BogoMIPS\t\t: %lu.%02lu\n",
+	seq_printf(m, "BogoMIPS\t\t: %u.%02u\n",
 	              cpu_data[n].udelay_val / (500000/HZ),
 	              (cpu_data[n].udelay_val / (5000/HZ)) % 100);
 	seq_printf(m, "wait instruction\t: %s\n", cpu_wait ? "yes" : "no");
diff --git a/arch/mips/lib/Makefile b/arch/mips/lib/Makefile
index c13c7ad2cda..2adead5a8a3 100644
--- a/arch/mips/lib/Makefile
+++ b/arch/mips/lib/Makefile
@@ -2,8 +2,8 @@
 # Makefile for MIPS-specific library files..
 #
 
-lib-y	+= csum_partial.o memcpy.o memcpy-inatomic.o memset.o strlen_user.o \
-	   strncpy_user.o strnlen_user.o uncached.o
+lib-y	+= csum_partial.o delay.o memcpy.o memcpy-inatomic.o memset.o \
+	   strlen_user.o strncpy_user.o strnlen_user.o uncached.o
 
 obj-y			+= iomap.o
 obj-$(CONFIG_PCI)	+= iomap-pci.o
diff --git a/arch/mips/lib/delay.c b/arch/mips/lib/delay.c
new file mode 100644
index 00000000000..f69c6b569eb
--- /dev/null
+++ b/arch/mips/lib/delay.c
@@ -0,0 +1,56 @@
+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 1994 by Waldorf Electronics
+ * Copyright (C) 1995 - 2000, 01, 03 by Ralf Baechle
+ * Copyright (C) 1999, 2000 Silicon Graphics, Inc.
+ * Copyright (C) 2007  Maciej W. Rozycki
+ */
+#include <linux/module.h>
+#include <linux/param.h>
+#include <linux/smp.h>
+
+#include <asm/compiler.h>
+#include <asm/war.h>
+
+inline void __delay(unsigned int loops)
+{
+	__asm__ __volatile__ (
+	"	.set	noreorder				\n"
+	"	.align	3					\n"
+	"1:	bnez	%0, 1b					\n"
+	"	subu	%0, 1					\n"
+	"	.set	reorder					\n"
+	: "=r" (loops)
+	: "0" (loops));
+}
+EXPORT_SYMBOL(__delay);
+
+/*
+ * Division by multiplication: you don't have to worry about
+ * loss of precision.
+ *
+ * Use only for very small delays ( < 1 msec).  Should probably use a
+ * lookup table, really, as the multiplications take much too long with
+ * short delays.  This is a "reasonable" implementation, though (and the
+ * first constant multiplications gets optimized away if the delay is
+ * a constant)
+ */
+
+void __udelay(unsigned long us)
+{
+	unsigned int lpj = current_cpu_data.udelay_val;
+
+	__delay((us * 0x000010c7 * HZ * lpj) >> 32);
+}
+EXPORT_SYMBOL(__udelay);
+
+void __ndelay(unsigned long ns)
+{
+	unsigned int lpj = current_cpu_data.udelay_val;
+
+	__delay((us * 0x00000005 * HZ * lpj) >> 32);
+}
+EXPORT_SYMBOL(__ndelay);