Merge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  sh: Fix up HAVE_PERF_COUNTERS typo.
  sh: Fix up more dma-mapping fallout.
  sh: SH7786 SMP support.
  sh: Wire up the uncached fixmap on sh64 as well.
  sh: Use local TLB flush in set_pte_phys().
  sh: Provide cpu_idle_wait() to fix up cpuidle/SMP build.

arch/sh/Kconfig

@@ -15,7 +15,7 @@ config SUPERH
 	select HAVE_IOREMAP_PROT if MMU
 	select HAVE_ARCH_TRACEHOOK
 	select HAVE_DMA_API_DEBUG
-	select HAVE_PERF_COUNTER
+	select HAVE_PERF_COUNTERS
 	select RTC_LIB
 	select GENERIC_ATOMIC64
 	help
@@ -71,6 +71,9 @@ config GENERIC_HARDIRQS_NO__DO_IRQ
 config GENERIC_IRQ_PROBE
 	def_bool y
 
+config IRQ_PER_CPU
+	def_bool y
+
 config GENERIC_GPIO
 	def_bool n
 
@@ -151,6 +154,9 @@ config ARCH_NO_VIRT_TO_BUS
 config ARCH_HAS_DEFAULT_IDLE
 	def_bool y
 
+config ARCH_HAS_CPU_IDLE_WAIT
+	def_bool y
+
 config IO_TRAPPED
 	bool
 
@@ -411,6 +417,8 @@ config CPU_SUBTYPE_SH7786
 	select CPU_HAS_PTEAEX
 	select ARCH_SPARSEMEM_ENABLE
 	select SYS_SUPPORTS_NUMA
+	select SYS_SUPPORTS_SMP
+	select GENERIC_CLOCKEVENTS_BROADCAST if SMP
 
 config CPU_SUBTYPE_SHX3
 	bool "Support SH-X3 processor"
@@ -648,7 +656,7 @@ config NR_CPUS
 	int "Maximum number of CPUs (2-32)"
 	range 2 32
 	depends on SMP
-	default "4" if CPU_SHX3
+	default "4" if CPU_SUBTYPE_SHX3
 	default "2"
 	help
 	  This allows you to specify the maximum number of CPUs which this

arch/sh/include/asm/dma-mapping.h

@@ -97,7 +97,7 @@ static inline void dma_unmap_page(struct device *dev, dma_addr_t dma_address,
 	dma_unmap_single(dev, dma_address, size, dir);
 }
 
-static inline void dma_sync_single(struct device *dev, dma_addr_t dma_handle,
+static inline void __dma_sync_single(struct device *dev, dma_addr_t dma_handle,
 				   size_t size, enum dma_data_direction dir)
 {
 #if defined(CONFIG_PCI) && !defined(CONFIG_SH_PCIDMA_NONCOHERENT)
@@ -119,7 +119,7 @@ static inline void dma_sync_single_range(struct device *dev,
 	dma_cache_sync(dev, phys_to_virt(dma_handle) + offset, size, dir);
 }
 
-static inline void dma_sync_sg(struct device *dev, struct scatterlist *sg,
+static inline void __dma_sync_sg(struct device *dev, struct scatterlist *sg,
 			       int nelems, enum dma_data_direction dir)
 {
 	int i;
@@ -137,7 +137,7 @@ static inline void dma_sync_single_for_cpu(struct device *dev,
 					   dma_addr_t dma_handle, size_t size,
 					   enum dma_data_direction dir)
 {
-	dma_sync_single(dev, dma_handle, size, dir);
+	__dma_sync_single(dev, dma_handle, size, dir);
 	debug_dma_sync_single_for_cpu(dev, dma_handle, size, dir);
 }
 
@@ -146,7 +146,7 @@ static inline void dma_sync_single_for_device(struct device *dev,
 					      size_t size,
 					      enum dma_data_direction dir)
 {
-	dma_sync_single(dev, dma_handle, size, dir);
+	__dma_sync_single(dev, dma_handle, size, dir);
 	debug_dma_sync_single_for_device(dev, dma_handle, size, dir);
 }
 
@@ -177,7 +177,7 @@ static inline void dma_sync_sg_for_cpu(struct device *dev,
 				       struct scatterlist *sg, int nelems,
 				       enum dma_data_direction dir)
 {
-	dma_sync_sg(dev, sg, nelems, dir);
+	__dma_sync_sg(dev, sg, nelems, dir);
 	debug_dma_sync_sg_for_cpu(dev, sg, nelems, dir);
 }
 
@@ -185,7 +185,7 @@ static inline void dma_sync_sg_for_device(struct device *dev,
 					  struct scatterlist *sg, int nelems,
 					  enum dma_data_direction dir)
 {
-	dma_sync_sg(dev, sg, nelems, dir);
+	__dma_sync_sg(dev, sg, nelems, dir);
 	debug_dma_sync_sg_for_device(dev, sg, nelems, dir);
 }
 

arch/sh/include/asm/system.h

@@ -154,6 +154,7 @@ extern struct dentry *sh_debugfs_root;
 
 void per_cpu_trap_init(void);
 void default_idle(void);
+void cpu_idle_wait(void);
 
 asmlinkage void break_point_trap(void);
 

arch/sh/kernel/cpu/sh4a/Makefile

@@ -16,7 +16,7 @@ obj-$(CONFIG_CPU_SUBTYPE_SH7366)	+= setup-sh7366.o
 obj-$(CONFIG_CPU_SUBTYPE_SHX3)		+= setup-shx3.o
 
 # SMP setup
-smp-$(CONFIG_CPU_SUBTYPE_SHX3)		:= smp-shx3.o
+smp-$(CONFIG_CPU_SHX3)			:= smp-shx3.o
 
 # Primary on-chip clocks (common)
 clock-$(CONFIG_CPU_SUBTYPE_SH7763)	:= clock-sh7763.o
@@ -38,6 +38,6 @@ pinmux-$(CONFIG_CPU_SUBTYPE_SH7724)	:= pinmux-sh7724.o
 pinmux-$(CONFIG_CPU_SUBTYPE_SH7785)	:= pinmux-sh7785.o
 pinmux-$(CONFIG_CPU_SUBTYPE_SH7786)	:= pinmux-sh7786.o
 
-obj-y			+= $(clock-y)
-obj-$(CONFIG_SMP)	+= $(smp-y)
+obj-y				+= $(clock-y)
+obj-$(CONFIG_SMP)		+= $(smp-y)
 obj-$(CONFIG_GENERIC_GPIO)	+= $(pinmux-y)

arch/sh/kernel/cpu/sh4a/setup-sh7786.c

@@ -595,9 +595,8 @@ enum {
 	HSPI,
 	GPIO0, GPIO1,
 	Thermal,
-	INTC0, INTC1, INTC2, INTC3, INTC4, INTC5, INTC6, INTC7,
-
-	/* interrupt groups */
+	INTICI0, INTICI1, INTICI2, INTICI3,
+	INTICI4, INTICI5, INTICI6, INTICI7,
 };
 
 static struct intc_vect vectors[] __initdata = {
@@ -638,10 +637,12 @@ static struct intc_vect vectors[] __initdata = {
 	INTC_VECT(HSPI, 0xe80),
 	INTC_VECT(GPIO0, 0xea0), INTC_VECT(GPIO1, 0xec0),
 	INTC_VECT(Thermal, 0xee0),
+	INTC_VECT(INTICI0, 0xf00), INTC_VECT(INTICI1, 0xf20),
+	INTC_VECT(INTICI2, 0xf40), INTC_VECT(INTICI3, 0xf60),
+	INTC_VECT(INTICI4, 0xf80), INTC_VECT(INTICI5, 0xfa0),
+	INTC_VECT(INTICI6, 0xfc0), INTC_VECT(INTICI7, 0xfe0),
 };
 
-/* FIXME: Main CPU support only now */
-#if 1 /* Main CPU */
 #define CnINTMSK0	0xfe410030
 #define CnINTMSK1	0xfe410040
 #define CnINTMSKCLR0	0xfe410050
@@ -654,21 +655,6 @@ static struct intc_vect vectors[] __initdata = {
 #define CnINT2MSKCR1	0xfe410a34
 #define CnINT2MSKCR2	0xfe410a38
 #define CnINT2MSKCR3	0xfe410a3c
-#else /* Sub CPU */
-#define CnINTMSK0	0xfe410034
-#define CnINTMSK1	0xfe410044
-#define CnINTMSKCLR0	0xfe410054
-#define CnINTMSKCLR1	0xfe410064
-#define CnINT2MSKR0	0xfe410b20
-#define CnINT2MSKR1	0xfe410b24
-#define CnINT2MSKR2	0xfe410b28
-#define CnINT2MSKR3	0xfe410b2c
-#define CnINT2MSKCR0	0xfe410b30
-#define CnINT2MSKCR1	0xfe410b34
-#define CnINT2MSKCR2	0xfe410b38
-#define CnINT2MSKCR3	0xfe410b3c
-#endif
-
 #define INTMSK2		0xfe410068
 #define INTMSKCLR2	0xfe41006c
 
@@ -753,6 +739,9 @@ static struct intc_prio_reg prio_registers[] __initdata = {
 						  GPIO1, Thermal } },
 	{ 0xfe41085c, 0, 32, 8, /* INT2PRI23 */ { 0, 0, 0, 0 } },
 	{ 0xfe410860, 0, 32, 8, /* INT2PRI24 */ { 0, 0, 0, 0 } },
+	{ 0xfe410090, 0xfe4100a0, 32, 4, /* CnICIPRI / CnICIPRICLR */
+	  { INTICI7, INTICI6, INTICI5, INTICI4,
+	    INTICI3, INTICI2, INTICI1, INTICI0 }, INTC_SMP(4, 2) },
 };
 
 static DECLARE_INTC_DESC(intc_desc, "sh7786", vectors, NULL,

arch/sh/kernel/idle.c

@@ -1,7 +1,7 @@
 /*
  * The idle loop for all SuperH platforms.
  *
- *  Copyright (C) 2002 - 2008  Paul Mundt
+ *  Copyright (C) 2002 - 2009  Paul Mundt
  *
  * 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
@@ -15,6 +15,7 @@
 #include <linux/preempt.h>
 #include <linux/thread_info.h>
 #include <linux/irqflags.h>
+#include <linux/smp.h>
 #include <asm/pgalloc.h>
 #include <asm/system.h>
 #include <asm/atomic.h>
@@ -79,3 +80,23 @@ void cpu_idle(void)
 		check_pgt_cache();
 	}
 }
+
+static void do_nothing(void *unused)
+{
+}
+
+/*
+ * cpu_idle_wait - Used to ensure that all the CPUs discard old value of
+ * pm_idle and update to new pm_idle value. Required while changing pm_idle
+ * handler on SMP systems.
+ *
+ * Caller must have changed pm_idle to the new value before the call. Old
+ * pm_idle value will not be used by any CPU after the return of this function.
+ */
+void cpu_idle_wait(void)
+{
+	smp_mb();
+	/* kick all the CPUs so that they exit out of pm_idle */
+	smp_call_function(do_nothing, NULL, 1);
+}
+EXPORT_SYMBOL_GPL(cpu_idle_wait);

arch/sh/mm/init.c

@@ -70,7 +70,7 @@ static void set_pte_phys(unsigned long addr, unsigned long phys, pgprot_t prot)
 	}
 
 	set_pte(pte, pfn_pte(phys >> PAGE_SHIFT, prot));
-	flush_tlb_one(get_asid(), addr);
+	local_flush_tlb_one(get_asid(), addr);
 }
 
 /*
@@ -177,10 +177,8 @@ void __init paging_init(void)
 
 	free_area_init_nodes(max_zone_pfns);
 
-#ifdef CONFIG_SUPERH32
 	/* Set up the uncached fixmap */
 	set_fixmap_nocache(FIX_UNCACHED, __pa(&__uncached_start));
-#endif
 }
 
 static struct kcore_list kcore_mem, kcore_vmalloc;