Merge branch 'highmem' into devel

Documentation/arm/memory.txt

@@ -29,7 +29,14 @@ ffff0000	ffff0fff	CPU vector page.
 				CPU supports vector relocation (control
 				register V bit.)
 
-ffc00000	fffeffff	DMA memory mapping region.  Memory returned
+fffe0000	fffeffff	XScale cache flush area.  This is used
+				in proc-xscale.S to flush the whole data
+				cache.  Free for other usage on non-XScale.
+
+fff00000	fffdffff	Fixmap mapping region.  Addresses provided
+				by fix_to_virt() will be located here.
+
+ffc00000	ffefffff	DMA memory mapping region.  Memory returned
 				by the dma_alloc_xxx functions will be
 				dynamically mapped here.
 

arch/arm/Kconfig

@@ -939,6 +939,23 @@ config NODES_SHIFT
 	default "2"
 	depends on NEED_MULTIPLE_NODES
 
+config HIGHMEM
+	bool "High Memory Support (EXPERIMENTAL)"
+	depends on MMU && EXPERIMENTAL
+	help
+	  The address space of ARM processors is only 4 Gigabytes large
+	  and it has to accommodate user address space, kernel address
+	  space as well as some memory mapped IO. That means that, if you
+	  have a large amount of physical memory and/or IO, not all of the
+	  memory can be "permanently mapped" by the kernel. The physical
+	  memory that is not permanently mapped is called "high memory".
+
+	  Depending on the selected kernel/user memory split, minimum
+	  vmalloc space and actual amount of RAM, you may not need this
+	  option which should result in a slightly faster kernel.
+
+	  If unsure, say n.
+
 source "mm/Kconfig"
 
 config LEDS

arch/arm/common/dmabounce.c

@@ -25,6 +25,7 @@
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/slab.h>
+#include <linux/page-flags.h>
 #include <linux/device.h>
 #include <linux/dma-mapping.h>
 #include <linux/dmapool.h>
@@ -349,6 +350,12 @@ dma_addr_t dma_map_page(struct device *dev, struct page *page,
 
 	BUG_ON(!valid_dma_direction(dir));
 
+	if (PageHighMem(page)) {
+		dev_err(dev, "DMA buffer bouncing of HIGHMEM pages "
+			     "is not supported\n");
+		return ~0;
+	}
+
 	return map_single(dev, page_address(page) + offset, size, dir);
 }
 EXPORT_SYMBOL(dma_map_page);

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

@@ -15,10 +15,20 @@
  * must not be used by drivers.
  */
 #ifndef __arch_page_to_dma
+
+#if !defined(CONFIG_HIGHMEM)
 static inline dma_addr_t page_to_dma(struct device *dev, struct page *page)
 {
 	return (dma_addr_t)__virt_to_bus((unsigned long)page_address(page));
 }
+#elif defined(__pfn_to_bus)
+static inline dma_addr_t page_to_dma(struct device *dev, struct page *page)
+{
+	return (dma_addr_t)__pfn_to_bus(page_to_pfn(page));
+}
+#else
+#error "this machine class needs to define __arch_page_to_dma to use HIGHMEM"
+#endif
 
 static inline void *dma_to_virt(struct device *dev, dma_addr_t addr)
 {
@@ -57,6 +67,8 @@ static inline dma_addr_t virt_to_dma(struct device *dev, void *addr)
  * Use the driver DMA support - see dma-mapping.h (dma_sync_*)
  */
 extern void dma_cache_maint(const void *kaddr, size_t size, int rw);
+extern void dma_cache_maint_page(struct page *page, unsigned long offset,
+				 size_t size, int rw);
 
 /*
  * Return whether the given device DMA address mask can be supported
@@ -316,7 +328,7 @@ static inline dma_addr_t dma_map_page(struct device *dev, struct page *page,
 	BUG_ON(!valid_dma_direction(dir));
 
 	if (!arch_is_coherent())
-		dma_cache_maint(page_address(page) + offset, size, dir);
+		dma_cache_maint_page(page, offset, size, dir);
 
 	return page_to_dma(dev, page) + offset;
 }

arch/arm/include/asm/fixmap.h

@@ -0,0 +1,41 @@
+#ifndef _ASM_FIXMAP_H
+#define _ASM_FIXMAP_H
+
+/*
+ * Nothing too fancy for now.
+ *
+ * On ARM we already have well known fixed virtual addresses imposed by
+ * the architecture such as the vector page which is located at 0xffff0000,
+ * therefore a second level page table is already allocated covering
+ * 0xfff00000 upwards.
+ *
+ * The cache flushing code in proc-xscale.S uses the virtual area between
+ * 0xfffe0000 and 0xfffeffff.
+ */
+
+#define FIXADDR_START		0xfff00000UL
+#define FIXADDR_TOP		0xfffe0000UL
+#define FIXADDR_SIZE		(FIXADDR_TOP - FIXADDR_START)
+
+#define FIX_KMAP_BEGIN		0
+#define FIX_KMAP_END		(FIXADDR_SIZE >> PAGE_SHIFT)
+
+#define __fix_to_virt(x)	(FIXADDR_START + ((x) << PAGE_SHIFT))
+#define __virt_to_fix(x)	(((x) - FIXADDR_START) >> PAGE_SHIFT)
+
+extern void __this_fixmap_does_not_exist(void);
+
+static inline unsigned long fix_to_virt(const unsigned int idx)
+{
+	if (idx >= FIX_KMAP_END)
+		__this_fixmap_does_not_exist();
+	return __fix_to_virt(idx);
+}
+
+static inline unsigned int virt_to_fix(const unsigned long vaddr)
+{
+	BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START);
+	return __virt_to_fix(vaddr);
+}
+
+#endif

arch/arm/include/asm/highmem.h

@@ -0,0 +1,31 @@
+#ifndef _ASM_HIGHMEM_H
+#define _ASM_HIGHMEM_H
+
+#include <asm/kmap_types.h>
+
+#define PKMAP_BASE		(PAGE_OFFSET - PMD_SIZE)
+#define LAST_PKMAP		PTRS_PER_PTE
+#define LAST_PKMAP_MASK		(LAST_PKMAP - 1)
+#define PKMAP_NR(virt)		(((virt) - PKMAP_BASE) >> PAGE_SHIFT)
+#define PKMAP_ADDR(nr)		(PKMAP_BASE + ((nr) << PAGE_SHIFT))
+
+#define kmap_prot		PAGE_KERNEL
+
+#define flush_cache_kmaps()	flush_cache_all()
+
+extern pte_t *pkmap_page_table;
+
+#define ARCH_NEEDS_KMAP_HIGH_GET
+
+extern void *kmap_high(struct page *page);
+extern void *kmap_high_get(struct page *page);
+extern void kunmap_high(struct page *page);
+
+extern void *kmap(struct page *page);
+extern void kunmap(struct page *page);
+extern void *kmap_atomic(struct page *page, enum km_type type);
+extern void kunmap_atomic(void *kvaddr, enum km_type type);
+extern void *kmap_atomic_pfn(unsigned long pfn, enum km_type type);
+extern struct page *kmap_atomic_to_page(const void *ptr);
+
+#endif

arch/arm/include/asm/kmap_types.h

@@ -18,6 +18,7 @@ enum km_type {
 	KM_IRQ1,
 	KM_SOFTIRQ0,
 	KM_SOFTIRQ1,
+	KM_L2_CACHE,
 	KM_TYPE_NR
 };
 

arch/arm/include/asm/memory.h

@@ -44,13 +44,20 @@
  * The module space lives between the addresses given by TASK_SIZE
  * and PAGE_OFFSET - it must be within 32MB of the kernel text.
  */
-#define MODULES_END		(PAGE_OFFSET)
-#define MODULES_VADDR		(MODULES_END - 16*1048576)
-
+#define MODULES_VADDR		(PAGE_OFFSET - 16*1024*1024)
 #if TASK_SIZE > MODULES_VADDR
 #error Top of user space clashes with start of module space
 #endif
 
+/*
+ * The highmem pkmap virtual space shares the end of the module area.
+ */
+#ifdef CONFIG_HIGHMEM
+#define MODULES_END		(PAGE_OFFSET - PMD_SIZE)
+#else
+#define MODULES_END		(PAGE_OFFSET)
+#endif
+
 /*
  * The XIP kernel gets mapped at the bottom of the module vm area.
  * Since we use sections to map it, this macro replaces the physical address
@@ -181,6 +188,7 @@ static inline void *phys_to_virt(unsigned long x)
 #ifndef __virt_to_bus
 #define __virt_to_bus	__virt_to_phys
 #define __bus_to_virt	__phys_to_virt
+#define __pfn_to_bus(x)	((x) << PAGE_SHIFT)
 #endif
 
 static inline __deprecated unsigned long virt_to_bus(void *x)

arch/arm/mach-iop13xx/include/mach/memory.h

@@ -59,7 +59,10 @@ static inline unsigned long __lbus_to_virt(dma_addr_t x)
 	})
 
 #define __arch_page_to_dma(dev, page)					\
-	__arch_virt_to_dma(dev, page_address(page))
+	({								\
+		/* __is_lbus_virt() can never be true for RAM pages */	\
+		(dma_addr_t)page_to_phys(page);				\
+	})
 
 #endif /* CONFIG_ARCH_IOP13XX */
 #endif /* !ASSEMBLY */

arch/arm/mach-ks8695/include/mach/memory.h

@@ -35,7 +35,11 @@ extern struct bus_type platform_bus_type;
 					__phys_to_virt(x) : __bus_to_virt(x)); })
 #define __arch_virt_to_dma(dev, x)	({ is_lbus_device(dev) ? \
 					(dma_addr_t)__virt_to_phys(x) : (dma_addr_t)__virt_to_bus(x); })
-#define __arch_page_to_dma(dev, x)	__arch_virt_to_dma(dev, page_address(x))
+#define __arch_page_to_dma(dev, x)	\
+	({ dma_addr_t __dma = page_to_phys(page); \
+	   if (!is_lbus_device(dev)) \
+		__dma = __dma - PHYS_OFFSET + KS8695_PCIMEM_PA; \
+	   __dma; })
 
 #endif
 

arch/arm/mm/Makefile

@@ -16,6 +16,7 @@ obj-$(CONFIG_MODULES)		+= proc-syms.o
 
 obj-$(CONFIG_ALIGNMENT_TRAP)	+= alignment.o
 obj-$(CONFIG_DISCONTIGMEM)	+= discontig.o
+obj-$(CONFIG_HIGHMEM)		+= highmem.o
 
 obj-$(CONFIG_CPU_ABRT_NOMMU)	+= abort-nommu.o
 obj-$(CONFIG_CPU_ABRT_EV4)	+= abort-ev4.o

arch/arm/mm/cache-feroceon-l2.c

@@ -14,8 +14,12 @@
 
 #include <linux/init.h>
 #include <asm/cacheflush.h>
+#include <asm/kmap_types.h>
+#include <asm/fixmap.h>
+#include <asm/pgtable.h>
+#include <asm/tlbflush.h>
 #include <plat/cache-feroceon-l2.h>
-
+#include "mm.h"
 
 /*
  * Low-level cache maintenance operations.
@@ -34,14 +38,36 @@
  * The range operations require two successive cp15 writes, in
  * between which we don't want to be preempted.
  */
+
+static inline unsigned long l2_start_va(unsigned long paddr)
+{
+#ifdef CONFIG_HIGHMEM
+	/*
+	 * Let's do our own fixmap stuff in a minimal way here.
+	 * Because range ops can't be done on physical addresses,
+	 * we simply install a virtual mapping for it only for the
+	 * TLB lookup to occur, hence no need to flush the untouched
+	 * memory mapping.  This is protected with the disabling of
+	 * interrupts by the caller.
+	 */
+	unsigned long idx = KM_L2_CACHE + KM_TYPE_NR * smp_processor_id();
+	unsigned long vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
+	set_pte_ext(TOP_PTE(vaddr), pfn_pte(paddr >> PAGE_SHIFT, PAGE_KERNEL), 0);
+	local_flush_tlb_kernel_page(vaddr);
+	return vaddr + (paddr & ~PAGE_MASK);
+#else
+	return __phys_to_virt(paddr);
+#endif
+}
+
 static inline void l2_clean_pa(unsigned long addr)
 {
 	__asm__("mcr p15, 1, %0, c15, c9, 3" : : "r" (addr));
 }
 
-static inline void l2_clean_mva_range(unsigned long start, unsigned long end)
+static inline void l2_clean_pa_range(unsigned long start, unsigned long end)
 {
-	unsigned long flags;
+	unsigned long va_start, va_end, flags;
 
 	/*
 	 * Make sure 'start' and 'end' reference the same page, as
@@ -51,17 +77,14 @@ static inline void l2_clean_mva_range(unsigned long start, unsigned long end)
 	BUG_ON((start ^ end) >> PAGE_SHIFT);
 
 	raw_local_irq_save(flags);
+	va_start = l2_start_va(start);
+	va_end = va_start + (end - start);
 	__asm__("mcr p15, 1, %0, c15, c9, 4\n\t"
 		"mcr p15, 1, %1, c15, c9, 5"
-		: : "r" (start), "r" (end));
+		: : "r" (va_start), "r" (va_end));
 	raw_local_irq_restore(flags);
 }
 
-static inline void l2_clean_pa_range(unsigned long start, unsigned long end)
-{
-	l2_clean_mva_range(__phys_to_virt(start), __phys_to_virt(end));
-}
-
 static inline void l2_clean_inv_pa(unsigned long addr)
 {
 	__asm__("mcr p15, 1, %0, c15, c10, 3" : : "r" (addr));
@@ -72,9 +95,9 @@ static inline void l2_inv_pa(unsigned long addr)
 	__asm__("mcr p15, 1, %0, c15, c11, 3" : : "r" (addr));
 }
 
-static inline void l2_inv_mva_range(unsigned long start, unsigned long end)
+static inline void l2_inv_pa_range(unsigned long start, unsigned long end)
 {
-	unsigned long flags;
+	unsigned long va_start, va_end, flags;
 
 	/*
 	 * Make sure 'start' and 'end' reference the same page, as
@@ -84,17 +107,14 @@ static inline void l2_inv_mva_range(unsigned long start, unsigned long end)
 	BUG_ON((start ^ end) >> PAGE_SHIFT);
 
 	raw_local_irq_save(flags);
+	va_start = l2_start_va(start);
+	va_end = va_start + (end - start);
 	__asm__("mcr p15, 1, %0, c15, c11, 4\n\t"
 		"mcr p15, 1, %1, c15, c11, 5"
-		: : "r" (start), "r" (end));
+		: : "r" (va_start), "r" (va_end));
 	raw_local_irq_restore(flags);
 }
 
-static inline void l2_inv_pa_range(unsigned long start, unsigned long end)
-{
-	l2_inv_mva_range(__phys_to_virt(start), __phys_to_virt(end));
-}
-
 
 /*
  * Linux primitives.