Add generic support for outer caches

The outer cache can be L2 as on RealView/EB MPCore platform or even L3
or further on ARMv7 cores. This patch adds the generic support for
flushing the outer cache in the DMA operations.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>

arch/arm/common/dmabounce.c

@@ -334,6 +334,7 @@ unmap_single(struct device *dev, dma_addr_t dma_addr, size_t size,
 			 */
 			ptr = (unsigned long)buf->ptr;
 			dmac_clean_range(ptr, ptr + size);
+			outer_clean_range(__pa(ptr), __pa(ptr) + size);
 		}
 		free_safe_buffer(device_info, buf);
 	}

arch/arm/kernel/setup.c

@@ -88,6 +88,9 @@ struct cpu_user_fns cpu_user;
 #ifdef MULTI_CACHE
 struct cpu_cache_fns cpu_cache;
 #endif
+#ifdef CONFIG_OUTER_CACHE
+struct outer_cache_fns outer_cache;
+#endif
 
 struct stack {
 	u32 irq[3];

arch/arm/mm/Kconfig

@@ -609,3 +609,6 @@ config NEEDS_SYSCALL_FOR_CMPXCHG
 	  Forget about fast user space cmpxchg support.
 	  It is just not possible.
 
+config OUTER_CACHE
+	bool
+	default n

arch/arm/mm/consistent.c

@@ -208,6 +208,7 @@ __dma_alloc(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp,
 		unsigned long kaddr = (unsigned long)page_address(page);
 		memset(page_address(page), 0, size);
 		dmac_flush_range(kaddr, kaddr + size);
+		outer_flush_range(__pa(kaddr), __pa(kaddr) + size);
 	}
 
 	/*
@@ -485,15 +486,20 @@ void consistent_sync(void *vaddr, size_t size, int direction)
 	unsigned long start = (unsigned long)vaddr;
 	unsigned long end   = start + size;
 
+	BUG_ON(!virt_addr_valid(start) || !virt_addr_valid(end - 1));
+
 	switch (direction) {
 	case DMA_FROM_DEVICE:		/* invalidate only */
 		dmac_inv_range(start, end);
+		outer_inv_range(__pa(start), __pa(end));
 		break;
 	case DMA_TO_DEVICE:		/* writeback only */
 		dmac_clean_range(start, end);
+		outer_clean_range(__pa(start), __pa(end));
 		break;
 	case DMA_BIDIRECTIONAL:		/* writeback and invalidate */
 		dmac_flush_range(start, end);
+		outer_flush_range(__pa(start), __pa(end));
 		break;
 	default:
 		BUG();

include/asm-arm/cacheflush.h

@@ -190,6 +190,12 @@ struct cpu_cache_fns {
 	void (*dma_flush_range)(unsigned long, unsigned long);
 };
 
+struct outer_cache_fns {
+	void (*inv_range)(unsigned long, unsigned long);
+	void (*clean_range)(unsigned long, unsigned long);
+	void (*flush_range)(unsigned long, unsigned long);
+};
+
 /*
  * Select the calling method
  */
@@ -246,6 +252,37 @@ extern void dmac_flush_range(unsigned long, unsigned long);
 
 #endif
 
+#ifdef CONFIG_OUTER_CACHE
+
+extern struct outer_cache_fns outer_cache;
+
+static inline void outer_inv_range(unsigned long start, unsigned long end)
+{
+	if (outer_cache.inv_range)
+		outer_cache.inv_range(start, end);
+}
+static inline void outer_clean_range(unsigned long start, unsigned long end)
+{
+	if (outer_cache.clean_range)
+		outer_cache.clean_range(start, end);
+}
+static inline void outer_flush_range(unsigned long start, unsigned long end)
+{
+	if (outer_cache.flush_range)
+		outer_cache.flush_range(start, end);
+}
+
+#else
+
+static inline void outer_inv_range(unsigned long start, unsigned long end)
+{ }
+static inline void outer_clean_range(unsigned long start, unsigned long end)
+{ }
+static inline void outer_flush_range(unsigned long start, unsigned long end)
+{ }
+
+#endif
+
 /*
  * flush_cache_vmap() is used when creating mappings (eg, via vmap,
  * vmalloc, ioremap etc) in kernel space for pages.  Since the