1 files changed, 181 insertions, 143 deletions

diff --git a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_2835_arm.c b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_2835_arm.c
index 1091b9f1dd07..2b500d85cebc 100644
--- a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_2835_arm.c
+++ b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_2835_arm.c
@@ -45,16 +45,8 @@
 #include <asm/pgtable.h>
 #include <soc/bcm2835/raspberrypi-firmware.h>
 
-#define dmac_map_area			__glue(_CACHE,_dma_map_area)
-#define dmac_unmap_area 		__glue(_CACHE,_dma_unmap_area)
-
-extern void dmac_map_area(const void *, size_t, int);
-extern void dmac_unmap_area(const void *, size_t, int);
-
 #define TOTAL_SLOTS (VCHIQ_SLOT_ZERO_SLOTS + 2 * 32)
 
-#define VCHIQ_ARM_ADDRESS(x) ((void *)((char *)x + g_virt_to_bus_offset))
-
 #include "vchiq_arm.h"
 #include "vchiq_2835.h"
 #include "vchiq_connected.h"
@@ -70,13 +62,25 @@ typedef struct vchiq_2835_state_struct {
    VCHIQ_ARM_STATE_T arm_state;
 } VCHIQ_2835_ARM_STATE_T;
 
+struct vchiq_pagelist_info {
+	PAGELIST_T *pagelist;
+	size_t pagelist_buffer_size;
+	dma_addr_t dma_addr;
+	enum dma_data_direction dma_dir;
+	unsigned int num_pages;
+	unsigned int pages_need_release;
+	struct page **pages;
+	struct scatterlist *scatterlist;
+	unsigned int scatterlist_mapped;
+};
+
 static void __iomem *g_regs;
 static unsigned int g_cache_line_size = sizeof(CACHE_LINE_SIZE);
 static unsigned int g_fragments_size;
 static char *g_fragments_base;
 static char *g_free_fragments;
 static struct semaphore g_free_fragments_sema;
-static unsigned long g_virt_to_bus_offset;
+static struct device *g_dev;
 
 extern int vchiq_arm_log_level;
 
@@ -85,12 +89,13 @@ static DEFINE_SEMAPHORE(g_free_fragments_mutex);
 static irqreturn_t
 vchiq_doorbell_irq(int irq, void *dev_id);
 
-static int
+static struct vchiq_pagelist_info *
 create_pagelist(char __user *buf, size_t count, unsigned short type,
-                struct task_struct *task, PAGELIST_T ** ppagelist);
+		struct task_struct *task);
 
 static void
-free_pagelist(PAGELIST_T *pagelist, int actual);
+free_pagelist(struct vchiq_pagelist_info *pagelistinfo,
+	      int actual);
 
 int vchiq_platform_init(struct platform_device *pdev, VCHIQ_STATE_T *state)
 {
@@ -104,7 +109,14 @@ int vchiq_platform_init(struct platform_device *pdev, VCHIQ_STATE_T *state)
 	int slot_mem_size, frag_mem_size;
 	int err, irq, i;
 
-	g_virt_to_bus_offset = virt_to_dma(dev, (void *)0);
+	/*
+	 * VCHI messages between the CPU and firmware use
+	 * 32-bit bus addresses.
+	 */
+	err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
+
+	if (err < 0)
+		return err;
 
 	(void)of_property_read_u32(dev->of_node, "cache-line-size",
 				   &g_cache_line_size);
@@ -121,7 +133,7 @@ int vchiq_platform_init(struct platform_device *pdev, VCHIQ_STATE_T *state)
 		return -ENOMEM;
 	}
 
-	WARN_ON(((int)slot_mem & (PAGE_SIZE - 1)) != 0);
+	WARN_ON(((unsigned long)slot_mem & (PAGE_SIZE - 1)) != 0);
 
 	vchiq_slot_zero = vchiq_init_slots(slot_mem, slot_mem_size);
 	if (!vchiq_slot_zero)
@@ -173,9 +185,10 @@ int vchiq_platform_init(struct platform_device *pdev, VCHIQ_STATE_T *state)
 		return err ? : -ENXIO;
 	}
 
+	g_dev = dev;
 	vchiq_log_info(vchiq_arm_log_level,
-		"vchiq_init - done (slots %x, phys %pad)",
-		(unsigned int)vchiq_slot_zero, &slot_phys);
+		"vchiq_init - done (slots %pK, phys %pad)",
+		vchiq_slot_zero, &slot_phys);
 
 	vchiq_call_connected_callbacks();
 
@@ -213,47 +226,37 @@ remote_event_signal(REMOTE_EVENT_T *event)
 
 	event->fired = 1;
 
-	dsb();         /* data barrier operation */
+	dsb(sy);         /* data barrier operation */
 
 	if (event->armed)
 		writel(0, g_regs + BELL2); /* trigger vc interrupt */
 }
 
-int
-vchiq_copy_from_user(void *dst, const void *src, int size)
-{
-	if ((uint32_t)src < TASK_SIZE) {
-		return copy_from_user(dst, src, size);
-	} else {
-		memcpy(dst, src, size);
-		return 0;
-	}
-}
-
 VCHIQ_STATUS_T
 vchiq_prepare_bulk_data(VCHIQ_BULK_T *bulk, VCHI_MEM_HANDLE_T memhandle,
 	void *offset, int size, int dir)
 {
-	PAGELIST_T *pagelist;
-	int ret;
+	struct vchiq_pagelist_info *pagelistinfo;
 
 	WARN_ON(memhandle != VCHI_MEM_HANDLE_INVALID);
 
-	ret = create_pagelist((char __user *)offset, size,
-			(dir == VCHIQ_BULK_RECEIVE)
-			? PAGELIST_READ
-			: PAGELIST_WRITE,
-			current,
-			&pagelist);
-	if (ret != 0)
+	pagelistinfo = create_pagelist((char __user *)offset, size,
+				       (dir == VCHIQ_BULK_RECEIVE)
+				       ? PAGELIST_READ
+				       : PAGELIST_WRITE,
+				       current);
+
+	if (!pagelistinfo)
 		return VCHIQ_ERROR;
 
 	bulk->handle = memhandle;
-	bulk->data = VCHIQ_ARM_ADDRESS(pagelist);
+	bulk->data = (void *)(unsigned long)pagelistinfo->dma_addr;
 
-	/* Store the pagelist address in remote_data, which isn't used by the
-	   slave. */
-	bulk->remote_data = pagelist;
+	/*
+	 * Store the pagelistinfo address in remote_data,
+	 * which isn't used by the slave.
+	 */
+	bulk->remote_data = pagelistinfo;
 
 	return VCHIQ_SUCCESS;
 }
@@ -262,7 +265,8 @@ void
 vchiq_complete_bulk(VCHIQ_BULK_T *bulk)
 {
 	if (bulk && bulk->remote_data && bulk->actual)
-		free_pagelist((PAGELIST_T *)bulk->remote_data, bulk->actual);
+		free_pagelist((struct vchiq_pagelist_info *)bulk->remote_data,
+			      bulk->actual);
 }
 
 void
@@ -350,57 +354,93 @@ vchiq_doorbell_irq(int irq, void *dev_id)
 	return ret;
 }
 
+static void
+cleaup_pagelistinfo(struct vchiq_pagelist_info *pagelistinfo)
+{
+	if (pagelistinfo->scatterlist_mapped) {
+		dma_unmap_sg(g_dev, pagelistinfo->scatterlist,
+			     pagelistinfo->num_pages, pagelistinfo->dma_dir);
+	}
+
+	if (pagelistinfo->pages_need_release) {
+		unsigned int i;
+
+		for (i = 0; i < pagelistinfo->num_pages; i++)
+			put_page(pagelistinfo->pages[i]);
+	}
+
+	dma_free_coherent(g_dev, pagelistinfo->pagelist_buffer_size,
+			  pagelistinfo->pagelist, pagelistinfo->dma_addr);
+}
+
 /* There is a potential problem with partial cache lines (pages?)
 ** at the ends of the block when reading. If the CPU accessed anything in
 ** the same line (page?) then it may have pulled old data into the cache,
 ** obscuring the new data underneath. We can solve this by transferring the
 ** partial cache lines separately, and allowing the ARM to copy into the
 ** cached area.
-
-** N.B. This implementation plays slightly fast and loose with the Linux
-** driver programming rules, e.g. its use of dmac_map_area instead of
-** dma_map_single, but it isn't a multi-platform driver and it benefits
-** from increased speed as a result.
 */
 
-static int
+static struct vchiq_pagelist_info *
 create_pagelist(char __user *buf, size_t count, unsigned short type,
-	struct task_struct *task, PAGELIST_T ** ppagelist)
+		struct task_struct *task)
 {
 	PAGELIST_T *pagelist;
+	struct vchiq_pagelist_info *pagelistinfo;
 	struct page **pages;
-	unsigned long *addrs;
-	unsigned int num_pages, offset, i;
-	char *addr, *base_addr, *next_addr;
-	int run, addridx, actual_pages;
-        unsigned long *need_release;
-
-	offset = (unsigned int)buf & (PAGE_SIZE - 1);
+	u32 *addrs;
+	unsigned int num_pages, offset, i, k;
+	int actual_pages;
+	size_t pagelist_size;
+	struct scatterlist *scatterlist, *sg;
+	int dma_buffers;
+	dma_addr_t dma_addr;
+
+	offset = ((unsigned int)(unsigned long)buf & (PAGE_SIZE - 1));
 	num_pages = (count + offset + PAGE_SIZE - 1) / PAGE_SIZE;
 
-	*ppagelist = NULL;
+	pagelist_size = sizeof(PAGELIST_T) +
+			(num_pages * sizeof(u32)) +
+			(num_pages * sizeof(pages[0]) +
+			(num_pages * sizeof(struct scatterlist))) +
+			sizeof(struct vchiq_pagelist_info);
 
 	/* Allocate enough storage to hold the page pointers and the page
 	** list
 	*/
-	pagelist = kmalloc(sizeof(PAGELIST_T) +
-                           (num_pages * sizeof(unsigned long)) +
-                           sizeof(unsigned long) +
-                           (num_pages * sizeof(pages[0])),
-                           GFP_KERNEL);
-
-	vchiq_log_trace(vchiq_arm_log_level,
-		"create_pagelist - %x", (unsigned int)pagelist);
+	pagelist = dma_zalloc_coherent(g_dev,
+				       pagelist_size,
+				       &dma_addr,
+				       GFP_KERNEL);
+
+	vchiq_log_trace(vchiq_arm_log_level, "create_pagelist - %pK",
+			pagelist);
 	if (!pagelist)
-		return -ENOMEM;
+		return NULL;
 
-	addrs = pagelist->addrs;
-        need_release = (unsigned long *)(addrs + num_pages);
-	pages = (struct page **)(addrs + num_pages + 1);
+	addrs		= pagelist->addrs;
+	pages		= (struct page **)(addrs + num_pages);
+	scatterlist	= (struct scatterlist *)(pages + num_pages);
+	pagelistinfo	= (struct vchiq_pagelist_info *)
+			  (scatterlist + num_pages);
+
+	pagelist->length = count;
+	pagelist->type = type;
+	pagelist->offset = offset;
+
+	/* Populate the fields of the pagelistinfo structure */
+	pagelistinfo->pagelist = pagelist;
+	pagelistinfo->pagelist_buffer_size = pagelist_size;
+	pagelistinfo->dma_addr = dma_addr;
+	pagelistinfo->dma_dir =  (type == PAGELIST_WRITE) ?
+				  DMA_TO_DEVICE : DMA_FROM_DEVICE;
+	pagelistinfo->num_pages = num_pages;
+	pagelistinfo->pages_need_release = 0;
+	pagelistinfo->pages = pages;
+	pagelistinfo->scatterlist = scatterlist;
+	pagelistinfo->scatterlist_mapped = 0;
 
 	if (is_vmalloc_addr(buf)) {
-		int dir = (type == PAGELIST_WRITE) ?
-			DMA_TO_DEVICE : DMA_FROM_DEVICE;
 		unsigned long length = count;
 		unsigned int off = offset;
 
@@ -413,14 +453,13 @@ create_pagelist(char __user *buf, size_t count, unsigned short type,
 			if (bytes > length)
 				bytes = length;
 			pages[actual_pages] = pg;
-			dmac_map_area(page_address(pg) + off, bytes, dir);
 			length -= bytes;
 			off = 0;
 		}
-		*need_release = 0; /* do not try and release vmalloc pages */
+		/* do not try and release vmalloc pages */
 	} else {
 		down_read(&task->mm->mmap_sem);
-		actual_pages = get_user_pages(task, task->mm,
+		actual_pages = get_user_pages(
 				          (unsigned long)buf & ~(PAGE_SIZE - 1),
 					  num_pages,
 					  (type == PAGELIST_READ) ? FOLL_WRITE : 0,
@@ -438,44 +477,59 @@ create_pagelist(char __user *buf, size_t count, unsigned short type,
 			while (actual_pages > 0)
 			{
 				actual_pages--;
-				page_cache_release(pages[actual_pages]);
+				put_page(pages[actual_pages]);
 			}
-			kfree(pagelist);
-			if (actual_pages == 0)
-				actual_pages = -ENOMEM;
-			return actual_pages;
+			cleaup_pagelistinfo(pagelistinfo);
+			return NULL;
 		}
-		*need_release = 1; /* release user pages */
+		 /* release user pages */
+		pagelistinfo->pages_need_release = 1;
 	}
 
-	pagelist->length = count;
-	pagelist->type = type;
-	pagelist->offset = offset;
-
-	/* Group the pages into runs of contiguous pages */
-
-	base_addr = VCHIQ_ARM_ADDRESS(page_address(pages[0]));
-	next_addr = base_addr + PAGE_SIZE;
-	addridx = 0;
-	run = 0;
+	/*
+	 * Initialize the scatterlist so that the magic cookie
+	 *  is filled if debugging is enabled
+	 */
+	sg_init_table(scatterlist, num_pages);
+	/* Now set the pages for each scatterlist */
+	for (i = 0; i < num_pages; i++)
+		sg_set_page(scatterlist + i, pages[i], PAGE_SIZE, 0);
+
+	dma_buffers = dma_map_sg(g_dev,
+				 scatterlist,
+				 num_pages,
+				 pagelistinfo->dma_dir);
+
+	if (dma_buffers == 0) {
+		cleaup_pagelistinfo(pagelistinfo);
+		return NULL;
+	}
 
-	for (i = 1; i < num_pages; i++) {
-		addr = VCHIQ_ARM_ADDRESS(page_address(pages[i]));
-		if ((addr == next_addr) && (run < (PAGE_SIZE - 1))) {
-			next_addr += PAGE_SIZE;
-			run++;
+	pagelistinfo->scatterlist_mapped = 1;
+
+	/* Combine adjacent blocks for performance */
+	k = 0;
+	for_each_sg(scatterlist, sg, dma_buffers, i) {
+		u32 len = sg_dma_len(sg);
+		u32 addr = sg_dma_address(sg);
+
+		/* Note: addrs is the address + page_count - 1
+		 * The firmware expects the block to be page
+		 * aligned and a multiple of the page size
+		 */
+		WARN_ON(len == 0);
+		WARN_ON(len & ~PAGE_MASK);
+		WARN_ON(addr & ~PAGE_MASK);
+		if (k > 0 &&
+		    ((addrs[k - 1] & PAGE_MASK) |
+			((addrs[k - 1] & ~PAGE_MASK) + 1) << PAGE_SHIFT)
+		    == addr) {
+			addrs[k - 1] += (len >> PAGE_SHIFT);
 		} else {
-			addrs[addridx] = (unsigned long)base_addr + run;
-			addridx++;
-			base_addr = addr;
-			next_addr = addr + PAGE_SIZE;
-			run = 0;
+			addrs[k++] = addr | ((len >> PAGE_SHIFT) - 1);
 		}
 	}
 
-	addrs[addridx] = (unsigned long)base_addr + run;
-	addridx++;
-
 	/* Partial cache lines (fragments) require special measures */
 	if ((type == PAGELIST_READ) &&
 		((pagelist->offset & (g_cache_line_size - 1)) ||
@@ -484,8 +538,8 @@ create_pagelist(char __user *buf, size_t count, unsigned short type,
 		char *fragments;
 
 		if (down_interruptible(&g_free_fragments_sema) != 0) {
-			kfree(pagelist);
-			return -EINTR;
+			cleaup_pagelistinfo(pagelistinfo);
+			return NULL;
 		}
 
 		WARN_ON(g_free_fragments == NULL);
@@ -499,29 +553,28 @@ create_pagelist(char __user *buf, size_t count, unsigned short type,
 			(fragments - g_fragments_base) / g_fragments_size;
 	}
 
-	dmac_flush_range(pagelist, addrs + num_pages);
-
-	*ppagelist = pagelist;
-
-	return 0;
+	return pagelistinfo;
 }
 
 static void
-free_pagelist(PAGELIST_T *pagelist, int actual)
+free_pagelist(struct vchiq_pagelist_info *pagelistinfo,
+	      int actual)
 {
-        unsigned long *need_release;
-	struct page **pages;
-	unsigned int num_pages, i;
-
-	vchiq_log_trace(vchiq_arm_log_level,
-		"free_pagelist - %x, %d", (unsigned int)pagelist, actual);
+	unsigned int i;
+	PAGELIST_T *pagelist   = pagelistinfo->pagelist;
+	struct page **pages    = pagelistinfo->pages;
+	unsigned int num_pages = pagelistinfo->num_pages;
 
-	num_pages =
-		(pagelist->length + pagelist->offset + PAGE_SIZE - 1) /
-		PAGE_SIZE;
+	vchiq_log_trace(vchiq_arm_log_level, "free_pagelist - %pK, %d",
+			pagelistinfo->pagelist, actual);
 
-        need_release = (unsigned long *)(pagelist->addrs + num_pages);
-	pages = (struct page **)(pagelist->addrs + num_pages + 1);
+	/*
+	 * NOTE: dma_unmap_sg must be called before the
+	 * cpu can touch any of the data/pages.
+	 */
+	dma_unmap_sg(g_dev, pagelistinfo->scatterlist,
+		     pagelistinfo->num_pages, pagelistinfo->dma_dir);
+	pagelistinfo->scatterlist_mapped = 0;
 
 	/* Deal with any partial cache lines (fragments) */
 	if (pagelist->type >= PAGELIST_READ_WITH_FRAGMENTS) {
@@ -559,27 +612,12 @@ free_pagelist(PAGELIST_T *pagelist, int actual)
 		up(&g_free_fragments_sema);
 	}
 
-	if (*need_release) {
-		unsigned int length = pagelist->length;
-		unsigned int offset = pagelist->offset;
-
-		for (i = 0; i < num_pages; i++) {
-			struct page *pg = pages[i];
-
-			if (pagelist->type != PAGELIST_WRITE) {
-				unsigned int bytes = PAGE_SIZE - offset;
-
-				if (bytes > length)
-					bytes = length;
-				dmac_unmap_area(page_address(pg) + offset,
-						bytes, DMA_FROM_DEVICE);
-				length -= bytes;
-				offset = 0;
-				set_page_dirty(pg);
-			}
-			page_cache_release(pg);
-		}
+	/* Need to mark all the pages dirty. */
+	if (pagelist->type != PAGELIST_WRITE &&
+	    pagelistinfo->pages_need_release) {
+		for (i = 0; i < num_pages; i++)
+			set_page_dirty(pages[i]);
 	}
 
-	kfree(pagelist);
+	cleaup_pagelistinfo(pagelistinfo);
 }