1 files changed, 102 insertions, 66 deletions

diff --git a/drivers/gpu/drm/gma500/gtt.c b/drivers/gpu/drm/gma500/gtt.c
index aff194fbe9f..592d205a008 100644
--- a/drivers/gpu/drm/gma500/gtt.c
+++ b/drivers/gpu/drm/gma500/gtt.c
@@ -22,6 +22,7 @@
 #include <drm/drmP.h>
 #include <linux/shmem_fs.h>
 #include "psb_drv.h"
+#include "blitter.h"
 
 
 /*
@@ -39,6 +40,10 @@ static inline uint32_t psb_gtt_mask_pte(uint32_t pfn, int type)
 {
 	uint32_t mask = PSB_PTE_VALID;
 
+	/* Ensure we explode rather than put an invalid low mapping of
+	   a high mapping page into the gtt */
+	BUG_ON(pfn & ~(0xFFFFFFFF >> PAGE_SHIFT));
+
 	if (type & PSB_MMU_CACHED_MEMORY)
 		mask |= PSB_PTE_CACHED;
 	if (type & PSB_MMU_RO_MEMORY)
@@ -57,7 +62,7 @@ static inline uint32_t psb_gtt_mask_pte(uint32_t pfn, int type)
  *	Given a gtt_range object return the GTT offset of the page table
  *	entries for this gtt_range
  */
-u32 *psb_gtt_entry(struct drm_device *dev, struct gtt_range *r)
+static u32 __iomem *psb_gtt_entry(struct drm_device *dev, struct gtt_range *r)
 {
 	struct drm_psb_private *dev_priv = dev->dev_private;
 	unsigned long offset;
@@ -76,9 +81,11 @@ u32 *psb_gtt_entry(struct drm_device *dev, struct gtt_range *r)
  *	the GTT. This is protected via the gtt mutex which the caller
  *	must hold.
  */
-static int psb_gtt_insert(struct drm_device *dev, struct gtt_range *r)
+static int psb_gtt_insert(struct drm_device *dev, struct gtt_range *r,
+			  int resume)
 {
-	u32 *gtt_slot, pte;
+	u32 __iomem *gtt_slot;
+	u32 pte;
 	struct page **pages;
 	int i;
 
@@ -92,16 +99,20 @@ static int psb_gtt_insert(struct drm_device *dev, struct gtt_range *r)
 	gtt_slot = psb_gtt_entry(dev, r);
 	pages = r->pages;
 
-	/* Make sure changes are visible to the GPU */
-	set_pages_array_uc(pages, r->npage);
+	if (!resume) {
+		/* Make sure changes are visible to the GPU */
+		set_pages_array_wc(pages, r->npage);
+	}
 
 	/* Write our page entries into the GTT itself */
 	for (i = r->roll; i < r->npage; i++) {
-		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]), 0);
+		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]),
+				       PSB_MMU_CACHED_MEMORY);
 		iowrite32(pte, gtt_slot++);
 	}
 	for (i = 0; i < r->roll; i++) {
-		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]), 0);
+		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]),
+				       PSB_MMU_CACHED_MEMORY);
 		iowrite32(pte, gtt_slot++);
 	}
 	/* Make sure all the entries are set before we return */
@@ -119,16 +130,18 @@ static int psb_gtt_insert(struct drm_device *dev, struct gtt_range *r)
  *	page table entries with the dummy page. This is protected via the gtt
  *	mutex which the caller must hold.
  */
-static void psb_gtt_remove(struct drm_device *dev, struct gtt_range *r)
+void psb_gtt_remove(struct drm_device *dev, struct gtt_range *r)
 {
 	struct drm_psb_private *dev_priv = dev->dev_private;
-	u32 *gtt_slot, pte;
+	u32 __iomem *gtt_slot;
+	u32 pte;
 	int i;
 
 	WARN_ON(r->stolen);
 
 	gtt_slot = psb_gtt_entry(dev, r);
-	pte = psb_gtt_mask_pte(page_to_pfn(dev_priv->scratch_page), 0);
+	pte = psb_gtt_mask_pte(page_to_pfn(dev_priv->scratch_page),
+			       PSB_MMU_CACHED_MEMORY);
 
 	for (i = 0; i < r->npage; i++)
 		iowrite32(pte, gtt_slot++);
@@ -148,7 +161,8 @@ static void psb_gtt_remove(struct drm_device *dev, struct gtt_range *r)
  */
 void psb_gtt_roll(struct drm_device *dev, struct gtt_range *r, int roll)
 {
-	u32 *gtt_slot, pte;
+	u32 __iomem *gtt_slot;
+	u32 pte;
 	int i;
 
 	if (roll >= r->npage) {
@@ -166,11 +180,13 @@ void psb_gtt_roll(struct drm_device *dev, struct gtt_range *r, int roll)
 	gtt_slot = psb_gtt_entry(dev, r);
 
 	for (i = r->roll; i < r->npage; i++) {
-		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]), 0);
+		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]),
+				       PSB_MMU_CACHED_MEMORY);
 		iowrite32(pte, gtt_slot++);
 	}
 	for (i = 0; i < r->roll; i++) {
-		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]), 0);
+		pte = psb_gtt_mask_pte(page_to_pfn(r->pages[i]),
+				       PSB_MMU_CACHED_MEMORY);
 		iowrite32(pte, gtt_slot++);
 	}
 	ioread32(gtt_slot - 1);
@@ -186,37 +202,18 @@ void psb_gtt_roll(struct drm_device *dev, struct gtt_range *r, int roll)
  */
 static int psb_gtt_attach_pages(struct gtt_range *gt)
 {
-	struct inode *inode;
-	struct address_space *mapping;
-	int i;
-	struct page *p;
-	int pages = gt->gem.size / PAGE_SIZE;
+	struct page **pages;
 
 	WARN_ON(gt->pages);
 
-	/* This is the shared memory object that backs the GEM resource */
-	inode = gt->gem.filp->f_path.dentry->d_inode;
-	mapping = inode->i_mapping;
+	pages = drm_gem_get_pages(&gt->gem, 0);
+	if (IS_ERR(pages))
+		return PTR_ERR(pages);
 
-	gt->pages = kmalloc(pages * sizeof(struct page *), GFP_KERNEL);
-	if (gt->pages == NULL)
-		return -ENOMEM;
-	gt->npage = pages;
+	gt->npage = gt->gem.size / PAGE_SIZE;
+	gt->pages = pages;
 
-	for (i = 0; i < pages; i++) {
-		p = shmem_read_mapping_page(mapping, i);
-		if (IS_ERR(p))
-			goto err;
-		gt->pages[i] = p;
-	}
 	return 0;
-
-err:
-	while (i--)
-		page_cache_release(gt->pages[i]);
-	kfree(gt->pages);
-	gt->pages = NULL;
-	return PTR_ERR(p);
 }
 
 /**
@@ -230,13 +227,7 @@ err:
  */
 static void psb_gtt_detach_pages(struct gtt_range *gt)
 {
-	int i;
-	for (i = 0; i < gt->npage; i++) {
-		/* FIXME: do we need to force dirty */
-		set_page_dirty(gt->pages[i]);
-		page_cache_release(gt->pages[i]);
-	}
-	kfree(gt->pages);
+	drm_gem_put_pages(&gt->gem, gt->pages, true, false);
 	gt->pages = NULL;
 }
 
@@ -255,6 +246,7 @@ int psb_gtt_pin(struct gtt_range *gt)
 	int ret = 0;
 	struct drm_device *dev = gt->gem.dev;
 	struct drm_psb_private *dev_priv = dev->dev_private;
+	u32 gpu_base = dev_priv->gtt.gatt_start;
 
 	mutex_lock(&dev_priv->gtt_mutex);
 
@@ -262,11 +254,14 @@ int psb_gtt_pin(struct gtt_range *gt)
 		ret = psb_gtt_attach_pages(gt);
 		if (ret < 0)
 			goto out;
-		ret = psb_gtt_insert(dev, gt);
+		ret = psb_gtt_insert(dev, gt, 0);
 		if (ret < 0) {
 			psb_gtt_detach_pages(gt);
 			goto out;
 		}
+		psb_mmu_insert_pages(psb_mmu_get_default_pd(dev_priv->mmu),
+				     gt->pages, (gpu_base + gt->offset),
+				     gt->npage, 0, 0, PSB_MMU_CACHED_MEMORY);
 	}
 	gt->in_gart++;
 out:
@@ -289,16 +284,30 @@ void psb_gtt_unpin(struct gtt_range *gt)
 {
 	struct drm_device *dev = gt->gem.dev;
 	struct drm_psb_private *dev_priv = dev->dev_private;
+	u32 gpu_base = dev_priv->gtt.gatt_start;
+	int ret;
 
+	/* While holding the gtt_mutex no new blits can be initiated */
 	mutex_lock(&dev_priv->gtt_mutex);
 
+	/* Wait for any possible usage of the memory to be finished */
+	ret = gma_blt_wait_idle(dev_priv);
+	if (ret) {
+		DRM_ERROR("Failed to idle the blitter, unpin failed!");
+		goto out;
+	}
+
 	WARN_ON(!gt->in_gart);
 
 	gt->in_gart--;
 	if (gt->in_gart == 0 && gt->stolen == 0) {
+		psb_mmu_remove_pages(psb_mmu_get_default_pd(dev_priv->mmu),
+				     (gpu_base + gt->offset), gt->npage, 0, 0);
 		psb_gtt_remove(dev, gt);
 		psb_gtt_detach_pages(gt);
 	}
+
+out:
 	mutex_unlock(&dev_priv->gtt_mutex);
 }
 
@@ -321,7 +330,7 @@ void psb_gtt_unpin(struct gtt_range *gt)
  *	as in use.
  */
 struct gtt_range *psb_gtt_alloc_range(struct drm_device *dev, int len,
-						const char *name, int backed)
+				      const char *name, int backed, u32 align)
 {
 	struct drm_psb_private *dev_priv = dev->dev_private;
 	struct gtt_range *gt;
@@ -349,7 +358,7 @@ struct gtt_range *psb_gtt_alloc_range(struct drm_device *dev, int len,
 	/* Ensure this is set for non GEM objects */
 	gt->gem.dev = dev;
 	ret = allocate_resource(dev_priv->gtt_mem, &gt->resource,
-				len, start, end, PAGE_SIZE, NULL, NULL);
+				len, start, end, align, NULL, NULL);
 	if (ret == 0) {
 		gt->offset = gt->resource.start - r->start;
 		return gt;
@@ -378,7 +387,7 @@ void psb_gtt_free_range(struct drm_device *dev, struct gtt_range *gt)
 	kfree(gt);
 }
 
-void psb_gtt_alloc(struct drm_device *dev)
+static void psb_gtt_alloc(struct drm_device *dev)
 {
 	struct drm_psb_private *dev_priv = dev->dev_private;
 	init_rwsem(&dev_priv->gtt.sem);
@@ -409,16 +418,16 @@ int psb_gtt_init(struct drm_device *dev, int resume)
 	unsigned long stolen_size, vram_stolen_size;
 	unsigned i, num_pages;
 	unsigned pfn_base;
-	uint32_t vram_pages;
-	uint32_t dvmt_mode = 0;
 	struct psb_gtt *pg;
 
 	int ret = 0;
 	uint32_t pte;
 
-	mutex_init(&dev_priv->gtt_mutex);
+	if (!resume) {
+		mutex_init(&dev_priv->gtt_mutex);
+		psb_gtt_alloc(dev);
+	}
 
-	psb_gtt_alloc(dev);
 	pg = &dev_priv->gtt;
 
 	/* Enable the GTT */
@@ -483,13 +492,8 @@ int psb_gtt_init(struct drm_device *dev, int resume)
 
 	stolen_size = vram_stolen_size;
 
-	printk(KERN_INFO "Stolen memory information\n");
-	printk(KERN_INFO "       base in RAM: 0x%x\n", dev_priv->stolen_base);
-	printk(KERN_INFO "       size: %luK, calculated by (GTT RAM base) - (Stolen base), seems wrong\n",
-		vram_stolen_size/1024);
-	dvmt_mode = (dev_priv->gmch_ctrl >> 4) & 0x7;
-	printk(KERN_INFO "      the correct size should be: %dM(dvmt mode=%d)\n",
-		(dvmt_mode == 1) ? 1 : (2 << (dvmt_mode - 1)), dvmt_mode);
+	dev_dbg(dev->dev, "Stolen memory base 0x%x, size %luK\n",
+			dev_priv->stolen_base, vram_stolen_size / 1024);
 
 	if (resume && (gtt_pages != pg->gtt_pages) &&
 	    (stolen_size != pg->stolen_size)) {
@@ -505,7 +509,8 @@ int psb_gtt_init(struct drm_device *dev, int resume)
 	/*
 	 *	Map the GTT and the stolen memory area
 	 */
-	dev_priv->gtt_map = ioremap_nocache(pg->gtt_phys_start,
+	if (!resume)
+		dev_priv->gtt_map = ioremap_nocache(pg->gtt_phys_start,
 						gtt_pages << PAGE_SHIFT);
 	if (!dev_priv->gtt_map) {
 		dev_err(dev->dev, "Failure to map gtt.\n");
@@ -513,7 +518,10 @@ int psb_gtt_init(struct drm_device *dev, int resume)
 		goto out_err;
 	}
 
-	dev_priv->vram_addr = ioremap_wc(dev_priv->stolen_base, stolen_size);
+	if (!resume)
+		dev_priv->vram_addr = ioremap_wc(dev_priv->stolen_base,
+						 stolen_size);
+
 	if (!dev_priv->vram_addr) {
 		dev_err(dev->dev, "Failure to map stolen base.\n");
 		ret = -ENOMEM;
@@ -525,11 +533,11 @@ int psb_gtt_init(struct drm_device *dev, int resume)
 	 */
 
 	pfn_base = dev_priv->stolen_base >> PAGE_SHIFT;
-	vram_pages = num_pages = vram_stolen_size >> PAGE_SHIFT;
-	printk(KERN_INFO"Set up %d stolen pages starting at 0x%08x, GTT offset %dK\n",
+	num_pages = vram_stolen_size >> PAGE_SHIFT;
+	dev_dbg(dev->dev, "Set up %d stolen pages starting at 0x%08x, GTT offset %dK\n",
 		num_pages, pfn_base << PAGE_SHIFT, 0);
 	for (i = 0; i < num_pages; ++i) {
-		pte = psb_gtt_mask_pte(pfn_base + i, 0);
+		pte = psb_gtt_mask_pte(pfn_base + i, PSB_MMU_CACHED_MEMORY);
 		iowrite32(pte, dev_priv->gtt_map + i);
 	}
 
@@ -538,7 +546,7 @@ int psb_gtt_init(struct drm_device *dev, int resume)
 	 */
 
 	pfn_base = page_to_pfn(dev_priv->scratch_page);
-	pte = psb_gtt_mask_pte(pfn_base, 0);
+	pte = psb_gtt_mask_pte(pfn_base, PSB_MMU_CACHED_MEMORY);
 	for (; i < gtt_pages; ++i)
 		iowrite32(pte, dev_priv->gtt_map + i);
 
@@ -549,3 +557,31 @@ out_err:
 	psb_gtt_takedown(dev);
 	return ret;
 }
+
+int psb_gtt_restore(struct drm_device *dev)
+{
+	struct drm_psb_private *dev_priv = dev->dev_private;
+	struct resource *r = dev_priv->gtt_mem->child;
+	struct gtt_range *range;
+	unsigned int restored = 0, total = 0, size = 0;
+
+	/* On resume, the gtt_mutex is already initialized */
+	mutex_lock(&dev_priv->gtt_mutex);
+	psb_gtt_init(dev, 1);
+
+	while (r != NULL) {
+		range = container_of(r, struct gtt_range, resource);
+		if (range->pages) {
+			psb_gtt_insert(dev, range, 1);
+			size += range->resource.end - range->resource.start;
+			restored++;
+		}
+		r = r->sibling;
+		total++;
+	}
+	mutex_unlock(&dev_priv->gtt_mutex);
+	DRM_DEBUG_DRIVER("Restored %u of %u gtt ranges (%u KB)", restored,
+			 total, (size / 1024));
+
+	return 0;
+}