1 files changed, 354 insertions, 180 deletions

diff --git a/drivers/gpu/drm/etnaviv/etnaviv_mmu.c b/drivers/gpu/drm/etnaviv/etnaviv_mmu.c
index f103e787de94..a992be2ede88 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_mmu.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_mmu.c
@@ -1,117 +1,149 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
- * Copyright (C) 2015 Etnaviv Project
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 as published by
- * the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program.  If not, see <http://www.gnu.org/licenses/>.
+ * Copyright (C) 2015-2018 Etnaviv Project
  */
 
+#include <linux/dma-mapping.h>
+#include <linux/scatterlist.h>
+
+#include <drm/drm_print.h>
+
 #include "common.xml.h"
 #include "etnaviv_cmdbuf.h"
 #include "etnaviv_drv.h"
 #include "etnaviv_gem.h"
 #include "etnaviv_gpu.h"
-#include "etnaviv_iommu.h"
 #include "etnaviv_mmu.h"
 
-static int etnaviv_fault_handler(struct iommu_domain *iommu, struct device *dev,
-		unsigned long iova, int flags, void *arg)
+static void etnaviv_context_unmap(struct etnaviv_iommu_context *context,
+				 unsigned long iova, size_t size)
 {
-	DBG("*** fault: iova=%08lx, flags=%d", iova, flags);
-	return 0;
+	size_t unmapped_page, unmapped = 0;
+	size_t pgsize = SZ_4K;
+
+	while (unmapped < size) {
+		unmapped_page = context->global->ops->unmap(context, iova,
+							    pgsize);
+		if (!unmapped_page)
+			break;
+
+		iova += unmapped_page;
+		unmapped += unmapped_page;
+	}
+}
+
+static int etnaviv_context_map(struct etnaviv_iommu_context *context,
+			      unsigned long iova, phys_addr_t paddr,
+			      size_t size, int prot)
+{
+	unsigned long orig_iova = iova;
+	size_t pgsize = SZ_4K;
+	size_t orig_size = size;
+	int ret = 0;
+
+	while (size) {
+		ret = context->global->ops->map(context, iova, paddr, pgsize,
+						prot);
+		if (ret)
+			break;
+
+		iova += pgsize;
+		paddr += pgsize;
+		size -= pgsize;
+	}
+
+	/* unroll mapping in case something went wrong */
+	if (ret)
+		etnaviv_context_unmap(context, orig_iova, orig_size - size);
+
+	return ret;
 }
 
-int etnaviv_iommu_map(struct etnaviv_iommu *iommu, u32 iova,
-		struct sg_table *sgt, unsigned len, int prot)
+static int etnaviv_iommu_map(struct etnaviv_iommu_context *context,
+			     u32 iova, unsigned int va_len,
+			     struct sg_table *sgt, int prot)
 {
-	struct iommu_domain *domain = iommu->domain;
 	struct scatterlist *sg;
 	unsigned int da = iova;
-	unsigned int i, j;
+	unsigned int i;
 	int ret;
 
-	if (!domain || !sgt)
+	if (!context || !sgt)
 		return -EINVAL;
 
-	for_each_sg(sgt->sgl, sg, sgt->nents, i) {
-		u32 pa = sg_dma_address(sg) - sg->offset;
-		size_t bytes = sg_dma_len(sg) + sg->offset;
+	for_each_sgtable_dma_sg(sgt, sg, i) {
+		phys_addr_t pa = sg_dma_address(sg);
+		unsigned int da_len = sg_dma_len(sg);
+		unsigned int bytes = min_t(unsigned int, da_len, va_len);
 
-		VERB("map[%d]: %08x %08x(%zx)", i, iova, pa, bytes);
+		VERB("map[%d]: %08x %pap(%x)", i, da, &pa, bytes);
+
+		if (!IS_ALIGNED(iova | pa | bytes, SZ_4K)) {
+			dev_err(context->global->dev,
+				"unaligned: iova 0x%x pa %pa size 0x%x\n",
+				iova, &pa, bytes);
+			ret = -EINVAL;
+			goto fail;
+		}
 
-		ret = iommu_map(domain, da, pa, bytes, prot);
+		ret = etnaviv_context_map(context, da, pa, bytes, prot);
 		if (ret)
 			goto fail;
 
+		va_len -= bytes;
 		da += bytes;
 	}
 
+	context->flush_seq++;
+
 	return 0;
 
 fail:
-	da = iova;
-
-	for_each_sg(sgt->sgl, sg, i, j) {
-		size_t bytes = sg_dma_len(sg) + sg->offset;
-
-		iommu_unmap(domain, da, bytes);
-		da += bytes;
-	}
+	etnaviv_context_unmap(context, iova, da - iova);
 	return ret;
 }
 
-int etnaviv_iommu_unmap(struct etnaviv_iommu *iommu, u32 iova,
-		struct sg_table *sgt, unsigned len)
+static void etnaviv_iommu_unmap(struct etnaviv_iommu_context *context, u32 iova,
+				struct sg_table *sgt, unsigned len)
 {
-	struct iommu_domain *domain = iommu->domain;
-	struct scatterlist *sg;
-	unsigned int da = iova;
-	int i;
-
-	for_each_sg(sgt->sgl, sg, sgt->nents, i) {
-		size_t bytes = sg_dma_len(sg) + sg->offset;
-		size_t unmapped;
-
-		unmapped = iommu_unmap(domain, da, bytes);
-		if (unmapped < bytes)
-			return unmapped;
-
-		VERB("unmap[%d]: %08x(%zx)", i, iova, bytes);
+	etnaviv_context_unmap(context, iova, len);
 
-		BUG_ON(!PAGE_ALIGNED(bytes));
-
-		da += bytes;
-	}
-
-	return 0;
+	context->flush_seq++;
 }
 
-static void etnaviv_iommu_remove_mapping(struct etnaviv_iommu *mmu,
+static void etnaviv_iommu_remove_mapping(struct etnaviv_iommu_context *context,
 	struct etnaviv_vram_mapping *mapping)
 {
 	struct etnaviv_gem_object *etnaviv_obj = mapping->object;
 
-	etnaviv_iommu_unmap(mmu, mapping->vram_node.start,
-			    etnaviv_obj->sgt, etnaviv_obj->base.size);
+	lockdep_assert_held(&context->lock);
+
+	etnaviv_iommu_unmap(context, mapping->vram_node.start,
+			    etnaviv_obj->sgt, etnaviv_obj->size);
 	drm_mm_remove_node(&mapping->vram_node);
 }
 
-static int etnaviv_iommu_find_iova(struct etnaviv_iommu *mmu,
+void etnaviv_iommu_reap_mapping(struct etnaviv_vram_mapping *mapping)
+{
+	struct etnaviv_iommu_context *context = mapping->context;
+
+	lockdep_assert_held(&context->lock);
+	WARN_ON(mapping->use);
+
+	etnaviv_iommu_remove_mapping(context, mapping);
+	etnaviv_iommu_context_put(mapping->context);
+	mapping->context = NULL;
+	list_del_init(&mapping->mmu_node);
+}
+
+static int etnaviv_iommu_find_iova(struct etnaviv_iommu_context *context,
 				   struct drm_mm_node *node, size_t size)
 {
 	struct etnaviv_vram_mapping *free = NULL;
 	enum drm_mm_insert_mode mode = DRM_MM_INSERT_LOW;
 	int ret;
 
-	lockdep_assert_held(&mmu->lock);
+	lockdep_assert_held(&context->lock);
 
 	while (1) {
 		struct etnaviv_vram_mapping *m, *n;
@@ -119,29 +151,17 @@ static int etnaviv_iommu_find_iova(struct etnaviv_iommu *mmu,
 		struct list_head list;
 		bool found;
 
-		ret = drm_mm_insert_node_in_range(&mmu->mm, node,
-						  size, 0, 0,
-						  mmu->last_iova, U64_MAX,
-						  mode);
+		ret = drm_mm_insert_node_in_range(&context->mm, node,
+						  size, 0, 0, 0, U64_MAX, mode);
 		if (ret != -ENOSPC)
 			break;
 
-		/*
-		 * If we did not search from the start of the MMU region,
-		 * try again in case there are free slots.
-		 */
-		if (mmu->last_iova) {
-			mmu->last_iova = 0;
-			mmu->need_flush = true;
-			continue;
-		}
-
 		/* Try to retire some entries */
-		drm_mm_scan_init(&scan, &mmu->mm, size, 0, 0, mode);
+		drm_mm_scan_init(&scan, &context->mm, size, 0, 0, mode);
 
 		found = 0;
 		INIT_LIST_HEAD(&list);
-		list_for_each_entry(free, &mmu->mappings, mmu_node) {
+		list_for_each_entry(free, &context->mappings, mmu_node) {
 			/* If this vram node has not been used, skip this. */
 			if (!free->vram_node.mm)
 				continue;
@@ -183,9 +203,7 @@ static int etnaviv_iommu_find_iova(struct etnaviv_iommu *mmu,
 		 * this mapping.
 		 */
 		list_for_each_entry_safe(m, n, &list, scan_node) {
-			etnaviv_iommu_remove_mapping(mmu, m);
-			m->mmu = NULL;
-			list_del_init(&m->mmu_node);
+			etnaviv_iommu_reap_mapping(m);
 			list_del_init(&m->scan_node);
 		}
 
@@ -200,9 +218,54 @@ static int etnaviv_iommu_find_iova(struct etnaviv_iommu *mmu,
 	return ret;
 }
 
-int etnaviv_iommu_map_gem(struct etnaviv_iommu *mmu,
+static int etnaviv_iommu_insert_exact(struct etnaviv_iommu_context *context,
+		   struct drm_mm_node *node, size_t size, u64 va)
+{
+	struct etnaviv_vram_mapping *m, *n;
+	struct drm_mm_node *scan_node;
+	LIST_HEAD(scan_list);
+	int ret;
+
+	lockdep_assert_held(&context->lock);
+
+	ret = drm_mm_insert_node_in_range(&context->mm, node, size, 0, 0, va,
+					  va + size, DRM_MM_INSERT_LOWEST);
+	if (ret != -ENOSPC)
+		return ret;
+
+	/*
+	 * When we can't insert the node, due to a existing mapping blocking
+	 * the address space, there are two possible reasons:
+	 * 1. Userspace genuinely messed up and tried to reuse address space
+	 * before the last job using this VMA has finished executing.
+	 * 2. The existing buffer mappings are idle, but the buffers are not
+	 * destroyed yet (likely due to being referenced by another context) in
+	 * which case the mappings will not be cleaned up and we must reap them
+	 * here to make space for the new mapping.
+	 */
+
+	drm_mm_for_each_node_in_range(scan_node, &context->mm, va, va + size) {
+		m = container_of(scan_node, struct etnaviv_vram_mapping,
+				 vram_node);
+
+		if (m->use)
+			return -ENOSPC;
+
+		list_add(&m->scan_node, &scan_list);
+	}
+
+	list_for_each_entry_safe(m, n, &scan_list, scan_node) {
+		etnaviv_iommu_reap_mapping(m);
+		list_del_init(&m->scan_node);
+	}
+
+	return drm_mm_insert_node_in_range(&context->mm, node, size, 0, 0, va,
+					   va + size, DRM_MM_INSERT_LOWEST);
+}
+
+int etnaviv_iommu_map_gem(struct etnaviv_iommu_context *context,
 	struct etnaviv_gem_object *etnaviv_obj, u32 memory_base,
-	struct etnaviv_vram_mapping *mapping)
+	struct etnaviv_vram_mapping *mapping, u64 va)
 {
 	struct sg_table *sgt = etnaviv_obj->sgt;
 	struct drm_mm_node *node;
@@ -210,177 +273,288 @@ int etnaviv_iommu_map_gem(struct etnaviv_iommu *mmu,
 
 	lockdep_assert_held(&etnaviv_obj->lock);
 
-	mutex_lock(&mmu->lock);
+	mutex_lock(&context->lock);
 
 	/* v1 MMU can optimize single entry (contiguous) scatterlists */
-	if (mmu->version == ETNAVIV_IOMMU_V1 &&
+	if (context->global->version == ETNAVIV_IOMMU_V1 &&
 	    sgt->nents == 1 && !(etnaviv_obj->flags & ETNA_BO_FORCE_MMU)) {
 		u32 iova;
 
 		iova = sg_dma_address(sgt->sgl) - memory_base;
 		if (iova < 0x80000000 - sg_dma_len(sgt->sgl)) {
 			mapping->iova = iova;
-			list_add_tail(&mapping->mmu_node, &mmu->mappings);
-			mutex_unlock(&mmu->lock);
-			return 0;
+			mapping->context = etnaviv_iommu_context_get(context);
+			list_add_tail(&mapping->mmu_node, &context->mappings);
+			ret = 0;
+			goto unlock;
 		}
 	}
 
 	node = &mapping->vram_node;
 
-	ret = etnaviv_iommu_find_iova(mmu, node, etnaviv_obj->base.size);
-	if (ret < 0) {
-		mutex_unlock(&mmu->lock);
-		return ret;
-	}
+	if (va)
+		ret = etnaviv_iommu_insert_exact(context, node, etnaviv_obj->size, va);
+	else
+		ret = etnaviv_iommu_find_iova(context, node, etnaviv_obj->size);
+	if (ret < 0)
+		goto unlock;
 
-	mmu->last_iova = node->start + etnaviv_obj->base.size;
 	mapping->iova = node->start;
-	ret = etnaviv_iommu_map(mmu, node->start, sgt, etnaviv_obj->base.size,
-				IOMMU_READ | IOMMU_WRITE);
+	ret = etnaviv_iommu_map(context, node->start, etnaviv_obj->size, sgt,
+				ETNAVIV_PROT_READ | ETNAVIV_PROT_WRITE);
 
 	if (ret < 0) {
 		drm_mm_remove_node(node);
-		mutex_unlock(&mmu->lock);
-		return ret;
+		goto unlock;
 	}
 
-	list_add_tail(&mapping->mmu_node, &mmu->mappings);
-	mmu->need_flush = true;
-	mutex_unlock(&mmu->lock);
+	mapping->context = etnaviv_iommu_context_get(context);
+	list_add_tail(&mapping->mmu_node, &context->mappings);
+unlock:
+	mutex_unlock(&context->lock);
 
 	return ret;
 }
 
-void etnaviv_iommu_unmap_gem(struct etnaviv_iommu *mmu,
+void etnaviv_iommu_unmap_gem(struct etnaviv_iommu_context *context,
 	struct etnaviv_vram_mapping *mapping)
 {
 	WARN_ON(mapping->use);
 
-	mutex_lock(&mmu->lock);
+	mutex_lock(&context->lock);
+
+	/* Bail if the mapping has been reaped by another thread */
+	if (!mapping->context) {
+		mutex_unlock(&context->lock);
+		return;
+	}
 
 	/* If the vram node is on the mm, unmap and remove the node */
-	if (mapping->vram_node.mm == &mmu->mm)
-		etnaviv_iommu_remove_mapping(mmu, mapping);
+	if (mapping->vram_node.mm == &context->mm)
+		etnaviv_iommu_remove_mapping(context, mapping);
 
 	list_del(&mapping->mmu_node);
-	mmu->need_flush = true;
-	mutex_unlock(&mmu->lock);
+	mutex_unlock(&context->lock);
+	etnaviv_iommu_context_put(context);
 }
 
-void etnaviv_iommu_destroy(struct etnaviv_iommu *mmu)
+static void etnaviv_iommu_context_free(struct kref *kref)
 {
-	drm_mm_takedown(&mmu->mm);
-	iommu_domain_free(mmu->domain);
-	kfree(mmu);
-}
+	struct etnaviv_iommu_context *context =
+		container_of(kref, struct etnaviv_iommu_context, refcount);
 
-struct etnaviv_iommu *etnaviv_iommu_new(struct etnaviv_gpu *gpu)
+	etnaviv_cmdbuf_suballoc_unmap(context, &context->cmdbuf_mapping);
+	mutex_destroy(&context->lock);
+	context->global->ops->free(context);
+}
+void etnaviv_iommu_context_put(struct etnaviv_iommu_context *context)
 {
-	enum etnaviv_iommu_version version;
-	struct etnaviv_iommu *mmu;
+	kref_put(&context->refcount, etnaviv_iommu_context_free);
+}
 
-	mmu = kzalloc(sizeof(*mmu), GFP_KERNEL);
-	if (!mmu)
-		return ERR_PTR(-ENOMEM);
+struct etnaviv_iommu_context *
+etnaviv_iommu_context_init(struct etnaviv_iommu_global *global,
+			   struct etnaviv_cmdbuf_suballoc *suballoc)
+{
+	struct etnaviv_iommu_context *ctx;
+	int ret;
 
-	if (!(gpu->identity.minor_features1 & chipMinorFeatures1_MMU_VERSION)) {
-		mmu->domain = etnaviv_iommuv1_domain_alloc(gpu);
-		version = ETNAVIV_IOMMU_V1;
-	} else {
-		mmu->domain = etnaviv_iommuv2_domain_alloc(gpu);
-		version = ETNAVIV_IOMMU_V2;
-	}
+	if (global->version == ETNAVIV_IOMMU_V1)
+		ctx = etnaviv_iommuv1_context_alloc(global);
+	else
+		ctx = etnaviv_iommuv2_context_alloc(global);
 
-	if (!mmu->domain) {
-		dev_err(gpu->dev, "Failed to allocate GPU IOMMU domain\n");
-		kfree(mmu);
-		return ERR_PTR(-ENOMEM);
-	}
+	if (!ctx)
+		return NULL;
 
-	mmu->gpu = gpu;
-	mmu->version = version;
-	mutex_init(&mmu->lock);
-	INIT_LIST_HEAD(&mmu->mappings);
+	ret = etnaviv_cmdbuf_suballoc_map(suballoc, ctx, &ctx->cmdbuf_mapping,
+					  global->memory_base);
+	if (ret)
+		goto out_free;
 
-	drm_mm_init(&mmu->mm, mmu->domain->geometry.aperture_start,
-		    mmu->domain->geometry.aperture_end -
-		    mmu->domain->geometry.aperture_start + 1);
+	if (global->version == ETNAVIV_IOMMU_V1 &&
+	    ctx->cmdbuf_mapping.iova > 0x80000000) {
+		dev_err(global->dev,
+		        "command buffer outside valid memory window\n");
+		goto out_unmap;
+	}
 
-	iommu_set_fault_handler(mmu->domain, etnaviv_fault_handler, gpu->dev);
+	return ctx;
 
-	return mmu;
+out_unmap:
+	etnaviv_cmdbuf_suballoc_unmap(ctx, &ctx->cmdbuf_mapping);
+out_free:
+	global->ops->free(ctx);
+	return NULL;
 }
 
-void etnaviv_iommu_restore(struct etnaviv_gpu *gpu)
+void etnaviv_iommu_restore(struct etnaviv_gpu *gpu,
+			   struct etnaviv_iommu_context *context)
 {
-	if (gpu->mmu->version == ETNAVIV_IOMMU_V1)
-		etnaviv_iommuv1_restore(gpu);
-	else
-		etnaviv_iommuv2_restore(gpu);
+	context->global->ops->restore(gpu, context);
 }
 
-int etnaviv_iommu_get_suballoc_va(struct etnaviv_gpu *gpu, dma_addr_t paddr,
-				  struct drm_mm_node *vram_node, size_t size,
-				  u32 *iova)
+int etnaviv_iommu_get_suballoc_va(struct etnaviv_iommu_context *context,
+				  struct etnaviv_vram_mapping *mapping,
+				  u32 memory_base, dma_addr_t paddr,
+				  size_t size)
 {
-	struct etnaviv_iommu *mmu = gpu->mmu;
+	mutex_lock(&context->lock);
 
-	if (mmu->version == ETNAVIV_IOMMU_V1) {
-		*iova = paddr - gpu->memory_base;
+	if (mapping->use > 0) {
+		mapping->use++;
+		mutex_unlock(&context->lock);
 		return 0;
+	}
+
+	/*
+	 * For MMUv1 we don't add the suballoc region to the pagetables, as
+	 * those GPUs can only work with cmdbufs accessed through the linear
+	 * window. Instead we manufacture a mapping to make it look uniform
+	 * to the upper layers.
+	 */
+	if (context->global->version == ETNAVIV_IOMMU_V1) {
+		mapping->iova = paddr - memory_base;
 	} else {
+		struct drm_mm_node *node = &mapping->vram_node;
 		int ret;
 
-		mutex_lock(&mmu->lock);
-		ret = etnaviv_iommu_find_iova(mmu, vram_node, size);
+		ret = etnaviv_iommu_find_iova(context, node, size);
 		if (ret < 0) {
-			mutex_unlock(&mmu->lock);
+			mutex_unlock(&context->lock);
 			return ret;
 		}
-		ret = iommu_map(mmu->domain, vram_node->start, paddr, size,
-				IOMMU_READ);
+
+		mapping->iova = node->start;
+		ret = etnaviv_context_map(context, node->start, paddr, size,
+					  ETNAVIV_PROT_READ);
 		if (ret < 0) {
-			drm_mm_remove_node(vram_node);
-			mutex_unlock(&mmu->lock);
+			drm_mm_remove_node(node);
+			mutex_unlock(&context->lock);
 			return ret;
 		}
-		mmu->last_iova = vram_node->start + size;
-		gpu->mmu->need_flush = true;
-		mutex_unlock(&mmu->lock);
 
-		*iova = (u32)vram_node->start;
-		return 0;
+		context->flush_seq++;
 	}
+
+	list_add_tail(&mapping->mmu_node, &context->mappings);
+	mapping->use = 1;
+
+	mutex_unlock(&context->lock);
+
+	return 0;
 }
 
-void etnaviv_iommu_put_suballoc_va(struct etnaviv_gpu *gpu,
-				   struct drm_mm_node *vram_node, size_t size,
-				   u32 iova)
+void etnaviv_iommu_put_suballoc_va(struct etnaviv_iommu_context *context,
+		  struct etnaviv_vram_mapping *mapping)
 {
-	struct etnaviv_iommu *mmu = gpu->mmu;
+	struct drm_mm_node *node = &mapping->vram_node;
+
+	mutex_lock(&context->lock);
+	mapping->use--;
 
-	if (mmu->version == ETNAVIV_IOMMU_V2) {
-		mutex_lock(&mmu->lock);
-		iommu_unmap(mmu->domain,iova, size);
-		drm_mm_remove_node(vram_node);
-		mutex_unlock(&mmu->lock);
+	if (mapping->use > 0 || context->global->version == ETNAVIV_IOMMU_V1) {
+		mutex_unlock(&context->lock);
+		return;
 	}
+
+	etnaviv_context_unmap(context, node->start, node->size);
+	drm_mm_remove_node(node);
+	mutex_unlock(&context->lock);
 }
-size_t etnaviv_iommu_dump_size(struct etnaviv_iommu *iommu)
+
+size_t etnaviv_iommu_dump_size(struct etnaviv_iommu_context *context)
+{
+	return context->global->ops->dump_size(context);
+}
+
+void etnaviv_iommu_dump(struct etnaviv_iommu_context *context, void *buf)
 {
-	struct etnaviv_iommu_ops *ops;
+	context->global->ops->dump(context, buf);
+}
+
+int etnaviv_iommu_global_init(struct etnaviv_gpu *gpu)
+{
+	enum etnaviv_iommu_version version = ETNAVIV_IOMMU_V1;
+	struct etnaviv_drm_private *priv = gpu->drm->dev_private;
+	struct etnaviv_iommu_global *global;
+	struct device *dev = gpu->drm->dev;
+
+	if (gpu->identity.minor_features1 & chipMinorFeatures1_MMU_VERSION)
+		version = ETNAVIV_IOMMU_V2;
+
+	if (priv->mmu_global) {
+		if (priv->mmu_global->version != version) {
+			dev_err(gpu->dev,
+				"MMU version doesn't match global version\n");
+			return -ENXIO;
+		}
+
+		priv->mmu_global->use++;
+		return 0;
+	}
+
+	global = kzalloc(sizeof(*global), GFP_KERNEL);
+	if (!global)
+		return -ENOMEM;
 
-	ops = container_of(iommu->domain->ops, struct etnaviv_iommu_ops, ops);
+	global->bad_page_cpu = dma_alloc_wc(dev, SZ_4K, &global->bad_page_dma,
+					    GFP_KERNEL);
+	if (!global->bad_page_cpu)
+		goto free_global;
 
-	return ops->dump_size(iommu->domain);
+	memset32(global->bad_page_cpu, 0xdead55aa, SZ_4K / sizeof(u32));
+
+	if (version == ETNAVIV_IOMMU_V2) {
+		global->v2.pta_cpu = dma_alloc_wc(dev, ETNAVIV_PTA_SIZE,
+					       &global->v2.pta_dma, GFP_KERNEL);
+		if (!global->v2.pta_cpu)
+			goto free_bad_page;
+	}
+
+	global->dev = dev;
+	global->version = version;
+	global->use = 1;
+	mutex_init(&global->lock);
+
+	if (version == ETNAVIV_IOMMU_V1)
+		global->ops = &etnaviv_iommuv1_ops;
+	else
+		global->ops = &etnaviv_iommuv2_ops;
+
+	priv->mmu_global = global;
+
+	return 0;
+
+free_bad_page:
+	dma_free_wc(dev, SZ_4K, global->bad_page_cpu, global->bad_page_dma);
+free_global:
+	kfree(global);
+
+	return -ENOMEM;
 }
 
-void etnaviv_iommu_dump(struct etnaviv_iommu *iommu, void *buf)
+void etnaviv_iommu_global_fini(struct etnaviv_gpu *gpu)
 {
-	struct etnaviv_iommu_ops *ops;
+	struct etnaviv_drm_private *priv = gpu->drm->dev_private;
+	struct etnaviv_iommu_global *global = priv->mmu_global;
+
+	if (!global)
+		return;
+
+	if (--global->use > 0)
+		return;
+
+	if (global->v2.pta_cpu)
+		dma_free_wc(global->dev, ETNAVIV_PTA_SIZE,
+			    global->v2.pta_cpu, global->v2.pta_dma);
+
+	if (global->bad_page_cpu)
+		dma_free_wc(global->dev, SZ_4K,
+			    global->bad_page_cpu, global->bad_page_dma);
 
-	ops = container_of(iommu->domain->ops, struct etnaviv_iommu_ops, ops);
+	mutex_destroy(&global->lock);
+	kfree(global);
 
-	ops->dump(iommu->domain, buf);
+	priv->mmu_global = NULL;
 }