1 files changed, 173 insertions, 78 deletions

diff --git a/drivers/accel/ivpu/ivpu_mmu_context.c b/drivers/accel/ivpu/ivpu_mmu_context.c
index 12a8c09d4547..f0267efa55aa 100644
--- a/drivers/accel/ivpu/ivpu_mmu_context.c
+++ b/drivers/accel/ivpu/ivpu_mmu_context.c
@@ -6,6 +6,7 @@
 #include <linux/bitfield.h>
 #include <linux/highmem.h>
 #include <linux/set_memory.h>
+#include <linux/vmalloc.h>
 
 #include <drm/drm_cache.h>
 
@@ -23,6 +24,7 @@
 #define IVPU_MMU_ENTRY_FLAG_CONT         BIT(52)
 #define IVPU_MMU_ENTRY_FLAG_NG           BIT(11)
 #define IVPU_MMU_ENTRY_FLAG_AF           BIT(10)
+#define IVPU_MMU_ENTRY_FLAG_RO           BIT(7)
 #define IVPU_MMU_ENTRY_FLAG_USER         BIT(6)
 #define IVPU_MMU_ENTRY_FLAG_LLC_COHERENT BIT(2)
 #define IVPU_MMU_ENTRY_FLAG_TYPE_PAGE    BIT(1)
@@ -88,19 +90,6 @@ static void ivpu_pgtable_free_page(struct ivpu_device *vdev, u64 *cpu_addr, dma_
 	}
 }
 
-static int ivpu_mmu_pgtable_init(struct ivpu_device *vdev, struct ivpu_mmu_pgtable *pgtable)
-{
-	dma_addr_t pgd_dma;
-
-	pgtable->pgd_dma_ptr = ivpu_pgtable_alloc_page(vdev, &pgd_dma);
-	if (!pgtable->pgd_dma_ptr)
-		return -ENOMEM;
-
-	pgtable->pgd_dma = pgd_dma;
-
-	return 0;
-}
-
 static void ivpu_mmu_pgtables_free(struct ivpu_device *vdev, struct ivpu_mmu_pgtable *pgtable)
 {
 	int pgd_idx, pud_idx, pmd_idx;
@@ -138,6 +127,27 @@ static void ivpu_mmu_pgtables_free(struct ivpu_device *vdev, struct ivpu_mmu_pgt
 	}
 
 	ivpu_pgtable_free_page(vdev, pgtable->pgd_dma_ptr, pgtable->pgd_dma);
+	pgtable->pgd_dma_ptr = NULL;
+	pgtable->pgd_dma = 0;
+}
+
+static u64*
+ivpu_mmu_ensure_pgd(struct ivpu_device *vdev, struct ivpu_mmu_pgtable *pgtable)
+{
+	u64 *pgd_dma_ptr = pgtable->pgd_dma_ptr;
+	dma_addr_t pgd_dma;
+
+	if (pgd_dma_ptr)
+		return pgd_dma_ptr;
+
+	pgd_dma_ptr = ivpu_pgtable_alloc_page(vdev, &pgd_dma);
+	if (!pgd_dma_ptr)
+		return NULL;
+
+	pgtable->pgd_dma_ptr = pgd_dma_ptr;
+	pgtable->pgd_dma = pgd_dma;
+
+	return pgd_dma_ptr;
 }
 
 static u64*
@@ -235,6 +245,12 @@ ivpu_mmu_context_map_page(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx
 	int pmd_idx = FIELD_GET(IVPU_MMU_PMD_INDEX_MASK, vpu_addr);
 	int pte_idx = FIELD_GET(IVPU_MMU_PTE_INDEX_MASK, vpu_addr);
 
+	drm_WARN_ON(&vdev->drm, ctx->id == IVPU_RESERVED_CONTEXT_MMU_SSID);
+
+	/* Allocate PGD - first level page table if needed */
+	if (!ivpu_mmu_ensure_pgd(vdev, &ctx->pgtable))
+		return -ENOMEM;
+
 	/* Allocate PUD - second level page table if needed */
 	if (!ivpu_mmu_ensure_pud(vdev, &ctx->pgtable, pgd_idx))
 		return -ENOMEM;
@@ -318,6 +334,91 @@ ivpu_mmu_context_map_pages(struct ivpu_device *vdev, struct ivpu_mmu_context *ct
 	return 0;
 }
 
+static void ivpu_mmu_context_set_page_ro(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
+					 u64 vpu_addr)
+{
+	int pgd_idx = FIELD_GET(IVPU_MMU_PGD_INDEX_MASK, vpu_addr);
+	int pud_idx = FIELD_GET(IVPU_MMU_PUD_INDEX_MASK, vpu_addr);
+	int pmd_idx = FIELD_GET(IVPU_MMU_PMD_INDEX_MASK, vpu_addr);
+	int pte_idx = FIELD_GET(IVPU_MMU_PTE_INDEX_MASK, vpu_addr);
+
+	ctx->pgtable.pte_ptrs[pgd_idx][pud_idx][pmd_idx][pte_idx] |= IVPU_MMU_ENTRY_FLAG_RO;
+}
+
+static void ivpu_mmu_context_split_page(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
+					u64 vpu_addr)
+{
+	int pgd_idx = FIELD_GET(IVPU_MMU_PGD_INDEX_MASK, vpu_addr);
+	int pud_idx = FIELD_GET(IVPU_MMU_PUD_INDEX_MASK, vpu_addr);
+	int pmd_idx = FIELD_GET(IVPU_MMU_PMD_INDEX_MASK, vpu_addr);
+	int pte_idx = FIELD_GET(IVPU_MMU_PTE_INDEX_MASK, vpu_addr);
+
+	ctx->pgtable.pte_ptrs[pgd_idx][pud_idx][pmd_idx][pte_idx] &= ~IVPU_MMU_ENTRY_FLAG_CONT;
+}
+
+static void ivpu_mmu_context_split_64k_page(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
+					    u64 vpu_addr)
+{
+	u64 start = ALIGN_DOWN(vpu_addr, IVPU_MMU_CONT_PAGES_SIZE);
+	u64 end = ALIGN(vpu_addr, IVPU_MMU_CONT_PAGES_SIZE);
+	u64 offset = 0;
+
+	ivpu_dbg(vdev, MMU_MAP, "Split 64K page ctx: %u vpu_addr: 0x%llx\n", ctx->id, vpu_addr);
+
+	while (start + offset < end) {
+		ivpu_mmu_context_split_page(vdev, ctx, start + offset);
+		offset += IVPU_MMU_PAGE_SIZE;
+	}
+}
+
+int
+ivpu_mmu_context_set_pages_ro(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx, u64 vpu_addr,
+			      size_t size)
+{
+	u64 end = vpu_addr + size;
+	size_t size_left = size;
+	int ret;
+
+	if (size == 0)
+		return 0;
+
+	if (drm_WARN_ON(&vdev->drm, !IS_ALIGNED(vpu_addr | size, IVPU_MMU_PAGE_SIZE)))
+		return -EINVAL;
+
+	mutex_lock(&ctx->lock);
+
+	ivpu_dbg(vdev, MMU_MAP, "Set read-only pages ctx: %u vpu_addr: 0x%llx size: %lu\n",
+		 ctx->id, vpu_addr, size);
+
+	if (!ivpu_disable_mmu_cont_pages) {
+		/* Split 64K contiguous page at the beginning if needed */
+		if (!IS_ALIGNED(vpu_addr, IVPU_MMU_CONT_PAGES_SIZE))
+			ivpu_mmu_context_split_64k_page(vdev, ctx, vpu_addr);
+
+		/* Split 64K contiguous page at the end if needed */
+		if (!IS_ALIGNED(vpu_addr + size, IVPU_MMU_CONT_PAGES_SIZE))
+			ivpu_mmu_context_split_64k_page(vdev, ctx, vpu_addr + size);
+	}
+
+	while (size_left) {
+		if (vpu_addr < end)
+			ivpu_mmu_context_set_page_ro(vdev, ctx, vpu_addr);
+
+		vpu_addr += IVPU_MMU_PAGE_SIZE;
+		size_left -= IVPU_MMU_PAGE_SIZE;
+	}
+
+	/* Ensure page table modifications are flushed from wc buffers to memory */
+	wmb();
+
+	mutex_unlock(&ctx->lock);
+	ret = ivpu_mmu_invalidate_tlb(vdev, ctx->id);
+	if (ret)
+		ivpu_err(vdev, "Failed to invalidate TLB for ctx %u: %d\n", ctx->id, ret);
+
+	return 0;
+}
+
 static void ivpu_mmu_context_unmap_pages(struct ivpu_mmu_context *ctx, u64 vpu_addr, size_t size)
 {
 	while (size) {
@@ -331,6 +432,7 @@ int
 ivpu_mmu_context_map_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 			 u64 vpu_addr, struct sg_table *sgt,  bool llc_coherent)
 {
+	size_t start_vpu_addr = vpu_addr;
 	struct scatterlist *sg;
 	int ret;
 	u64 prot;
@@ -355,22 +457,42 @@ ivpu_mmu_context_map_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx,
 		dma_addr_t dma_addr = sg_dma_address(sg) - sg->offset;
 		size_t size = sg_dma_len(sg) + sg->offset;
 
+		ivpu_dbg(vdev, MMU_MAP, "Map ctx: %u dma_addr: 0x%llx vpu_addr: 0x%llx size: %lu\n",
+			 ctx->id, dma_addr, vpu_addr, size);
+
 		ret = ivpu_mmu_context_map_pages(vdev, ctx, vpu_addr, dma_addr, size, prot);
 		if (ret) {
 			ivpu_err(vdev, "Failed to map context pages\n");
-			mutex_unlock(&ctx->lock);
-			return ret;
+			goto err_unmap_pages;
 		}
 		vpu_addr += size;
 	}
 
+	if (!ctx->is_cd_valid) {
+		ret = ivpu_mmu_cd_set(vdev, ctx->id, &ctx->pgtable);
+		if (ret) {
+			ivpu_err(vdev, "Failed to set context descriptor for context %u: %d\n",
+				 ctx->id, ret);
+			goto err_unmap_pages;
+		}
+		ctx->is_cd_valid = true;
+	}
+
 	/* Ensure page table modifications are flushed from wc buffers to memory */
 	wmb();
-	mutex_unlock(&ctx->lock);
 
 	ret = ivpu_mmu_invalidate_tlb(vdev, ctx->id);
-	if (ret)
+	if (ret) {
 		ivpu_err(vdev, "Failed to invalidate TLB for ctx %u: %d\n", ctx->id, ret);
+		goto err_unmap_pages;
+	}
+
+	mutex_unlock(&ctx->lock);
+	return 0;
+
+err_unmap_pages:
+	ivpu_mmu_context_unmap_pages(ctx, start_vpu_addr, vpu_addr - start_vpu_addr);
+	mutex_unlock(&ctx->lock);
 	return ret;
 }
 
@@ -388,14 +510,19 @@ ivpu_mmu_context_unmap_sgt(struct ivpu_device *vdev, struct ivpu_mmu_context *ct
 	mutex_lock(&ctx->lock);
 
 	for_each_sgtable_dma_sg(sgt, sg, i) {
+		dma_addr_t dma_addr = sg_dma_address(sg) - sg->offset;
 		size_t size = sg_dma_len(sg) + sg->offset;
 
+		ivpu_dbg(vdev, MMU_MAP, "Unmap ctx: %u dma_addr: 0x%llx vpu_addr: 0x%llx size: %lu\n",
+			 ctx->id, dma_addr, vpu_addr, size);
+
 		ivpu_mmu_context_unmap_pages(ctx, vpu_addr, size);
 		vpu_addr += size;
 	}
 
 	/* Ensure page table modifications are flushed from wc buffers to memory */
 	wmb();
+
 	mutex_unlock(&ctx->lock);
 
 	ret = ivpu_mmu_invalidate_tlb(vdev, ctx->id);
@@ -434,109 +561,77 @@ ivpu_mmu_context_remove_node(struct ivpu_mmu_context *ctx, struct drm_mm_node *n
 	mutex_unlock(&ctx->lock);
 }
 
-static int
-ivpu_mmu_context_init(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx, u32 context_id)
+void ivpu_mmu_context_init(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx, u32 context_id)
 {
 	u64 start, end;
-	int ret;
 
 	mutex_init(&ctx->lock);
 
-	ret = ivpu_mmu_pgtable_init(vdev, &ctx->pgtable);
-	if (ret) {
-		ivpu_err(vdev, "Failed to initialize pgtable for ctx %u: %d\n", context_id, ret);
-		return ret;
-	}
-
 	if (!context_id) {
 		start = vdev->hw->ranges.global.start;
 		end = vdev->hw->ranges.shave.end;
 	} else {
-		start = vdev->hw->ranges.user.start;
-		end = vdev->hw->ranges.dma.end;
+		start = min_t(u64, vdev->hw->ranges.user.start, vdev->hw->ranges.shave.start);
+		end = max_t(u64, vdev->hw->ranges.user.end, vdev->hw->ranges.dma.end);
 	}
 
 	drm_mm_init(&ctx->mm, start, end - start);
 	ctx->id = context_id;
-
-	return 0;
 }
 
-static void ivpu_mmu_context_fini(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx)
+void ivpu_mmu_context_fini(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx)
 {
-	if (drm_WARN_ON(&vdev->drm, !ctx->pgtable.pgd_dma_ptr))
-		return;
+	if (ctx->is_cd_valid) {
+		ivpu_mmu_cd_clear(vdev, ctx->id);
+		ctx->is_cd_valid = false;
+	}
 
 	mutex_destroy(&ctx->lock);
 	ivpu_mmu_pgtables_free(vdev, &ctx->pgtable);
 	drm_mm_takedown(&ctx->mm);
-
-	ctx->pgtable.pgd_dma_ptr = NULL;
-	ctx->pgtable.pgd_dma = 0;
 }
 
-int ivpu_mmu_global_context_init(struct ivpu_device *vdev)
+void ivpu_mmu_global_context_init(struct ivpu_device *vdev)
 {
-	return ivpu_mmu_context_init(vdev, &vdev->gctx, IVPU_GLOBAL_CONTEXT_MMU_SSID);
+	ivpu_mmu_context_init(vdev, &vdev->gctx, IVPU_GLOBAL_CONTEXT_MMU_SSID);
 }
 
 void ivpu_mmu_global_context_fini(struct ivpu_device *vdev)
 {
-	return ivpu_mmu_context_fini(vdev, &vdev->gctx);
+	ivpu_mmu_context_fini(vdev, &vdev->gctx);
 }
 
 int ivpu_mmu_reserved_context_init(struct ivpu_device *vdev)
 {
-	return ivpu_mmu_user_context_init(vdev, &vdev->rctx, IVPU_RESERVED_CONTEXT_MMU_SSID);
-}
-
-void ivpu_mmu_reserved_context_fini(struct ivpu_device *vdev)
-{
-	return ivpu_mmu_user_context_fini(vdev, &vdev->rctx);
-}
-
-void ivpu_mmu_user_context_mark_invalid(struct ivpu_device *vdev, u32 ssid)
-{
-	struct ivpu_file_priv *file_priv;
-
-	xa_lock(&vdev->context_xa);
-
-	file_priv = xa_load(&vdev->context_xa, ssid);
-	if (file_priv)
-		file_priv->has_mmu_faults = true;
-
-	xa_unlock(&vdev->context_xa);
-}
-
-int ivpu_mmu_user_context_init(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx, u32 ctx_id)
-{
 	int ret;
 
-	drm_WARN_ON(&vdev->drm, !ctx_id);
+	ivpu_mmu_context_init(vdev, &vdev->rctx, IVPU_RESERVED_CONTEXT_MMU_SSID);
 
-	ret = ivpu_mmu_context_init(vdev, ctx, ctx_id);
-	if (ret) {
-		ivpu_err(vdev, "Failed to initialize context %u: %d\n", ctx_id, ret);
-		return ret;
+	mutex_lock(&vdev->rctx.lock);
+
+	if (!ivpu_mmu_ensure_pgd(vdev, &vdev->rctx.pgtable)) {
+		ivpu_err(vdev, "Failed to allocate root page table for reserved context\n");
+		ret = -ENOMEM;
+		goto err_ctx_fini;
 	}
 
-	ret = ivpu_mmu_set_pgtable(vdev, ctx_id, &ctx->pgtable);
+	ret = ivpu_mmu_cd_set(vdev, vdev->rctx.id, &vdev->rctx.pgtable);
 	if (ret) {
-		ivpu_err(vdev, "Failed to set page table for context %u: %d\n", ctx_id, ret);
-		goto err_context_fini;
+		ivpu_err(vdev, "Failed to set context descriptor for reserved context\n");
+		goto err_ctx_fini;
 	}
 
-	return 0;
+	mutex_unlock(&vdev->rctx.lock);
+	return ret;
 
-err_context_fini:
-	ivpu_mmu_context_fini(vdev, ctx);
+err_ctx_fini:
+	mutex_unlock(&vdev->rctx.lock);
+	ivpu_mmu_context_fini(vdev, &vdev->rctx);
 	return ret;
 }
 
-void ivpu_mmu_user_context_fini(struct ivpu_device *vdev, struct ivpu_mmu_context *ctx)
+void ivpu_mmu_reserved_context_fini(struct ivpu_device *vdev)
 {
-	drm_WARN_ON(&vdev->drm, !ctx->id);
-
-	ivpu_mmu_clear_pgtable(vdev, ctx->id);
-	ivpu_mmu_context_fini(vdev, ctx);
+	ivpu_mmu_cd_clear(vdev, vdev->rctx.id);
+	ivpu_mmu_context_fini(vdev, &vdev->rctx);
 }