Merge tag 'misc-habanalabs-next-2020-09-22' of git://people.freedeskt…

…op.org/~gabbayo/linux into char-misc-next

Oded writes:

This tag contains the following changes for kernel 5.10-rc1:

- Stop using the DRM's dma-fence module and instead use kernel completions.
- Support PCIe AER
- Use dma_mmap_coherent for memory allocated using dma_alloc_coherent
- Use smallest possible alignment when allocating virtual addresses in our
  MMU driver.
- Refactor MMU driver code to be device-oriented
- Allow user to check CS status without any sleep
- Add an option to map a Command Buffer to the Device's MMU
- Expose sync manager resource allocation to user through INFO IOCTL
- Convert code to use standard BIT(), GENMASK() and FIELD_PREP()
- Many small fixes (casting, better error messages, remove unused
  defines, h/w configuration fixes, etc.)

* tag 'misc-habanalabs-next-2020-09-22' of git://people.freedesktop.org/~gabbayo/linux: (46 commits)
  habanalabs: update scratchpad register map
  habanalabs: add indication of security-enabled F/W
  habanalabs/gaudi: fix DMA completions max outstanding to 15
  habanalabs/gaudi: remove axi drain support
  habanalabs: update firmware interface file
  habanalabs: Add an option to map CB to device MMU
  habanalabs: Save context in a command buffer object
  habanalabs: no need for DMA_SHARED_BUFFER
  habanalabs: allow to wait on CS without sleep
  habanalabs/gaudi: increase timeout for boot fit load
  habanalabs: add debugfs support for MMU with 6 HOPs
  habanalabs: add num_hops to hl_mmu_properties
  habanalabs: refactor MMU as device-oriented
  habanalabs: rename mmu.c to mmu_v1.c
  habanalabs: use smallest possible alignment for virtual addresses
  habanalabs: check flag before reset because of f/w event
  habanalabs: increase PQ COMP_OFFSET by one nibble
  habanalabs: Fix alignment issue in cpucp_info structure
  habanalabs: remove unused define
  habanalabs: remove unused ASIC function pointer
  ...

Documentation/ABI/testing/sysfs-driver-habanalabs

@@ -2,13 +2,17 @@ What:           /sys/class/habanalabs/hl<n>/armcp_kernel_ver
 Date:           Jan 2019
 KernelVersion:  5.1
 Contact:        [email protected]
-Description:    Version of the Linux kernel running on the device's CPU
+Description:    Version of the Linux kernel running on the device's CPU.
+                Will be DEPRECATED in Linux kernel version 5.10, and be
+                replaced with cpucp_kernel_ver
 
 What:           /sys/class/habanalabs/hl<n>/armcp_ver
 Date:           Jan 2019
 KernelVersion:  5.1
 Contact:        [email protected]
 Description:    Version of the application running on the device's CPU
+                Will be DEPRECATED in Linux kernel version 5.10, and be
+                replaced with cpucp_ver
 
 What:           /sys/class/habanalabs/hl<n>/clk_max_freq_mhz
 Date:           Jun 2019
@@ -33,6 +37,18 @@ KernelVersion:  5.1
 Contact:        [email protected]
 Description:    Version of the Device's CPLD F/W
 
+What:           /sys/class/habanalabs/hl<n>/cpucp_kernel_ver
+Date:           Oct 2020
+KernelVersion:  5.10
+Contact:        [email protected]
+Description:    Version of the Linux kernel running on the device's CPU
+
+What:           /sys/class/habanalabs/hl<n>/cpucp_ver
+Date:           Oct 2020
+KernelVersion:  5.10
+Contact:        [email protected]
+Description:    Version of the application running on the device's CPU
+
 What:           /sys/class/habanalabs/hl<n>/device_type
 Date:           Jan 2019
 KernelVersion:  5.1

drivers/misc/habanalabs/Kconfig

@@ -7,7 +7,6 @@ config HABANA_AI
 	tristate "HabanaAI accelerators (habanalabs)"
 	depends on PCI && HAS_IOMEM
 	select FRAME_VECTOR
-	select DMA_SHARED_BUFFER
 	select GENERIC_ALLOCATOR
 	select HWMON
 	help

drivers/misc/habanalabs/common/Makefile

@@ -3,5 +3,5 @@ HL_COMMON_FILES := common/habanalabs_drv.o common/device.o common/context.o \
 		common/asid.o common/habanalabs_ioctl.o \
 		common/command_buffer.o common/hw_queue.o common/irq.o \
 		common/sysfs.o common/hwmon.o common/memory.o \
-		common/command_submission.o common/mmu.o common/firmware_if.o \
-		common/pci.o
+		common/command_submission.o common/mmu.o common/mmu_v1.o \
+		common/firmware_if.o common/pci.o

drivers/misc/habanalabs/common/command_buffer.c

@@ -13,6 +13,131 @@
 #include <linux/uaccess.h>
 #include <linux/genalloc.h>
 
+static int cb_map_mem(struct hl_ctx *ctx, struct hl_cb *cb)
+{
+	struct hl_device *hdev = ctx->hdev;
+	struct asic_fixed_properties *prop = &hdev->asic_prop;
+	struct hl_vm_va_block *va_block, *tmp;
+	dma_addr_t bus_addr;
+	u64 virt_addr;
+	u32 page_size = prop->pmmu.page_size;
+	s32 offset;
+	int rc;
+
+	if (!hdev->supports_cb_mapping) {
+		dev_err_ratelimited(hdev->dev,
+				"Cannot map CB because no VA range is allocated for CB mapping\n");
+		return -EINVAL;
+	}
+
+	if (!hdev->mmu_enable) {
+		dev_err_ratelimited(hdev->dev,
+				"Cannot map CB because MMU is disabled\n");
+		return -EINVAL;
+	}
+
+	INIT_LIST_HEAD(&cb->va_block_list);
+
+	for (bus_addr = cb->bus_address;
+			bus_addr < cb->bus_address + cb->size;
+			bus_addr += page_size) {
+
+		virt_addr = (u64) gen_pool_alloc(ctx->cb_va_pool, page_size);
+		if (!virt_addr) {
+			dev_err(hdev->dev,
+				"Failed to allocate device virtual address for CB\n");
+			rc = -ENOMEM;
+			goto err_va_pool_free;
+		}
+
+		va_block = kzalloc(sizeof(*va_block), GFP_KERNEL);
+		if (!va_block) {
+			rc = -ENOMEM;
+			gen_pool_free(ctx->cb_va_pool, virt_addr, page_size);
+			goto err_va_pool_free;
+		}
+
+		va_block->start = virt_addr;
+		va_block->end = virt_addr + page_size;
+		va_block->size = page_size;
+		list_add_tail(&va_block->node, &cb->va_block_list);
+	}
+
+	mutex_lock(&ctx->mmu_lock);
+
+	bus_addr = cb->bus_address;
+	offset = 0;
+	list_for_each_entry(va_block, &cb->va_block_list, node) {
+		rc = hl_mmu_map(ctx, va_block->start, bus_addr, va_block->size,
+				list_is_last(&va_block->node,
+						&cb->va_block_list));
+		if (rc) {
+			dev_err(hdev->dev, "Failed to map VA %#llx to CB\n",
+				va_block->start);
+			goto err_va_umap;
+		}
+
+		bus_addr += va_block->size;
+		offset += va_block->size;
+	}
+
+	hdev->asic_funcs->mmu_invalidate_cache(hdev, false, VM_TYPE_USERPTR);
+
+	mutex_unlock(&ctx->mmu_lock);
+
+	cb->is_mmu_mapped = true;
+
+	return 0;
+
+err_va_umap:
+	list_for_each_entry(va_block, &cb->va_block_list, node) {
+		if (offset <= 0)
+			break;
+		hl_mmu_unmap(ctx, va_block->start, va_block->size,
+				offset <= va_block->size);
+		offset -= va_block->size;
+	}
+
+	hdev->asic_funcs->mmu_invalidate_cache(hdev, true, VM_TYPE_USERPTR);
+
+	mutex_unlock(&ctx->mmu_lock);
+
+err_va_pool_free:
+	list_for_each_entry_safe(va_block, tmp, &cb->va_block_list, node) {
+		gen_pool_free(ctx->cb_va_pool, va_block->start, va_block->size);
+		list_del(&va_block->node);
+		kfree(va_block);
+	}
+
+	return rc;
+}
+
+static void cb_unmap_mem(struct hl_ctx *ctx, struct hl_cb *cb)
+{
+	struct hl_device *hdev = ctx->hdev;
+	struct hl_vm_va_block *va_block, *tmp;
+
+	mutex_lock(&ctx->mmu_lock);
+
+	list_for_each_entry(va_block, &cb->va_block_list, node)
+		if (hl_mmu_unmap(ctx, va_block->start, va_block->size,
+				list_is_last(&va_block->node,
+						&cb->va_block_list)))
+			dev_warn_ratelimited(hdev->dev,
+					"Failed to unmap CB's va 0x%llx\n",
+					va_block->start);
+
+	hdev->asic_funcs->mmu_invalidate_cache(hdev, true, VM_TYPE_USERPTR);
+
+	mutex_unlock(&ctx->mmu_lock);
+
+	list_for_each_entry_safe(va_block, tmp, &cb->va_block_list, node) {
+		gen_pool_free(ctx->cb_va_pool, va_block->start, va_block->size);
+		list_del(&va_block->node);
+		kfree(va_block);
+	}
+}
+
 static void cb_fini(struct hl_device *hdev, struct hl_cb *cb)
 {
 	if (cb->is_internal)
@@ -47,6 +172,11 @@ static void cb_release(struct kref *ref)
 
 	hl_debugfs_remove_cb(cb);
 
+	if (cb->is_mmu_mapped)
+		cb_unmap_mem(cb->ctx, cb);
+
+	hl_ctx_put(cb->ctx);
+
 	cb_do_release(hdev, cb);
 }
 
@@ -107,11 +237,12 @@ static struct hl_cb *hl_cb_alloc(struct hl_device *hdev, u32 cb_size,
 }
 
 int hl_cb_create(struct hl_device *hdev, struct hl_cb_mgr *mgr,
-			u32 cb_size, u64 *handle, int ctx_id, bool internal_cb)
+			struct hl_ctx *ctx, u32 cb_size, bool internal_cb,
+			bool map_cb, u64 *handle)
 {
 	struct hl_cb *cb;
 	bool alloc_new_cb = true;
-	int rc;
+	int rc, ctx_id = ctx->asid;
 
 	/*
 	 * Can't use generic function to check this because of special case
@@ -163,18 +294,32 @@ int hl_cb_create(struct hl_device *hdev, struct hl_cb_mgr *mgr,
 	}
 
 	cb->hdev = hdev;
-	cb->ctx_id = ctx_id;
+	cb->ctx = ctx;
+	hl_ctx_get(hdev, cb->ctx);
+
+	if (map_cb) {
+		if (ctx_id == HL_KERNEL_ASID_ID) {
+			dev_err(hdev->dev,
+				"CB mapping is not supported for kernel context\n");
+			rc = -EINVAL;
+			goto release_cb;
+		}
+
+		rc = cb_map_mem(ctx, cb);
+		if (rc)
+			goto release_cb;
+	}
 
 	spin_lock(&mgr->cb_lock);
 	rc = idr_alloc(&mgr->cb_handles, cb, 1, 0, GFP_ATOMIC);
 	spin_unlock(&mgr->cb_lock);
 
 	if (rc < 0) {
 		dev_err(hdev->dev, "Failed to allocate IDR for a new CB\n");
-		goto release_cb;
+		goto unmap_mem;
 	}
 
-	cb->id = rc;
+	cb->id = (u64) rc;
 
 	kref_init(&cb->refcount);
 	spin_lock_init(&cb->lock);
@@ -183,14 +328,18 @@ int hl_cb_create(struct hl_device *hdev, struct hl_cb_mgr *mgr,
 	 * idr is 32-bit so we can safely OR it with a mask that is above
 	 * 32 bit
 	 */
-	*handle = cb->id | HL_MMAP_CB_MASK;
+	*handle = cb->id | HL_MMAP_TYPE_CB;
 	*handle <<= PAGE_SHIFT;
 
 	hl_debugfs_add_cb(cb);
 
 	return 0;
 
+unmap_mem:
+	if (cb->is_mmu_mapped)
+		cb_unmap_mem(cb->ctx, cb);
 release_cb:
+	hl_ctx_put(cb->ctx);
 	cb_do_release(hdev, cb);
 out_err:
 	*handle = 0;
@@ -250,9 +399,10 @@ int hl_cb_ioctl(struct hl_fpriv *hpriv, void *data)
 				args->in.cb_size, HL_MAX_CB_SIZE);
 			rc = -EINVAL;
 		} else {
-			rc = hl_cb_create(hdev, &hpriv->cb_mgr,
-					args->in.cb_size, &handle,
-					hpriv->ctx->asid, false);
+			rc = hl_cb_create(hdev, &hpriv->cb_mgr, hpriv->ctx,
+					args->in.cb_size, false,
+					!!(args->in.flags & HL_CB_FLAGS_MAP),
+					&handle);
 		}
 
 		memset(args, 0, sizeof(*args));
@@ -300,11 +450,14 @@ int hl_cb_mmap(struct hl_fpriv *hpriv, struct vm_area_struct *vma)
 {
 	struct hl_device *hdev = hpriv->hdev;
 	struct hl_cb *cb;
-	phys_addr_t address;
 	u32 handle, user_cb_size;
 	int rc;
 
+	/* We use the page offset to hold the idr and thus we need to clear
+	 * it before doing the mmap itself
+	 */
 	handle = vma->vm_pgoff;
+	vma->vm_pgoff = 0;
 
 	/* reference was taken here */
 	cb = hl_cb_get(hdev, &hpriv->cb_mgr, handle);
@@ -356,12 +509,8 @@ int hl_cb_mmap(struct hl_fpriv *hpriv, struct vm_area_struct *vma)
 
 	vma->vm_private_data = cb;
 
-	/* Calculate address for CB */
-	address = virt_to_phys((void *) (uintptr_t) cb->kernel_address);
-
-	rc = hdev->asic_funcs->cb_mmap(hdev, vma, cb->kernel_address,
-					address, cb->size);
-
+	rc = hdev->asic_funcs->cb_mmap(hdev, vma, (void *) cb->kernel_address,
+					cb->bus_address, cb->size);
 	if (rc) {
 		spin_lock(&cb->lock);
 		cb->mmap = false;
@@ -425,7 +574,7 @@ void hl_cb_mgr_fini(struct hl_device *hdev, struct hl_cb_mgr *mgr)
 		if (kref_put(&cb->refcount, cb_release) != 1)
 			dev_err(hdev->dev,
 				"CB %d for CTX ID %d is still alive\n",
-				id, cb->ctx_id);
+				id, cb->ctx->asid);
 	}
 
 	idr_destroy(&mgr->cb_handles);
@@ -438,8 +587,8 @@ struct hl_cb *hl_cb_kernel_create(struct hl_device *hdev, u32 cb_size,
 	struct hl_cb *cb;
 	int rc;
 
-	rc = hl_cb_create(hdev, &hdev->kernel_cb_mgr, cb_size, &cb_handle,
-			HL_KERNEL_ASID_ID, internal_cb);
+	rc = hl_cb_create(hdev, &hdev->kernel_cb_mgr, hdev->kernel_ctx, cb_size,
+				internal_cb, false, &cb_handle);
 	if (rc) {
 		dev_err(hdev->dev,
 			"Failed to allocate CB for the kernel driver %d\n", rc);
@@ -495,3 +644,45 @@ int hl_cb_pool_fini(struct hl_device *hdev)
 
 	return 0;
 }
+
+int hl_cb_va_pool_init(struct hl_ctx *ctx)
+{
+	struct hl_device *hdev = ctx->hdev;
+	struct asic_fixed_properties *prop = &hdev->asic_prop;
+	int rc;
+
+	if (!hdev->supports_cb_mapping)
+		return 0;
+
+	ctx->cb_va_pool = gen_pool_create(__ffs(prop->pmmu.page_size), -1);
+	if (!ctx->cb_va_pool) {
+		dev_err(hdev->dev,
+			"Failed to create VA gen pool for CB mapping\n");
+		return -ENOMEM;
+	}
+
+	rc = gen_pool_add(ctx->cb_va_pool, prop->cb_va_start_addr,
+			prop->cb_va_end_addr - prop->cb_va_start_addr, -1);
+	if (rc) {
+		dev_err(hdev->dev,
+			"Failed to add memory to VA gen pool for CB mapping\n");
+		goto err_pool_destroy;
+	}
+
+	return 0;
+
+err_pool_destroy:
+	gen_pool_destroy(ctx->cb_va_pool);
+
+	return rc;
+}
+
+void hl_cb_va_pool_fini(struct hl_ctx *ctx)
+{
+	struct hl_device *hdev = ctx->hdev;
+
+	if (!hdev->supports_cb_mapping)
+		return;
+
+	gen_pool_destroy(ctx->cb_va_pool);
+}