Commit acbabe63 authored by Ohad Sharabi's avatar Ohad Sharabi Committed by Greg Kroah-Hartman
Browse files

habanalabs: add MMU prefetch to ASIC-specific code



This is necessary pre-requisite for future ASIC support, where MMU
TLB prefetch is supported.

Signed-off-by: default avatarOhad Sharabi <osharabi@habana.ai>
Reviewed-by: default avatarOded Gabbay <ogabbay@kernel.org>
Signed-off-by: default avatarOded Gabbay <ogabbay@kernel.org>
Signed-off-by: default avatarGreg Kroah-Hartman <gregkh@linuxfoundation.org>
parent 4d530e7d
Loading
Loading
Loading
Loading
+5 −2
Original line number Diff line number Diff line
@@ -1283,8 +1283,8 @@ struct fw_load_mgr {
 * @write_pte: write MMU page table entry to DRAM.
 * @mmu_invalidate_cache: flush MMU STLB host/DRAM cache, either with soft
 *                        (L1 only) or hard (L0 & L1) flush.
 * @mmu_invalidate_cache_range: flush specific MMU STLB cache lines with
 *                              ASID-VA-size mask.
 * @mmu_invalidate_cache_range: flush specific MMU STLB cache lines with ASID-VA-size mask.
 * @mmu_prefetch_cache_range: pre-fetch specific MMU STLB cache lines with ASID-VA-size mask.
 * @send_heartbeat: send is-alive packet to CPU-CP and verify response.
 * @debug_coresight: perform certain actions on Coresight for debugging.
 * @is_device_idle: return true if device is idle, false otherwise.
@@ -1416,6 +1416,8 @@ struct hl_asic_funcs {
					u32 flags);
	int (*mmu_invalidate_cache_range)(struct hl_device *hdev, bool is_hard,
				u32 flags, u32 asid, u64 va, u64 size);
	int (*mmu_prefetch_cache_range)(struct hl_device *hdev, u32 flags, u32 asid, u64 va,
				u64 size);
	int (*send_heartbeat)(struct hl_device *hdev);
	int (*debug_coresight)(struct hl_device *hdev, struct hl_ctx *ctx, void *data);
	bool (*is_device_idle)(struct hl_device *hdev, u64 *mask_arr,
@@ -3143,6 +3145,7 @@ int hl_mmu_unmap_contiguous(struct hl_ctx *ctx, u64 virt_addr, u32 size);
int hl_mmu_invalidate_cache(struct hl_device *hdev, bool is_hard, u32 flags);
int hl_mmu_invalidate_cache_range(struct hl_device *hdev, bool is_hard,
					u32 flags, u32 asid, u64 va, u64 size);
int hl_mmu_prefetch_cache_range(struct hl_device *hdev, u32 flags, u32 asid, u64 va, u64 size);
u64 hl_mmu_get_next_hop_addr(struct hl_ctx *ctx, u64 curr_pte);
u64 hl_mmu_get_hop_pte_phys_addr(struct hl_ctx *ctx, struct hl_mmu_properties *mmu_prop,
					u8 hop_idx, u64 hop_addr, u64 virt_addr);
+10 −5
Original line number Diff line number Diff line
@@ -1241,20 +1241,23 @@ static int map_device_va(struct hl_ctx *ctx, struct hl_mem_in *args,

	rc = map_phys_pg_pack(ctx, ret_vaddr, phys_pg_pack);
	if (rc) {
		mutex_unlock(&ctx->mmu_lock);
		dev_err(hdev->dev, "mapping page pack failed for handle %u\n",
				handle);
		dev_err(hdev->dev, "mapping page pack failed for handle %u\n", handle);
		goto map_err;
	}

	rc = hl_mmu_invalidate_cache_range(hdev, false, *vm_type | MMU_OP_SKIP_LOW_CACHE_INV,
				ctx->asid, ret_vaddr, phys_pg_pack->total_size);
	if (rc)
		goto map_err;

	mutex_unlock(&ctx->mmu_lock);

	/* already prefetch the relevant translations to the cache */
	rc = hl_mmu_prefetch_cache_range(hdev, *vm_type, ctx->asid, ret_vaddr,
						phys_pg_pack->total_size);
	if (rc)
		goto map_err;

	mutex_unlock(&ctx->mmu_lock);

	ret_vaddr += phys_pg_pack->offset;

	hnode->ptr = vm_type;
@@ -1272,6 +1275,8 @@ static int map_device_va(struct hl_ctx *ctx, struct hl_mem_in *args,
	return rc;

map_err:
	mutex_unlock(&ctx->mmu_lock);

	if (add_va_block(hdev, va_range, ret_vaddr,
				ret_vaddr + phys_pg_pack->total_size - 1))
		dev_warn(hdev->dev,
+11 −0
Original line number Diff line number Diff line
@@ -665,6 +665,17 @@ int hl_mmu_invalidate_cache_range(struct hl_device *hdev, bool is_hard,
	return rc;
}

int hl_mmu_prefetch_cache_range(struct hl_device *hdev, u32 flags, u32 asid, u64 va, u64 size)
{
	int rc;

	rc = hdev->asic_funcs->mmu_prefetch_cache_range(hdev, flags, asid, va, size);
	if (rc)
		dev_err_ratelimited(hdev->dev, "MMU cache range prefetch failed\n");

	return rc;
}

u64 hl_mmu_get_next_hop_addr(struct hl_ctx *ctx, u64 curr_pte)
{
	return (curr_pte & PAGE_PRESENT_MASK) ? (curr_pte & HOP_PHYS_ADDR_MASK) : ULLONG_MAX;
+7 −0
Original line number Diff line number Diff line
@@ -9377,6 +9377,12 @@ static void gaudi_get_valid_dram_page_orders(struct hl_info_dev_memalloc_page_si
	info->page_order_bitmask = 0;
}

static int gaudi_mmu_prefetch_cache_range(struct hl_device *hdev, u32 flags, u32 asid, u64 va,
						u64 size)
{
	return 0;
}

static ssize_t infineon_ver_show(struct device *dev, struct device_attribute *attr, char *buf)
{
	struct hl_device *hdev = dev_get_drvdata(dev);
@@ -9444,6 +9450,7 @@ static const struct hl_asic_funcs gaudi_funcs = {
	.write_pte = gaudi_write_pte,
	.mmu_invalidate_cache = gaudi_mmu_invalidate_cache,
	.mmu_invalidate_cache_range = gaudi_mmu_invalidate_cache_range,
	.mmu_prefetch_cache_range = gaudi_mmu_prefetch_cache_range,
	.send_heartbeat = gaudi_send_heartbeat,
	.debug_coresight = gaudi_debug_coresight,
	.is_device_idle = gaudi_is_device_idle,
+7 −0
Original line number Diff line number Diff line
@@ -5680,6 +5680,12 @@ static void goya_get_valid_dram_page_orders(struct hl_info_dev_memalloc_page_siz
	info->page_order_bitmask = 0;
}

static int goya_mmu_prefetch_cache_range(struct hl_device *hdev, u32 flags, u32 asid, u64 va,
					u64 size)
{
	return 0;
}

static const struct hl_asic_funcs goya_funcs = {
	.early_init = goya_early_init,
	.early_fini = goya_early_fini,
@@ -5724,6 +5730,7 @@ static const struct hl_asic_funcs goya_funcs = {
	.write_pte = goya_write_pte,
	.mmu_invalidate_cache = goya_mmu_invalidate_cache,
	.mmu_invalidate_cache_range = goya_mmu_invalidate_cache_range,
	.mmu_prefetch_cache_range = goya_mmu_prefetch_cache_range,
	.send_heartbeat = goya_send_heartbeat,
	.debug_coresight = goya_debug_coresight,
	.is_device_idle = goya_is_device_idle,