Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id F25ECEB64D7 for ; Fri, 16 Jun 2023 15:38:32 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id A6A5910E64B; Fri, 16 Jun 2023 15:38:32 +0000 (UTC) Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by gabe.freedesktop.org (Postfix) with ESMTPS id 5D36210E64B for ; Fri, 16 Jun 2023 15:38:31 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1686929911; x=1718465911; h=message-id:date:mime-version:subject:to:references:from: in-reply-to:content-transfer-encoding; bh=76J3uY0OCIfPym0DwaghfD7+3OmhMkpWR6rqKyYWfrM=; b=jhYJbWBzqkq9RUfHu+U0a1N8Klx6pzQz7TUuU0lx5pdP2OQUaZjvCbPc BOa+WiZAforMzjoS5gLgIi66pk6GO9H65yJ4N1h04yN9U0+6rDd+zQu/E 54ySl7pDpUv+Bj/i6pO4I1f/JgePIYTsyn40g6U1ZjYngFe+gr/LYwwm7 vqT9BpDSM2v5U7d8o3D0X8TkNJU4YBM4riDXoe7CcLqpOu7kq1QmYJTZL Zh6gdeDEfmrb5LN1spd2PrSL6mxyxaioXVeXW3zcttb4VTbhHrkBKBRtd 53x3XvbvjFJPtAWKRxzegrCGzHsUQfJMkzEbB5U9ByX9DpMRcfI13l3ci w==; X-IronPort-AV: E=McAfee;i="6600,9927,10742"; a="356732924" X-IronPort-AV: E=Sophos;i="6.00,247,1681196400"; d="scan'208";a="356732924" Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 16 Jun 2023 08:38:28 -0700 X-IronPort-AV: E=McAfee;i="6600,9927,10742"; a="746324048" X-IronPort-AV: E=Sophos;i="6.00,247,1681196400"; d="scan'208";a="746324048" Received: from nurha65x-mobl2.gar.corp.intel.com (HELO [10.249.254.148]) ([10.249.254.148]) by orsmga001-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 16 Jun 2023 08:38:27 -0700 Message-ID: Date: Fri, 16 Jun 2023 17:38:24 +0200 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101 Thunderbird/102.11.1 Content-Language: en-US To: Matthew Brost , intel-xe@lists.freedesktop.org References: <20230615183455.290494-1-matthew.brost@intel.com> From: =?UTF-8?Q?Thomas_Hellstr=c3=b6m?= In-Reply-To: <20230615183455.290494-1-matthew.brost@intel.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Subject: Re: [Intel-xe] [PATCH] drm/xe: NULL binding implementation X-BeenThere: intel-xe@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel Xe graphics driver List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" On 6/15/23 20:34, Matthew Brost wrote: > Add uAPI and implementation for NULL bindings. A NULL binding is defined > as writes dropped and read zero. A single bit in the uAPI has been added > which results in a single bit in the PTEs being set. > > NULL bindings are intendedd to be used to implement VK sparse bindings, > in particular residencyNonResidentStrict property. > > v2: Fix BUG_ON shown in VK testing, fix check patch warning, fix > xe_pt_scan_64K, update __gen8_pte_encode to understand NULL bindings, > remove else if vma_addr > > Suggested-by: Paulo Zanoni > Signed-off-by: Matthew Brost Reviewed-by: Thomas Hellström > --- > drivers/gpu/drm/xe/xe_bo.h | 1 + > drivers/gpu/drm/xe/xe_exec.c | 2 + > drivers/gpu/drm/xe/xe_gt_pagefault.c | 4 +- > drivers/gpu/drm/xe/xe_pt.c | 54 ++++++++++----- > drivers/gpu/drm/xe/xe_vm.c | 99 ++++++++++++++++++---------- > drivers/gpu/drm/xe/xe_vm.h | 12 +++- > drivers/gpu/drm/xe/xe_vm_types.h | 1 + > include/uapi/drm/xe_drm.h | 8 +++ > 8 files changed, 126 insertions(+), 55 deletions(-) > > diff --git a/drivers/gpu/drm/xe/xe_bo.h b/drivers/gpu/drm/xe/xe_bo.h > index dd3d448fee0b..3a148cc6e811 100644 > --- a/drivers/gpu/drm/xe/xe_bo.h > +++ b/drivers/gpu/drm/xe/xe_bo.h > @@ -61,6 +61,7 @@ > #define XE_PPGTT_PTE_LM BIT_ULL(11) > #define XE_PDE_64K BIT_ULL(6) > #define XE_PTE_PS64 BIT_ULL(8) > +#define XE_PTE_NULL BIT_ULL(9) > > #define XE_PAGE_PRESENT BIT_ULL(0) > #define XE_PAGE_RW BIT_ULL(1) > diff --git a/drivers/gpu/drm/xe/xe_exec.c b/drivers/gpu/drm/xe/xe_exec.c > index e44076ee2e11..4f7694a29348 100644 > --- a/drivers/gpu/drm/xe/xe_exec.c > +++ b/drivers/gpu/drm/xe/xe_exec.c > @@ -120,6 +120,8 @@ static int xe_exec_begin(struct xe_engine *e, struct ww_acquire_ctx *ww, > * to a location where the GPU can access it). > */ > list_for_each_entry(vma, &vm->rebind_list, rebind_link) { > + XE_WARN_ON(xe_vma_is_null(vma)); > + > if (xe_vma_is_userptr(vma)) > continue; > > diff --git a/drivers/gpu/drm/xe/xe_gt_pagefault.c b/drivers/gpu/drm/xe/xe_gt_pagefault.c > index 73db7f7c0381..6faebd02f3fb 100644 > --- a/drivers/gpu/drm/xe/xe_gt_pagefault.c > +++ b/drivers/gpu/drm/xe/xe_gt_pagefault.c > @@ -533,8 +533,8 @@ static int handle_acc(struct xe_gt *gt, struct acc *acc) > > trace_xe_vma_acc(vma); > > - /* Userptr can't be migrated, nothing to do */ > - if (xe_vma_is_userptr(vma)) > + /* Userptr or null can't be migrated, nothing to do */ > + if (xe_vma_has_no_bo(vma)) > goto unlock_vm; > > /* Lock VM and BOs dma-resv */ > diff --git a/drivers/gpu/drm/xe/xe_pt.c b/drivers/gpu/drm/xe/xe_pt.c > index 29c1b1f0bd7c..fe1c77b139e4 100644 > --- a/drivers/gpu/drm/xe/xe_pt.c > +++ b/drivers/gpu/drm/xe/xe_pt.c > @@ -81,6 +81,9 @@ u64 xe_pde_encode(struct xe_bo *bo, u64 bo_offset, > static dma_addr_t vma_addr(struct xe_vma *vma, u64 offset, > size_t page_size, bool *is_vram) > { > + if (xe_vma_is_null(vma)) > + return 0; > + > if (xe_vma_is_userptr(vma)) { > struct xe_res_cursor cur; > u64 page; > @@ -105,6 +108,9 @@ static u64 __pte_encode(u64 pte, enum xe_cache_level cache, u32 flags, > if (unlikely(flags & XE_PTE_FLAG_READ_ONLY)) > pte &= ~XE_PAGE_RW; > > + if (unlikely(flags & XE_PTE_FLAG_NULL)) > + pte |= XE_PTE_NULL; > + > /* FIXME: I don't think the PPAT handling is correct for MTL */ > > switch (cache) { > @@ -557,6 +563,10 @@ static bool xe_pt_hugepte_possible(u64 addr, u64 next, unsigned int level, > if (next - xe_walk->va_curs_start > xe_walk->curs->size) > return false; > > + /* null VMA's do not have dma addresses */ > + if (xe_walk->pte_flags & XE_PTE_FLAG_NULL) > + return true; > + > /* Is the DMA address huge PTE size aligned? */ > size = next - addr; > dma = addr - xe_walk->va_curs_start + xe_res_dma(xe_walk->curs); > @@ -579,6 +589,10 @@ xe_pt_scan_64K(u64 addr, u64 next, struct xe_pt_stage_bind_walk *xe_walk) > if (next > xe_walk->l0_end_addr) > return false; > > + /* null VMA's do not have dma addresses */ > + if (xe_walk->pte_flags & XE_PTE_FLAG_NULL) > + return true; > + > xe_res_next(&curs, addr - xe_walk->va_curs_start); > for (; addr < next; addr += SZ_64K) { > if (!IS_ALIGNED(xe_res_dma(&curs), SZ_64K) || curs.size < SZ_64K) > @@ -629,10 +643,12 @@ xe_pt_stage_bind_entry(struct xe_ptw *parent, pgoff_t offset, > /* Is this a leaf entry ?*/ > if (level == 0 || xe_pt_hugepte_possible(addr, next, level, xe_walk)) { > struct xe_res_cursor *curs = xe_walk->curs; > + bool is_null = xe_walk->pte_flags & XE_PTE_FLAG_NULL; > > XE_WARN_ON(xe_walk->va_curs_start != addr); > > - pte = __pte_encode(xe_res_dma(curs) + xe_walk->dma_offset, > + pte = __pte_encode(is_null ? 0 : > + xe_res_dma(curs) + xe_walk->dma_offset, > xe_walk->cache, xe_walk->pte_flags, > level); > pte |= xe_walk->default_pte; > @@ -652,7 +668,8 @@ xe_pt_stage_bind_entry(struct xe_ptw *parent, pgoff_t offset, > if (unlikely(ret)) > return ret; > > - xe_res_next(curs, next - addr); > + if (!is_null) > + xe_res_next(curs, next - addr); > xe_walk->va_curs_start = next; > *action = ACTION_CONTINUE; > > @@ -759,24 +776,29 @@ xe_pt_stage_bind(struct xe_tile *tile, struct xe_vma *vma, > xe_walk.dma_offset = vram_region_gpu_offset(bo->ttm.resource); > xe_walk.cache = XE_CACHE_WB; > } else { > - if (!xe_vma_is_userptr(vma) && bo->flags & XE_BO_SCANOUT_BIT) > + if (!xe_vma_has_no_bo(vma) && bo->flags & XE_BO_SCANOUT_BIT) > xe_walk.cache = XE_CACHE_WT; > else > xe_walk.cache = XE_CACHE_WB; > } > - if (!xe_vma_is_userptr(vma) && xe_bo_is_stolen(bo)) > + if (!xe_vma_has_no_bo(vma) && xe_bo_is_stolen(bo)) > xe_walk.dma_offset = xe_ttm_stolen_gpu_offset(xe_bo_device(bo)); > > xe_bo_assert_held(bo); > - if (xe_vma_is_userptr(vma)) > - xe_res_first_sg(vma->userptr.sg, 0, vma->end - vma->start + 1, > - &curs); > - else if (xe_bo_is_vram(bo) || xe_bo_is_stolen(bo)) > - xe_res_first(bo->ttm.resource, vma->bo_offset, > - vma->end - vma->start + 1, &curs); > - else > - xe_res_first_sg(xe_bo_get_sg(bo), vma->bo_offset, > - vma->end - vma->start + 1, &curs); > + > + if (!xe_vma_is_null(vma)) { > + if (xe_vma_is_userptr(vma)) > + xe_res_first_sg(vma->userptr.sg, 0, > + vma->end - vma->start + 1, &curs); > + else if (xe_bo_is_vram(bo) || xe_bo_is_stolen(bo)) > + xe_res_first(bo->ttm.resource, vma->bo_offset, > + vma->end - vma->start + 1, &curs); > + else > + xe_res_first_sg(xe_bo_get_sg(bo), vma->bo_offset, > + vma->end - vma->start + 1, &curs); > + } else { > + curs.size = vma->end - vma->start + 1; > + } > > ret = xe_pt_walk_range(&pt->base, pt->level, vma->start, vma->end + 1, > &xe_walk.base); > @@ -965,7 +987,7 @@ static void xe_pt_commit_locks_assert(struct xe_vma *vma) > > if (xe_vma_is_userptr(vma)) > lockdep_assert_held_read(&vm->userptr.notifier_lock); > - else > + else if (!xe_vma_is_null(vma)) > dma_resv_assert_held(vma->bo->ttm.base.resv); > > dma_resv_assert_held(&vm->resv); > @@ -1341,7 +1363,7 @@ __xe_pt_bind_vma(struct xe_tile *tile, struct xe_vma *vma, struct xe_engine *e, > DMA_RESV_USAGE_KERNEL : > DMA_RESV_USAGE_BOOKKEEP); > > - if (!xe_vma_is_userptr(vma) && !vma->bo->vm) > + if (!xe_vma_has_no_bo(vma) && !vma->bo->vm) > dma_resv_add_fence(vma->bo->ttm.base.resv, fence, > DMA_RESV_USAGE_BOOKKEEP); > xe_pt_commit_bind(vma, entries, num_entries, rebind, > @@ -1658,7 +1680,7 @@ __xe_pt_unbind_vma(struct xe_tile *tile, struct xe_vma *vma, struct xe_engine *e > DMA_RESV_USAGE_BOOKKEEP); > > /* This fence will be installed by caller when doing eviction */ > - if (!xe_vma_is_userptr(vma) && !vma->bo->vm) > + if (!xe_vma_has_no_bo(vma) && !vma->bo->vm) > dma_resv_add_fence(vma->bo->ttm.base.resv, fence, > DMA_RESV_USAGE_BOOKKEEP); > xe_pt_commit_unbind(vma, entries, num_entries, > diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c > index 56ed2cb75c8e..113853d496c7 100644 > --- a/drivers/gpu/drm/xe/xe_vm.c > +++ b/drivers/gpu/drm/xe/xe_vm.c > @@ -590,7 +590,7 @@ static void preempt_rebind_work_func(struct work_struct *w) > goto out_unlock; > > list_for_each_entry(vma, &vm->rebind_list, rebind_link) { > - if (xe_vma_is_userptr(vma) || vma->destroyed) > + if (xe_vma_has_no_bo(vma) || vma->destroyed) > continue; > > err = xe_bo_validate(vma->bo, vm, false); > @@ -843,6 +843,7 @@ static struct xe_vma *xe_vma_create(struct xe_vm *vm, > u64 bo_offset_or_userptr, > u64 start, u64 end, > bool read_only, > + bool is_null, > u64 tile_mask) > { > struct xe_vma *vma; > @@ -868,8 +869,11 @@ static struct xe_vma *xe_vma_create(struct xe_vm *vm, > vma->vm = vm; > vma->start = start; > vma->end = end; > + vma->pte_flags = 0; > if (read_only) > - vma->pte_flags = XE_PTE_FLAG_READ_ONLY; > + vma->pte_flags |= XE_PTE_FLAG_READ_ONLY; > + if (is_null) > + vma->pte_flags |= XE_PTE_FLAG_NULL; > > if (tile_mask) { > vma->tile_mask = tile_mask; > @@ -886,23 +890,26 @@ static struct xe_vma *xe_vma_create(struct xe_vm *vm, > vma->bo_offset = bo_offset_or_userptr; > vma->bo = xe_bo_get(bo); > list_add_tail(&vma->bo_link, &bo->vmas); > - } else /* userptr */ { > - u64 size = end - start + 1; > - int err; > + } else /* userptr or null */ { > + if (!is_null) { > + u64 size = end - start + 1; > + int err; > > - vma->userptr.ptr = bo_offset_or_userptr; > + vma->userptr.ptr = bo_offset_or_userptr; > > - err = mmu_interval_notifier_insert(&vma->userptr.notifier, > - current->mm, > - vma->userptr.ptr, size, > - &vma_userptr_notifier_ops); > - if (err) { > - kfree(vma); > - vma = ERR_PTR(err); > - return vma; > + err = mmu_interval_notifier_insert(&vma->userptr.notifier, > + current->mm, > + vma->userptr.ptr, size, > + &vma_userptr_notifier_ops); > + if (err) { > + kfree(vma); > + vma = ERR_PTR(err); > + return vma; > + } > + > + vma->userptr.notifier_seq = LONG_MAX; > } > > - vma->userptr.notifier_seq = LONG_MAX; > xe_vm_get(vm); > } > > @@ -942,6 +949,8 @@ static void xe_vma_destroy_late(struct xe_vma *vma) > */ > mmu_interval_notifier_remove(&vma->userptr.notifier); > xe_vm_put(vm); > + } else if (xe_vma_is_null(vma)) { > + xe_vm_put(vm); > } else { > xe_bo_put(vma->bo); > } > @@ -1024,7 +1033,7 @@ static void xe_vma_destroy(struct xe_vma *vma, struct dma_fence *fence) > list_del_init(&vma->userptr.invalidate_link); > spin_unlock(&vm->userptr.invalidated_lock); > list_del(&vma->userptr_link); > - } else { > + } else if (!xe_vma_is_null(vma)) { > xe_bo_assert_held(vma->bo); > list_del(&vma->bo_link); > > @@ -1393,7 +1402,7 @@ void xe_vm_close_and_put(struct xe_vm *vm) > while (vm->vmas.rb_node) { > struct xe_vma *vma = to_xe_vma(vm->vmas.rb_node); > > - if (xe_vma_is_userptr(vma)) { > + if (xe_vma_has_no_bo(vma)) { > down_read(&vm->userptr.notifier_lock); > vma->destroyed = true; > up_read(&vm->userptr.notifier_lock); > @@ -1402,7 +1411,7 @@ void xe_vm_close_and_put(struct xe_vm *vm) > rb_erase(&vma->vm_node, &vm->vmas); > > /* easy case, remove from VMA? */ > - if (xe_vma_is_userptr(vma) || vma->bo->vm) { > + if (xe_vma_has_no_bo(vma) || vma->bo->vm) { > xe_vma_destroy(vma, NULL); > continue; > } > @@ -2036,7 +2045,7 @@ static int xe_vm_prefetch(struct xe_vm *vm, struct xe_vma *vma, > > XE_BUG_ON(region > ARRAY_SIZE(region_to_mem_type)); > > - if (!xe_vma_is_userptr(vma)) { > + if (!xe_vma_has_no_bo(vma)) { > err = xe_bo_migrate(vma->bo, region_to_mem_type[region]); > if (err) > return err; > @@ -2645,6 +2654,8 @@ static struct xe_vma *vm_unbind_lookup_vmas(struct xe_vm *vm, > lookup->start - 1, > (first->pte_flags & > XE_PTE_FLAG_READ_ONLY), > + (first->pte_flags & > + XE_PTE_FLAG_NULL), > first->tile_mask); > if (first->bo) > xe_bo_unlock(first->bo, &ww); > @@ -2652,7 +2663,7 @@ static struct xe_vma *vm_unbind_lookup_vmas(struct xe_vm *vm, > err = -ENOMEM; > goto unwind; > } > - if (!first->bo) { > + if (xe_vma_is_userptr(first)) { > err = xe_vma_userptr_pin_pages(new_first); > if (err) > goto unwind; > @@ -2677,6 +2688,7 @@ static struct xe_vma *vm_unbind_lookup_vmas(struct xe_vm *vm, > last->end, > (last->pte_flags & > XE_PTE_FLAG_READ_ONLY), > + (last->pte_flags & XE_PTE_FLAG_NULL), > last->tile_mask); > if (last->bo) > xe_bo_unlock(last->bo, &ww); > @@ -2684,7 +2696,7 @@ static struct xe_vma *vm_unbind_lookup_vmas(struct xe_vm *vm, > err = -ENOMEM; > goto unwind; > } > - if (!last->bo) { > + if (xe_vma_is_userptr(last)) { > err = xe_vma_userptr_pin_pages(new_last); > if (err) > goto unwind; > @@ -2744,7 +2756,7 @@ static struct xe_vma *vm_prefetch_lookup_vmas(struct xe_vm *vm, > *next; > struct rb_node *node; > > - if (!xe_vma_is_userptr(vma)) { > + if (!xe_vma_has_no_bo(vma)) { > if (!xe_bo_can_migrate(vma->bo, region_to_mem_type[region])) > return ERR_PTR(-EINVAL); > } > @@ -2753,7 +2765,7 @@ static struct xe_vma *vm_prefetch_lookup_vmas(struct xe_vm *vm, > while ((node = rb_next(node))) { > if (!xe_vma_cmp_vma_cb(lookup, node)) { > __vma = to_xe_vma(node); > - if (!xe_vma_is_userptr(__vma)) { > + if (!xe_vma_has_no_bo(__vma)) { > if (!xe_bo_can_migrate(__vma->bo, region_to_mem_type[region])) > goto flush_list; > } > @@ -2767,7 +2779,7 @@ static struct xe_vma *vm_prefetch_lookup_vmas(struct xe_vm *vm, > while ((node = rb_prev(node))) { > if (!xe_vma_cmp_vma_cb(lookup, node)) { > __vma = to_xe_vma(node); > - if (!xe_vma_is_userptr(__vma)) { > + if (!xe_vma_has_no_bo(__vma)) { > if (!xe_bo_can_migrate(__vma->bo, region_to_mem_type[region])) > goto flush_list; > } > @@ -2826,21 +2838,23 @@ static struct xe_vma *vm_bind_ioctl_lookup_vma(struct xe_vm *vm, > > switch (VM_BIND_OP(op)) { > case XE_VM_BIND_OP_MAP: > - XE_BUG_ON(!bo); > - > - err = xe_bo_lock(bo, &ww, 0, true); > - if (err) > - return ERR_PTR(err); > + if (bo) { > + err = xe_bo_lock(bo, &ww, 0, true); > + if (err) > + return ERR_PTR(err); > + } > vma = xe_vma_create(vm, bo, bo_offset_or_userptr, addr, > addr + range - 1, > op & XE_VM_BIND_FLAG_READONLY, > + op & XE_VM_BIND_FLAG_NULL, > tile_mask); > - xe_bo_unlock(bo, &ww); > + if (bo) > + xe_bo_unlock(bo, &ww); > if (!vma) > return ERR_PTR(-ENOMEM); > > xe_vm_insert_vma(vm, vma); > - if (!bo->vm) { > + if (bo && !bo->vm) { > vm_insert_extobj(vm, vma); > err = add_preempt_fences(vm, bo); > if (err) { > @@ -2874,6 +2888,7 @@ static struct xe_vma *vm_bind_ioctl_lookup_vma(struct xe_vm *vm, > vma = xe_vma_create(vm, NULL, bo_offset_or_userptr, addr, > addr + range - 1, > op & XE_VM_BIND_FLAG_READONLY, > + op & XE_VM_BIND_FLAG_NULL, > tile_mask); > if (!vma) > return ERR_PTR(-ENOMEM); > @@ -2899,11 +2914,12 @@ static struct xe_vma *vm_bind_ioctl_lookup_vma(struct xe_vm *vm, > #ifdef TEST_VM_ASYNC_OPS_ERROR > #define SUPPORTED_FLAGS \ > (FORCE_ASYNC_OP_ERROR | XE_VM_BIND_FLAG_ASYNC | \ > - XE_VM_BIND_FLAG_READONLY | XE_VM_BIND_FLAG_IMMEDIATE | 0xffff) > + XE_VM_BIND_FLAG_READONLY | XE_VM_BIND_FLAG_IMMEDIATE | \ > + XE_VM_BIND_FLAG_NULL | 0xffff) > #else > #define SUPPORTED_FLAGS \ > (XE_VM_BIND_FLAG_ASYNC | XE_VM_BIND_FLAG_READONLY | \ > - XE_VM_BIND_FLAG_IMMEDIATE | 0xffff) > + XE_VM_BIND_FLAG_IMMEDIATE | XE_VM_BIND_FLAG_NULL | 0xffff) > #endif > #define XE_64K_PAGE_MASK 0xffffull > > @@ -2951,6 +2967,7 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, > u32 obj = (*bind_ops)[i].obj; > u64 obj_offset = (*bind_ops)[i].obj_offset; > u32 region = (*bind_ops)[i].region; > + bool is_null = op & XE_VM_BIND_FLAG_NULL; > > if (XE_IOCTL_ERR(xe, (*bind_ops)[i].pad) || > XE_IOCTL_ERR(xe, (*bind_ops)[i].reserved[0] || > @@ -2984,8 +3001,13 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, > if (XE_IOCTL_ERR(xe, VM_BIND_OP(op) > > XE_VM_BIND_OP_PREFETCH) || > XE_IOCTL_ERR(xe, op & ~SUPPORTED_FLAGS) || > + XE_IOCTL_ERR(xe, obj && is_null) || > + XE_IOCTL_ERR(xe, obj_offset && is_null) || > + XE_IOCTL_ERR(xe, VM_BIND_OP(op) != XE_VM_BIND_OP_MAP && > + is_null) || > XE_IOCTL_ERR(xe, !obj && > - VM_BIND_OP(op) == XE_VM_BIND_OP_MAP) || > + VM_BIND_OP(op) == XE_VM_BIND_OP_MAP && > + !is_null) || > XE_IOCTL_ERR(xe, !obj && > VM_BIND_OP(op) == XE_VM_BIND_OP_UNMAP_ALL) || > XE_IOCTL_ERR(xe, addr && > @@ -3390,6 +3412,7 @@ int xe_vm_invalidate_vma(struct xe_vma *vma) > int ret; > > XE_BUG_ON(!xe_vm_in_fault_mode(vma->vm)); > + XE_WARN_ON(xe_vma_is_null(vma)); > trace_xe_vma_usm_invalidate(vma); > > /* Check that we don't race with page-table updates */ > @@ -3452,8 +3475,11 @@ int xe_analyze_vm(struct drm_printer *p, struct xe_vm *vm, int gt_id) > for (node = rb_first(&vm->vmas); node; node = rb_next(node)) { > struct xe_vma *vma = to_xe_vma(node); > bool is_userptr = xe_vma_is_userptr(vma); > + bool is_null = xe_vma_is_null(vma); > > - if (is_userptr) { > + if (is_null) { > + addr = 0; > + } else if (is_userptr) { > struct xe_res_cursor cur; > > if (vma->userptr.sg) { > @@ -3468,7 +3494,8 @@ int xe_analyze_vm(struct drm_printer *p, struct xe_vm *vm, int gt_id) > } > drm_printf(p, " [%016llx-%016llx] S:0x%016llx A:%016llx %s\n", > vma->start, vma->end, vma->end - vma->start + 1ull, > - addr, is_userptr ? "USR" : is_vram ? "VRAM" : "SYS"); > + addr, is_null ? "NULL" : is_userptr ? "USR" : > + is_vram ? "VRAM" : "SYS"); > } > up_read(&vm->lock); > > diff --git a/drivers/gpu/drm/xe/xe_vm.h b/drivers/gpu/drm/xe/xe_vm.h > index bb2996856841..5edb7771629c 100644 > --- a/drivers/gpu/drm/xe/xe_vm.h > +++ b/drivers/gpu/drm/xe/xe_vm.h > @@ -115,11 +115,21 @@ static inline void xe_vm_reactivate_rebind(struct xe_vm *vm) > } > } > > -static inline bool xe_vma_is_userptr(struct xe_vma *vma) > +static inline bool xe_vma_is_null(struct xe_vma *vma) > +{ > + return vma->pte_flags & XE_PTE_FLAG_NULL; > +} > + > +static inline bool xe_vma_has_no_bo(struct xe_vma *vma) > { > return !vma->bo; > } > > +static inline bool xe_vma_is_userptr(struct xe_vma *vma) > +{ > + return xe_vma_has_no_bo(vma) && !xe_vma_is_null(vma); > +} > + > int xe_vma_userptr_pin_pages(struct xe_vma *vma); > > int xe_vma_userptr_check_repin(struct xe_vma *vma); > diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h > index a51e84e584b4..9b39c5f64afa 100644 > --- a/drivers/gpu/drm/xe/xe_vm_types.h > +++ b/drivers/gpu/drm/xe/xe_vm_types.h > @@ -31,6 +31,7 @@ struct xe_vma { > u64 end; > /** @pte_flags: pte flags for this VMA */ > #define XE_PTE_FLAG_READ_ONLY BIT(0) > +#define XE_PTE_FLAG_NULL BIT(1) > u32 pte_flags; > > /** @bo: BO if not a userptr, must be NULL is userptr */ > diff --git a/include/uapi/drm/xe_drm.h b/include/uapi/drm/xe_drm.h > index 7f9a5849e88b..9bdea8b81435 100644 > --- a/include/uapi/drm/xe_drm.h > +++ b/include/uapi/drm/xe_drm.h > @@ -418,6 +418,14 @@ struct drm_xe_vm_bind_op { > * than differing the MAP to the page fault handler. > */ > #define XE_VM_BIND_FLAG_IMMEDIATE (0x1 << 18) > + /* > + * When the NULL flag is set, the page tables are setup with a special > + * bit which indicates writes are dropped and all reads return zero. In > + * the future, the NULL flags will only be valid for XE_VM_BIND_OP_MAP > + * operations, the BO handle MBZ, and the BO offset MBZ. This flag is > + * intended to implement VK sparse bindings. > + */ > +#define XE_VM_BIND_FLAG_NULL (0x1 << 19) > > /** @reserved: Reserved */ > __u64 reserved[2];