From mboxrd@z Thu Jan 1 00:00:00 1970 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 672A5C021B0 for ; Wed, 19 Feb 2025 10:30:38 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 147BA10E7B4; Wed, 19 Feb 2025 10:30:38 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.b="QUIQ8eOM"; dkim-atps=neutral Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) by gabe.freedesktop.org (Postfix) with ESMTPS id EA86410E7B6; Wed, 19 Feb 2025 10:30:36 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1739961037; x=1771497037; h=message-id:subject:from:to:cc:date:in-reply-to: references:content-transfer-encoding:mime-version; bh=r7kOeQ4EAJQnOLbWGHQTB7nX8N81vpekWzcZ6id/7aU=; b=QUIQ8eOMsEvVO4NKoYSjP3w4oMP8hQqi9upF2rqoqhP4FxIbZi9VaL92 OfHNbYVZ4Ab0Qj7zppS/5gwMVjDkumtIvHHxgOuCJ+TjFi5lg4cZjSbBn rvRcsNkIYH1C9Lik3Y1hoh61xifkcA+3TJ6CBYnyXwuQFjC6YcynE6Cj6 ym3+oOBC2kkgL1O4X2nGcJ9C07Iod/1LxtCNm2Doovf3orfGlzK0XsC3V Zu/C1CpkdbrEq6pWCfSFmgqk+GWudahBA+FrDnni0A6VmfiQ+lNbcCVku WPM6QPgeeJjYvx2ntyrp8FyhIpbsLsv3gMsw8Ies02S6m/+Mo9lZ0xzb6 Q==; X-CSE-ConnectionGUID: 0Cq4kzk8TeuYwfhf2AE1lg== X-CSE-MsgGUID: CF4esyyzRyGPYyi5chw7Xw== X-IronPort-AV: E=McAfee;i="6700,10204,11348"; a="58230458" X-IronPort-AV: E=Sophos;i="6.13,298,1732608000"; d="scan'208";a="58230458" Received: from orviesa002.jf.intel.com ([10.64.159.142]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 19 Feb 2025 02:30:36 -0800 X-CSE-ConnectionGUID: fHFuazXMTdSDW3eA167gLQ== X-CSE-MsgGUID: r00N9BCbTImy/JAh0WnMFQ== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.13,298,1732608000"; d="scan'208";a="145526559" Received: from lfiedoro-mobl.ger.corp.intel.com (HELO [10.245.246.115]) ([10.245.246.115]) by orviesa002-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 19 Feb 2025 02:30:33 -0800 Message-ID: <55808e272d7d8e9302912f236c249fea11422061.camel@linux.intel.com> Subject: Re: [PATCH v5 27/32] drm/xe: Add SVM VRAM migration From: Thomas =?ISO-8859-1?Q?Hellstr=F6m?= To: Matthew Brost , intel-xe@lists.freedesktop.org, dri-devel@lists.freedesktop.org Cc: himal.prasad.ghimiray@intel.com, apopple@nvidia.com, airlied@gmail.com, simona.vetter@ffwll.ch, felix.kuehling@amd.com, dakr@kernel.org Date: Wed, 19 Feb 2025 11:30:31 +0100 In-Reply-To: <20250213021112.1228481-28-matthew.brost@intel.com> References: <20250213021112.1228481-1-matthew.brost@intel.com> <20250213021112.1228481-28-matthew.brost@intel.com> Organization: Intel Sweden AB, Registration Number: 556189-6027 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable User-Agent: Evolution 3.54.3 (3.54.3-1.fc41) MIME-Version: 1.0 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 Wed, 2025-02-12 at 18:11 -0800, Matthew Brost wrote: > Migration is implemented with range granularity, with VRAM backing > being > a VM private TTM BO (i.e., shares dma-resv with VM). The lifetime of > the > TTM BO is limited to when the SVM range is in VRAM (i.e., when a VRAM > SVM range is migrated to SRAM, the TTM BO is destroyed). >=20 > The design choice for using TTM BO for VRAM backing store, as opposed > to > direct buddy allocation, is as follows: >=20 > - DRM buddy allocations are not at page granularity, offering no > =C2=A0 advantage over a BO. > - Unified eviction is required (SVM VRAM and TTM BOs need to be able > to > =C2=A0 evict each other). > - For exhaustive eviction [1], SVM VRAM allocations will almost > certainly > =C2=A0 require a dma-resv. > - Likely allocation size is 2M which makes of size of BO (872) > =C2=A0 acceptable per allocation (872 / 2M =3D=3D .0004158). >=20 > With this, using TTM BO for VRAM backing store seems to be an obvious > choice as it allows leveraging of the TTM eviction code. >=20 > Current migration policy is migrate any SVM range greater than or > equal > to 64k once. >=20 > [1] https://patchwork.freedesktop.org/series/133643/ >=20 > v2: > =C2=A0- Rebase on latest GPU SVM > =C2=A0- Retry page fault on get pages returning mixed allocation > =C2=A0- Use drm_gpusvm_devmem > v3: > =C2=A0- Use new BO flags > =C2=A0- New range structure (Thomas) > =C2=A0- Hide migration behind Kconfig > =C2=A0- Kernel doc (Thomas) > =C2=A0- Use check_pages_threshold > v4: > =C2=A0- Don't evict partial unmaps in garbage collector (Thomas) > =C2=A0- Use %pe to print errors (Thomas) > =C2=A0- Use %p to print pointers (Thomas) > v5: > =C2=A0- Use range size helper (Thomas) > =C2=A0- Make BO external (Thomas) > =C2=A0- Set tile to NULL for BO creation (Thomas) > =C2=A0- Drop BO mirror flag (Thomas) > =C2=A0- Hold BO dma-resv lock across migration (Auld, Thomas) >=20 > Signed-off-by: Matthew Brost > --- > =C2=A0drivers/gpu/drm/xe/xe_svm.c | 111 > ++++++++++++++++++++++++++++++++++-- > =C2=A0drivers/gpu/drm/xe/xe_svm.h |=C2=A0=C2=A0 5 ++ > =C2=A02 files changed, 112 insertions(+), 4 deletions(-) >=20 > diff --git a/drivers/gpu/drm/xe/xe_svm.c > b/drivers/gpu/drm/xe/xe_svm.c > index 0a78a838508c..2e1e0f31c1a8 100644 > --- a/drivers/gpu/drm/xe/xe_svm.c > +++ b/drivers/gpu/drm/xe/xe_svm.c > @@ -32,6 +32,11 @@ static unsigned long xe_svm_range_end(struct > xe_svm_range *range) > =C2=A0 return drm_gpusvm_range_end(&range->base); > =C2=A0} > =C2=A0 > +static unsigned long xe_svm_range_size(struct xe_svm_range *range) > +{ > + return drm_gpusvm_range_size(&range->base); > +} > + > =C2=A0static void *xe_svm_devm_owner(struct xe_device *xe) > =C2=A0{ > =C2=A0 return xe; > @@ -512,7 +517,6 @@ static int xe_svm_populate_devmem_pfn(struct > drm_gpusvm_devmem *devmem_allocatio > =C2=A0 return 0; > =C2=A0} > =C2=A0 > -__maybe_unused > =C2=A0static const struct drm_gpusvm_devmem_ops gpusvm_devmem_ops =3D { > =C2=A0 .devmem_release =3D xe_svm_devmem_release, > =C2=A0 .populate_devmem_pfn =3D xe_svm_populate_devmem_pfn, > @@ -592,6 +596,71 @@ static bool xe_svm_range_is_valid(struct > xe_svm_range *range, > =C2=A0 return (range->tile_present & ~range->tile_invalidated) & > BIT(tile->id); > =C2=A0} > =C2=A0 > +static struct xe_vram_region *tile_to_vr(struct xe_tile *tile) > +{ > + return &tile->mem.vram; > +} > + > +static struct xe_bo *xe_svm_alloc_vram(struct xe_vm *vm, struct > xe_tile *tile, > + =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct xe_svm_range *range, > + =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 const struct drm_gpusvm_ctx > *ctx) > +{ > + struct mm_struct *mm =3D vm->svm.gpusvm.mm; > + struct xe_vram_region *vr =3D tile_to_vr(tile); > + struct drm_buddy_block *block; > + struct list_head *blocks; > + struct xe_bo *bo; > + ktime_t end =3D 0; > + int err; > + > + if (!mmget_not_zero(mm)) > + return ERR_PTR(-EFAULT); > + mmap_read_lock(mm); > + > +retry: > + bo =3D xe_bo_create_locked(tile_to_xe(tile), NULL, NULL, > + xe_svm_range_size(range), > + ttm_bo_type_device, > + XE_BO_FLAG_VRAM_IF_DGFX(tile)); > + if (IS_ERR(bo)) { > + err =3D PTR_ERR(bo); > + if (xe_vm_validate_should_retry(NULL, err, &end)) > + goto retry; > + goto unlock; > + } > + > + drm_gpusvm_devmem_init(&bo->devmem_allocation, > + =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 vm->xe->drm.dev, mm, > + =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 &gpusvm_devmem_ops, > + =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 &tile->mem.vram.dpagemap, > + =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 xe_svm_range_size(range)); The way this is done now looks good, although the corresponding multi- device commit https://gitlab.freedesktop.org/thomash/kernel/-/commit/1f204b5d05896e5cf63a= b54a679ee343e76600ff Goes one step further and removes the bo return of this function to favor an int error code. The caller doesn't really use the bo, right? > + > + blocks =3D &to_xe_ttm_vram_mgr_resource(bo->ttm.resource)- > >blocks; > + list_for_each_entry(block, blocks, link) > + block->private =3D vr; > + > + /* > + * Take ref because as soon as drm_gpusvm_migrate_to_devmem > succeeds the > + * creation ref can be dropped upon CPU fault or unmap. > + */ > + xe_bo_get(bo); > + > + err =3D drm_gpusvm_migrate_to_devmem(&vm->svm.gpusvm, &range- > >base, > + =C2=A0=C2=A0 &bo->devmem_allocation, > ctx); > + xe_bo_unlock(bo); > + if (err) { > + xe_bo_put(bo); /* Local ref */ > + xe_bo_put(bo); /* Creation ref */ Strictly, with the drm_gpusvm_devmem_init() succeeding, this bo put should be an xe_svm_devmem_release(), right? While equivalent here, if devmem_init moving forward sets up other resources that are expected to be released at devmem_release time, they'd be leaked. > + bo =3D ERR_PTR(err); > + } > + > +unlock: > + mmap_read_unlock(mm); > + mmput(mm); > + > + return bo; > +} > + > =C2=A0/** > =C2=A0 * xe_svm_handle_pagefault() - SVM handle page fault > =C2=A0 * @vm: The VM. > @@ -600,7 +669,8 @@ static bool xe_svm_range_is_valid(struct > xe_svm_range *range, > =C2=A0 * @fault_addr: The GPU fault address. > =C2=A0 * @atomic: The fault atomic access bit. > =C2=A0 * > - * Create GPU bindings for a SVM page fault. > + * Create GPU bindings for a SVM page fault. Optionally migrate to > device > + * memory. > =C2=A0 * > =C2=A0 * Return: 0 on success, negative error code on error. > =C2=A0 */ > @@ -608,11 +678,18 @@ int xe_svm_handle_pagefault(struct xe_vm *vm, > struct xe_vma *vma, > =C2=A0 =C2=A0=C2=A0=C2=A0 struct xe_tile *tile, u64 fault_addr, > =C2=A0 =C2=A0=C2=A0=C2=A0 bool atomic) > =C2=A0{ > - struct drm_gpusvm_ctx ctx =3D { .read_only =3D > xe_vma_read_only(vma), }; > + struct drm_gpusvm_ctx ctx =3D { > + .read_only =3D xe_vma_read_only(vma), > + .devmem_possible =3D IS_DGFX(vm->xe) && > + IS_ENABLED(CONFIG_DRM_XE_DEVMEM_MIRROR), > + .check_pages_threshold =3D IS_DGFX(vm->xe) && > + IS_ENABLED(CONFIG_DRM_XE_DEVMEM_MIRROR) ? > SZ_64K : 0, > + }; > =C2=A0 struct xe_svm_range *range; > =C2=A0 struct drm_gpusvm_range *r; > =C2=A0 struct drm_exec exec; > =C2=A0 struct dma_fence *fence; > + struct xe_bo *bo =3D NULL; > =C2=A0 ktime_t end =3D 0; > =C2=A0 int err; > =C2=A0 > @@ -620,6 +697,9 @@ int xe_svm_handle_pagefault(struct xe_vm *vm, > struct xe_vma *vma, > =C2=A0 xe_assert(vm->xe, xe_vma_is_cpu_addr_mirror(vma)); > =C2=A0 > =C2=A0retry: > + xe_bo_put(bo); > + bo =3D NULL; > + > =C2=A0 /* Always process UNMAPs first so view SVM ranges is current > */ > =C2=A0 err =3D xe_svm_garbage_collector(vm); > =C2=A0 if (err) > @@ -635,9 +715,31 @@ int xe_svm_handle_pagefault(struct xe_vm *vm, > struct xe_vma *vma, > =C2=A0 if (xe_svm_range_is_valid(range, tile)) > =C2=A0 return 0; > =C2=A0 > + /* XXX: Add migration policy, for now migrate range once */ > + if (!range->migrated && range->base.flags.migrate_devmem && > + =C2=A0=C2=A0=C2=A0 xe_svm_range_size(range) >=3D SZ_64K) { > + range->migrated =3D true; > + > + bo =3D xe_svm_alloc_vram(vm, tile, range, &ctx); > + if (IS_ERR(bo)) { > + drm_info(&vm->xe->drm, > + "VRAM allocation failed, falling > back to retrying, asid=3D%u, errno %pe\n", > + vm->usm.asid, bo); Typically we use drm_dbg() to print error messages to avoid spamming the logs and facilitate dynamic debug. Should we change this. /Thomas > + bo =3D NULL; > + goto retry; > + } > + } > + > =C2=A0 err =3D drm_gpusvm_range_get_pages(&vm->svm.gpusvm, r, &ctx); > - if (err =3D=3D -EFAULT || err =3D=3D -EPERM) /* Corner where CPU > mappings have changed */ > + /* Corner where CPU mappings have changed */ > + if (err =3D=3D -EOPNOTSUPP || err =3D=3D -EFAULT || err =3D=3D -EPERM) = { > + if (err =3D=3D -EOPNOTSUPP) > + drm_gpusvm_range_evict(&vm->svm.gpusvm, > &range->base); > + drm_info(&vm->xe->drm, > + "Get pages failed, falling back to > retrying, asid=3D%u, gpusvm=3D%p, errno %pe\n", > + vm->usm.asid, &vm->svm.gpusvm, > ERR_PTR(err)); > =C2=A0 goto retry; > + } > =C2=A0 if (err) > =C2=A0 goto err_out; > =C2=A0 > @@ -668,6 +770,7 @@ int xe_svm_handle_pagefault(struct xe_vm *vm, > struct xe_vma *vma, > =C2=A0 dma_fence_put(fence); > =C2=A0 > =C2=A0err_out: > + xe_bo_put(bo); > =C2=A0 > =C2=A0 return err; > =C2=A0} > diff --git a/drivers/gpu/drm/xe/xe_svm.h > b/drivers/gpu/drm/xe/xe_svm.h > index 0fa525d34987..984a61651d9e 100644 > --- a/drivers/gpu/drm/xe/xe_svm.h > +++ b/drivers/gpu/drm/xe/xe_svm.h > @@ -35,6 +35,11 @@ struct xe_svm_range { > =C2=A0 * range. Protected by GPU SVM notifier lock. > =C2=A0 */ > =C2=A0 u8 tile_invalidated; > + /** > + * @migrated: Range has been migrated to device memory, > protected by > + * GPU fault handler locking. > + */ > + u8 migrated :1; > =C2=A0}; > =C2=A0 > =C2=A0int xe_devm_add(struct xe_tile *tile, struct xe_vram_region *vr);