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 63769CD11C2 for ; Fri, 5 Apr 2024 08:05:54 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id D0BAB113A2A; Fri, 5 Apr 2024 08:05:53 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.b="RBi2AfNa"; dkim-atps=neutral Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.10]) by gabe.freedesktop.org (Postfix) with ESMTPS id 1626A113A2D for ; Fri, 5 Apr 2024 08:05:50 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1712304351; x=1743840351; h=message-id:date:subject:from:to:cc:references: in-reply-to:mime-version; bh=RdvmCHiv8gSUqfzm25eSFOazfKfosdtmD3QNybbtBdg=; b=RBi2AfNaL9vw3WpkOWcHbwmJfS1RVzmBfhlYtMter+ncfOrGPb0IMQXr sznsW/Vs4d9mec9d8sJw36Nb+EUbj52ULxIZ8Y1JayRKHfH9HDSbnszNC z/whxhbK23SvcfRxLJcHo/ao17B3HO/QaerO8RDh1mCgQzmpx4yGqjfoA tvBne5/kw5GvaaPHbAXtwHQkEDpEartKEEP36W+hVOZK8iXEj0sbeVQTl xcVRBZ9nZhlWNqy5penjfy2fwH2rcet0PJiZ/roE9WOY3M/FwrE3w03Jx P+142rnV3xLd0iSHz//9QgmntdiKl27XBvmQyvqdGyZdb+C+C3UCBCbAn g==; X-CSE-ConnectionGUID: XH5FRw5pRfWMWPSNPOcV3g== X-CSE-MsgGUID: 95+alm3PRsCB3WvYgl1dOQ== X-IronPort-AV: E=McAfee;i="6600,9927,11034"; a="25065686" X-IronPort-AV: E=Sophos;i="6.07,180,1708416000"; d="scan'208,217";a="25065686" Received: from orviesa004.jf.intel.com ([10.64.159.144]) by orvoesa102.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 05 Apr 2024 01:05:30 -0700 X-CSE-ConnectionGUID: VoHTJQfCT9+YXPI84fCi9Q== X-CSE-MsgGUID: Q4J1T4a8RSOZKgHl/Ahgdg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.07,180,1708416000"; d="scan'208,217";a="23809798" Received: from fmsmsx602.amr.corp.intel.com ([10.18.126.82]) by orviesa004.jf.intel.com with ESMTP/TLS/AES256-GCM-SHA384; 05 Apr 2024 01:05:30 -0700 Received: from fmsmsx603.amr.corp.intel.com (10.18.126.83) by fmsmsx602.amr.corp.intel.com (10.18.126.82) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.35; Fri, 5 Apr 2024 01:05:29 -0700 Received: from fmsedg601.ED.cps.intel.com (10.1.192.135) by fmsmsx603.amr.corp.intel.com (10.18.126.83) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.35 via Frontend Transport; Fri, 5 Apr 2024 01:05:29 -0700 Received: from NAM10-DM6-obe.outbound.protection.outlook.com (104.47.58.100) by edgegateway.intel.com (192.55.55.70) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.1.2507.35; Fri, 5 Apr 2024 01:05:28 -0700 ARC-Seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=nzJ8Zr3uN2kCh89BiBl61RNErnxJDOCJj8Cx1FWY8xW/YXQrMrs66xjRQk0qI1TholvAzR4KzUPKEM68hsGbKNzPViNyHcqsEXwjPBg72Trltho11N2Pg4Dym/Xo+wSA3JycZSAifxadrgBJj3qNPbmgsFjfkxNDbO5jK5+68g/bR3JwoNh5OcQ5Tu7IA3g2IFOaULBWpXHsuR+yM+sB4TOu80v+Tu+9b0pT1L6S96KPIH9E/nMzOlZ3t+cTzp1PhlM5QLevbuCDcS9REfTiWvf14z575FNmm+t7xncGBgVsGZt7RWzcT8eSMZ2oLlxuv6xu87Hr/F/lb6FSjZZwfQ== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector9901; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-AntiSpam-MessageData-ChunkCount:X-MS-Exchange-AntiSpam-MessageData-0:X-MS-Exchange-AntiSpam-MessageData-1; bh=M2fOWHUN6VeZ+4+bO4KFDnpLkalGoWo2ubCoPrUQal0=; b=jCc0wO1e9pvksQhOewtHR+j0zV+YUdg74h7v3iDFZ51+3E3u5jzTS2bp8mSEcR5ay7DLGKsaTbBgDSAokoWzRsJ5iCKAQFCe01PGWzX7EfZiPB449prcIMI4rrYayX0i3qY12jS4O00Vv0BxqbVuT7Gv5NkEGAB6M6ItPLn9LnlPD++UNuMZw+1fT6dL2sxlDj1IB8o8854rIKSaCmi/3PMNBI+AuJmN/KCwJzX95v9IJzsGN0Jz4MmwyynfUpQZ1PYoW/r1H3CmQc5s5HebBAYgeqk1eQ3Kfcmf7U+0TxpPvfBk2JeZpVFK08TJnEQYyDctB1D9YeLKkrZWvJq1yQ== ARC-Authentication-Results: i=1; mx.microsoft.com 1; spf=pass smtp.mailfrom=intel.com; dmarc=pass action=none header.from=intel.com; dkim=pass header.d=intel.com; arc=none Received: from MW4PR11MB7056.namprd11.prod.outlook.com (2603:10b6:303:21a::12) by SN7PR11MB7042.namprd11.prod.outlook.com (2603:10b6:806:299::5) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.7452.26; Fri, 5 Apr 2024 08:05:26 +0000 Received: from MW4PR11MB7056.namprd11.prod.outlook.com ([fe80::8664:8749:8357:f11a]) by MW4PR11MB7056.namprd11.prod.outlook.com ([fe80::8664:8749:8357:f11a%7]) with mapi id 15.20.7452.019; Fri, 5 Apr 2024 08:05:26 +0000 Content-Type: multipart/alternative; boundary="------------hLCJLUtY4ssM0iUvcKFtGG19" Message-ID: Date: Fri, 5 Apr 2024 13:35:20 +0530 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH v3 1/2] drm/xe: Introduce helper to populate userptr From: "Ghimiray, Himal Prasad" To: Matthew Brost CC: , Oak Zeng , "Niranjana Vishwanathapura" , =?UTF-8?Q?Thomas_Hellstr=C3=B6m?= , Brian Welty References: <20240404061631.4150858-1-himal.prasad.ghimiray@intel.com> <20240404061631.4150858-2-himal.prasad.ghimiray@intel.com> <1b73dc81-01f2-4aa6-a13b-f3f890b35d42@intel.com> Content-Language: en-US In-Reply-To: <1b73dc81-01f2-4aa6-a13b-f3f890b35d42@intel.com> X-ClientProxiedBy: PN2PR01CA0068.INDPRD01.PROD.OUTLOOK.COM (2603:1096:c01:23::13) To MW4PR11MB7056.namprd11.prod.outlook.com (2603:10b6:303:21a::12) MIME-Version: 1.0 X-MS-PublicTrafficType: Email X-MS-TrafficTypeDiagnostic: MW4PR11MB7056:EE_|SN7PR11MB7042:EE_ X-MS-Exchange-SenderADCheck: 1 X-MS-Exchange-AntiSpam-Relay: 0 X-Microsoft-Antispam: BCL:0; X-Microsoft-Antispam-Message-Info: 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 X-Forefront-Antispam-Report: CIP:255.255.255.255; CTRY:; LANG:en; SCL:1; SRV:; IPV:NLI; SFV:NSPM; H:MW4PR11MB7056.namprd11.prod.outlook.com; PTR:; CAT:NONE; SFS:(13230031)(1800799015)(376005)(366007); DIR:OUT; SFP:1102; X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1 X-MS-Exchange-AntiSpam-MessageData-0: =?utf-8?B?SXgya0lMaUtaSWpoOHVxV3R2WUVpbUNCMnRERVZSSFo1M3FWYWtOanZTU2VX?= =?utf-8?B?eUl3cVp0Q3hrQWswTkU3MktZRmt3MkJaQlFtMkN1SFFOUTJxSnNCQUlVakxa?= =?utf-8?B?cFJ1OUl6UTFrRDIyR1NRaEhHVk9IVWRDYkJKLys0Q0RMNHNldGVLSWI5OWwz?= =?utf-8?B?b3NLVDR0ai81b1E4WXRuaFBGUUhWVG9Rc0NLZ0JQWU83MGJHc3ZUcTV3aWF6?= =?utf-8?B?Lzl6SjM0Ui90Um9iSjZWd1JQeHFRamVaUFhTMEJPSE1tZFpKUllGN2ZzdTkx?= =?utf-8?B?ME1Ca0M2MkNiTmFMMjBXMDBoSkgxVmRkNkZ4NGdHU0F5c0xaSk1CaUxBczBX?= =?utf-8?B?bkRDaUs1SDRENFlDZHpGdWdiai9YWDFjSTRiNnpaSFRRSkJJL0NnTnI0Z05h?= =?utf-8?B?OVQvNjkySTZ6MzlZS01ZYzcrVDl1YTRHbkhXSXU1MUgyS205SmFwcXFYcnF1?= =?utf-8?B?ZnBZSXBFWVBvSVRERVloV09HOEh0d1ZWVjlrUEJDVTF0Vmdkd2tWMVY5MEcr?= =?utf-8?B?T3hqaVA0dE5OdFVNV2xlYnloVmY0SG16N1g5MlNPRWtWRXViQnFGVEZ3VXF2?= =?utf-8?B?SmZpTGdjTFloWk1hNXhKOXl5MWppeGtpRDhFUXFiLzJFU0xicTdydkF0cUxV?= =?utf-8?B?VTQ2M2RDaU83UjlEV2d4YVZPVVJsWVpQUnI2dm03aC9xWGw4eDFOUmw5SkNx?= =?utf-8?B?TWUrTENURnZVZ1cwK3lXR1c5RlJEU1ZJa0ZjeE9qSk92Q1VrdXJEWlM0Z1I1?= =?utf-8?B?R2U5WVhNYjNuVmhJZGYxa0d4eXFjdE9WZVdnZHIzbTFONG85Q3g1S2pPdUE1?= =?utf-8?B?V2owcVRteUNDbzRYcVNIampjMEx1bGxoWkNrM3NoR0dxUC9ldWN5OXByT1hw?= =?utf-8?B?ZHpYVTNHYlJLakJ3UFIyeVRCTnBVQnBxMGpHV05tcGxSTXVuNnhXWGlVWHda?= =?utf-8?B?ekNhNEE3YXJjcmdxOGZ3WmVKT3IyUHIrMmhlUTNQNXRQbzB4RXdsaTlhZGoz?= =?utf-8?B?R0RHVThocUxsR0Z0SFhiWHo4ZEZoWmJWaTNEcFJqRWpIc1RsZEZieFY1NTFB?= =?utf-8?B?REl1ZSszUDY1ZGhDUWxCSzJqNU9LQ25yNWZjd3Z5emNTdFhEb0xBNWg1SHVV?= =?utf-8?B?UlJ0a2lXc1lXZ0ZBZG1KK0VUcXkyV3BkWlBIdzZqWEhid3Q1ZGtkaDFqU2Ew?= =?utf-8?B?TzhqaTYrUHdQMHJOc2Rybmtjb2VTaHdQdFZBUER0aEpkdmpxMDlWYjhLbnJZ?= =?utf-8?B?MW56SjJqcEtzcjBFOHoxY1VqWFFXQXNMbkhkc0FIS1VXNXNCWnpNMFNkSUVG?= =?utf-8?B?Q24xaFpYbk56YnIzV2FxM2YwdVYyOEpCVlJ2RnhoMmVlMW02czJVdkN3cnJK?= =?utf-8?B?cmVsNmNsV2RXYjE1QmxpSmpwV29DajNweXJycTNIN1lUZktINjhCU0ltT2FT?= =?utf-8?B?eVlMTHlOWGZDWFBQUGVZL3czeXA1RW45MjhKKzJOVnJqSmo2a0ZJRFhDVDVu?= =?utf-8?B?eHJNVnJ2cm5NQm1xQ2dBNTE5ZkhCYmFvUHExSHZqK1lQZlNndXlNY0pJMXN4?= =?utf-8?B?MHVSa1hKZWN2V0x0eVcwL3pUeXdwNFlUTHZaVmNXcjk4NHVOUHpWVHp6dU8v?= =?utf-8?B?M25Yb2FSV2hqQWdhekZQcnJXb3lBaldOT08xTEdiR2oxcmFCZ1pVbE83ZHpT?= =?utf-8?B?SEFHMHZmV1hPSnQvNzIvNGl2MWtrZU9zdU5CNVJiYjZkZkpYTzdUNzVvdks2?= =?utf-8?B?emVJMEwxd2oybHZibElVOXlCR2w4R0JzSUVuaGFOcWNjZUhKbjlPa3hkR2t4?= =?utf-8?B?Ynd2N1VhY2d1Vzk0WFpGTStNYzNmb1lmVEErOFJ0K0NVeDVsb2RueEFSZmVX?= =?utf-8?B?ZXdNSGtVTUZBSzdtR1MraGQzSlZyVk1DeklTcmM0Y2QxSHdUUUFQdE9HZWZv?= =?utf-8?B?NmdacGF6b1lmRmxFNSt2QlpWR054RWZFWmlqTXBLanBzditXT2grRzNCUEpP?= =?utf-8?B?LzVBZ2NXOUQxZC9UaEZaeEovWndSKzE4TEJPeEVDZFVDZGNkeUFjMHowMkFv?= =?utf-8?B?RDJoV3R3L0RGeWo5dVhLR0JZSmttYWZjSFZ6Z2J6bTFaSDlmU094L2VRRHBP?= =?utf-8?B?TkFMNVppaFVTbjYvU3lBUSsxWVJJSkJMVXVXUWYrdDBzQlFRWFRoUkpENHM4?= =?utf-8?B?bkE9PQ==?= X-MS-Exchange-CrossTenant-Network-Message-Id: 6d45a387-94e9-4a80-77fb-08dc554726e1 X-MS-Exchange-CrossTenant-AuthSource: MW4PR11MB7056.namprd11.prod.outlook.com X-MS-Exchange-CrossTenant-AuthAs: Internal X-MS-Exchange-CrossTenant-OriginalArrivalTime: 05 Apr 2024 08:05:26.6754 (UTC) X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted X-MS-Exchange-CrossTenant-Id: 46c98d88-e344-4ed4-8496-4ed7712e255d X-MS-Exchange-CrossTenant-MailboxType: HOSTED X-MS-Exchange-CrossTenant-UserPrincipalName: xqzg2r779Olf2SDPSzyp6wio14tC4e2u/hVM9SVCHydCJYBhIdj3I+66Y82pkH/RyLXD6BxjFqXizdf1hzYe7lEThkBQTgogDHl+zcgZOwM= X-MS-Exchange-Transport-CrossTenantHeadersStamped: SN7PR11MB7042 X-OriginatorOrg: intel.com 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" --------------hLCJLUtY4ssM0iUvcKFtGG19 Content-Type: text/plain; charset="UTF-8"; format=flowed Content-Transfer-Encoding: 8bit On 05-04-2024 11:42, Ghimiray, Himal Prasad wrote: > > > On 05-04-2024 06:37, Matthew Brost wrote: >> On Thu, Apr 04, 2024 at 11:46:30AM +0530, Himal Prasad Ghimiray wrote: >>> From: Oak Zeng >>> >>> Introduce a helper function xe_userptr_populate_range to populate >>> a a userptr range. This functions calls hmm_range_fault to read >>> CPU page tables and populate all pfns/pages of this virtual address >>> range. >>> >>> If the populated page is system memory page, dma-mapping is performed >>> to get a dma-address which can be used later for GPU to access pages. >>> >>> If the populated page is device private page, we calculate the dpa ( >>> device physical address) of the page. This will be handled in future >>> patches. >>> >>> The dma-address or dpa is then saved in userptr's sg table. This is >>> prepare work to replace the get_user_pages_fast code in userptr code >>> path. >>> >>> v1: Address review comments: >>> separate a npage_in_range function (Matt) >>> reparameterize function xe_userptr_populate_range function (Matt) >>> move mmu_interval_read_begin() call into while loop (Thomas) >>> s/mark_range_accessed/xe_mark_range_accessed (Thomas) >>> use set_page_dirty_lock (vs set_page_dirty) (Thomas) >>> move a few checking in xe_vma_userptr_pin_pages to hmm.c (Matt) >>> v2: Remove device private page support. Only support system >>> pages for now. use dma-map-sg rather than dma-map-page (Matt/Thomas) >>> v3: Address review comments: >>> Squash patch "drm/xe: Introduce a helper to free sg table" to current >>> patch (Matt) >>> start and end addresses are already page aligned (Matt) >>> Do mmap_read_lock and mmap_read_unlock for hmm_range_fault incase of >>> non system allocator call. (Matt) >>> Drop kthread_use_mm and kthread_unuse_mm. (Matt) >>> No need of kernel-doc for static functions.(Matt) >>> Modify function names. (Matt) >>> Free sgtable incase of dma_map_sgtable failure.(Matt) >>> Modify loop for hmm_range_fault.(Matt) >>> >>> Signed-off-by: Oak Zeng >>> Co-developed-by: Niranjana Vishwanathapura >>> Signed-off-by: Niranjana Vishwanathapura >>> Cc: Matthew Brost >>> Cc: Thomas Hellström >>> Cc: Brian Welty >>> Signed-off-by: Himal Prasad Ghimiray >>> --- >>> drivers/gpu/drm/xe/Kconfig | 1 + >>> drivers/gpu/drm/xe/Makefile | 2 + >>> drivers/gpu/drm/xe/xe_hmm.c | 253 ++++++++++++++++++++++++++++++++++++ >>> drivers/gpu/drm/xe/xe_hmm.h | 18 +++ >>> 4 files changed, 274 insertions(+) >>> create mode 100644 drivers/gpu/drm/xe/xe_hmm.c >>> create mode 100644 drivers/gpu/drm/xe/xe_hmm.h >>> >>> diff --git a/drivers/gpu/drm/xe/Kconfig b/drivers/gpu/drm/xe/Kconfig >>> index 1a556d087e63..449a1ecbc92a 100644 >>> --- a/drivers/gpu/drm/xe/Kconfig >>> +++ b/drivers/gpu/drm/xe/Kconfig >>> @@ -41,6 +41,7 @@ config DRM_XE >>> select MMU_NOTIFIER >>> select WANT_DEV_COREDUMP >>> select AUXILIARY_BUS >>> + select HMM_MIRROR >>> help >>> Experimental driver for Intel Xe series GPUs >>> >>> diff --git a/drivers/gpu/drm/xe/Makefile b/drivers/gpu/drm/xe/Makefile >>> index 21316ee47026..07031b5ba977 100644 >>> --- a/drivers/gpu/drm/xe/Makefile >>> +++ b/drivers/gpu/drm/xe/Makefile >>> @@ -146,6 +146,8 @@ xe-y += xe_bb.o \ >>> xe_wa.o \ >>> xe_wopcm.o >>> >>> +xe-$(CONFIG_HMM_MIRROR) += xe_hmm.o >>> + >>> # graphics hardware monitoring (HWMON) support >>> xe-$(CONFIG_HWMON) += xe_hwmon.o >>> >>> diff --git a/drivers/gpu/drm/xe/xe_hmm.c b/drivers/gpu/drm/xe/xe_hmm.c >>> new file mode 100644 >>> index 000000000000..d7a11896ad72 >>> --- /dev/null >>> +++ b/drivers/gpu/drm/xe/xe_hmm.c >>> @@ -0,0 +1,253 @@ >>> +// SPDX-License-Identifier: MIT >>> +/* >>> + * Copyright © 2024 Intel Corporation >>> + */ >>> + >>> +#include >>> +#include >>> +#include >>> +#include >>> +#include >>> +#include >>> +#include >>> +#include "xe_hmm.h" >>> +#include "xe_vm.h" >>> +#include "xe_bo.h" >>> + >>> +static u64 xe_npages_in_range(unsigned long start, unsigned long end) >>> +{ >>> + return (end - start) >> PAGE_SHIFT; >>> +} >>> + >>> +/* >>> + * xe_mark_range_accessed() - mark a range is accessed, so core mm >>> + * have such information for memory eviction or write back to >>> + * hard disk >>> + * >>> + * @range: the range to mark >>> + * @write: if write to this range, we mark pages in this range >>> + * as dirty >>> + */ >>> +static void xe_mark_range_accessed(struct hmm_range *range, bool write) >>> +{ >>> + struct page *page; >>> + u64 i, npages; >>> + >>> + npages = xe_npages_in_range(range->start, range->end); >>> + for (i = 0; i < npages; i++) { >>> + page = hmm_pfn_to_page(range->hmm_pfns[i]); >>> + if (write) >>> + set_page_dirty_lock(page); >>> + >>> + mark_page_accessed(page); >>> + } >>> +} >>> + >>> +/* >>> + * xe_build_sg() - build a scatter gather table for all the physical pages/pfn >>> + * in a hmm_range. dma-map pages if necessary. dma-address is save in sg table >>> + * and will be used to program GPU page table later. >>> + * >>> + * @xe: the xe device who will access the dma-address in sg table >>> + * @range: the hmm range that we build the sg table from. range->hmm_pfns[] >>> + * has the pfn numbers of pages that back up this hmm address range. >>> + * @st: pointer to the sg table. >>> + * @write: whether we write to this range. This decides dma map direction >>> + * for system pages. If write we map it bi-diretional; otherwise >>> + * DMA_TO_DEVICE >>> + * >>> + * All the contiguous pfns will be collapsed into one entry in >>> + * the scatter gather table. This is for the purpose of efficiently >>> + * programming GPU page table. >>> + * >>> + * The dma_address in the sg table will later be used by GPU to >>> + * access memory. So if the memory is system memory, we need to >>> + * do a dma-mapping so it can be accessed by GPU/DMA. >>> + * >>> + * FIXME: This function currently only support pages in system >>> + * memory. If the memory is GPU local memory (of the GPU who >>> + * is going to access memory), we need gpu dpa (device physical >>> + * address), and there is no need of dma-mapping. This is TBD. >>> + * >>> + * FIXME: dma-mapping for peer gpu device to access remote gpu's >>> + * memory. Add this when you support p2p >>> + * >>> + * This function allocates the storage of the sg table. It is >>> + * caller's responsibility to free it calling sg_free_table. >>> + * >>> + * Returns 0 if successful; -ENOMEM if fails to allocate memory >>> + */ >>> +static int xe_build_sg(struct xe_device *xe, struct hmm_range *range, >>> + struct sg_table *st, bool write) >>> +{ >>> + struct device *dev = xe->drm.dev; >>> + struct page **pages; >>> + u64 i, npages; >>> + int ret; >>> + >>> + npages = xe_npages_in_range(range->start, range->end); >>> + pages = kvmalloc_array(npages, sizeof(*pages), GFP_KERNEL); >>> + if (!pages) >>> + return -ENOMEM; >>> + >>> + for (i = 0; i < npages; i++) { >>> + pages[i] = hmm_pfn_to_page(range->hmm_pfns[i]); >>> + xe_assert(xe, !is_device_private_page(pages[i])); >>> + } >>> + >>> + ret = sg_alloc_table_from_pages_segment(st, pages, npages, 0, npages << PAGE_SHIFT, >>> + xe_sg_segment_size(dev), GFP_KERNEL); >>> + if (ret) >>> + goto free_pages; >>> + >>> + ret = dma_map_sgtable(dev, st, write ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE, >>> + DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_NO_KERNEL_MAPPING); >>> + if (ret) { >>> + sg_free_table(st); >>> + st = NULL; >>> + } >>> + >>> +free_pages: >>> + kvfree(pages); >>> + return ret; >>> +} >>> + >>> +/* >>> + * xe_hmm_userptr_free_sg() - Free the scatter gather table of userptr >>> + * >>> + * @uvma: the userptr vma which hold the scatter gather table >>> + * >>> + * With function xe_userptr_populate_range, we allocate storage of >>> + * the userptr sg table. This is a helper function to free this >>> + * sg table, and dma unmap the address in the table. >>> + */ >>> +void xe_hmm_userptr_free_sg(struct xe_userptr_vma *uvma) >>> +{ >>> + struct xe_userptr *userptr = &uvma->userptr; >>> + struct xe_vma *vma = &uvma->vma; >>> + bool write = !xe_vma_read_only(vma); >>> + struct xe_vm *vm = xe_vma_vm(vma); >>> + struct xe_device *xe = vm->xe; >>> + struct device *dev = xe->drm.dev; >>> + >>> + xe_assert(xe, userptr->sg); >>> + dma_unmap_sgtable(dev, userptr->sg, >>> + write ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE, 0); >>> + >>> + sg_free_table(userptr->sg); >>> + userptr->sg = NULL; >>> +} >>> + >>> +/** >>> + * xe_hmm_userptr_populate_range() - Populate physical pages of a virtual >>> + * address range >>> + * >>> + * @uvma: userptr vma which has information of the range to populate. >>> + * @is_mm_mmap_locked: True if mmap_read_lock is already acquired by caller. >>> + * >>> + * This function populate the physical pages of a virtual >>> + * address range. The populated physical pages is saved in >>> + * userptr's sg table. It is similar to get_user_pages but call >>> + * hmm_range_fault. >>> + * >>> + * This function also read mmu notifier sequence # ( >>> + * mmu_interval_read_begin), for the purpose of later >>> + * comparison (through mmu_interval_read_retry). >>> + * >>> + * This must be called with mmap read or write lock held. >>> + * >>> + * This function allocates the storage of the userptr sg table. >>> + * It is caller's responsibility to free it calling sg_free_table. >>> + * >>> + * returns: 0 for succuss; negative error no on failure >>> + */ >>> +int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvma, >>> + bool is_mm_mmap_locked) >>> +{ >>> + unsigned long timeout = >>> + jiffies + msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT); >>> + unsigned long *pfns, flags = HMM_PFN_REQ_FAULT; >>> + struct xe_userptr *userptr; >>> + struct xe_vma *vma = &uvma->vma; >>> + u64 start = xe_vma_userptr(vma); >>> + u64 end = start + xe_vma_size(vma); >> Nit: xe_vma_end > Sure. Will address in next patch. >>> + struct xe_vm *vm = xe_vma_vm(vma); >>> + struct hmm_range hmm_range; >>> + bool write = !xe_vma_read_only(vma); >>> + unsigned long notifier_seq; >>> + u64 npages; >>> + int ret; >>> + >>> + userptr = &uvma->userptr; >>> + >>> + if (is_mm_mmap_locked) >>> + mmap_assert_locked(userptr->notifier.mm); >>> + >>> + if (vma->gpuva.flags & XE_VMA_DESTROYED) >>> + return 0; >>> + >>> + notifier_seq = mmu_interval_read_begin(&userptr->notifier); >>> + if (notifier_seq == userptr->notifier_seq) >>> + return 0; >>> + >>> + if (userptr->sg) >>> + xe_hmm_userptr_free_sg(uvma); >>> + >>> + npages = xe_npages_in_range(start, end); >>> + pfns = kvmalloc_array(npages, sizeof(*pfns), GFP_KERNEL); >>> + if (unlikely(!pfns)) >>> + return -ENOMEM; >>> + >>> + if (write) >>> + flags |= HMM_PFN_REQ_WRITE; >>> + >>> + if (!mmget_not_zero(userptr->notifier.mm)) { >>> + ret = -EFAULT; >>> + goto free_pfns; >>> + } >>> + >>> + hmm_range.default_flags = flags; >>> + hmm_range.hmm_pfns = pfns; >>> + hmm_range.notifier = &userptr->notifier; >>> + hmm_range.start = start; >>> + hmm_range.end = end; >>> + hmm_range.dev_private_owner = vm->xe; >>> + >>> + while (true) { >>> + hmm_range.notifier_seq = mmu_interval_read_begin(&userptr->notifier); >> Nit: This could be set above in this code: >> >> notifier_seq = mmu_interval_read_begin(&userptr->notifier); >> if (notifier_seq == userptr->notifier_seq) >> return 0; >> >> i.e. drop local notifier_seq and set hmm_range.notifier_seq... >> >> Then reset before continue in the if (ret == -EBUSY) function. > I chose to retain |*mmu_interval_read_begin*| before |hmm_range_fault| > because I anticipated a scenario where the notifier sequence might be > invalid  between the initial call and just before |hmm_range_fault|. > In such a case, |hmm_range_fault| would return EBUSY, requiring us to > unlock the mmap_lock, use a |time_after| routine, lock again, and call > |hmm_range_fault| once more. > > However, by utilizing |mmu_interval_read_begin| before > |hmm_range_fault|, we might avoid this issue. Any change in sequence > between the initial call and this point wouldn't impact the process. > Even if there's no change in sequence, we're only making an extra call > to |mmu_interval_read_begin|. > > Both approaches seem to work, but if you believe my concern about > changes in the notifier sequence is unfounded, I will proceed with the > suggested changes. > >>> + >>> + if (!is_mm_mmap_locked) >>> + mmap_read_lock(userptr->notifier.mm); >>> + >>> + ret = hmm_range_fault(&hmm_range); >>> + >>> + if (!is_mm_mmap_locked) >>> + mmap_read_unlock(userptr->notifier.mm); >>> + >>> + if (ret == -EBUSY) { >>> + if (time_after(jiffies, timeout)) >>> + break; >>> + >>> + continue; >>> + } >>> + break; >>> + } >>> + >>> + mmput(userptr->notifier.mm); >>> + >>> + if (ret) >>> + goto free_pfns; >>> + >>> + ret = xe_build_sg(vm->xe, &hmm_range, &userptr->sgt, write); >>> + if (ret) >>> + goto free_pfns; >>> + >>> + xe_mark_range_accessed(&hmm_range, write); >>> + userptr->sg = &userptr->sgt; >>> + userptr->notifier_seq = hmm_range.notifier_seq; >>> + >>> +free_pfns: >>> + kvfree(pfns); >>> + return ret; >>> +} >>> + >>> diff --git a/drivers/gpu/drm/xe/xe_hmm.h b/drivers/gpu/drm/xe/xe_hmm.h >>> new file mode 100644 >>> index 000000000000..40250e3f84d1 >>> --- /dev/null >>> +++ b/drivers/gpu/drm/xe/xe_hmm.h >>> @@ -0,0 +1,18 @@ >>> +/* SPDX-License-Identifier: MIT >>> + * >>> + * Copyright © 2024 Intel Corporation >>> + */ >>> + >>> +#include >>> + >>> +struct xe_userptr_vma; >>> + >>> +#if IS_ENABLED(CONFIG_HMM_MIRROR) >>> +int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvma, bool is_mm_mmap_locked); >>> +#else >>> +static inline int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvma, bool is_mm_mmap_locked) >>> +{ >>> + return -ENODEV; >>> +} >>> +#endif >> Mentioned that I don't think is needed in last rev as our driver require >> CONFIG_HMM_MIRROR to be built. > > I noticed some KUnit failures in earlier patches, so I opted not to > remove it. The KUnit configuration file doesn't include the HMM_MIRROR > config. If we want to remove this we need to modify the KUnit configs > to include this config, as it is essential for our driver. > > Please confirm it is OK to add the config to kunit configuration file ? > My bad in earlier patch HMM_MIRROR was under config DRM_XE_SVM , this patch is selecting it under DRM_XE itself which is used by kunit conf file. This should be safe to remove. Will remove it in next version. >> Matt >> >>> +void xe_hmm_userptr_free_sg(struct xe_userptr_vma *uvma); >>> -- >>> 2.25.1 >>> --------------hLCJLUtY4ssM0iUvcKFtGG19 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable


On 05-04-2024 11:42, Ghimiray, Himal Prasad wrote:
=20


On 05-04-2024 06:37, Matthew Brost wrote:
On Thu, Apr 04, 2024 at 11:4=
6:30AM +0530, Himal Prasad Ghimiray wrote:

From: Oak Zeng <oak.zeng@intel.com>

Introduce a helper function xe_userptr_populate_range to populate
a a userptr range. This functions calls hmm_range_fault to read
CPU page tables and populate all pfns/pages of this virtual address
range.

If the populated page is system memory page, dma-mapping is performed
to get a dma-address which can be used later for GPU to access pages.

If the populated page is device private page, we calculate the dpa (
device physical address) of the page. This will be handled in future
patches.

The dma-address or dpa is then saved in userptr's sg table. This is
prepare work to replace the get_user_pages_fast code in userptr code
path.

v1: Address review comments:
    separate a npage_in_range function (Matt)
    reparameterize function xe_userptr_populate_range function (Matt)
    move mmu_interval_read_begin() call into while loop (Thomas)
    s/mark_range_accessed/xe_mark_range_accessed (Thomas)
    use set_page_dirty_lock (vs set_page_dirty) (Thomas)
    move a few checking in xe_vma_userptr_pin_pages to hmm.c (Matt)
v2: Remove device private page support. Only support system
    pages for now. use dma-map-sg rather than dma-map-page (Matt/Thomas)
v3: Address review comments:
    Squash patch "drm/xe: Introduce a helper to free sg table" to=
 current
    patch (Matt)
    start and end addresses are already page aligned (Matt)
    Do mmap_read_lock and mmap_read_unlock for hmm_range_fault incase of
    non system allocator call. (Matt)
    Drop kthread_use_mm and kthread_unuse_mm. (Matt)
    No need of kernel-doc for static functions.(Matt)
    Modify function names. (Matt)
    Free sgtable incase of dma_map_sgtable failure.(Matt)
    Modify loop for hmm_range_fault.(Matt)

Signed-off-by: Oak Zeng <oak.zeng@intel.com>
Co-developed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com><=
/a>
Cc: Thomas Hellstr=C3=B6m <thomas.hellstrom@=
intel.com>
Cc: Brian Welty <brian.welty@intel.com>
Signed-off-by: Himal Prasad Ghimiray <h=
imal.prasad.ghimiray@intel.com>
---
 drivers/gpu/drm/xe/Kconfig  |   1 +
 drivers/gpu/drm/xe/Makefile |   2 +
 drivers/gpu/drm/xe/xe_hmm.c | 253 ++++++++++++++++++++++++++++++++++++
 drivers/gpu/drm/xe/xe_hmm.h |  18 +++
 4 files changed, 274 insertions(+)
 create mode 100644 drivers/gpu/drm/xe/xe_hmm.c
 create mode 100644 drivers/gpu/drm/xe/xe_hmm.h

diff --git a/drivers/gpu/drm/xe/Kconfig b/drivers/gpu/drm/xe/Kconfig
index 1a556d087e63..449a1ecbc92a 100644
--- a/drivers/gpu/drm/xe/Kconfig
+++ b/drivers/gpu/drm/xe/Kconfig
@@ -41,6 +41,7 @@ config DRM_XE
 	select MMU_NOTIFIER
 	select WANT_DEV_COREDUMP
 	select AUXILIARY_BUS
+	select HMM_MIRROR
 	help
 	  Experimental driver for Intel Xe series GPUs
=20
diff --git a/drivers/gpu/drm/xe/Makefile b/drivers/gpu/drm/xe/Makefile
index 21316ee47026..07031b5ba977 100644
--- a/drivers/gpu/drm/xe/Makefile
+++ b/drivers/gpu/drm/xe/Makefile
@@ -146,6 +146,8 @@ xe-y +=3D xe_bb.o \
 	xe_wa.o \
 	xe_wopcm.o
=20
+xe-$(CONFIG_HMM_MIRROR) +=3D xe_hmm.o
+
 # graphics hardware monitoring (HWMON) support
 xe-$(CONFIG_HWMON) +=3D xe_hwmon.o
=20
diff --git a/drivers/gpu/drm/xe/xe_hmm.c b/drivers/gpu/drm/xe/xe_hmm.c
new file mode 100644
index 000000000000..d7a11896ad72
--- /dev/null
+++ b/drivers/gpu/drm/xe/xe_hmm.c
@@ -0,0 +1,253 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright =C2=A9 2024 Intel Corporation
+ */
+
+#include <linux/scatterlist.h>
+#include <linux/mmu_notifier.h>
+#include <linux/dma-mapping.h>
+#include <linux/memremap.h>
+#include <linux/swap.h>
+#include <linux/hmm.h>
+#include <linux/mm.h>
+#include "xe_hmm.h"
+#include "xe_vm.h"
+#include "xe_bo.h"
+
+static u64 xe_npages_in_range(unsigned long start, unsigned long end)
+{
+	return (end - start) >> PAGE_SHIFT;
+}
+
+/*
+ * xe_mark_range_accessed() - mark a range is accessed, so core mm
+ * have such information for memory eviction or write back to
+ * hard disk
+ *
+ * @range: the range to mark
+ * @write: if write to this range, we mark pages in this range
+ * as dirty
+ */
+static void xe_mark_range_accessed(struct hmm_range *range, bool write)
+{
+	struct page *page;
+	u64 i, npages;
+
+	npages =3D xe_npages_in_range(range->start, range->end);
+	for (i =3D 0; i < npages; i++) {
+		page =3D hmm_pfn_to_page(range->hmm_pfns[i]);
+		if (write)
+			set_page_dirty_lock(page);
+
+		mark_page_accessed(page);
+	}
+}
+
+/*
+ * xe_build_sg() - build a scatter gather table for all the physical pages=
/pfn
+ * in a hmm_range. dma-map pages if necessary. dma-address is save in sg t=
able
+ * and will be used to program GPU page table later.
+ *
+ * @xe: the xe device who will access the dma-address in sg table
+ * @range: the hmm range that we build the sg table from. range->hmm_pf=
ns[]
+ * has the pfn numbers of pages that back up this hmm address range.
+ * @st: pointer to the sg table.
+ * @write: whether we write to this range. This decides dma map direction
+ * for system pages. If write we map it bi-diretional; otherwise
+ * DMA_TO_DEVICE
+ *
+ * All the contiguous pfns will be collapsed into one entry in
+ * the scatter gather table. This is for the purpose of efficiently
+ * programming GPU page table.
+ *
+ * The dma_address in the sg table will later be used by GPU to
+ * access memory. So if the memory is system memory, we need to
+ * do a dma-mapping so it can be accessed by GPU/DMA.
+ *
+ * FIXME: This function currently only support pages in system
+ * memory. If the memory is GPU local memory (of the GPU who
+ * is going to access memory), we need gpu dpa (device physical
+ * address), and there is no need of dma-mapping. This is TBD.
+ *
+ * FIXME: dma-mapping for peer gpu device to access remote gpu's
+ * memory. Add this when you support p2p
+ *
+ * This function allocates the storage of the sg table. It is
+ * caller's responsibility to free it calling sg_free_table.
+ *
+ * Returns 0 if successful; -ENOMEM if fails to allocate memory
+ */
+static int xe_build_sg(struct xe_device *xe, struct hmm_range *range,
+		       struct sg_table *st, bool write)
+{
+	struct device *dev =3D xe->drm.dev;
+	struct page **pages;
+	u64 i, npages;
+	int ret;
+
+	npages =3D xe_npages_in_range(range->start, range->end);
+	pages =3D kvmalloc_array(npages, sizeof(*pages), GFP_KERNEL);
+	if (!pages)
+		return -ENOMEM;
+
+	for (i =3D 0; i < npages; i++) {
+		pages[i] =3D hmm_pfn_to_page(range->hmm_pfns[i]);
+		xe_assert(xe, !is_device_private_page(pages[i]));
+	}
+
+	ret =3D sg_alloc_table_from_pages_segment(st, pages, npages, 0, npages &l=
t;< PAGE_SHIFT,
+						xe_sg_segment_size(dev), GFP_KERNEL);
+	if (ret)
+		goto free_pages;
+
+	ret =3D dma_map_sgtable(dev, st, write ? DMA_BIDIRECTIONAL : DMA_TO_DEVIC=
E,
+			      DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_NO_KERNEL_MAPPING);
+	if (ret) {
+		sg_free_table(st);
+		st =3D NULL;
+	}
+
+free_pages:
+	kvfree(pages);
+	return ret;
+}
+
+/*
+ * xe_hmm_userptr_free_sg() - Free the scatter gather table of userptr
+ *
+ * @uvma: the userptr vma which hold the scatter gather table
+ *
+ * With function xe_userptr_populate_range, we allocate storage of
+ * the userptr sg table. This is a helper function to free this
+ * sg table, and dma unmap the address in the table.
+ */
+void xe_hmm_userptr_free_sg(struct xe_userptr_vma *uvma)
+{
+	struct xe_userptr *userptr =3D &uvma->userptr;
+	struct xe_vma *vma =3D &uvma->vma;
+	bool write =3D !xe_vma_read_only(vma);
+	struct xe_vm *vm =3D xe_vma_vm(vma);
+	struct xe_device *xe =3D vm->xe;
+	struct device *dev =3D xe->drm.dev;
+
+	xe_assert(xe, userptr->sg);
+	dma_unmap_sgtable(dev, userptr->sg,
+			  write ? DMA_BIDIRECTIONAL : DMA_TO_DEVICE, 0);
+
+	sg_free_table(userptr->sg);
+	userptr->sg =3D NULL;
+}
+
+/**
+ * xe_hmm_userptr_populate_range() - Populate physical pages of a virtual
+ * address range
+ *
+ * @uvma: userptr vma which has information of the range to populate.
+ * @is_mm_mmap_locked: True if mmap_read_lock is already acquired by calle=
r.
+ *
+ * This function populate the physical pages of a virtual
+ * address range. The populated physical pages is saved in
+ * userptr's sg table. It is similar to get_user_pages but call
+ * hmm_range_fault.
+ *
+ * This function also read mmu notifier sequence # (
+ * mmu_interval_read_begin), for the purpose of later
+ * comparison (through mmu_interval_read_retry).
+ *
+ * This must be called with mmap read or write lock held.
+ *
+ * This function allocates the storage of the userptr sg table.
+ * It is caller's responsibility to free it calling sg_free_table.
+ *
+ * returns: 0 for succuss; negative error no on failure
+ */
+int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvma,
+				  bool is_mm_mmap_locked)
+{
+	unsigned long timeout =3D
+		jiffies + msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
+	unsigned long *pfns, flags =3D HMM_PFN_REQ_FAULT;
+	struct xe_userptr *userptr;
+	struct xe_vma *vma =3D &uvma->vma;
+	u64 start =3D xe_vma_userptr(vma);
+	u64 end =3D start + xe_vma_size(vma);

Nit: xe_vma_end
Sure. Will address in next patch. 
+	struct xe_vm *vm =3D xe_=
vma_vm(vma);
+	struct hmm_range hmm_range;
+	bool write =3D !xe_vma_read_only(vma);
+	unsigned long notifier_seq;
+	u64 npages;
+	int ret;
+
+	userptr =3D &uvma->userptr;
+
+	if (is_mm_mmap_locked)
+		mmap_assert_locked(userptr->notifier.mm);
+
+	if (vma->gpuva.flags & XE_VMA_DESTROYED)
+		return 0;
+
+	notifier_seq =3D mmu_interval_read_begin(&userptr->notifier);
+	if (notifier_seq =3D=3D userptr->notifier_seq)
+		return 0;
+
+	if (userptr->sg)
+		xe_hmm_userptr_free_sg(uvma);
+
+	npages =3D xe_npages_in_range(start, end);
+	pfns =3D kvmalloc_array(npages, sizeof(*pfns), GFP_KERNEL);
+	if (unlikely(!pfns))
+		return -ENOMEM;
+
+	if (write)
+		flags |=3D HMM_PFN_REQ_WRITE;
+
+	if (!mmget_not_zero(userptr->notifier.mm)) {
+		ret =3D -EFAULT;
+		goto free_pfns;
+	}
+
+	hmm_range.default_flags =3D flags;
+	hmm_range.hmm_pfns =3D pfns;
+	hmm_range.notifier =3D &userptr->notifier;
+	hmm_range.start =3D start;
+	hmm_range.end =3D end;
+	hmm_range.dev_private_owner =3D vm->xe;
+
+	while (true) {
+		hmm_range.notifier_seq =3D mmu_interval_read_begin(&userptr->noti=
fier);

Nit: This could be set above=
 in this code:

	notifier_seq =3D mmu_interval_read_begin(&userptr->notifier);
	if (notifier_seq =3D=3D userptr->notifier_seq)
		return 0;
=20
i.e. drop local notifier_seq and set hmm_range.notifier_seq...

Then reset before continue in the if (ret =3D=3D -EBUSY) function.
I chose to retain mmu_interval_read_begin before hmm_range_fault because I anticipated a scenario where the notifier sequence might be invalid  between the initial call and just before hmm_range_fault. In such a case, hmm_range_fault would return EBUSY, requiring us to unlock the mmap_lock, use a time_after routine, lock again, and call hmm_range_fault once more.

However, by utilizing mmu_i= nterval_read_begin before hmm_range_fault, we might avoid this issue. Any change in sequence between the initial c= all and this point wouldn't impact the process. Even if there's no change i= n sequence, we're only making an extra call to mmu_in= terval_read_begin.

Both approaches seem to work, but if you beli= eve my concern about changes in the notifier sequence is unfounded, I will = proceed with the suggested changes.

+
+		if (!is_mm_mmap_locked)
+			mmap_read_lock(userptr->notifier.mm);
+
+		ret =3D hmm_range_fault(&hmm_range);
+
+		if (!is_mm_mmap_locked)
+			mmap_read_unlock(userptr->notifier.mm);
+
+		if (ret =3D=3D -EBUSY) {
+			if (time_after(jiffies, timeout))
+				break;
+
+			continue;
+		}
+		break;
+	}
+
+	mmput(userptr->notifier.mm);
+
+	if (ret)
+		goto free_pfns;
+
+	ret =3D xe_build_sg(vm->xe, &hmm_range, &userptr->sgt, writ=
e);
+	if (ret)
+		goto free_pfns;
+
+	xe_mark_range_accessed(&hmm_range, write);
+	userptr->sg =3D &userptr->sgt;
+	userptr->notifier_seq =3D hmm_range.notifier_seq;
+
+free_pfns:
+	kvfree(pfns);
+	return ret;
+}
+
diff --git a/drivers/gpu/drm/xe/xe_hmm.h b/drivers/gpu/drm/xe/xe_hmm.h
new file mode 100644
index 000000000000..40250e3f84d1
--- /dev/null
+++ b/drivers/gpu/drm/xe/xe_hmm.h
@@ -0,0 +1,18 @@
+/* SPDX-License-Identifier: MIT
+ *
+ * Copyright =C2=A9 2024 Intel Corporation
+ */
+
+#include <linux/types.h>
+
+struct xe_userptr_vma;
+
+#if IS_ENABLED(CONFIG_HMM_MIRROR)
+int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvma, bool is_mm_=
mmap_locked);
+#else
+static inline int xe_hmm_userptr_populate_range(struct xe_userptr_vma *uvm=
a, bool is_mm_mmap_locked)
+{
+	return -ENODEV;
+}
+#endif

Mentioned that I don't think=
 is needed in last rev as our driver require
CONFIG_HMM_MIRROR to be built.

= I noticed some KUnit failures in earlier patches, so I opted not to remove = it. The KUnit configuration file doesn't include the HMM_MIRROR config. If = we want to remove this we need to modify the KUnit configs to include this = config, as it is essential for our driver.

= Please confirm it is OK to add the config to kunit configuration file ?

My bad in earlier patch HMM_MIRROR was under config DRM_XE_SVM , this patch is selecting it under DRM_XE itself which is used by kunit conf file. This should be safe to remove. Will remove it in next version.

Matt


+void xe_hmm_userptr_free_=
sg(struct xe_userptr_vma *uvma);
--=20
2.25.1
--------------hLCJLUtY4ssM0iUvcKFtGG19--