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 D4E69C4167B for ; Fri, 8 Dec 2023 19:25:18 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 65A3710EAF9; Fri, 8 Dec 2023 19:25:18 +0000 (UTC) Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.13]) by gabe.freedesktop.org (Postfix) with ESMTPS id 50EB910EAF9 for ; Fri, 8 Dec 2023 19:25:16 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1702063517; x=1733599517; h=date:from:to:cc:subject:message-id:references: content-transfer-encoding:in-reply-to:mime-version; bh=DTRqgD4pOFVpUlstnGK++kdHmaZ6JrADmBmOLv/a7Uw=; b=MKsU1Sqct3UsvKU1wZHLDOijlnnQQLfz9mFxM+hSKIfAMHd5+x7Fn4PF aCseQGBd15ii1+dW8ewv8RYFAZC65J/66fXxFbuOnm+rCAY5/VYDU1cXW 1NhN9fJhe9A8859Kf2Z9RIwixNlZA37Z1oFK6A8fEl4Ku/k7Qcg6/uaLP SIlK0yq4zZqI5vewpqXwwDYjp+vSf6iZaH5eHNyPo/tGosrsmaoRrAvZ4 cZSNPgjPMn175PUJEKMQpZFj98+46SaVV6Kl5bImiwyQ/7/9WNtWgRJvb NMkZfkjIm0yWJONyZ4EAE7Jb8TkZH3AyVo1ekaICbnR1NeK1qm7G+pYbd A==; X-IronPort-AV: E=McAfee;i="6600,9927,10918"; a="1519728" X-IronPort-AV: E=Sophos;i="6.04,261,1695711600"; d="scan'208";a="1519728" Received: from orsmga007.jf.intel.com ([10.7.209.58]) by orvoesa105.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 08 Dec 2023 11:25:16 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=McAfee;i="6600,9927,10918"; a="765579507" X-IronPort-AV: E=Sophos;i="6.04,261,1695711600"; d="scan'208";a="765579507" Received: from fmsmsx603.amr.corp.intel.com ([10.18.126.83]) by orsmga007.jf.intel.com with ESMTP/TLS/AES256-GCM-SHA384; 08 Dec 2023 11:25:15 -0800 Received: from fmsmsx611.amr.corp.intel.com (10.18.126.91) 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; Fri, 8 Dec 2023 11:25:14 -0800 Received: from fmsmsx610.amr.corp.intel.com (10.18.126.90) by fmsmsx611.amr.corp.intel.com (10.18.126.91) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.35; Fri, 8 Dec 2023 11:25:14 -0800 Received: from FMSEDG603.ED.cps.intel.com (10.1.192.133) by fmsmsx610.amr.corp.intel.com (10.18.126.90) 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, 8 Dec 2023 11:25:14 -0800 Received: from NAM02-BN1-obe.outbound.protection.outlook.com (104.47.51.41) by edgegateway.intel.com (192.55.55.68) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.1.2507.35; Fri, 8 Dec 2023 11:25:13 -0800 ARC-Seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=ejADHXSdw4ONhEQdyYZyhdnA18nMASVLLwebmqNqAQEoOvyE+cnoWut6aHoPR1xwYnBLL/AXB3hjmHjSAjycP21IaxuOIu4WjSAtioN6TLJzn1zNqFtODQMOsot2nh+WWf1Z/0nDZpWZtjDcDUuVMx6yblJvYqwErWVRe5ATfltY1qopcjHMUmjelI8hxdF/x5j5mSNaeu/2VhLLW63qUjrAUJ0NvVixF8JWhQWkyiZLZLE9UublQcoZSXit1vqVLXbCsQkO1f+8KRXzNnoFZrgJKh9p6CxOV071e77LEBB2T7s5U/KR7u9BB5/Mo6DEZklSjWKO85bE1IsdkyhJPw== 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=4kXRhEmGND0LinpXouU10MvaFCnG0BYIOn9lrkLIjxE=; b=LoAArPOS0GfrBqc/fyYGOuGHxvbr0h8ASplWNXXnbDsxEJRW4/Avm2FR1DmTFxjFaLt0NMJwgm9WN0OAuyDPZW8s2liuzJDu+gVCeL33QSATOFc0cx4pCPf0gsLd9tCnEyhSfOSVxj/d8PZLbTCjCwvBogSZasuZKY8hDnCv2KZEOB+cbs58xIbflk/+LFuDw70Z2tM4bHpZHUyOkJ8XsOpGEOFAzNcRxMX1nL0QRCZplCcNJyR/2g8Z5AW+Gz0eCBE04eQCZx8SlObWnpqO6sk/pd8xY/uOHzEuBIarNyTcqtWqC3rPyaEJ1ElrOBcANZeS6FI6q12yNCjrOFln5A== 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 Authentication-Results: dkim=none (message not signed) header.d=none;dmarc=none action=none header.from=intel.com; Received: from PH7PR11MB6522.namprd11.prod.outlook.com (2603:10b6:510:212::12) by PH7PR11MB6524.namprd11.prod.outlook.com (2603:10b6:510:210::12) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.20.7068.27; Fri, 8 Dec 2023 19:25:11 +0000 Received: from PH7PR11MB6522.namprd11.prod.outlook.com ([fe80::b9a8:8221:e4a1:4cda]) by PH7PR11MB6522.namprd11.prod.outlook.com ([fe80::b9a8:8221:e4a1:4cda%4]) with mapi id 15.20.7068.028; Fri, 8 Dec 2023 19:25:11 +0000 Date: Fri, 8 Dec 2023 12:24:12 +0000 From: Matthew Brost To: Thomas =?iso-8859-1?Q?Hellstr=F6m?= Subject: Re: [RFC PATCH 7/7] drm/xe/uapi: Uniform async vs sync handling Message-ID: References: <20231207055729.438642-1-matthew.brost@intel.com> <20231207055729.438642-8-matthew.brost@intel.com> Content-Type: text/plain; charset="iso-8859-1" Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: X-ClientProxiedBy: BYAPR07CA0105.namprd07.prod.outlook.com (2603:10b6:a03:12b::46) To PH7PR11MB6522.namprd11.prod.outlook.com (2603:10b6:510:212::12) MIME-Version: 1.0 X-MS-PublicTrafficType: Email X-MS-TrafficTypeDiagnostic: PH7PR11MB6522:EE_|PH7PR11MB6524:EE_ X-MS-Office365-Filtering-Correlation-Id: b7db558f-d205-4273-23bf-08dbf8236552 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:PH7PR11MB6522.namprd11.prod.outlook.com; PTR:; CAT:NONE; SFS:(13230031)(136003)(39860400002)(396003)(376002)(346002)(366004)(230922051799003)(230173577357003)(230273577357003)(1800799012)(186009)(451199024)(64100799003)(41300700001)(30864003)(2906002)(5660300002)(8676002)(8936002)(4326008)(53546011)(966005)(6512007)(6486002)(86362001)(6506007)(478600001)(38100700002)(6916009)(66556008)(54906003)(66476007)(82960400001)(66946007)(44832011)(83380400001)(66574015)(316002)(26005); DIR:OUT; SFP:1102; X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1 X-MS-Exchange-AntiSpam-MessageData-0: =?iso-8859-1?Q?u8v50CiZuIrRKIQLogUMZpUplRPOwG4u9mwdb35gTDZCFIF3VM2Qf8Sl0O?= =?iso-8859-1?Q?ddAV2H+2QBrTuZm2jIZTvrNkKMkIB8F0KejPTz3TZ7k5OGaFLSqOcEvOcw?= =?iso-8859-1?Q?rjOdhDmo/e/KwbiRV58wGwGASo1a9XKyMFbnypwLG0Ktu9chkoS/5feAoV?= =?iso-8859-1?Q?ilZjeHemIkKLEQI0tPoGumoYJ/HlwhYc8SVinycb5i6QjgEjRjjItbNAMQ?= =?iso-8859-1?Q?7Ymuso8ytpYUXcu09LgwyohqzoZvDNqvT4m1SVuK1KfFdTaAx40MBM6Z+N?= =?iso-8859-1?Q?6w31fcqP7QFnjCoVNc7PfF6DaBu/vMmFbXVF4OirGTxs/nUyO8GqMDxkVb?= =?iso-8859-1?Q?v/i+rPZnI9iLJUoybijft5Zmut3SWe53RdRnWLmPb4tezkn7r3LhFAHJya?= =?iso-8859-1?Q?pKBY1CmyBNnbxPHxNM0pWdWmPgpApgyQ0vGuu7XbWO+3kgVfYRXMfi4vrV?= =?iso-8859-1?Q?gBUPfjjuzRDp+iQ9HgVsKrHbGBgLnyCQHVaxAtlmbilBs3ydcnhdTGFAit?= =?iso-8859-1?Q?HUtkaeaqs5rKmbWfsBxyFNvis55PhgrVy/TPCAos5sdQ7n7S7caE/jEDgg?= =?iso-8859-1?Q?jRaZBK36qHGuU2GhW/tS2HpYikc095AX3YpsgiG8t6w5pTII+tvb0AaJ/D?= =?iso-8859-1?Q?8lqSOEOeJvNdcF1MXjIkLK0m6SV+wx49Lhn/+4ZqIKI+t18bokFxfzV6uv?= =?iso-8859-1?Q?s064DDYUUz2gz4WcnMsvpPEYn+mxGlqRaPY0Q2tBhkaubPD6wvMPYYWnta?= =?iso-8859-1?Q?pVqrQ/s3AaWJncCGOekLVUPMGJJKKDrco9mUKIOQKBNfoXemnxRqfV4k0L?= =?iso-8859-1?Q?sPW8+MwzCW3kD2UsDCULiZvAn+8gsQZV3gRUBpelc8quJMycZaka593FnF?= =?iso-8859-1?Q?Jf6D5ynZAUNtiJ0n7Ap0aCP3oCRxjS9Hkn+1CKk3iXhTGK+MzDVeWevTb5?= =?iso-8859-1?Q?RmAkeYAgfMhUaGFlzpMxE1T7dDokNggehH/DFzaND1mmAjpa1Epdp3HHD3?= =?iso-8859-1?Q?cfAd0e6pqLJvwBY4SyxxIMtFKweit4ycdMxEg59FH291FKoNm0lm1JSomm?= =?iso-8859-1?Q?ujGrSGCiLD7aDcJLPhdXVBgUaJst8/dWBelN5j+aAC2wOdaZkl2P1JZjAh?= =?iso-8859-1?Q?3KGNLBq728jrHijfYBvVtpIJzuzaHiecKRYgRlNYSURB/ObsohpNoWRnW/?= =?iso-8859-1?Q?k4EoeJvfW2s/QzcUuMLj69GUq/iIpkkS2awFZrhmLPr9ZQGeoloAQO0fvv?= =?iso-8859-1?Q?Un9FSiPRZNlJ6BsNxWsspaYGRgKz7thlOXwpJQQKJd+CI3pBt0l+V3zvEQ?= =?iso-8859-1?Q?cxnYbT7OuFWyFXMdN25B4zC4bsZZ3jNsoFhbh1/4oPQgriZUtxnH8cUJ45?= =?iso-8859-1?Q?lJ+ENN1vbqnnbsWw8CPtH/pM7jSATACy5IWmH+BElljDl75hUR851h4uEF?= =?iso-8859-1?Q?79Hic3idT2id7zWTE7/+HRL7e940Cmtf+qcfQWZajQlzridcyQmKlEvcbv?= =?iso-8859-1?Q?WMdmFHIuKhLElP3c2p+5DSaptHA4r0hNCtYhWUZuoEtQ0sn/NQvYbAdGMw?= =?iso-8859-1?Q?/jek5py7ua7BqTnzSSIlLuPkxfJp4NdlEL4zwwaOhp5CLt1ig5p5Ul4ZEm?= =?iso-8859-1?Q?o7xmm3w5Xc7MxtigE9i/go5taXj+3B/Q94z5k/8nJfdK90lQyVefEtmw?= =?iso-8859-1?Q?=3D=3D?= X-MS-Exchange-CrossTenant-Network-Message-Id: b7db558f-d205-4273-23bf-08dbf8236552 X-MS-Exchange-CrossTenant-AuthSource: PH7PR11MB6522.namprd11.prod.outlook.com X-MS-Exchange-CrossTenant-AuthAs: Internal X-MS-Exchange-CrossTenant-OriginalArrivalTime: 08 Dec 2023 19:25:11.2789 (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: Vc7sZiecCFRJ8GRB/avhpVXAHveONQDOwzczLhYJwt/sJqKzcksDH9daJGAL1R1bjmLYRWqUFwGlpiCT+n0Hig== X-MS-Exchange-Transport-CrossTenantHeadersStamped: PH7PR11MB6524 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: , Cc: Francois Dugast , intel-xe@lists.freedesktop.org, Rodrigo Vivi Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" On Fri, Dec 08, 2023 at 04:00:37PM +0100, Thomas Hellström wrote: > Missed a comment, addressing below. > On 12/7/23 06:57, Matthew Brost wrote: > > Remove concept of async vs sync VM bind queues, rather make async vs > > sync a per IOCTL choice. Since this is per IOCTL, it makes sense to have > > a singular flag IOCTL rather than per VM bind op flag too. Add > > DRM_XE_SYNCS_FLAG_WAIT_FOR_OP which is an input sync flag to support > > this. Support this new flag for both the VM bind IOCTL and the exec > > IOCTL to match behavior. > > > > Cc: Rodrigo Vivi > > Cc: Thomas Hellström > > Cc: Francois Dugast > > Signed-off-by: Matthew Brost > > --- > > drivers/gpu/drm/xe/xe_exec.c | 58 ++++++++---- > > drivers/gpu/drm/xe/xe_exec_queue.c | 7 +- > > drivers/gpu/drm/xe/xe_exec_queue_types.h | 2 - > > drivers/gpu/drm/xe/xe_vm.c | 110 ++++++++++------------- > > drivers/gpu/drm/xe/xe_vm_types.h | 15 ++-- > > include/uapi/drm/xe_drm.h | 56 +++++++----- > > 6 files changed, 129 insertions(+), 119 deletions(-) > > > > diff --git a/drivers/gpu/drm/xe/xe_exec.c b/drivers/gpu/drm/xe/xe_exec.c > > index 92b0da6580e8..c62cabfaa112 100644 > > --- a/drivers/gpu/drm/xe/xe_exec.c > > +++ b/drivers/gpu/drm/xe/xe_exec.c > > @@ -130,12 +130,15 @@ static int xe_exec_begin(struct drm_exec *exec, struct xe_vm *vm) > > return err; > > } > > +#define ALL_DRM_XE_SYNCS_FLAGS (DRM_XE_SYNCS_FLAG_WAIT_FOR_OP) > > + > > int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > { > > struct xe_device *xe = to_xe_device(dev); > > struct xe_file *xef = to_xe_file(file); > > struct drm_xe_exec *args = data; > > - struct drm_xe_sync __user *syncs_user = u64_to_user_ptr(args->syncs); > > + struct drm_xe_sync __user *syncs_user = > > + u64_to_user_ptr(args->syncs.syncs); > > u64 __user *addresses_user = u64_to_user_ptr(args->address); > > struct xe_exec_queue *q; > > struct xe_sync_entry *syncs = NULL; > > @@ -143,15 +146,18 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > struct drm_exec exec; > > u32 i, num_syncs = 0; > > struct xe_sched_job *job; > > - struct dma_fence *rebind_fence; > > + struct dma_fence *rebind_fence, *job_fence; > > struct xe_vm *vm; > > - bool write_locked; > > + bool write_locked, skip_job_put = false; > > + bool wait = args->syncs.flags & DRM_XE_SYNCS_FLAG_WAIT_FOR_OP; > > ktime_t end = 0; > > int err = 0; > > if (XE_IOCTL_DBG(xe, args->extensions) || > > - XE_IOCTL_DBG(xe, args->pad[0] || args->pad[1] || args->pad[2]) || > > - XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1])) > > + XE_IOCTL_DBG(xe, args->pad || args->pad2[0] || args->pad2[1] || args->pad2[2]) || > > + XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1]) || > > + XE_IOCTL_DBG(xe, args->syncs.flags & ~ALL_DRM_XE_SYNCS_FLAGS) || > > + XE_IOCTL_DBG(xe, wait && args->syncs.num_syncs)) > > return -EINVAL; > > q = xe_exec_queue_lookup(xef, args->exec_queue_id); > > @@ -170,8 +176,9 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > goto err_exec_queue; > > } > > - if (args->num_syncs) { > > - syncs = kcalloc(args->num_syncs, sizeof(*syncs), GFP_KERNEL); > > + if (args->syncs.num_syncs) { > > + syncs = kcalloc(args->syncs.num_syncs, sizeof(*syncs), > > + GFP_KERNEL); > > if (!syncs) { > > err = -ENOMEM; > > goto err_exec_queue; > > @@ -180,7 +187,7 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > vm = q->vm; > > - for (i = 0; i < args->num_syncs; i++) { > > + for (i = 0; i < args->syncs.num_syncs; i++) { > > err = xe_sync_entry_parse(xe, xef, &syncs[num_syncs++], > > &syncs_user[i], SYNC_PARSE_FLAG_EXEC | > > (xe_vm_in_lr_mode(vm) ? > > @@ -245,9 +252,17 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > err = PTR_ERR(fence); > > goto err_exec; > > } > > + > > for (i = 0; i < num_syncs; i++) > > xe_sync_entry_signal(&syncs[i], NULL, fence); > > + > > xe_exec_queue_last_fence_set(q, vm, fence); > > + if (wait) { > > + long timeout = dma_fence_wait(fence, true); > > + > > + if (timeout < 0) > > + err = -EINTR; > > + } > > Here it looks like we will rerun the same IOCTL again if we return -EINTR. > The user-space expected action on -EINTR is to just restart the IOCTL > without any argument changes. Solution is to add an ioctl argument cookie > (or to skip sync vm binds and have the user just use the 0 batch buffers / > vm_binds calls or wait for an out-fence). If you go for the cookie solution > then IMO we should keep the -ERESTARTSYS returned from dma_fence_wait() > since it's converted to -EINTR on return-to-user-space, and the kernel > restarts the IOCTL automatically if there was no requested-for-delivery > signal pending. > > I think the simplest solution at this point is to skip the sync behaviour, > in particular if we enable the 0 batch / bind possibility. > > If we still want to provide it, we could add a cookie address as an > extension to the ioctl and activate sync if present? (Just throwing up ideas > here). > > > dma_fence_put(fence); > > } > > @@ -331,42 +346,51 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > * the job and let the DRM scheduler / backend clean up the job. > > */ > > xe_sched_job_arm(job); > > + job_fence = &job->drm.s_fence->finished; > > + if (wait) > > + dma_fence_get(job_fence); > > if (!xe_vm_in_lr_mode(vm)) { > > /* Block userptr invalidations / BO eviction */ > > - dma_resv_add_fence(&vm->resv, > > - &job->drm.s_fence->finished, > > + dma_resv_add_fence(&vm->resv, job_fence, > > DMA_RESV_USAGE_BOOKKEEP); > > /* > > * Make implicit sync work across drivers, assuming all external > > * BOs are written as we don't pass in a read / write list. > > */ > > - xe_vm_fence_all_extobjs(vm, &job->drm.s_fence->finished, > > - DMA_RESV_USAGE_WRITE); > > + xe_vm_fence_all_extobjs(vm, job_fence, DMA_RESV_USAGE_WRITE); > > } > > for (i = 0; i < num_syncs; i++) > > - xe_sync_entry_signal(&syncs[i], job, > > - &job->drm.s_fence->finished); > > + xe_sync_entry_signal(&syncs[i], job, job_fence); > > if (xe_exec_queue_is_lr(q)) > > q->ring_ops->emit_job(job); > > if (!xe_vm_in_lr_mode(vm)) > > - xe_exec_queue_last_fence_set(q, vm, &job->drm.s_fence->finished); > > + xe_exec_queue_last_fence_set(q, vm, job_fence); > > xe_sched_job_push(job); > > xe_vm_reactivate_rebind(vm); > > - if (!err && !xe_vm_in_lr_mode(vm)) { > > + if (!xe_vm_in_lr_mode(vm)) { > > spin_lock(&xe->ttm.lru_lock); > > ttm_lru_bulk_move_tail(&vm->lru_bulk_move); > > spin_unlock(&xe->ttm.lru_lock); > > } > > + skip_job_put = true; > > + if (wait) { > > + long timeout = dma_fence_wait(job_fence, true); > > + > > + dma_fence_put(job_fence); > > + if (timeout < 0) > > + err = -EINTR; > > + } > > + > > err_repin: > > if (!xe_vm_in_lr_mode(vm)) > > up_read(&vm->userptr.notifier_lock); > > err_put_job: > > - if (err) > > + if (err && !skip_job_put) > > xe_sched_job_put(job); > > err_exec: > > drm_exec_fini(&exec); > > diff --git a/drivers/gpu/drm/xe/xe_exec_queue.c b/drivers/gpu/drm/xe/xe_exec_queue.c > > index 3911d14522ee..98776d02d634 100644 > > --- a/drivers/gpu/drm/xe/xe_exec_queue.c > > +++ b/drivers/gpu/drm/xe/xe_exec_queue.c > > @@ -625,10 +625,7 @@ int xe_exec_queue_create_ioctl(struct drm_device *dev, void *data, > > if (XE_IOCTL_DBG(xe, eci[0].gt_id >= xe->info.gt_count)) > > return -EINVAL; > > - if (eci[0].engine_class >= DRM_XE_ENGINE_CLASS_VM_BIND_ASYNC) { > > - bool sync = eci[0].engine_class == > > - DRM_XE_ENGINE_CLASS_VM_BIND_SYNC; > > - > > + if (eci[0].engine_class == DRM_XE_ENGINE_CLASS_VM_BIND) { > > for_each_gt(gt, xe, id) { > > struct xe_exec_queue *new; > > @@ -654,8 +651,6 @@ int xe_exec_queue_create_ioctl(struct drm_device *dev, void *data, > > args->width, hwe, > > EXEC_QUEUE_FLAG_PERSISTENT | > > EXEC_QUEUE_FLAG_VM | > > - (sync ? 0 : > > - EXEC_QUEUE_FLAG_VM_ASYNC) | > > (id ? > > EXEC_QUEUE_FLAG_BIND_ENGINE_CHILD : > > 0)); > > diff --git a/drivers/gpu/drm/xe/xe_exec_queue_types.h b/drivers/gpu/drm/xe/xe_exec_queue_types.h > > index 52f0927d0d9b..c78f6e8b41c4 100644 > > --- a/drivers/gpu/drm/xe/xe_exec_queue_types.h > > +++ b/drivers/gpu/drm/xe/xe_exec_queue_types.h > > @@ -74,8 +74,6 @@ struct xe_exec_queue { > > #define EXEC_QUEUE_FLAG_VM BIT(4) > > /* child of VM queue for multi-tile VM jobs */ > > #define EXEC_QUEUE_FLAG_BIND_ENGINE_CHILD BIT(5) > > -/* VM jobs for this queue are asynchronous */ > > -#define EXEC_QUEUE_FLAG_VM_ASYNC BIT(6) > > /** > > * @flags: flags for this exec queue, should statically setup aside from ban > > diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c > > index cf2eb44a71db..4b0c976c003a 100644 > > --- a/drivers/gpu/drm/xe/xe_vm.c > > +++ b/drivers/gpu/drm/xe/xe_vm.c > > @@ -1433,9 +1433,7 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags) > > struct xe_gt *gt = tile->primary_gt; > > struct xe_vm *migrate_vm; > > struct xe_exec_queue *q; > > - u32 create_flags = EXEC_QUEUE_FLAG_VM | > > - ((flags & XE_VM_FLAG_ASYNC_DEFAULT) ? > > - EXEC_QUEUE_FLAG_VM_ASYNC : 0); > > + u32 create_flags = EXEC_QUEUE_FLAG_VM; > > if (!vm->pt_root[id]) > > continue; > > @@ -1835,16 +1833,10 @@ xe_vm_bind_vma(struct xe_vma *vma, struct xe_exec_queue *q, > > return ERR_PTR(err); > > } > > -static bool xe_vm_sync_mode(struct xe_vm *vm, struct xe_exec_queue *q) > > -{ > > - return q ? !(q->flags & EXEC_QUEUE_FLAG_VM_ASYNC) : > > - !(vm->flags & XE_VM_FLAG_ASYNC_DEFAULT); > > -} > > - > > static int __xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma, > > struct xe_exec_queue *q, struct xe_sync_entry *syncs, > > u32 num_syncs, bool immediate, bool first_op, > > - bool last_op) > > + bool last_op, bool async) > > { > > struct dma_fence *fence; > > struct xe_exec_queue *wait_exec_queue = to_wait_exec_queue(vm, q); > > @@ -1870,7 +1862,7 @@ static int __xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma, > > if (last_op) > > xe_exec_queue_last_fence_set(wait_exec_queue, vm, fence); > > - if (last_op && xe_vm_sync_mode(vm, q)) > > + if (last_op && !async) > > dma_fence_wait(fence, true); > > dma_fence_put(fence); > > @@ -1880,7 +1872,7 @@ static int __xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma, > > static int xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma, struct xe_exec_queue *q, > > struct xe_bo *bo, struct xe_sync_entry *syncs, > > u32 num_syncs, bool immediate, bool first_op, > > - bool last_op) > > + bool last_op, bool async) > > { > > int err; > > @@ -1894,12 +1886,12 @@ static int xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma, struct xe_exec_queue > > } > > return __xe_vm_bind(vm, vma, q, syncs, num_syncs, immediate, first_op, > > - last_op); > > + last_op, async); > > } > > static int xe_vm_unbind(struct xe_vm *vm, struct xe_vma *vma, > > struct xe_exec_queue *q, struct xe_sync_entry *syncs, > > - u32 num_syncs, bool first_op, bool last_op) > > + u32 num_syncs, bool first_op, bool last_op, bool async) > > { > > struct dma_fence *fence; > > struct xe_exec_queue *wait_exec_queue = to_wait_exec_queue(vm, q); > > @@ -1914,7 +1906,7 @@ static int xe_vm_unbind(struct xe_vm *vm, struct xe_vma *vma, > > xe_vma_destroy(vma, fence); > > if (last_op) > > xe_exec_queue_last_fence_set(wait_exec_queue, vm, fence); > > - if (last_op && xe_vm_sync_mode(vm, q)) > > + if (last_op && !async) > > dma_fence_wait(fence, true); > > It looks like we're dropping the error return code here. > I am aware of this. This is fixed in the larger refactor of the VM bind error handling [1]. The idea with this series is land the uAPI and get the implementation 100% correct in the larger follow up series. Matt [1] https://patchwork.freedesktop.org/series/125608/ > > > dma_fence_put(fence); > > @@ -1923,7 +1915,6 @@ static int xe_vm_unbind(struct xe_vm *vm, struct xe_vma *vma, > > #define ALL_DRM_XE_VM_CREATE_FLAGS (DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE | \ > > DRM_XE_VM_CREATE_FLAG_LR_MODE | \ > > - DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT | \ > > DRM_XE_VM_CREATE_FLAG_FAULT_MODE) > > int xe_vm_create_ioctl(struct drm_device *dev, void *data, > > @@ -1977,8 +1968,6 @@ int xe_vm_create_ioctl(struct drm_device *dev, void *data, > > flags |= XE_VM_FLAG_SCRATCH_PAGE; > > if (args->flags & DRM_XE_VM_CREATE_FLAG_LR_MODE) > > flags |= XE_VM_FLAG_LR_MODE; > > - if (args->flags & DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT) > > - flags |= XE_VM_FLAG_ASYNC_DEFAULT; > > if (args->flags & DRM_XE_VM_CREATE_FLAG_FAULT_MODE) > > flags |= XE_VM_FLAG_FAULT_MODE; > > @@ -2062,7 +2051,7 @@ static const u32 region_to_mem_type[] = { > > static int xe_vm_prefetch(struct xe_vm *vm, struct xe_vma *vma, > > struct xe_exec_queue *q, u32 region, > > struct xe_sync_entry *syncs, u32 num_syncs, > > - bool first_op, bool last_op) > > + bool first_op, bool last_op, bool async) > > { > > struct xe_exec_queue *wait_exec_queue = to_wait_exec_queue(vm, q); > > int err; > > @@ -2077,7 +2066,7 @@ static int xe_vm_prefetch(struct xe_vm *vm, struct xe_vma *vma, > > if (vma->tile_mask != (vma->tile_present & ~vma->usm.tile_invalidated)) { > > return xe_vm_bind(vm, vma, q, xe_vma_bo(vma), syncs, num_syncs, > > - true, first_op, last_op); > > + true, first_op, last_op, async); > > } else { > > int i; > > @@ -2400,6 +2389,8 @@ static int vm_bind_ioctl_ops_parse(struct xe_vm *vm, struct xe_exec_queue *q, > > } > > op->q = q; > > + if (async) > > + op->flags |= XE_VMA_OP_ASYNC; > > switch (op->base.op) { > > case DRM_GPUVA_OP_MAP: > > @@ -2538,7 +2529,8 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm, > > op->syncs, op->num_syncs, > > op->map.immediate || !xe_vm_in_fault_mode(vm), > > op->flags & XE_VMA_OP_FIRST, > > - op->flags & XE_VMA_OP_LAST); > > + op->flags & XE_VMA_OP_LAST, > > + op->flags & XE_VMA_OP_ASYNC); > > break; > > case DRM_GPUVA_OP_REMAP: > > { > > @@ -2552,7 +2544,8 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm, > > op->num_syncs, > > op->flags & XE_VMA_OP_FIRST, > > op->flags & XE_VMA_OP_LAST && > > - !prev && !next); > > + !prev && !next, > > + op->flags & XE_VMA_OP_ASYNC); > > if (err) > > break; > > op->remap.unmap_done = true; > > @@ -2563,7 +2556,8 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm, > > err = xe_vm_bind(vm, op->remap.prev, op->q, > > xe_vma_bo(op->remap.prev), op->syncs, > > op->num_syncs, true, false, > > - op->flags & XE_VMA_OP_LAST && !next); > > + op->flags & XE_VMA_OP_LAST && !next, > > + op->flags & XE_VMA_OP_ASYNC); > > op->remap.prev->gpuva.flags &= ~XE_VMA_LAST_REBIND; > > if (err) > > break; > > @@ -2576,7 +2570,8 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm, > > xe_vma_bo(op->remap.next), > > op->syncs, op->num_syncs, > > true, false, > > - op->flags & XE_VMA_OP_LAST); > > + op->flags & XE_VMA_OP_LAST, > > + op->flags & XE_VMA_OP_ASYNC); > > op->remap.next->gpuva.flags &= ~XE_VMA_LAST_REBIND; > > if (err) > > break; > > @@ -2588,13 +2583,15 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm, > > case DRM_GPUVA_OP_UNMAP: > > err = xe_vm_unbind(vm, vma, op->q, op->syncs, > > op->num_syncs, op->flags & XE_VMA_OP_FIRST, > > - op->flags & XE_VMA_OP_LAST); > > + op->flags & XE_VMA_OP_LAST, > > + op->flags & XE_VMA_OP_ASYNC); > > break; > > case DRM_GPUVA_OP_PREFETCH: > > err = xe_vm_prefetch(vm, vma, op->q, op->prefetch.region, > > op->syncs, op->num_syncs, > > op->flags & XE_VMA_OP_FIRST, > > - op->flags & XE_VMA_OP_LAST); > > + op->flags & XE_VMA_OP_LAST, > > + op->flags & XE_VMA_OP_ASYNC); > > break; > > default: > > drm_warn(&vm->xe->drm, "NOT POSSIBLE"); > > @@ -2808,16 +2805,16 @@ static int vm_bind_ioctl_ops_execute(struct xe_vm *vm, > > #ifdef TEST_VM_ASYNC_OPS_ERROR > > #define SUPPORTED_FLAGS \ > > - (FORCE_ASYNC_OP_ERROR | DRM_XE_VM_BIND_FLAG_ASYNC | \ > > - DRM_XE_VM_BIND_FLAG_READONLY | DRM_XE_VM_BIND_FLAG_IMMEDIATE | \ > > - DRM_XE_VM_BIND_FLAG_NULL | 0xffff) > > + (FORCE_ASYNC_OP_ERROR | DRM_XE_VM_BIND_FLAG_READONLY | \ > > + DRM_XE_VM_BIND_FLAG_IMMEDIATE | DRM_XE_VM_BIND_FLAG_NULL | 0xffff) > > #else > > #define SUPPORTED_FLAGS \ > > - (DRM_XE_VM_BIND_FLAG_ASYNC | DRM_XE_VM_BIND_FLAG_READONLY | \ > > + (DRM_XE_VM_BIND_FLAG_READONLY | \ > > DRM_XE_VM_BIND_FLAG_IMMEDIATE | DRM_XE_VM_BIND_FLAG_NULL | \ > > 0xffff) > > #endif > > #define XE_64K_PAGE_MASK 0xffffull > > +#define ALL_DRM_XE_SYNCS_FLAGS (DRM_XE_SYNCS_FLAG_WAIT_FOR_OP) > > #define MAX_BINDS 512 /* FIXME: Picking random upper limit */ > > @@ -2829,7 +2826,7 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, > > int err; > > int i; > > - if (XE_IOCTL_DBG(xe, args->pad || args->pad2) || > > + if (XE_IOCTL_DBG(xe, args->pad) || > > XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1])) > > return -EINVAL; > > @@ -2857,6 +2854,14 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, > > *bind_ops = &args->bind; > > } > > + *async = !(args->syncs.flags & DRM_XE_SYNCS_FLAG_WAIT_FOR_OP); > > + > > + if (XE_IOCTL_DBG(xe, args->syncs.flags & ~ALL_DRM_XE_SYNCS_FLAGS) || > > + XE_IOCTL_DBG(xe, !*async && args->syncs.num_syncs)) { > > + err = -EINVAL; > > + goto free_bind_ops; > > + } > > + > > for (i = 0; i < args->num_binds; ++i) { > > u64 range = (*bind_ops)[i].range; > > u64 addr = (*bind_ops)[i].addr; > > @@ -2887,18 +2892,6 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, > > goto free_bind_ops; > > } > > - if (i == 0) { > > - *async = !!(flags & DRM_XE_VM_BIND_FLAG_ASYNC); > > - if (XE_IOCTL_DBG(xe, !*async && args->num_syncs)) { > > - err = -EINVAL; > > - goto free_bind_ops; > > - } > > - } else if (XE_IOCTL_DBG(xe, *async != > > - !!(flags & DRM_XE_VM_BIND_FLAG_ASYNC))) { > > - err = -EINVAL; > > - goto free_bind_ops; > > - } > > - > > if (XE_IOCTL_DBG(xe, op > DRM_XE_VM_BIND_OP_PREFETCH) || > > XE_IOCTL_DBG(xe, flags & ~SUPPORTED_FLAGS) || > > XE_IOCTL_DBG(xe, obj && is_null) || > > @@ -2951,7 +2944,7 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe, > > static int vm_bind_ioctl_signal_fences(struct xe_vm *vm, > > struct xe_exec_queue *q, > > struct xe_sync_entry *syncs, > > - int num_syncs) > > + int num_syncs, bool async) > > { > > struct dma_fence *fence; > > int i, err = 0; > > @@ -2967,7 +2960,7 @@ static int vm_bind_ioctl_signal_fences(struct xe_vm *vm, > > xe_exec_queue_last_fence_set(to_wait_exec_queue(vm, q), vm, > > fence); > > - if (xe_vm_sync_mode(vm, q)) { > > + if (!async) { > > long timeout = dma_fence_wait(fence, true); > > if (timeout < 0) > > @@ -3001,7 +2994,7 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > if (err) > > return err; > > - if (XE_IOCTL_DBG(xe, args->pad || args->pad2) || > > + if (XE_IOCTL_DBG(xe, args->pad) || > > XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1])) > > return -EINVAL; > > @@ -3016,12 +3009,6 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > err = -EINVAL; > > goto put_exec_queue; > > } > > - > > - if (XE_IOCTL_DBG(xe, args->num_binds && async != > > - !!(q->flags & EXEC_QUEUE_FLAG_VM_ASYNC))) { > > - err = -EINVAL; > > - goto put_exec_queue; > > - } > > } > > vm = xe_vm_lookup(xef, args->vm_id); > > @@ -3030,14 +3017,6 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > goto put_exec_queue; > > } > > - if (!args->exec_queue_id) { > > - if (XE_IOCTL_DBG(xe, args->num_binds && async != > > - !!(vm->flags & XE_VM_FLAG_ASYNC_DEFAULT))) { > > - err = -EINVAL; > > - goto put_vm; > > - } > > - } > > - > > err = down_write_killable(&vm->lock); > > if (err) > > goto put_vm; > > @@ -3127,16 +3106,16 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > } > > } > > - if (args->num_syncs) { > > - syncs = kcalloc(args->num_syncs, sizeof(*syncs), GFP_KERNEL); > > + if (args->syncs.num_syncs) { > > + syncs = kcalloc(args->syncs.num_syncs, sizeof(*syncs), GFP_KERNEL); > > if (!syncs) { > > err = -ENOMEM; > > goto put_obj; > > } > > } > > - syncs_user = u64_to_user_ptr(args->syncs); > > - for (num_syncs = 0; num_syncs < args->num_syncs; num_syncs++) { > > + syncs_user = u64_to_user_ptr(args->syncs.syncs); > > + for (num_syncs = 0; num_syncs < args->syncs.num_syncs; num_syncs++) { > > err = xe_sync_entry_parse(xe, xef, &syncs[num_syncs], > > &syncs_user[num_syncs], > > (xe_vm_in_lr_mode(vm) ? > > @@ -3210,7 +3189,8 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file) > > vm_bind_ioctl_ops_unwind(vm, ops, args->num_binds); > > free_syncs: > > if (err == -ENODATA) > > - err = vm_bind_ioctl_signal_fences(vm, q, syncs, num_syncs); > > + err = vm_bind_ioctl_signal_fences(vm, q, syncs, num_syncs, > > + async); > > while (num_syncs--) > > xe_sync_entry_cleanup(&syncs[num_syncs]); > > diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h > > index 23abdfd8622f..ce8b9bde7e9c 100644 > > --- a/drivers/gpu/drm/xe/xe_vm_types.h > > +++ b/drivers/gpu/drm/xe/xe_vm_types.h > > @@ -167,13 +167,12 @@ struct xe_vm { > > */ > > #define XE_VM_FLAG_64K BIT(0) > > #define XE_VM_FLAG_LR_MODE BIT(1) > > -#define XE_VM_FLAG_ASYNC_DEFAULT BIT(2) > > -#define XE_VM_FLAG_MIGRATION BIT(3) > > -#define XE_VM_FLAG_SCRATCH_PAGE BIT(4) > > -#define XE_VM_FLAG_FAULT_MODE BIT(5) > > -#define XE_VM_FLAG_BANNED BIT(6) > > -#define XE_VM_FLAG_TILE_ID(flags) FIELD_GET(GENMASK(8, 7), flags) > > -#define XE_VM_FLAG_SET_TILE_ID(tile) FIELD_PREP(GENMASK(8, 7), (tile)->id) > > +#define XE_VM_FLAG_MIGRATION BIT(2) > > +#define XE_VM_FLAG_SCRATCH_PAGE BIT(3) > > +#define XE_VM_FLAG_FAULT_MODE BIT(4) > > +#define XE_VM_FLAG_BANNED BIT(5) > > +#define XE_VM_FLAG_TILE_ID(flags) FIELD_GET(GENMASK(7, 6), flags) > > +#define XE_VM_FLAG_SET_TILE_ID(tile) FIELD_PREP(GENMASK(7, 6), (tile)->id) > > unsigned long flags; > > /** @composite_fence_ctx: context composite fence */ > > @@ -385,6 +384,8 @@ enum xe_vma_op_flags { > > XE_VMA_OP_PREV_COMMITTED = BIT(3), > > /** @XE_VMA_OP_NEXT_COMMITTED: Next VMA operation committed */ > > XE_VMA_OP_NEXT_COMMITTED = BIT(4), > > + /** @XE_VMA_OP_ASYNC: operation is async */ > > + XE_VMA_OP_ASYNC = BIT(5), > > }; > > /** struct xe_vma_op - VMA operation */ > > diff --git a/include/uapi/drm/xe_drm.h b/include/uapi/drm/xe_drm.h > > index eb03a49c17a1..fd8172fe2d9a 100644 > > --- a/include/uapi/drm/xe_drm.h > > +++ b/include/uapi/drm/xe_drm.h > > @@ -141,8 +141,7 @@ struct drm_xe_engine_class_instance { > > * Kernel only classes (not actual hardware engine class). Used for > > * creating ordered queues of VM bind operations. > > */ > > -#define DRM_XE_ENGINE_CLASS_VM_BIND_ASYNC 5 > > -#define DRM_XE_ENGINE_CLASS_VM_BIND_SYNC 6 > > +#define DRM_XE_ENGINE_CLASS_VM_BIND 5 > > __u16 engine_class; > > __u16 engine_instance; > > @@ -660,7 +659,6 @@ struct drm_xe_vm_create { > > * still enable recoverable pagefaults if supported by the device. > > */ > > #define DRM_XE_VM_CREATE_FLAG_LR_MODE (1 << 1) > > -#define DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT (1 << 2) > > /* > > * DRM_XE_VM_CREATE_FLAG_FAULT_MODE requires also > > * DRM_XE_VM_CREATE_FLAG_LR_MODE. It allows memory to be allocated > > @@ -668,7 +666,7 @@ struct drm_xe_vm_create { > > * The xe driver internally uses recoverable pagefaults to implement > > * this. > > */ > > -#define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 3) > > +#define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 2) > > /** @flags: Flags */ > > __u32 flags; > > @@ -776,12 +774,11 @@ struct drm_xe_vm_bind_op { > > __u32 op; > > #define DRM_XE_VM_BIND_FLAG_READONLY (1 << 0) > > -#define DRM_XE_VM_BIND_FLAG_ASYNC (1 << 1) > > /* > > * Valid on a faulting VM only, do the MAP operation immediately rather > > * than deferring the MAP to the page fault handler. > > */ > > -#define DRM_XE_VM_BIND_FLAG_IMMEDIATE (1 << 2) > > +#define DRM_XE_VM_BIND_FLAG_IMMEDIATE (1 << 1) > > /* > > * 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 > > @@ -789,7 +786,7 @@ struct drm_xe_vm_bind_op { > > * operations, the BO handle MBZ, and the BO offset MBZ. This flag is > > * intended to implement VK sparse bindings. > > */ > > -#define DRM_XE_VM_BIND_FLAG_NULL (1 << 3) > > +#define DRM_XE_VM_BIND_FLAG_NULL (1 << 2) > > /** @flags: Bind flags */ > > __u32 flags; > > @@ -807,6 +804,27 @@ struct drm_xe_vm_bind_op { > > __u64 reserved[3]; > > }; > > +/** > > + * struct drm_xe_syncs - In / out syncs for IOCTLs. > > + */ > > +struct drm_xe_syncs { > > + /** @num_syncs: amount of syncs to wait on */ > > + __u32 num_syncs; > > + > > + /* > > + * Block in IOCTL until operation complete, num_syncs MBZ if set. > > + */ > > +#define DRM_XE_SYNCS_FLAG_WAIT_FOR_OP (1 << 0) > > + /** @flags: Sync flags */ > > + __u32 flags; > > + > > + /** @syncs: pointer to struct drm_xe_sync array */ > > + __u64 syncs; > > + > > + /** @reserved: Reserved */ > > + __u64 reserved[2]; > > +}; > > + > > struct drm_xe_vm_bind { > > /** @extensions: Pointer to the first extension struct, if any */ > > __u64 extensions; > > @@ -838,14 +856,8 @@ struct drm_xe_vm_bind { > > __u64 vector_of_binds; > > }; > > - /** @pad: MBZ */ > > - __u32 pad2; > > - > > - /** @num_syncs: amount of syncs to wait on */ > > - __u32 num_syncs; > > - > > - /** @syncs: pointer to struct drm_xe_sync array */ > > - __u64 syncs; > > + /** @syncs: syncs for bind */ > > + struct drm_xe_syncs syncs; > > /** @reserved: Reserved */ > > __u64 reserved[2]; > > @@ -974,14 +986,14 @@ struct drm_xe_exec { > > /** @extensions: Pointer to the first extension struct, if any */ > > __u64 extensions; > > + /** @pad: MBZ */ > > + __u32 pad; > > + > > /** @exec_queue_id: Exec queue ID for the batch buffer */ > > __u32 exec_queue_id; > > - /** @num_syncs: Amount of struct drm_xe_sync in array. */ > > - __u32 num_syncs; > > - > > - /** @syncs: Pointer to struct drm_xe_sync array. */ > > - __u64 syncs; > > + /** @syncs: syncs for exec */ > > + struct drm_xe_syncs syncs; > > /** > > * @address: address of batch buffer if num_batch_buffer == 1 or an > > @@ -995,8 +1007,8 @@ struct drm_xe_exec { > > */ > > __u16 num_batch_buffer; > > - /** @pad: MBZ */ > > - __u16 pad[3]; > > + /** @pad2: MBZ */ > > + __u16 pad2[3]; > > /** @reserved: Reserved */ > > __u64 reserved[2];