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 11132C197A0 for ; Mon, 20 Nov 2023 15:39:05 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id D420B10E40D; Mon, 20 Nov 2023 15:39:04 +0000 (UTC) Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.10]) by gabe.freedesktop.org (Postfix) with ESMTPS id 6251D10E40C for ; Mon, 20 Nov 2023 15:39:02 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1700494742; x=1732030742; h=message-id:date:mime-version:subject:to:cc:references: from:in-reply-to:content-transfer-encoding; bh=V+CEb+xuR8SOd2/QvbP9xkI4u58qGDK0v3ufMnPzWJg=; b=eMdsX+IppzOgdfrqGIqhLZJD5+IPXVJFYOimlgE8ttrlKOjmH3yVldd4 YlnKv21UurxmlEEXK4h72dvIJjhet8UrmtYa3quD+knYDYlYK9A7VXlLB yKD8iAcg9m2CJ5Zyc3fWZrzXUaAiVWzkkbwaDI0iZM1bl+FTPLWlknJ9T 1EYbZfF3ItcIAKv/JoZsAmb7VAm4oxY1QTc6CHxVu4l3ScEo4umHvc+4o Kaxz1Mo9g5V9kJ14OYS+fPEFkS/zoUmIKPCkr+ISMRk7xOyptdqFlEMBZ 5w4S65h4zA3tp/nB6brUYfXdic4NLfkXLVcArpplSwhB0okjKWpuA7wWs w==; X-IronPort-AV: E=McAfee;i="6600,9927,10900"; a="4828580" X-IronPort-AV: E=Sophos;i="6.04,214,1695711600"; d="scan'208";a="4828580" Received: from orviesa001.jf.intel.com ([10.64.159.141]) by orvoesa102.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 20 Nov 2023 07:39:02 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.04,214,1695711600"; d="scan'208";a="14195132" Received: from mgrott-mobl1.ger.corp.intel.com (HELO [10.252.6.126]) ([10.252.6.126]) by smtpauth.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 20 Nov 2023 07:39:01 -0800 Message-ID: <4da6214d-96c6-4aca-a48e-91d6dbd3e107@intel.com> Date: Mon, 20 Nov 2023 15:38:58 +0000 MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Content-Language: en-GB To: =?UTF-8?Q?Thomas_Hellstr=C3=B6m?= , intel-xe@lists.freedesktop.org References: <20231009090037.29132-2-matthew.auld@intel.com> From: Matthew Auld In-Reply-To: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Subject: Re: [Intel-xe] [PATCH] drm/xe/bo: don't hold dma-resv lock over drm_gem_handle_create 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: Rodrigo Vivi Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" On 20/11/2023 15:31, Thomas Hellström wrote: > > On 10/9/23 11:00, Matthew Auld wrote: >> This seems to create a locking inversion with object_name_lock. The lock >> is held by drm_prime_fd_to_handle when calling our xe_gem_prime_import >> hook, which might eventually go on to grab the dma-resv lock during the >> attach. However we also have the opposite locking order in >> xe_gem_create_ioctl which is holding the dma-resv lock when calling >> drm_gem_handle_create, which wants to eventually grab object_name_lock: >> >> -> #1 (reservation_ww_class_mutex){+.+.}-{3:3}: >> <4> [635.739288]        lock_acquire+0x169/0x3d0 >> <4> [635.739294]        __ww_mutex_lock.constprop.0+0x164/0x1e60 >> <4> [635.739300]        ww_mutex_lock_interruptible+0x42/0x1a0 >> <4> [635.739305]        drm_gem_shmem_pin+0x4b/0x140 [drm_shmem_helper] >> <4> [635.739317]        dma_buf_dynamic_attach+0x101/0x430 >> <4> [635.739323]        xe_gem_prime_import+0xcc/0x2e0 [xe] >> <4> [635.739499]        drm_prime_fd_to_handle_ioctl+0x184/0x2e0 [drm] >> <4> [635.739594]        drm_ioctl_kernel+0x16f/0x250 [drm] >> <4> [635.739693]        drm_ioctl+0x35e/0x620 [drm] >> <4> [635.739789]        __x64_sys_ioctl+0xb7/0xf0 >> <4> [635.739794]        do_syscall_64+0x3c/0x90 >> <4> [635.739799]        entry_SYSCALL_64_after_hwframe+0x6e/0xd8 >> <4> [635.739805] >> -> #0 (&dev->object_name_lock){+.+.}-{3:3}: >> <4> [635.739813]        check_prev_add+0x1ba/0x14a0 >> <4> [635.739818]        __lock_acquire+0x203e/0x2ff0 >> <4> [635.739823]        lock_acquire+0x169/0x3d0 >> <4> [635.739827]        __mutex_lock+0x124/0x1310 >> <4> [635.739832]        drm_gem_handle_create+0x32/0x50 [drm] >> <4> [635.739927]        xe_gem_create_ioctl+0x1d3/0x550 [xe] >> <4> [635.740102]        drm_ioctl_kernel+0x16f/0x250 [drm] >> <4> [635.740197]        drm_ioctl+0x35e/0x620 [drm] >> <4> [635.740293]        __x64_sys_ioctl+0xb7/0xf0 >> <4> [635.740297]        do_syscall_64+0x3c/0x90 >> <4> [635.740302]        entry_SYSCALL_64_after_hwframe+0x6e/0xd8 >> <4> [635.740307] >> >> It looks like it should be safe to simply drop the dma-resv lock prior >> to publishing the object when calling drm_gem_handle_create. >> >> Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/743 >> Signed-off-by: Matthew Auld >> Cc: Thomas Hellström >> Cc: Rodrigo Vivi > > LGTM. Reviewed-by: Thomas Hellström Thanks. > > > >> --- >>   drivers/gpu/drm/xe/xe_bo.c | 20 ++++++++++++-------- >>   1 file changed, 12 insertions(+), 8 deletions(-) >> >> diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c >> index 61789c0e88fb..ff116d3a8861 100644 >> --- a/drivers/gpu/drm/xe/xe_bo.c >> +++ b/drivers/gpu/drm/xe/xe_bo.c >> @@ -1815,14 +1815,16 @@ int xe_gem_create_ioctl(struct drm_device >> *dev, void *data, >>           if (XE_IOCTL_DBG(xe, !vm)) >>               return -ENOENT; >>           err = xe_vm_lock(vm, true); >> -        if (err) { >> -            xe_vm_put(vm); >> -            return err; >> -        } >> +        if (err) >> +            goto out_vm; >>       } >>       bo = xe_bo_create(xe, NULL, vm, args->size, ttm_bo_type_device, >>                 bo_flags); >> + >> +    if (vm) >> +        xe_vm_unlock(vm); >> + >>       if (IS_ERR(bo)) { >>           err = PTR_ERR(bo); >>           goto out_vm; >> @@ -1836,15 +1838,17 @@ int xe_gem_create_ioctl(struct drm_device >> *dev, void *data, >>       goto out_put; >>   out_bulk: >> -    if (vm && !xe_vm_in_fault_mode(vm)) >> +    if (vm && !xe_vm_in_fault_mode(vm)) { >> +        xe_vm_lock(vm, false); >>           __xe_bo_unset_bulk_move(bo); >> +        xe_vm_unlock(vm); >> +    } >>   out_put: >>       xe_bo_put(bo); >>   out_vm: >> -    if (vm) { >> -        xe_vm_unlock(vm); >> +    if (vm) >>           xe_vm_put(vm); >> -    } >> + >>       return err; >>   }