From patchwork Mon Jul 4 13:52:48 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Thomas Hellstrom X-Patchwork-Id: 12905667 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 955FEC433EF for ; Mon, 4 Jul 2022 16:23:43 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 16F8A10F8C7; Mon, 4 Jul 2022 16:16:35 +0000 (UTC) Received: from mga07.intel.com (mga07.intel.com [134.134.136.100]) by gabe.freedesktop.org (Postfix) with ESMTPS id 965D010E14A; Mon, 4 Jul 2022 13:53:26 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1656942804; x=1688478804; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=gi+steff+c6wtID9tUVp9+ZOlW7suAdBL8AT0jTG1sY=; b=mGfbNx5F8Ocx13KerJMtbCiQXhqoTrH1lfRx+s48+NeRmq1kDnhDKuBX qrim/q+3d90Xy/ndGhhFKa0r3y1xuHwlQSytExfk0Cg0pqeJ2/oWNlQQ4 8fVCNQq4J1PkfxMMWVPE5kd1v0FnDgqHBL90DdSfdGtEaM9jZBGBthncB mmz3QZ5aRKmz45TsIY3wm9EYNyzZMwyZ6n3ltArWaIMSiyAAL4gM0zbdk irLso6c+qAWdPKuhscSf8vU/BwsSAlaay4lH2gy0StT7IV1mP/ObMN44K w5VpR9Tu6rpovJcbswtSqhSfvYq31uBkEZJCjXtL0P9WQ/FuVpvg1EvYt A==; X-IronPort-AV: E=McAfee;i="6400,9594,10397"; a="347112573" X-IronPort-AV: E=Sophos;i="5.92,243,1650956400"; d="scan'208";a="347112573" Received: from orsmga003.jf.intel.com ([10.7.209.27]) by orsmga105.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Jul 2022 06:53:24 -0700 X-IronPort-AV: E=Sophos;i="5.92,243,1650956400"; d="scan'208";a="542595777" Received: from acagidia-mobl.ger.corp.intel.com (HELO thellstr-mobl1.intel.com) ([10.249.254.34]) by orsmga003-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Jul 2022 06:53:24 -0700 From: =?utf-8?q?Thomas_Hellstr=C3=B6m?= To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Subject: [PATCH 1/2] drm/i915: Take the object lock when destroying vmas from vm destruction Date: Mon, 4 Jul 2022 15:52:48 +0200 Message-Id: <20220704135249.8241-2-thomas.hellstrom@linux.intel.com> X-Mailer: git-send-email 2.36.1 In-Reply-To: <20220704135249.8241-1-thomas.hellstrom@linux.intel.com> References: <20220704135249.8241-1-thomas.hellstrom@linux.intel.com> MIME-Version: 1.0 X-BeenThere: dri-devel@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Direct Rendering Infrastructure - Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: =?utf-8?q?Thomas_Hellstr=C3=B6m?= Errors-To: dri-devel-bounces@lists.freedesktop.org Sender: "dri-devel" Currently we guarantee that vmas stay alive when the object lock is held only if we also hold a private vm reference. In order to relax the latter requirement, take the object lock also when destroying vmas from the vm destruction path. Signed-off-by: Thomas Hellström --- drivers/gpu/drm/i915/gt/intel_gtt.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/drivers/gpu/drm/i915/gt/intel_gtt.c b/drivers/gpu/drm/i915/gt/intel_gtt.c index b67831833c9a..92e22f727f88 100644 --- a/drivers/gpu/drm/i915/gt/intel_gtt.c +++ b/drivers/gpu/drm/i915/gt/intel_gtt.c @@ -112,7 +112,10 @@ int map_pt_dma_locked(struct i915_address_space *vm, struct drm_i915_gem_object static void clear_vm_list(struct list_head *list) { struct i915_vma *vma, *vn; + bool unlocked; +restart: + unlocked = false; list_for_each_entry_safe(vma, vn, list, vm_link) { struct drm_i915_gem_object *obj = vma->obj; @@ -138,8 +141,22 @@ static void clear_vm_list(struct list_head *list) i915_vm_resv_get(vma->vm); vma->vm_ddestroy = true; } else { + if (!i915_gem_object_trylock(obj, NULL)) { + unlocked = true; + mutex_unlock(&vma->vm->mutex); + i915_gem_object_lock(obj, NULL); + mutex_lock(&vma->vm->mutex); + /* + * The vma may now be on a different list, + * but not destroyed. We don't care. + * destroy it. + */ + } i915_vma_destroy_locked(vma); + i915_gem_object_unlock(obj); i915_gem_object_put(obj); + if (unlocked) + goto restart; } } From patchwork Mon Jul 4 13:52:49 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Thomas Hellstrom X-Patchwork-Id: 12905652 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 C8271C433EF for ; Mon, 4 Jul 2022 16:20:24 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id E25D810F0B5; Mon, 4 Jul 2022 16:15:44 +0000 (UTC) Received: from mga07.intel.com (mga07.intel.com [134.134.136.100]) by gabe.freedesktop.org (Postfix) with ESMTPS id 01D5210E14A; Mon, 4 Jul 2022 13:53:27 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1656942806; x=1688478806; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=1Js1Vns6LXoJe4GOCxHyTP8R3HC+/JHvhojp773kbGA=; b=lJdRPyuzZlBUCqyzZzUz5A6cClwGFbsdZXEeVYRVRHWtYbiSUuss4QRk S3gkx74F0gdpYyzPqCnIyuKZaDrskR6Q39SmBRmsoM4wIDKOzXas1iNaQ 78ON7ERT1Z1w8wFRXhlH9XYfdcLR/21SwNi+HHZj9B1cWn3dsRmBqziVF p1jNNMHDJ6tjXHAPGMnHbJDtazGThSxJKkkvCyUVQ4SshQITqQmsRpGiY tobSqOdtJmLE+gjybwjn1bLyv9owgHI4TQjNtxZneLaoIT7XJ2kJPbPjz pVzCHIK3U4ocqIsNeIcOOPhF4FxfyqCUMci+GKcjkhYexVj/xuuPjznUy w==; X-IronPort-AV: E=McAfee;i="6400,9594,10397"; a="347112578" X-IronPort-AV: E=Sophos;i="5.92,243,1650956400"; d="scan'208";a="347112578" Received: from orsmga003.jf.intel.com ([10.7.209.27]) by orsmga105.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Jul 2022 06:53:26 -0700 X-IronPort-AV: E=Sophos;i="5.92,243,1650956400"; d="scan'208";a="542595781" Received: from acagidia-mobl.ger.corp.intel.com (HELO thellstr-mobl1.intel.com) ([10.249.254.34]) by orsmga003-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 04 Jul 2022 06:53:26 -0700 From: =?utf-8?q?Thomas_Hellstr=C3=B6m?= To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Subject: [PATCH 2/2] drm/i915/gem: Rework i915_gem_object_unbind to rely on the object lock Date: Mon, 4 Jul 2022 15:52:49 +0200 Message-Id: <20220704135249.8241-3-thomas.hellstrom@linux.intel.com> X-Mailer: git-send-email 2.36.1 In-Reply-To: <20220704135249.8241-1-thomas.hellstrom@linux.intel.com> References: <20220704135249.8241-1-thomas.hellstrom@linux.intel.com> MIME-Version: 1.0 X-BeenThere: dri-devel@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Direct Rendering Infrastructure - Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: =?utf-8?q?Thomas_Hellstr=C3=B6m?= Errors-To: dri-devel-bounces@lists.freedesktop.org Sender: "dri-devel" Since vma destruction from the vm destruction path now takes the object lock, we no longer need to take a vm reference from i915_gem_object_unbind to make sure vmas are kept alive, the required object lock taken across i915_gem_object_unbind is sufficient. Also skip retrying on I915_GEM_OBJECT_UNBIND_BARRIER, since we can't race with vm destruction anymore. Signed-off-by: Thomas Hellström --- drivers/gpu/drm/i915/gem/i915_gem_domain.c | 4 +--- drivers/gpu/drm/i915/i915_drv.h | 7 +++---- drivers/gpu/drm/i915/i915_gem.c | 18 ------------------ 3 files changed, 4 insertions(+), 25 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c index 1674b0c5802b..cf1f39f80603 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c @@ -279,9 +279,7 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj, } /* The cache-level will be applied when each vma is rebound. */ - return i915_gem_object_unbind(obj, - I915_GEM_OBJECT_UNBIND_ACTIVE | - I915_GEM_OBJECT_UNBIND_BARRIER); + return i915_gem_object_unbind(obj, I915_GEM_OBJECT_UNBIND_ACTIVE); } int i915_gem_get_caching_ioctl(struct drm_device *dev, void *data, diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index c22f29c3faa0..6034e6319b6c 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -1415,10 +1415,9 @@ i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj, int i915_gem_object_unbind(struct drm_i915_gem_object *obj, unsigned long flags); #define I915_GEM_OBJECT_UNBIND_ACTIVE BIT(0) -#define I915_GEM_OBJECT_UNBIND_BARRIER BIT(1) -#define I915_GEM_OBJECT_UNBIND_TEST BIT(2) -#define I915_GEM_OBJECT_UNBIND_VM_TRYLOCK BIT(3) -#define I915_GEM_OBJECT_UNBIND_ASYNC BIT(4) +#define I915_GEM_OBJECT_UNBIND_TEST BIT(1) +#define I915_GEM_OBJECT_UNBIND_VM_TRYLOCK BIT(2) +#define I915_GEM_OBJECT_UNBIND_ASYNC BIT(3) void i915_gem_runtime_suspend(struct drm_i915_private *dev_priv); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 702e5b89be22..df16a5e8f5c7 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -137,7 +137,6 @@ int i915_gem_object_unbind(struct drm_i915_gem_object *obj, */ wakeref = intel_runtime_pm_get(rpm); -try_again: ret = 0; spin_lock(&obj->vma.lock); while (!ret && (vma = list_first_entry_or_null(&obj->vma.list, @@ -152,17 +151,6 @@ int i915_gem_object_unbind(struct drm_i915_gem_object *obj, break; } - /* - * Requiring the vm destructor to take the object lock - * before destroying a vma would help us eliminate the - * i915_vm_tryget() here, AND thus also the barrier stuff - * at the end. That's an easy fix, but sleeping locks in - * a kthread should generally be avoided. - */ - ret = -EAGAIN; - if (!i915_vm_tryget(vma->vm)) - break; - spin_unlock(&obj->vma.lock); /* @@ -189,17 +177,11 @@ int i915_gem_object_unbind(struct drm_i915_gem_object *obj, } } - i915_vm_put(vma->vm); spin_lock(&obj->vma.lock); } list_splice_init(&still_in_list, &obj->vma.list); spin_unlock(&obj->vma.lock); - if (ret == -EAGAIN && flags & I915_GEM_OBJECT_UNBIND_BARRIER) { - rcu_barrier(); /* flush the i915_vm_release() */ - goto try_again; - } - intel_runtime_pm_put(rpm, wakeref); return ret;