From patchwork Thu Jan 9 05:31:02 2014 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: akash.goel@intel.com X-Patchwork-Id: 3457421 Return-Path: X-Original-To: patchwork-intel-gfx@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork2.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.19.201]) by patchwork2.web.kernel.org (Postfix) with ESMTP id E012CC02DC for ; Thu, 9 Jan 2014 05:29:25 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id F11A920165 for ; Thu, 9 Jan 2014 05:29:24 +0000 (UTC) Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) by mail.kernel.org (Postfix) with ESMTP id 08E532015F for ; Thu, 9 Jan 2014 05:29:24 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 9E2E0106793; Wed, 8 Jan 2014 21:29:22 -0800 (PST) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mga02.intel.com (mga02.intel.com [134.134.136.20]) by gabe.freedesktop.org (Postfix) with ESMTP id C05761060EF for ; Wed, 8 Jan 2014 21:29:15 -0800 (PST) Received: from orsmga002.jf.intel.com ([10.7.209.21]) by orsmga101.jf.intel.com with ESMTP; 08 Jan 2014 21:29:15 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.95,628,1384329600"; d="scan'208";a="463923089" Received: from akashgoe-desktop.iind.intel.com ([10.223.82.34]) by orsmga002.jf.intel.com with ESMTP; 08 Jan 2014 21:29:13 -0800 From: akash.goel@intel.com To: intel-gfx@lists.freedesktop.org Date: Thu, 9 Jan 2014 11:01:02 +0530 Message-Id: <1389245462-11083-1-git-send-email-akash.goel@intel.com> X-Mailer: git-send-email 1.8.5.2 Cc: Akash Goel Subject: [Intel-gfx] [PATCH 5/7] drm/i915/vlv: Increase the utilization of stolen memory on VLV. X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.13 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: intel-gfx-bounces@lists.freedesktop.org Errors-To: intel-gfx-bounces@lists.freedesktop.org X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_MED, RP_MATCHES_RCVD, UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP From: Akash Goel On VLV, 64MB of system memory was being reserved for stolen area, but ~8MB of it was being utilized. Increased the utilization of Stolen area by allocating User created Frame buffers(only X Tiled) from it. User Frame buffers are suitable for allocation from stolen area, as its very unlikely that they are not accessed from CPU side. And its even more unlikely that the Tiled(X) buffers will be accessed directly from the CPU side. And any allocation from stolen area is not directly CPU accessible, but accessible only through the aperture space. With 1080p sized frame buffers (32 bpp), the allocation of 6-7 frame buffers itself gives almost the full utilization of stolen area. Signed-off-by: Akash Goel --- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 21 ++++++++ drivers/gpu/drm/i915/i915_gem_stolen.c | 93 ++++++++++++++++++++++++++++++++++ 3 files changed, 115 insertions(+) diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 1bcc543..db29537 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -2376,6 +2376,7 @@ i915_gem_object_create_stolen_for_preallocated(struct drm_device *dev, u32 stolen_offset, u32 gtt_offset, u32 size); +void i915_gem_object_move_to_stolen(struct drm_i915_gem_object *obj); void i915_gem_object_release_stolen(struct drm_i915_gem_object *obj); /* i915_gem_tiling.c */ diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 656406d..24f93ef 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -3915,6 +3915,27 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj, } } + /* Try to allocate the physical space for the GEM object, + * representing the User frame buffer, from the stolen area. + * But if there is no sufficient free space left in stolen + * area, will fallback to shmem. + */ + if (obj->user_fb == 1) { + if (obj->pages == NULL) { + if (obj->tiling_mode == I915_TILING_X) { + /* Tiled(X) Scanout buffers are more suitable + * for allocation from stolen area, as its very + * unlikely that they will be accessed directly + * from the CPU side and any allocation from + * stolen area is not directly CPU accessible, + * but accessible only through the aperture + * space. + */ + i915_gem_object_move_to_stolen(obj); + } + } + } + if (!i915_gem_obj_bound(obj, vm)) { ret = i915_gem_object_bind_to_vm(obj, vm, alignment, map_and_fenceable, diff --git a/drivers/gpu/drm/i915/i915_gem_stolen.c b/drivers/gpu/drm/i915/i915_gem_stolen.c index 5cf97d6..29c22f9 100644 --- a/drivers/gpu/drm/i915/i915_gem_stolen.c +++ b/drivers/gpu/drm/i915/i915_gem_stolen.c @@ -29,6 +29,7 @@ #include #include #include "i915_drv.h" +#include /* * The BIOS typically reserves some of the system's memory for the exclusive @@ -370,6 +371,98 @@ i915_gem_object_create_stolen(struct drm_device *dev, u32 size) return NULL; } +void +i915_gem_object_move_to_stolen(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_mm_node *stolen; + u32 size = obj->base.size; + int ret = 0; + + if (!IS_VALLEYVIEW(dev)) { + return; + } + + if (obj->stolen) { + BUG_ON(obj->pages == NULL); + return; + } + + if (!drm_mm_initialized(&dev_priv->mm.stolen)) + return; + + if (size == 0) + return; + + /* Check if already shmem space has been allocated for the object + * or not. We cannot rely upon on the value of 'pages' field for this. + * As even though if the 'pages' field is NULL, it does not actually + * indicate that the backing physical space (shmem) is currently not + * reserved for the object, as the object may not get purged/truncated + * on the calll to 'put_pages_gtt'. + */ + if (obj->base.filp) { + struct inode *inode = file_inode(obj->base.filp); + struct shmem_inode_info *info = SHMEM_I(inode); + if (!inode) + return; + spin_lock(&info->lock); + /* The alloced field stores how many data pages are + * allocated to the file. + */ + ret = info->alloced; + spin_unlock(&info->lock); + if (ret > 0) { + DRM_DEBUG_DRIVER( + "Already shmem space alloced, %d pges\n", ret); + return; + } + } + + stolen = kzalloc(sizeof(*stolen), GFP_KERNEL); + if (!stolen) + return; + + ret = drm_mm_insert_node(&dev_priv->mm.stolen, stolen, size, + 4096, DRM_MM_SEARCH_DEFAULT); + if (ret) { + kfree(stolen); + DRM_DEBUG_DRIVER("ran out of stolen space\n"); + return; + } + + /* Set up the object to use the stolen memory, + * backing store no longer managed by shmem layer */ + drm_gem_object_release(&(obj->base)); + obj->base.filp = NULL; + obj->ops = &i915_gem_object_stolen_ops; + + obj->pages = i915_pages_create_for_stolen(dev, + stolen->start, stolen->size); + if (obj->pages == NULL) + goto cleanup; + + i915_gem_object_pin_pages(obj); + list_add_tail(&obj->global_list, &dev_priv->mm.unbound_list); + obj->has_dma_mapping = true; + obj->stolen = stolen; + + DRM_DEBUG_DRIVER("Obj moved to stolen, ptr = %p, size = %x\n", + obj, size); + + obj->base.read_domains = I915_GEM_DOMAIN_CPU | I915_GEM_DOMAIN_GTT; + obj->cache_level = HAS_LLC(dev) ? I915_CACHE_LLC : I915_CACHE_NONE; + + /* No zeroing-out of buffers allocated from stolen area */ + return; + +cleanup: + drm_mm_remove_node(stolen); + kfree(stolen); + return; +} + struct drm_i915_gem_object * i915_gem_object_create_stolen_for_preallocated(struct drm_device *dev, u32 stolen_offset,