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 08EFFC6FD18 for ; Fri, 31 Mar 2023 08:46:59 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id DDFA010F136; Fri, 31 Mar 2023 08:46:58 +0000 (UTC) Received: from mga18.intel.com (mga18.intel.com [134.134.136.126]) by gabe.freedesktop.org (Postfix) with ESMTPS id DC4AD10F13E for ; Fri, 31 Mar 2023 08:46:55 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1680252415; x=1711788415; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=hW7Uk1nhINX12bQaodsLR9jTtAbhMBcQfta4EgF/Jdk=; b=bxC51yf3F+CpelJ4x4WajmJfuwTBmBHxF5qpgTolAqJw1gnrzdxVHGA0 mS0mql1EtQuP+OwzBTfQfyvWcECTGml7zldn4oYyo3rvHqJyE0eQ0/b+G qUU7PoUyhrT3iCjUmZ6D1odtcNFnJALL7ggPEG3wurLwk5U+ulyAfEXJe wB0RrNALpBOk1X6SgjjJIWbFdLO74UyYw4tkXNQvvPDWzVjj/HcnZy+Fp RI4BoSTbijWatn0dsSqP7YVbzyQ44updo+XETI7OrY4jVhc5jFTVVgi7f uPgKqcSQB8VonDsaQ7c3vQh2JWA7nS2MOpMF+TuOKAO61j7USAZkucjRc g==; X-IronPort-AV: E=McAfee;i="6600,9927,10665"; a="325364368" X-IronPort-AV: E=Sophos;i="5.98,307,1673942400"; d="scan'208";a="325364368" Received: from orsmga007.jf.intel.com ([10.7.209.58]) by orsmga106.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2023 01:46:55 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=McAfee;i="6600,9927,10665"; a="678544918" X-IronPort-AV: E=Sophos;i="5.98,307,1673942400"; d="scan'208";a="678544918" Received: from kuhnbrex-mobl1.ger.corp.intel.com (HELO mwauld-desk1.intel.com) ([10.252.28.81]) by orsmga007-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2023 01:46:53 -0700 From: Matthew Auld To: intel-xe@lists.freedesktop.org Date: Fri, 31 Mar 2023 09:46:27 +0100 Message-Id: <20230331084628.421814-5-matthew.auld@intel.com> X-Mailer: git-send-email 2.39.2 In-Reply-To: <20230331084628.421814-1-matthew.auld@intel.com> References: <20230331084628.421814-1-matthew.auld@intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Subject: [Intel-xe] [PATCH v4 4/5] drm/xe/uapi: add the userspace bits for small-bar 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: Filip Hazubski , Lucas De Marchi , Effie Yu , Carl Zhang Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" Mostly the same as i915. We add a new hint for userspace to force an object into the mappable part of vram. We also need to tell userspace how large the mappable part is. In Vulkan for example, there will be two vram heaps for small-bar systems. And here the size of each heap needs to be known. Likewise the used/avail tracking needs to account for the mappable part. We also limit the available tracking going forward, such that we limit to privileged users only, since these values are system wide and are technically considered an info leak. v2 (Maarten): - s/NEEDS_CPU_ACCESS/NEEDS_VISIBLE_VRAM/ in the uapi. We also no longer require smem as an extra placement. This is more flexible, and lets us use this for clear-color surfaces, since we need CPU access there but we don't want to attach smem, since that effectively disables CCS from kernel pov. - Reject clear-color CCS buffers where NEEDS_VISIBLE_VRAM is not set, instead of migrating it behind the scenes. v3 (José): - Split the changes that limit the accounting for perfmon_capable() into a separate patch. - Use XE_BO_CREATE_VRAM_MASK. v4 (Gwan-gyeong Mun): - Add some kernel-doc for the query bits. v5: - One small kernel-doc correction. The cpu_visible_size and corresponding used tracking are always zero for non XE_MEM_REGION_CLASS_VRAM. Testcase: igt/tests/xe_query Testcase: igt/tests/xe_mmap@small-bar Signed-off-by: Matthew Auld Cc: Maarten Lankhorst Cc: Thomas Hellström Cc: Gwan-gyeong Mun Cc: Lucas De Marchi Cc: José Roberto de Souza Cc: Filip Hazubski Cc: Carl Zhang Cc: Effie Yu Reviewed-by: José Roberto de Souza Reviewed-by: Gwan-gyeong Mun --- drivers/gpu/drm/xe/display/xe_fb_pin.c | 13 +++++++ drivers/gpu/drm/xe/xe_bo.c | 13 +++++-- drivers/gpu/drm/xe/xe_query.c | 13 ++++--- drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 18 ++++++++++ drivers/gpu/drm/xe/xe_ttm_vram_mgr.h | 4 +++ include/uapi/drm/xe_drm.h | 50 +++++++++++++++++++++++++- 6 files changed, 104 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/xe/display/xe_fb_pin.c b/drivers/gpu/drm/xe/display/xe_fb_pin.c index 65c0bc28a3d1..2a0edf9401da 100644 --- a/drivers/gpu/drm/xe/display/xe_fb_pin.c +++ b/drivers/gpu/drm/xe/display/xe_fb_pin.c @@ -195,6 +195,19 @@ static struct i915_vma *__xe_pin_fb_vma(struct intel_framebuffer *fb, goto err; } + /* + * If we need to able to access the clear-color value stored in the + * buffer, then we require that such buffers are also CPU accessible. + * This is important on small-bar systems where only some subset of VRAM + * is CPU accessible. + */ + if (IS_DGFX(to_xe_device(bo->ttm.base.dev)) && + intel_fb_rc_ccs_cc_plane(&fb->base) >= 0 && + !(bo->flags & XE_BO_NEEDS_CPU_ACCESS)) { + ret = -EINVAL; + goto err; + } + /* * Pin the framebuffer, we can't use xe_bo_(un)pin functions as the * assumptions are incorrect for framebuffers diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c index 2f2b6a89851e..2efb4dd42b6f 100644 --- a/drivers/gpu/drm/xe/xe_bo.c +++ b/drivers/gpu/drm/xe/xe_bo.c @@ -906,7 +906,6 @@ static vm_fault_t xe_gem_fault(struct vm_fault *vmf) ret = ttm_bo_vm_fault_reserved(vmf, vmf->vma->vm_page_prot, TTM_BO_VM_NUM_PREFAULT); - drm_dev_exit(idx); } else { ret = ttm_bo_vm_dummy_page(vmf, vmf->vma->vm_page_prot); @@ -1531,6 +1530,7 @@ int xe_gem_create_ioctl(struct drm_device *dev, void *data, if (XE_IOCTL_ERR(xe, args->flags & ~(XE_GEM_CREATE_FLAG_DEFER_BACKING | XE_GEM_CREATE_FLAG_SCANOUT | + XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM | xe->info.mem_region_mask))) return -EINVAL; @@ -1568,6 +1568,14 @@ int xe_gem_create_ioctl(struct drm_device *dev, void *data, bo_flags |= XE_BO_SCANOUT_BIT; bo_flags |= args->flags << (ffs(XE_BO_CREATE_SYSTEM_BIT) - 1); + + if (args->flags & XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM) { + if (XE_IOCTL_ERR(xe, !(bo_flags & XE_BO_CREATE_VRAM_MASK))) + return -EINVAL; + + bo_flags |= XE_BO_NEEDS_CPU_ACCESS; + } + bo = xe_bo_create(xe, NULL, vm, args->size, ttm_bo_type_device, bo_flags); if (vm) { @@ -1831,7 +1839,8 @@ int xe_bo_dumb_create(struct drm_file *file_priv, bo = xe_bo_create(xe, NULL, NULL, args->size, ttm_bo_type_device, XE_BO_CREATE_VRAM_IF_DGFX(to_gt(xe)) | - XE_BO_CREATE_USER_BIT | XE_BO_SCANOUT_BIT); + XE_BO_CREATE_USER_BIT | XE_BO_SCANOUT_BIT | + XE_BO_NEEDS_CPU_ACCESS); if (IS_ERR(bo)) return PTR_ERR(bo); diff --git a/drivers/gpu/drm/xe/xe_query.c b/drivers/gpu/drm/xe/xe_query.c index c134450dd0b0..fd38ff27cc84 100644 --- a/drivers/gpu/drm/xe/xe_query.c +++ b/drivers/gpu/drm/xe/xe_query.c @@ -17,6 +17,7 @@ #include "xe_gt.h" #include "xe_guc_hwconfig.h" #include "xe_macros.h" +#include "xe_ttm_vram_mgr.h" static const enum xe_engine_class xe_to_user_engine_class[] = { [XE_ENGINE_CLASS_RENDER] = DRM_XE_ENGINE_CLASS_RENDER, @@ -150,13 +151,17 @@ static int query_memory_usage(struct xe_device *xe, man->size; if (perfmon_capable()) { - usage->regions[usage->num_regions].used = - ttm_resource_manager_usage(man); + xe_ttm_vram_get_used(man, + &usage->regions[usage->num_regions].used, + &usage->regions[usage->num_regions].cpu_visible_used); } else { - usage->regions[usage->num_regions].used = - man->size; + usage->regions[usage->num_regions].used = man->size; + usage->regions[usage->num_regions].cpu_visible_used = + xe_ttm_vram_get_cpu_visible_size(man); } + usage->regions[usage->num_regions].cpu_visible_size = + xe_ttm_vram_get_cpu_visible_size(man); usage->num_regions++; } } diff --git a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c index cf081e4aedf6..654c5ae6516b 100644 --- a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +++ b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c @@ -458,3 +458,21 @@ void xe_ttm_vram_mgr_free_sgt(struct device *dev, enum dma_data_direction dir, sg_free_table(sgt); kfree(sgt); } + +u64 xe_ttm_vram_get_cpu_visible_size(struct ttm_resource_manager *man) +{ + struct xe_ttm_vram_mgr *mgr = to_xe_ttm_vram_mgr(man); + + return mgr->visible_size; +} + +void xe_ttm_vram_get_used(struct ttm_resource_manager *man, + u64 *used, u64 *used_visible) +{ + struct xe_ttm_vram_mgr *mgr = to_xe_ttm_vram_mgr(man); + + mutex_lock(&mgr->lock); + *used = mgr->mm.size - mgr->mm.avail; + *used_visible = mgr->visible_size - mgr->visible_avail; + mutex_unlock(&mgr->lock); +} diff --git a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.h b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.h index 35e5367a79fb..27f43490fa11 100644 --- a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.h +++ b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.h @@ -25,6 +25,10 @@ int xe_ttm_vram_mgr_alloc_sgt(struct xe_device *xe, void xe_ttm_vram_mgr_free_sgt(struct device *dev, enum dma_data_direction dir, struct sg_table *sgt); +u64 xe_ttm_vram_get_cpu_visible_size(struct ttm_resource_manager *man); +void xe_ttm_vram_get_used(struct ttm_resource_manager *man, + u64 *used, u64 *used_visible); + static inline struct xe_ttm_vram_mgr_resource * to_xe_ttm_vram_mgr_resource(struct ttm_resource *res) { diff --git a/include/uapi/drm/xe_drm.h b/include/uapi/drm/xe_drm.h index 8cb3324221f2..2d578e6e578f 100644 --- a/include/uapi/drm/xe_drm.h +++ b/include/uapi/drm/xe_drm.h @@ -231,8 +231,40 @@ struct drm_xe_query_mem_usage { * the @total_size. */ __u64 used; + /* + * @cpu_visible_size: How much of this region we can CPU access, + * in bytes. + * + * This will be always be <= @total_size, and the remainder (if + * there is any) will not be CPU accessible. If smaller then + * this system is considered a small BAR system. + * + * On systems without small BAR, the probed_size will always + * equal the @total_size, since all of it will be CPU + * accessible. + * + * Note this is only tracked for XE_MEM_REGION_CLASS_VRAM + * regions (for other types the value here will always equal + * zero). + */ + __u64 cpu_visible_size; + /** + * @cpu_visible_used: Estimate of CPU visible memory used, in + * bytes. + * + * Note this is only tracked for XE_MEM_REGION_CLASS_VRAM + * regions (for other types the value here will always equal the + * @cpu_visible_size). + * + * Requires CAP_PERFMON or CAP_SYS_ADMIN to get reliable + * accounting. Without this the value here will always equal the + * @cpu_visible_used. Note this is only currently tracked for + * XE_MEM_REGION_CLASS_VRAM regions (for other types the value + * here will also always be zero). + */ + __u64 cpu_visible_used; /** @reserved: MBZ */ - __u64 reserved[8]; + __u64 reserved[6]; } regions[]; }; @@ -334,6 +366,22 @@ struct drm_xe_gem_create { */ #define XE_GEM_CREATE_FLAG_DEFER_BACKING (0x1 << 24) #define XE_GEM_CREATE_FLAG_SCANOUT (0x1 << 25) +/* + * When using VRAM as a possible placement, ensure that the corresponding VRAM + * allocation will always use the CPU accessible part of VRAM. This is important + * for small-bar systems (on full-bar systems this gets turned into a noop). + * + * Note: System memory can be used as an extra placement if the kernel should + * spill the allocation to system memory, if space can't be made available in + * the CPU accessible part of VRAM (giving the same behaviour as the i915 + * interface, see I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS). + * + * Note: For clear-color CCS surfaces the kernel needs to read the clear-color + * value stored in the buffer, and on discrete platforms we need to use VRAM for + * display surfaces, therefore the kernel requires setting this flag for such + * objects, otherwise an error is thrown. + */ +#define XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM (0x1 << 26) __u32 flags; /** -- 2.39.2