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 DB491C00528 for ; Tue, 8 Aug 2023 01:32:06 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 9E83210E391; Tue, 8 Aug 2023 01:32:06 +0000 (UTC) Received: from mgamail.intel.com (mgamail.intel.com [192.55.52.136]) by gabe.freedesktop.org (Postfix) with ESMTPS id 9A8D610E152 for ; Tue, 8 Aug 2023 01:32:04 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1691458324; x=1722994324; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=z5RdCFJwd/gBHJ+5dngnHHdfciZkf+4jtu3CwcMzSV0=; b=huSjDLkyn2XjqEYeYtkzsfWjxZ9bpqVHdqqnae7VHUC4cNrowsvV3q1b tnGpwSJ50Mm7exXEW9GVJueDaNFyHFWkwdGa87/wq+6060woLmxdLLcFs YTS/jt9t+0chQOv7FJyqND0KwCAcJbWzkTIGsG4vmsCwF8Yx5bOUjZywl xxd2o50qPjcfJbpG2h1uAlGsBJGbppzaRwxNckwF3K6vV3t88OxELDDfm q3GT1OOFYrECY2nTljJP1CK6SrOurfnBTbRH/YhEVyMHwfAE+fhB0dxl5 wZY6kttpRnHI0joBmUYdcGPiDm0RyMCUFBn6fZFQRgyIqtLfUiRDgo2hk A==; X-IronPort-AV: E=McAfee;i="6600,9927,10795"; a="350284317" X-IronPort-AV: E=Sophos;i="6.01,263,1684825200"; d="scan'208";a="350284317" Received: from orsmga006.jf.intel.com ([10.7.209.51]) by fmsmga106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 07 Aug 2023 18:32:03 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=McAfee;i="6600,9927,10795"; a="708075773" X-IronPort-AV: E=Sophos;i="6.01,263,1684825200"; d="scan'208";a="708075773" Received: from orsosgc001.jf.intel.com ([10.165.21.138]) by orsmga006-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 07 Aug 2023 18:32:03 -0700 From: Ashutosh Dixit To: intel-xe@lists.freedesktop.org Date: Mon, 7 Aug 2023 18:31:55 -0700 Message-ID: <20230808013159.38811-7-ashutosh.dixit@intel.com> X-Mailer: git-send-email 2.41.0 In-Reply-To: <20230808013159.38811-1-ashutosh.dixit@intel.com> References: <20230808013159.38811-1-ashutosh.dixit@intel.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Subject: [Intel-xe] [PATCH 06/10] drm/xe/oa: Start implementing OA stream open ioctl 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: , Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" Start implementing OA stream open ioctl and parse properties passed in as part of OA stream open. The remaining operations associated with OA stream open continue in subsequent patches. Signed-off-by: Ashutosh Dixit --- drivers/gpu/drm/xe/xe_device.c | 1 + drivers/gpu/drm/xe/xe_oa.c | 240 +++++++++++++++++++++++++++++++++ drivers/gpu/drm/xe/xe_oa.h | 2 + 3 files changed, 243 insertions(+) diff --git a/drivers/gpu/drm/xe/xe_device.c b/drivers/gpu/drm/xe/xe_device.c index 3772ce1dd234b..37fb9e6772e40 100644 --- a/drivers/gpu/drm/xe/xe_device.c +++ b/drivers/gpu/drm/xe/xe_device.c @@ -115,6 +115,7 @@ static const struct drm_ioctl_desc xe_ioctls[] = { DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(XE_VM_MADVISE, xe_vm_madvise_ioctl, DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(XE_OA_OPEN, xe_oa_stream_open_ioctl, DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(XE_OA_ADD_CONFIG, xe_oa_add_config_ioctl, DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(XE_OA_REMOVE_CONFIG, xe_oa_remove_config_ioctl, DRM_RENDER_ALLOW), diff --git a/drivers/gpu/drm/xe/xe_oa.c b/drivers/gpu/drm/xe/xe_oa.c index b98eeab8573d7..0ea5af549bdaa 100644 --- a/drivers/gpu/drm/xe/xe_oa.c +++ b/drivers/gpu/drm/xe/xe_oa.c @@ -11,11 +11,16 @@ #include #include +#include "regs/xe_gt_regs.h" #include "regs/xe_oa_regs.h" #include "xe_device.h" #include "xe_gt.h" +#include "xe_mmio.h" #include "xe_oa.h" +#define DEFAULT_POLL_FREQUENCY_HZ 200 +#define DEFAULT_POLL_PERIOD_NS (NSEC_PER_SEC / DEFAULT_POLL_FREQUENCY_HZ) + static u32 xe_oa_stream_paranoid = true; static int xe_oa_sample_rate_hard_limit; static u32 xe_oa_max_sample_rate = 100000; @@ -31,6 +36,21 @@ static const struct xe_oa_format oa_formats[] = { [XE_OAM_FORMAT_MPEC8u32_B8_C8] = { 2, 128, TYPE_OAM, HDR_64_BIT }, }; +struct xe_oa_open_properties { + bool sample; + bool single_exec_q; + u64 exec_q_id; + + int metrics_set; + int oa_format; + bool oa_periodic; + int oa_period_exponent; + + struct xe_hw_engine *hwe; + + u64 poll_oa_period; +}; + static struct ctl_table_header *sysctl_header; static void xe_oa_config_release(struct kref *ref) @@ -53,6 +73,226 @@ static void xe_oa_config_put(struct xe_oa_config *oa_config) kref_put(&oa_config->ref, xe_oa_config_release); } +/* + * OA timestamp frequency = CS timestamp frequency in most platforms. On some + * platforms OA unit ignores the CTC_SHIFT and the 2 timestamps differ. In such + * cases, return the adjusted CS timestamp frequency to the user. + */ +u32 xe_oa_timestamp_frequency(struct xe_device *xe) +{ + struct xe_gt *gt = xe_root_mmio_gt(xe); + u32 reg, shift; + + /* + * Wa_18013179988:dg2 + * Wa_14015846243:mtl + */ + switch (xe->info.platform) { + case XE_DG2: + case XE_METEORLAKE: + xe_device_mem_access_get(xe); + XE_WARN_ON(xe_force_wake_get(gt_to_fw(gt), XE_FORCEWAKE_ALL)); + reg = xe_mmio_read32(xe_root_mmio_gt(xe), RPM_CONFIG0); + XE_WARN_ON(xe_force_wake_put(gt_to_fw(gt), XE_FORCEWAKE_ALL)); + xe_device_mem_access_put(xe); + + shift = REG_FIELD_GET(RPM_CONFIG0_CTC_SHIFT_PARAMETER_MASK, reg); + return xe_root_mmio_gt(xe)->info.clock_freq << (3 - shift); + + default: + return xe_root_mmio_gt(xe)->info.clock_freq; + } +} + +static u64 oa_exponent_to_ns(struct xe_oa *oa, int exponent) +{ + u64 nom = (2ULL << exponent) * NSEC_PER_SEC; + u32 den = xe_oa_timestamp_frequency(oa->xe); + + return div_u64(nom + den - 1, den); +} + +static bool oa_format_valid(struct xe_oa *oa, u64 format) +{ + if (format >= XE_OA_FORMAT_MAX) + return false; + return test_bit(format, oa->format_mask); +} + +static bool engine_supports_oa(const struct xe_hw_engine *hwe) +{ + return hwe->oa_group; +} + +static bool engine_supports_oa_format(const struct xe_hw_engine *hwe, int type) +{ + return hwe->oa_group && hwe->oa_group->type == type; +} + +#define OA_EXPONENT_MAX 31 + +static int xe_oa_read_properties_unlocked(struct xe_oa *oa, u64 __user *uprops, + u32 n_props, + struct xe_oa_open_properties *props) +{ + const struct xe_oa_format *f; + u64 __user *uprop = uprops; + bool config_instance = false; + bool config_class = false; + u8 class, instance; + struct xe_gt *gt; + u32 i; + int ret; + + if (!n_props || n_props >= DRM_XE_OA_PROP_MAX) { + drm_dbg(&oa->xe->drm, "Invalid number of xe perf properties given\n"); + return -EINVAL; + } + + props->poll_oa_period = DEFAULT_POLL_PERIOD_NS; + + /* Defaults when class:instance is not passed */ + class = XE_ENGINE_CLASS_RENDER; + instance = 0; + + for (i = 0; i < n_props; i++) { + u64 oa_period, oa_freq_hz; + u64 id, value; + + ret = get_user(id, uprop); + if (ret) + return ret; + + ret = get_user(value, uprop + 1); + if (ret) + return ret; + + switch ((enum drm_xe_oa_property_id)id) { + case DRM_XE_OA_PROP_EXEC_QUEUE_ID: + props->single_exec_q = true; + props->exec_q_id = value; + break; + case DRM_XE_OA_PROP_SAMPLE_OA: + props->sample = value; + break; + case DRM_XE_OA_PROP_OA_METRICS_SET: + if (!value) { + drm_dbg(&oa->xe->drm, "Unknown OA metric set ID\n"); + return -EINVAL; + } + props->metrics_set = value; + break; + case DRM_XE_OA_PROP_OA_FORMAT: + if (!oa_format_valid(oa, value)) { + drm_dbg(&oa->xe->drm, "Unsupported OA report format %llu\n", + value); + return -EINVAL; + } + props->oa_format = value; + break; + case DRM_XE_OA_PROP_OA_EXPONENT: + if (value > OA_EXPONENT_MAX) { + drm_dbg(&oa->xe->drm, "OA timer exponent too high (> %u)\n", + OA_EXPONENT_MAX); + return -EINVAL; + } + + BUILD_BUG_ON(sizeof(oa_period) != 8); + oa_period = oa_exponent_to_ns(oa, value); + + oa_freq_hz = div64_u64(NSEC_PER_SEC, oa_period); + if (oa_freq_hz > xe_oa_max_sample_rate && !perfmon_capable()) { + drm_dbg(&oa->xe->drm, + "OA exponent would exceed the max sampling frequency (sysctl dev.xe.oa_max_sample_rate) %uHz without CAP_PERFMON or CAP_SYS_ADMIN privileges\n", + xe_oa_max_sample_rate); + return -EACCES; + } + + props->oa_periodic = true; + props->oa_period_exponent = value; + break; + case DRM_XE_OA_PROP_POLL_OA_PERIOD: + if (value < 100000 /* 100us */) { + drm_dbg(&oa->xe->drm, "OA timer too small (%lluns < 100us)\n", + value); + return -EINVAL; + } + props->poll_oa_period = value; + break; + case DRM_XE_OA_PROP_OA_ENGINE_CLASS: + class = (u8)value; + config_class = true; + break; + case DRM_XE_OA_PROP_OA_ENGINE_INSTANCE: + instance = (u8)value; + config_instance = true; + break; + default: + drm_dbg(&oa->xe->drm, "Unknown xe oa property ID\n"); + return -EINVAL; + } + + uprop += 2; + } + + if ((config_class && !config_instance) || + (config_instance && !config_class)) { + drm_dbg(&oa->xe->drm, "OA engine class/instance parameters must be passed together\n"); + return -EINVAL; + } + + for_each_gt(gt, oa->xe, i) { + props->hwe = xe_gt_hw_engine(gt, class, instance, false); + if (props->hwe) + break; + } + if (!props->hwe) { + drm_dbg(&oa->xe->drm, "OA engine class and instance invalid %d:%d\n", + class, instance); + return -EINVAL; + } + + if (!engine_supports_oa(props->hwe)) { + drm_dbg(&oa->xe->drm, "Engine not supported by OA %d:%d\n", + class, instance); + return -EINVAL; + } + + f = &oa->oa_formats[props->oa_format]; + if (!props->oa_format || !f->size || + !engine_supports_oa_format(props->hwe, f->type)) { + drm_dbg(&oa->xe->drm, "Invalid OA format %d type %d size %d for class %d\n", + props->oa_format, f->type, f->size, props->hwe->class); + return -EINVAL; + } + + return 0; +} + +int xe_oa_stream_open_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct xe_oa *oa = &to_xe_device(dev)->oa; + struct drm_xe_oa_open_param *param = data; + struct xe_oa_open_properties props = {}; + u32 known_open_flags; + + if (!oa->xe) { + drm_dbg(&oa->xe->drm, "xe oa interface not available for this system\n"); + return -ENODEV; + } + + known_open_flags = XE_OA_FLAG_FD_CLOEXEC | XE_OA_FLAG_FD_NONBLOCK | XE_OA_FLAG_DISABLED; + if (param->flags & ~known_open_flags) { + drm_dbg(&oa->xe->drm, "Unknown drm_xe_oa_open_param flag\n"); + return -EINVAL; + } + + return xe_oa_read_properties_unlocked(oa, u64_to_user_ptr(param->properties_ptr), + param->num_properties, + &props); +} + static bool xe_oa_is_valid_flex_addr(struct xe_oa *oa, u32 addr) { static const struct xe_reg flex_eu_regs[] = { diff --git a/drivers/gpu/drm/xe/xe_oa.h b/drivers/gpu/drm/xe/xe_oa.h index 79f77f445deb0..fd6caf652047a 100644 --- a/drivers/gpu/drm/xe/xe_oa.h +++ b/drivers/gpu/drm/xe/xe_oa.h @@ -16,6 +16,8 @@ int xe_oa_ioctl_version(struct xe_device *xe); int xe_oa_sysctl_register(void); void xe_oa_sysctl_unregister(void); +int xe_oa_stream_open_ioctl(struct drm_device *dev, void *data, + struct drm_file *file); int xe_oa_add_config_ioctl(struct drm_device *dev, void *data, struct drm_file *file); int xe_oa_remove_config_ioctl(struct drm_device *dev, void *data, -- 2.41.0