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 13891C27C50 for ; Mon, 27 May 2024 01:43:46 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 5B58210EED8; Mon, 27 May 2024 01:43:45 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.b="QFUILrq9"; dkim-atps=neutral Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.16]) by gabe.freedesktop.org (Postfix) with ESMTPS id 3548310EBD0 for ; Mon, 27 May 2024 01:43:40 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1716774220; x=1748310220; h=from:to:subject:date:message-id:in-reply-to:references: mime-version:content-transfer-encoding; bh=MCkS7fffINpnhuXW0PYk1nv3YduWFZ/53p0ndxHDK4Q=; b=QFUILrq9ESomPcE6Sm7bAYGXQYktrt7j1AWPll4d8GAFeLLeJPs2wo+x +8ynpwVlcBfowL6RHExo2cLp5EYHsTONdnvReSx/+ocG+ioMGlFErnqxM LMJokgPRjDc9UJejGjfi4zecS1WWjsRPFtBzGtF6lSHLswR21JxwG28tI DsNOSpgC6eIC9F2mPpRqV/3NACDljbKrsz8Qb+Sp4BzEqjhA+KmaQ/Ecl LxeqzlcygdExci7TYWTUfzbSg+GZLtl2c9doUamyn7BL966LNK502Zk1r Gq1uC1wmlFtpVs+P+W6vsBF5THD5sW+7P74klgcD50cW1KsLHrgIm9LXv Q==; X-CSE-ConnectionGUID: RTQuzq7oSti4kmZD0jrchw== X-CSE-MsgGUID: WUDu1pewQjeYXAqTwnMlTA== X-IronPort-AV: E=McAfee;i="6600,9927,11084"; a="11707478" X-IronPort-AV: E=Sophos;i="6.08,191,1712646000"; d="scan'208";a="11707478" Received: from fmviesa001.fm.intel.com ([10.60.135.141]) by fmvoesa110.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 26 May 2024 18:43:39 -0700 X-CSE-ConnectionGUID: 8BtfwKIRRems8DPIWgellg== X-CSE-MsgGUID: UaAJFTsdRpe6FBwiUZiaKQ== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.08,191,1712646000"; d="scan'208";a="65821719" Received: from orsosgc001.jf.intel.com ([10.165.21.138]) by smtpauth.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 26 May 2024 18:43:39 -0700 From: Ashutosh Dixit To: intel-xe@lists.freedesktop.org Subject: [PATCH 17/17] drm/xe/oa: Enable Xe2+ overrun mode Date: Sun, 26 May 2024 18:43:33 -0700 Message-ID: <20240527014333.603914-18-ashutosh.dixit@intel.com> X-Mailer: git-send-email 2.41.0 In-Reply-To: <20240527014333.603914-1-ashutosh.dixit@intel.com> References: <20240527014333.603914-1-ashutosh.dixit@intel.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit 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" Enable Xe2+ overrun mode. For Xe2+, when overrun mode is enabled, there are no partial reports at the end of buffer, making the OA buffer effectively a non-power-of-2 size circular buffer whose size, circ_size, is a multiple of the report size. v2: Fix implementation of xe_oa_circ_diff/xe_oa_circ_incr (Umesh) Reviewed-by: Umesh Nerlige Ramappa Signed-off-by: Ashutosh Dixit --- drivers/gpu/drm/xe/xe_oa.c | 35 ++++++++++++++++++++++++-------- drivers/gpu/drm/xe/xe_oa_types.h | 3 +++ 2 files changed, 30 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_oa.c b/drivers/gpu/drm/xe/xe_oa.c index 5ef4c4abea32..3ed4ca19f9cd 100644 --- a/drivers/gpu/drm/xe/xe_oa.c +++ b/drivers/gpu/drm/xe/xe_oa.c @@ -111,7 +111,14 @@ static const struct xe_oa_format oa_formats[] = { static u32 xe_oa_circ_diff(struct xe_oa_stream *stream, u32 tail, u32 head) { - return (tail - head) & (XE_OA_BUFFER_SIZE - 1); + return tail >= head ? tail - head : + tail + stream->oa_buffer.circ_size - head; +} + +static u32 xe_oa_circ_incr(struct xe_oa_stream *stream, u32 ptr, u32 n) +{ + return ptr + n >= stream->oa_buffer.circ_size ? + ptr + n - stream->oa_buffer.circ_size : ptr + n; } static void xe_oa_config_release(struct kref *ref) @@ -285,7 +292,7 @@ static int xe_oa_append_report(struct xe_oa_stream *stream, char __user *buf, buf += *offset; - oa_buf_end = stream->oa_buffer.vaddr + XE_OA_BUFFER_SIZE; + oa_buf_end = stream->oa_buffer.vaddr + stream->oa_buffer.circ_size; report_size_partial = oa_buf_end - report; if (report_size_partial < report_size) { @@ -311,7 +318,6 @@ static int xe_oa_append_reports(struct xe_oa_stream *stream, char __user *buf, int report_size = stream->oa_buffer.format->size; u8 *oa_buf_base = stream->oa_buffer.vaddr; u32 gtt_offset = xe_bo_ggtt_addr(stream->oa_buffer.bo); - u32 mask = (XE_OA_BUFFER_SIZE - 1); size_t start_offset = *offset; unsigned long flags; u32 head, tail; @@ -322,21 +328,23 @@ static int xe_oa_append_reports(struct xe_oa_stream *stream, char __user *buf, tail = stream->oa_buffer.tail; spin_unlock_irqrestore(&stream->oa_buffer.ptr_lock, flags); - xe_assert(stream->oa->xe, head < XE_OA_BUFFER_SIZE && tail < XE_OA_BUFFER_SIZE); + xe_assert(stream->oa->xe, + head < stream->oa_buffer.circ_size && tail < stream->oa_buffer.circ_size); - for (; xe_oa_circ_diff(stream, tail, head); head = (head + report_size) & mask) { + for (; xe_oa_circ_diff(stream, tail, head); + head = xe_oa_circ_incr(stream, head, report_size)) { u8 *report = oa_buf_base + head; ret = xe_oa_append_report(stream, buf, count, offset, report); if (ret) break; - if (is_power_of_2(report_size)) { + if (!(stream->oa_buffer.circ_size % report_size)) { /* Clear out report id and timestamp to detect unlanded reports */ oa_report_id_clear(stream, (void *)report); oa_timestamp_clear(stream, (void *)report); } else { - u8 *oa_buf_end = stream->oa_buffer.vaddr + XE_OA_BUFFER_SIZE; + u8 *oa_buf_end = stream->oa_buffer.vaddr + stream->oa_buffer.circ_size; u32 part = oa_buf_end - report; /* Zero out the entire report */ @@ -374,7 +382,6 @@ static void xe_oa_init_oa_buffer(struct xe_oa_stream *stream) xe_mmio_write32(stream->gt, __oa_regs(stream)->oa_head_ptr, gtt_offset & OAG_OAHEADPTR_MASK); stream->oa_buffer.head = 0; - /* * PRM says: "This MMIO must be set before the OATAILPTR register and after the * OAHEADPTR register. This is to enable proper functionality of the overflow bit". @@ -1294,6 +1301,18 @@ static int xe_oa_stream_init(struct xe_oa_stream *stream, stream->periodic = param->period_exponent > 0; stream->period_exponent = param->period_exponent; + /* + * For Xe2+, when overrun mode is enabled, there are no partial reports at the end + * of buffer, making the OA buffer effectively a non-power-of-2 size circular + * buffer whose size, circ_size, is a multiple of the report size + */ + if (GRAPHICS_VER(stream->oa->xe) >= 20 && + stream->hwe->oa_unit->type == DRM_XE_OA_UNIT_TYPE_OAG && stream->sample) + stream->oa_buffer.circ_size = + XE_OA_BUFFER_SIZE - XE_OA_BUFFER_SIZE % stream->oa_buffer.format->size; + else + stream->oa_buffer.circ_size = XE_OA_BUFFER_SIZE; + if (stream->exec_q && engine_supports_mi_query(stream->hwe)) { /* If we don't find the context offset, just return error */ ret = xe_oa_set_ctx_ctrl_offset(stream); diff --git a/drivers/gpu/drm/xe/xe_oa_types.h b/drivers/gpu/drm/xe/xe_oa_types.h index 7775fe91616f..c62811482934 100644 --- a/drivers/gpu/drm/xe/xe_oa_types.h +++ b/drivers/gpu/drm/xe/xe_oa_types.h @@ -163,6 +163,9 @@ struct xe_oa_buffer { /** @tail: The last verified cached tail where HW has completed writing */ u32 tail; + + /** @circ_size: The effective circular buffer size, for Xe2+ */ + u32 circ_size; }; /** -- 2.41.0