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 34506CA0EE4 for ; Mon, 18 Aug 2025 07:54:52 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id D4CE910E3FD; Mon, 18 Aug 2025 07:54:51 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.b="ji/RtbDo"; dkim-atps=neutral Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.12]) by gabe.freedesktop.org (Postfix) with ESMTPS id D18AE10E40B for ; Mon, 18 Aug 2025 07:54:50 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1755503691; x=1787039691; h=from:to:subject:date:message-id:in-reply-to:references: mime-version:content-transfer-encoding; bh=UFzXrxeIOF9071Xlh/M7Iy0c7dZq4z54eFa8rINXdvY=; b=ji/RtbDoViSsuqMhBPmNykcu9Nic2bWrNBS9SMBIhauhvKn8wa/iNQey +0ZRP26TTmpB7lOIokUpISRlG4KcxwaOaQ8TrGTMZFfp14PFC96ye90rM /tIqW5cjP3k9aBXHCQQhqIKCqPknrE8cAuKDMR+TcFB7vpa/dfo5fi7cQ joKqdre6iPNXzcekxxgJ+DyvCAjY9gQp29xhYdIwsZmZHNfh+jSUBo8cZ CjfQtMmVk5vsq3Q3j3hdtmDOVqPN6bH8Ml1a9m6a+MqjLmiHkASOth/vy u+CIb0iYr+6h5APteMno3HCqRjGEDsWGYH5dEW5OdBxxPTCZ3dpPQ6iIE g==; X-CSE-ConnectionGUID: Mnj5mb2MQveeKF9Gh1/byA== X-CSE-MsgGUID: rlIH8NZfTe+Z768++nlvKQ== X-IronPort-AV: E=McAfee;i="6800,10657,11524"; a="69170844" X-IronPort-AV: E=Sophos;i="6.17,293,1747724400"; d="scan'208";a="69170844" Received: from fmviesa007.fm.intel.com ([10.60.135.147]) by orvoesa104.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 18 Aug 2025 00:54:51 -0700 X-CSE-ConnectionGUID: wJaAEDnASZWXytJj3+CroA== X-CSE-MsgGUID: Ewjo6hN9TmyqJ8W7PgL8Ig== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.17,293,1747724400"; d="scan'208";a="167010987" Received: from dut2943pvc.iind.intel.com ([10.190.215.44]) by fmviesa007.fm.intel.com with ESMTP; 18 Aug 2025 00:54:49 -0700 From: nishit.sharma@intel.com To: igt-dev@lists.freedesktop.org, zbigniew.kempczynski@intel.com, nishit.sharma@intel.com Subject: [PATCH i-g-t 1/3] lib/intel_compute: loop_kernel_duration in Pipeline Date: Mon, 18 Aug 2025 07:54:44 +0000 Message-ID: <20250818075447.15230-2-nishit.sharma@intel.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20250818075447.15230-1-nishit.sharma@intel.com> References: <20250818075447.15230-1-nishit.sharma@intel.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-BeenThere: igt-dev@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Development mailing list for IGT GPU Tools List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: igt-dev-bounces@lists.freedesktop.org Sender: "igt-dev" From: Nishit Sharma To execute KERNEL in GPU for specific duration a brief sleep is required. New variable loop_kernel_duration is introduced in struct user_execenv which holds duration for sleep and during this duration the GPU will be running kernel. xe_run_intel_compute_xxx() calls are synchronous hence to execute workload on GPU for sometime sleep has been called. loop_kernel_duration acts as sleep duration. loop_kernel_duration is checked for all pipelines. Currently xe2lpg and xe3lpg pipeline support loop_kernel shader hence loop_kernel_duration works in these pipelines only. Signed-off-by: Nishit Sharma --- lib/intel_compute.c | 21 ++++++++++++++++++++- lib/intel_compute.h | 2 ++ 2 files changed, 22 insertions(+), 1 deletion(-) diff --git a/lib/intel_compute.c b/lib/intel_compute.c index 147dd2916..125331c53 100644 --- a/lib/intel_compute.c +++ b/lib/intel_compute.c @@ -849,6 +849,9 @@ static void compute_exec(int fd, const unsigned char *kernel, uint16_t devid = intel_get_drm_devid(fd); int entries = ARRAY_SIZE(bo_dict); + if (user && (user->kernel || user->loop_kernel_duration)) + igt_skip("Pipeline doesn't support loop_kernel\n"); + bo_execenv_create(fd, &execenv, eci, user); /* Set dynamic sizes */ @@ -1131,6 +1134,9 @@ static void xehp_compute_exec(int fd, const unsigned char *kernel, uint64_t bind_output_addr = (user && user->output_addr) ? user->output_addr : ADDR_OUTPUT; int entries = ARRAY_SIZE(bo_dict); + if (user && (user->kernel || user->loop_kernel_duration)) + igt_skip("Pipeline doesn't support loop_kernel\n"); + bo_execenv_create(fd, &execenv, eci, user); /* Set dynamic sizes */ @@ -1343,6 +1349,9 @@ static void xehpc_compute_exec(int fd, const unsigned char *kernel, uint64_t bind_output_addr = (user && user->output_addr) ? user->output_addr : ADDR_OUTPUT; int entries = ARRAY_SIZE(bo_dict); + if (user && (user->kernel || user->loop_kernel_duration)) + igt_skip("Pipeline doesn't support loop_kernel\n"); + bo_execenv_create(fd, &execenv, eci, user); /* Set dynamic sizes */ @@ -1731,6 +1740,9 @@ static void xelpg_compute_exec(int fd, const unsigned char *kernel, uint64_t bind_output_addr = (user && user->output_addr) ? user->output_addr : ADDR_OUTPUT; int entries = ARRAY_SIZE(bo_dict); + if (user && (user->kernel || user->loop_kernel_duration)) + igt_skip("Pipeline doesn't support loop_kernel\n"); + bo_execenv_create(fd, &execenv, eci, user); /* Set dynamic sizes */ @@ -1849,7 +1861,14 @@ static void xe2lpg_compute_exec(int fd, const unsigned char *kernel, OFFSET_KERNEL, 0, false, execenv.array_size); - bo_execenv_exec(&execenv, ADDR_BATCH); + if (user && user->loop_kernel_duration) { + bo_execenv_exec_async(&execenv, ADDR_BATCH); + igt_measured_usleep(user->loop_kernel_duration); + ((int *)bo_dict[4].data)[0] = MAGIC_LOOP_STOP; + bo_execenv_sync(&execenv); + user->skip_results_check = 1; + } else + bo_execenv_exec(&execenv, ADDR_BATCH); if (!user || (user && !user->skip_results_check)) bo_check_square(input_data, output_data, execenv.array_size); diff --git a/lib/intel_compute.h b/lib/intel_compute.h index 412791d07..54a1c7f82 100644 --- a/lib/intel_compute.h +++ b/lib/intel_compute.h @@ -63,6 +63,8 @@ struct user_execenv { uint64_t input_addr; /** @output_addr: override default address of the output array if provided */ uint64_t output_addr; + /** @loop_kernel_duration: duration till kernel should execute in gpu **/ + uint64_t loop_kernel_duration; }; enum execenv_alloc_prefs { -- 2.43.0