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 642BBC3DA61 for ; Mon, 29 Jul 2024 16:05:38 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 2428010E436; Mon, 29 Jul 2024 16:05:38 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.b="D3KTxY8o"; dkim-atps=neutral Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.12]) by gabe.freedesktop.org (Postfix) with ESMTPS id 484CE10E425 for ; Mon, 29 Jul 2024 16:05:37 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1722269137; x=1753805137; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=S6MZv2oNV+hK5npuqSLPcIrP3/1xThRj25OywL2QLjo=; b=D3KTxY8oi4WDlago/FWwD0i4fZ2cPSrqO0gM92YYhRvNZWeaoBkB6JK8 fUqj/ucdfI4zSWSIsLdUAhMDe6arfjq73CIf1JabEjtGz5PhL614DVRHT OMs3vbgNCPcoHwUJN7nDIj/3rzuzJRGBayGTQZBL0CQyTDBEyKlJuMmAO dsj6A+1qmRrIhqt08NPY7bhXZXzAu9nkP0ErGg002cbE9suMQt6ITN7/t oXWylRTvewRdxHLmwDC1mBz88xmlCVpNFXi/igQz4bpNbBkMQyxkQK1eF eOCJkctAG5091pf7oPWXFbWfaWIaQEl5wS2oYCgg4wrMHWGjS3n/CRu+W g==; X-CSE-ConnectionGUID: zGpeJy9jSLScBq8AZQrEbg== X-CSE-MsgGUID: cY8qsIRPTF2aWUO4gBoYdw== X-IronPort-AV: E=McAfee;i="6700,10204,11148"; a="31427821" X-IronPort-AV: E=Sophos;i="6.09,246,1716274800"; d="scan'208";a="31427821" Received: from fmviesa007.fm.intel.com ([10.60.135.147]) by orvoesa104.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 29 Jul 2024 09:05:24 -0700 X-CSE-ConnectionGUID: sPftz1ivTl2rBoAvXkhf/g== X-CSE-MsgGUID: FpU2EwOCQ86F8e3lCDvHKA== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,246,1716274800"; d="scan'208";a="53739794" Received: from sschumil-mobl2.ger.corp.intel.com (HELO localhost.localdomain) ([10.245.246.217]) by fmviesa007-auth.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 29 Jul 2024 09:05:21 -0700 From: Christoph Manszewski To: igt-dev@lists.freedesktop.org Cc: =?UTF-8?q?Zbigniew=20Kempczy=C5=84ski?= , Kamil Konieczny , Dominik Grzegorzek , Maciej Patelczyk , =?UTF-8?q?Dominik=20Karol=20Pi=C4=85tkowski?= , Pawel Sikora , Andrzej Hajda , Kolanupaka Naveena , Mika Kuoppala , Gwan-gyeong Mun , Karolina Stolarek , Christoph Manszewski Subject: [PATCH 58/66] tests/xe_eudebug_online: Add interrupt-reconnect test Date: Mon, 29 Jul 2024 18:01:51 +0200 Message-Id: <20240729160159.37036-59-christoph.manszewski@intel.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240729160159.37036-1-christoph.manszewski@intel.com> References: <20240729160159.37036-1-christoph.manszewski@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: Karolina Stolarek Introduce interrupt-reconnect test case where the debugger is closed and reopened on attention event. Check if the workload is reset when there is no active debugger detected. Signed-off-by: Karolina Stolarek Cc: Christoph Manszewski Cc: Dominik Grzegorzek --- tests/intel/xe_eudebug_online.c | 126 +++++++++++++++++++++++++++++++- 1 file changed, 123 insertions(+), 3 deletions(-) diff --git a/tests/intel/xe_eudebug_online.c b/tests/intel/xe_eudebug_online.c index c101bde2f..9f55cec74 100644 --- a/tests/intel/xe_eudebug_online.c +++ b/tests/intel/xe_eudebug_online.c @@ -22,11 +22,14 @@ #define SHADER_BREAKPOINT (1 << 0) #define SHADER_LOOP (1 << 1) +#define TRIGGER_RECONNECT (1 << 27) #define TRIGGER_RESUME_SET_BP (1 << 28) #define TRIGGER_RESUME_DELAYED (1 << 29) #define TRIGGER_RESUME_DSS (1 << 30) #define TRIGGER_RESUME_ONE (1 << 31) +#define DEBUGGER_REATTACHED 1 + #define STEERING_END_LOOP 0xdeadca11 #define SHADER_CANARY 0x01010101 @@ -682,9 +685,12 @@ static void run_online_client(struct xe_eudebug_client *c) intel_bb_sync(ibb); - /* Make sure it wasn't the timeout. */ - igt_assert(igt_nsec_elapsed(&ts) < - XE_EUDEBUG_DEFAULT_TIMEOUT_MS / MSEC_PER_SEC * NSEC_PER_SEC); + if (c->flags & TRIGGER_RECONNECT) + xe_eudebug_client_wait_stage(c, DEBUGGER_REATTACHED); + else + /* Make sure it wasn't the timeout. */ + igt_assert(igt_nsec_elapsed(&ts) < + XE_EUDEBUG_DEFAULT_TIMEOUT_MS / MSEC_PER_SEC * NSEC_PER_SEC); ptr = xe_bo_mmap_ext(fd, buf->handle, buf->size, PROT_READ); data->threads_count = count_canaries_neq(ptr, w_dim, 0); @@ -1158,6 +1164,117 @@ static void test_tdctl_parameters(int fd, struct drm_xe_engine_class_instance *h online_debug_data_destroy(data); } +static void eu_attention_debugger_detach_trigger(struct xe_eudebug_debugger *d, + struct drm_xe_eudebug_event *event) +{ + struct online_debug_data *data = d->ptr; + unsigned int max_size; + uint64_t c_pid; + int ret; + + c_pid = d->target_pid; + + /* Reset VM data so the re-triggered VM open handler works properly */ + data->vm_fd = -1; + + xe_eudebug_debugger_dettach(d); + + /* Let the KMD scan function notice unhandled EU attention */ + sleep(1); + + /* + * New session that is created by EU debugger on reconnect restarts + * seqno, causing isses with log sorting. To avoid that, create + * a new event log. + */ + max_size = d->log->max_size; + xe_eudebug_event_log_destroy(d->log); + d->log = xe_eudebug_event_log_create("debugger-reconnect", max_size); + + ret = xe_eudebug_connect(d->master_fd, c_pid, 0); + igt_assert(ret >= 0); + d->fd = ret; + d->target_pid = c_pid; + + /* Let the discovery worker discover resources */ + sleep(2); + + xe_eudebug_debugger_signal_stage(d, DEBUGGER_REATTACHED); +} + +/** + * SUBTEST: interrupt-reconnect + * Description: + * Schedules EU workload which should last about a few seconds, + * interrupts all threads and detaches debugger when attention is + * raised. The test checks if KMD resets the workload when there's + * no debugger attached and does the event playback on discovery. + */ +static void test_interrupt_reconnect(int fd, struct drm_xe_engine_class_instance *hwe, int flags) +{ + struct drm_xe_eudebug_event *e = NULL; + struct online_debug_data *data; + struct xe_eudebug_session *s; + uint32_t val; + + data = online_debug_data_create(hwe); + s = xe_eudebug_session_create(fd, run_online_client, flags, data); + + xe_eudebug_debugger_add_trigger(s->d, DRM_XE_EUDEBUG_EVENT_OPEN, + open_trigger); + xe_eudebug_debugger_add_trigger(s->d, DRM_XE_EUDEBUG_EVENT_EXEC_QUEUE, + exec_queue_trigger); + xe_eudebug_debugger_add_trigger(s->d, DRM_XE_EUDEBUG_EVENT_EU_ATTENTION, + eu_attention_debug_trigger); + xe_eudebug_debugger_add_trigger(s->d, DRM_XE_EUDEBUG_EVENT_EU_ATTENTION, + eu_attention_debugger_detach_trigger); + xe_eudebug_debugger_add_trigger(s->d, DRM_XE_EUDEBUG_EVENT_VM, vm_open_trigger); + xe_eudebug_debugger_add_trigger(s->d, DRM_XE_EUDEBUG_EVENT_METADATA, + create_metadata_trigger); + xe_eudebug_debugger_add_trigger(s->d, DRM_XE_EUDEBUG_EVENT_VM_BIND_UFENCE, + ufence_ack_trigger); + + igt_assert_eq(xe_eudebug_debugger_attach(s->d, s->c), 0); + xe_eudebug_debugger_start_worker(s->d); + xe_eudebug_client_start(s->c); + + /* wait for workload to start */ + igt_for_milliseconds(STARTUP_TIMEOUT_MS) { + /* collect needed data from triggers */ + if (READ_ONCE(data->vm_fd) == -1 || READ_ONCE(data->target_size) == 0) + continue; + + if (pread(data->vm_fd, &val, sizeof(val), data->target_offset) == sizeof(val)) + if (val != 0) + break; + } + + pthread_mutex_lock(&data->mutex); + igt_assert(data->client_handle != -1); + igt_assert(data->exec_queue_handle != -1); + eu_ctl_interrupt_all(s->d->fd, data->client_handle, + data->exec_queue_handle, data->lrc_handle); + pthread_mutex_unlock(&data->mutex); + + xe_eudebug_client_wait_done(s->c); + + xe_eudebug_debugger_stop_worker(s->d, 1); + + xe_eudebug_event_log_print(s->d->log, true); + xe_eudebug_event_log_print(s->c->log, true); + + xe_eudebug_session_check(s, true, XE_EUDEBUG_FILTER_EVENT_VM_BIND | + XE_EUDEBUG_FILTER_EVENT_VM_BIND_OP | + XE_EUDEBUG_FILTER_EVENT_VM_BIND_UFENCE); + + /* We expect workload reset, so no attention should be raised */ + xe_eudebug_for_each_event(e, s->d->log) + igt_assert(e->type != DRM_XE_EUDEBUG_EVENT_EU_ATTENTION); + + xe_eudebug_session_destroy(s); + online_debug_data_destroy(data); +} + static struct drm_xe_engine_class_instance *pick_compute(int fd, int gt) { struct drm_xe_engine_class_instance *hwe; @@ -1220,6 +1337,9 @@ igt_main test_gt_render_or_compute("reset-with-attention", fd, hwe) test_reset_with_attention_online(fd, hwe, SHADER_BREAKPOINT); + test_gt_render_or_compute("interrupt-reconnect", fd, hwe) + test_interrupt_reconnect(fd, hwe, SHADER_LOOP | TRIGGER_RECONNECT); + igt_fixture { xe_eudebug_enable(fd, was_enabled); -- 2.34.1