From patchwork Wed Jun 28 23:10:26 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Thiago Jung Bauermann X-Patchwork-Id: 781961 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from lists.ozlabs.org (lists.ozlabs.org [IPv6:2401:3900:2:1::3]) (using TLSv1.2 with cipher ADH-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 3wyf2J4R92z9s72 for ; Thu, 29 Jun 2017 09:22:00 +1000 (AEST) Received: from lists.ozlabs.org (lists.ozlabs.org [IPv6:2401:3900:2:1::3]) by lists.ozlabs.org (Postfix) with ESMTP id 3wyf2J3ccmzDrPK for ; Thu, 29 Jun 2017 09:22:00 +1000 (AEST) X-Original-To: linuxppc-dev@lists.ozlabs.org Delivered-To: linuxppc-dev@lists.ozlabs.org Received: from mx0a-001b2d01.pphosted.com (mx0a-001b2d01.pphosted.com [148.163.156.1]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.ozlabs.org (Postfix) with ESMTPS id 3wydnc4W9YzDr4n for ; Thu, 29 Jun 2017 09:11:00 +1000 (AEST) Received: from pps.filterd (m0098409.ppops.net [127.0.0.1]) by mx0a-001b2d01.pphosted.com (8.16.0.20/8.16.0.20) with SMTP id v5SN5JYn007625 for ; Wed, 28 Jun 2017 19:10:58 -0400 Received: from e24smtp04.br.ibm.com (e24smtp04.br.ibm.com [32.104.18.25]) by mx0a-001b2d01.pphosted.com with ESMTP id 2bcgxubrr9-1 (version=TLSv1.2 cipher=AES256-SHA bits=256 verify=NOT) for ; Wed, 28 Jun 2017 19:10:58 -0400 Received: from localhost by e24smtp04.br.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Wed, 28 Jun 2017 20:10:55 -0300 Received: from d24relay03.br.ibm.com (9.13.39.225) by e24smtp04.br.ibm.com (10.172.0.140) with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted; Wed, 28 Jun 2017 20:10:53 -0300 Received: from d24av01.br.ibm.com (d24av01.br.ibm.com [9.8.31.91]) by d24relay03.br.ibm.com (8.14.9/8.14.9/NCO v10.0) with ESMTP id v5SN6aI226345520 for ; Wed, 28 Jun 2017 20:06:37 -0300 Received: from d24av01.br.ibm.com (localhost [127.0.0.1]) by d24av01.br.ibm.com (8.14.4/8.14.4/NCO v10.0 AVout) with ESMTP id v5SNArM0018648 for ; Wed, 28 Jun 2017 20:10:53 -0300 Received: from morokweng.ibm.com ([9.80.228.26]) by d24av01.br.ibm.com (8.14.4/8.14.4/NCO v10.0 AVin) with ESMTP id v5SNAZT9018415; Wed, 28 Jun 2017 20:10:51 -0300 From: Thiago Jung Bauermann To: linuxppc-dev@lists.ozlabs.org Subject: [PATCH v2 8/8] powerpc/perf/hv-24x7: Aggregate result elements on POWER9 SMT8 Date: Wed, 28 Jun 2017 20:10:26 -0300 X-Mailer: git-send-email 2.13.0 In-Reply-To: <20170628231026.23190-1-bauerman@linux.vnet.ibm.com> References: <20170628231026.23190-1-bauerman@linux.vnet.ibm.com> X-TM-AS-MML: disable x-cbid: 17062823-0028-0000-0000-000001C5B3A3 X-IBM-AV-DETECTION: SAVI=unused REMOTE=unused XFE=unused x-cbparentid: 17062823-0029-0000-0000-000014C71ADE Message-Id: <20170628231026.23190-9-bauerman@linux.vnet.ibm.com> X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:, , definitions=2017-06-28_14:, , signatures=0 X-Proofpoint-Spam-Details: rule=outbound_notspam policy=outbound score=0 spamscore=0 suspectscore=1 malwarescore=0 phishscore=0 adultscore=0 bulkscore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.0.1-1703280000 definitions=main-1706280369 X-BeenThere: linuxppc-dev@lists.ozlabs.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Thiago Jung Bauermann , Sukadev Bhattiprolu Errors-To: linuxppc-dev-bounces+patchwork-incoming=ozlabs.org@lists.ozlabs.org Sender: "Linuxppc-dev" On POWER9 SMT8 the 24x7 API returns two result elements for physical core and virtual CPU events and we need to add their counts to get the final result. Signed-off-by: Thiago Jung Bauermann --- arch/powerpc/perf/hv-24x7.c | 54 +++++++++++++++++++++++++++++++++++---------- 1 file changed, 42 insertions(+), 12 deletions(-) diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c index 38f6a81683a8..8a61b2415186 100644 --- a/arch/powerpc/perf/hv-24x7.c +++ b/arch/powerpc/perf/hv-24x7.c @@ -31,6 +31,9 @@ /* Version of the 24x7 hypervisor API that we should use in this machine. */ static int interface_version; +/* Whether we have to aggregate result data for some domains. */ +static bool aggregate_result_elements; + static bool domain_is_valid(unsigned domain) { switch (domain) { @@ -58,6 +61,15 @@ static bool is_physical_domain(unsigned domain) } } +/* Domains for which more than one result element are returned for each event. */ +static bool domain_needs_aggregation(unsigned int domain) +{ + return aggregate_result_elements && + (domain == HV_PERF_DOMAIN_PHYS_CORE || + (domain >= HV_PERF_DOMAIN_VCPU_HOME_CORE && + domain <= HV_PERF_DOMAIN_VCPU_REMOTE_NODE)); +} + static const char *domain_name(unsigned domain) { if (!domain_is_valid(domain)) @@ -1145,17 +1157,23 @@ static int add_event_to_24x7_request(struct perf_event *event, req->starting_ix = cpu_to_be16(idx); req->max_ix = cpu_to_be16(1); - if (request_buffer->interface_version > 1 && - req->performance_domain != HV_PERF_DOMAIN_PHYS_CHIP) { - req->starting_thread_group_ix = idx % 2; - req->max_num_thread_groups = 1; + if (request_buffer->interface_version > 1) { + if (domain_needs_aggregation(req->performance_domain)) + req->max_num_thread_groups = -1; + else if (req->performance_domain != HV_PERF_DOMAIN_PHYS_CHIP) { + req->starting_thread_group_ix = idx % 2; + req->max_num_thread_groups = 1; + } } return 0; } /** - * get_count_from_result - get event count from the given result + * get_count_from_result - get event count from all result elements in result + * + * If the event corresponding to this result needs aggregation of the result + * element values, then this function does that. * * @event: Event associated with @res. * @resb: Result buffer containing @res. @@ -1172,6 +1190,8 @@ static int get_count_from_result(struct perf_event *event, u16 data_size = be16_to_cpu(res->result_element_data_size); unsigned int data_offset; void *element_data; + int i; + u64 count; /* * We can bail out early if the result is empty. @@ -1189,8 +1209,10 @@ static int get_count_from_result(struct perf_event *event, /* * Since we always specify 1 as the maximum for the smallest resource * we're requesting, there should to be only one element per result. + * Except when an event needs aggregation, in which case there are more. */ - if (num_elements != 1) { + if (num_elements != 1 && + !domain_needs_aggregation(event_get_domain(event))) { pr_err("Error: result of request %hhu has %hu elements\n", res->result_ix, num_elements); @@ -1211,13 +1233,17 @@ static int get_count_from_result(struct perf_event *event, data_offset = offsetof(struct hv_24x7_result_element_v2, element_data); - element_data = res->elements + data_offset; + /* Go through the result elements in the result. */ + for (i = count = 0, element_data = res->elements + data_offset; + i < num_elements; + i++, element_data += data_size + data_offset) + count += be64_to_cpu(*((u64 *) element_data)); - *countp = be64_to_cpu(*((u64 *) element_data)); + *countp = count; - /* The next result is after the result element. */ + /* The next result is after the last result element. */ if (next) - *next = element_data + data_size; + *next = element_data - data_offset; return 0; } @@ -1565,9 +1591,13 @@ static int hv_24x7_init(void) } /* POWER8 only supports v1, while POWER9 only supports v2. */ - if (cpu_has_feature(CPU_FTR_ARCH_300)) + if (cpu_has_feature(CPU_FTR_ARCH_300)) { interface_version = 2; - else + + /* SMT8 in POWER9 needs to aggregate result elements. */ + if (threads_per_core == 8) + aggregate_result_elements = true; + } else interface_version = 1; hret = hv_perf_caps_get(&caps);