From patchwork Fri Mar 24 17:13:09 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jonathan Cameron X-Patchwork-Id: 74655 Return-Path: Delivered-To: ouuuleilei@gmail.com Received: by 2002:a59:b0ea:0:b0:3b6:4342:cba0 with SMTP id b10csp795563vqo; Fri, 24 Mar 2023 10:30:23 -0700 (PDT) X-Google-Smtp-Source: AKy350axb0kzOoSGMFbcYSgsfG8ZEj1uyoBSL0iq8ay/I9dfY4V3+QM5sI5lM5mztRIG/WvLwpd7 X-Received: by 2002:a17:90b:1e4e:b0:23f:5c60:67b with SMTP id pi14-20020a17090b1e4e00b0023f5c60067bmr3951709pjb.5.1679679023125; Fri, 24 Mar 2023 10:30:23 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1679679023; cv=none; d=google.com; s=arc-20160816; b=G3/7K7LLqDNcEvvvZYjT9siThtevT/PA/35Of95XqWbLEjWFvXwbZrSIp8FT7r69eA 8P/c+pLBZnTmu8S1eeC2Nciw7G+hPBfON/Yq8jGlxeU10ZRnR3A30T6X7+oTFvZfgHUt ckpDiygG6Uhb7Vfus4IWD3RHiBdA66ybueABfATN038mwFfGlpVKx1rNet8g/nnb52is GL/8OlJNUHAdsIJ2oRfgodTFfP7iTRlJ8Osxfzx9cf7OHHPpkaEXk2LTIKArJtxRPrkV uSEvFZ8o/J9Nz9jm5nZL/LpYaBfuSnBj2oZvpknNezLRPxfL1acLRjLSOwZATIUv1VoG 6DSw== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:content-transfer-encoding:mime-version :references:in-reply-to:message-id:date:subject:cc:to:from; bh=r/TQSANF7jqM0XmVHsyfHBesJNE0jGUJPn2XtyZ9Gy4=; b=eV28piH2KrHsucFfmJ8MOHVqm09yonSKvxKxCql96pMw4Vhv9PvpgzxUOmwLi/xaE7 9exzRtED8CIhgoNSiREblHQEJwkBunfmToVEZNB2EMYrz8JjOrPdU4sPJl8nXBZjt+Ju F/w6zHdl2deB34y+kq/WMpohiFvavQpbHchIfuJjequtAfQEnx8MP//jmTfP9Jek7due EN4ZwKgizz6SeD7n6pOj+Vrqm5FnNe5Xwlm/332fMsvHefcwiCB7DyerGvqu1t3lLDWO 1fmFUl0MmF3pJLtQ4sSKksmEqk2wsvRPHsW+CXhjmq3/FhAR2tHkubSgwx9l6dA4LFK5 aYgw== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: from out1.vger.email (out1.vger.email. [2620:137:e000::1:20]) by mx.google.com with ESMTP id a5-20020a17090a854500b0023a147a7892si4717795pjw.64.2023.03.24.10.30.10; Fri, 24 Mar 2023 10:30:23 -0700 (PDT) Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20; Authentication-Results: mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232123AbjCXROT (ORCPT + 99 others); Fri, 24 Mar 2023 13:14:19 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:46592 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S231596AbjCXROL (ORCPT ); Fri, 24 Mar 2023 13:14:11 -0400 Received: from frasgout.his.huawei.com (frasgout.his.huawei.com [185.176.79.56]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 0AE427AA5; Fri, 24 Mar 2023 10:13:46 -0700 (PDT) Received: from lhrpeml500005.china.huawei.com (unknown [172.18.147.200]) by frasgout.his.huawei.com (SkyGuard) with ESMTP id 4Pjpgl3MgDz6J7D9; Sat, 25 Mar 2023 01:13:07 +0800 (CST) Received: from SecurePC-101-06.china.huawei.com (10.122.247.231) by lhrpeml500005.china.huawei.com (7.191.163.240) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.21; Fri, 24 Mar 2023 17:13:37 +0000 From: Jonathan Cameron To: Liang Kan , , CC: , , , , , , , , Davidlohr Bueso , Dave Jiang Subject: [PATCH v2 1/5] cxl: Add function to count regblocks of a given type. Date: Fri, 24 Mar 2023 17:13:09 +0000 Message-ID: <20230324171313.18448-2-Jonathan.Cameron@huawei.com> X-Mailer: git-send-email 2.37.2 In-Reply-To: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> References: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> MIME-Version: 1.0 X-Originating-IP: [10.122.247.231] X-ClientProxiedBy: lhrpeml500002.china.huawei.com (7.191.160.78) To lhrpeml500005.china.huawei.com (7.191.163.240) X-CFilter-Loop: Reflected X-Spam-Status: No, score=-2.3 required=5.0 tests=RCVD_IN_DNSWL_MED, RCVD_IN_MSPIKE_H2,SPF_HELO_NONE,SPF_PASS autolearn=unavailable autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org X-getmail-retrieved-from-mailbox: =?utf-8?q?INBOX?= X-GMAIL-THRID: =?utf-8?q?1761271111093298625?= X-GMAIL-MSGID: =?utf-8?q?1761271111093298625?= Until the recently release CXL 3.0 specification, there was only ever one instance of any given register block pointed to by the Register Block Locator DVSEC. Now, the specification allows for multiple CXL PMU instances, each with their own register block. To enable this add an index parameter to cxl_find_regblock() and use that to implement cxl_count_regblock(). Reviewed-by: Davidlohr Bueso Reviewed-by: Dave Jiang Signed-off-by: Jonathan Cameron --- v2: Gathered tags. --- drivers/cxl/core/pci.c | 2 +- drivers/cxl/core/port.c | 2 +- drivers/cxl/core/regs.c | 34 +++++++++++++++++++++++++++++++--- drivers/cxl/cxl.h | 3 ++- drivers/cxl/pci.c | 2 +- 5 files changed, 36 insertions(+), 7 deletions(-) diff --git a/drivers/cxl/core/pci.c b/drivers/cxl/core/pci.c index 7328a2552411..c90251f60771 100644 --- a/drivers/cxl/core/pci.c +++ b/drivers/cxl/core/pci.c @@ -50,7 +50,7 @@ static int match_add_dports(struct pci_dev *pdev, void *data) &lnkcap)) return 0; - rc = cxl_find_regblock(pdev, CXL_REGLOC_RBI_COMPONENT, &map); + rc = cxl_find_regblock(pdev, CXL_REGLOC_RBI_COMPONENT, &map, 0); if (rc) dev_dbg(&port->dev, "failed to find component registers\n"); diff --git a/drivers/cxl/core/port.c b/drivers/cxl/core/port.c index 8ee6b6e2e2a4..97cc03dbceee 100644 --- a/drivers/cxl/core/port.c +++ b/drivers/cxl/core/port.c @@ -1333,7 +1333,7 @@ static resource_size_t find_component_registers(struct device *dev) pdev = to_pci_dev(dev); - cxl_find_regblock(pdev, CXL_REGLOC_RBI_COMPONENT, &map); + cxl_find_regblock(pdev, CXL_REGLOC_RBI_COMPONENT, &map, 0); return map.resource; } diff --git a/drivers/cxl/core/regs.c b/drivers/cxl/core/regs.c index 1476a0299c9b..7389dd1af967 100644 --- a/drivers/cxl/core/regs.c +++ b/drivers/cxl/core/regs.c @@ -290,6 +290,7 @@ static bool cxl_decode_regblock(struct pci_dev *pdev, u32 reg_lo, u32 reg_hi, * @pdev: The CXL PCI device to enumerate. * @type: Register Block Indicator id * @map: Enumeration output, clobbered on error + * @index: Index into which particular instance of a regblock we want. * * Return: 0 if register block enumerated, negative error code otherwise * @@ -297,9 +298,10 @@ static bool cxl_decode_regblock(struct pci_dev *pdev, u32 reg_lo, u32 reg_hi, * by @type. */ int cxl_find_regblock(struct pci_dev *pdev, enum cxl_regloc_type type, - struct cxl_register_map *map) + struct cxl_register_map *map, int index) { u32 regloc_size, regblocks; + int instance = 0; int regloc, i; map->resource = CXL_RESOURCE_NONE; @@ -323,8 +325,11 @@ int cxl_find_regblock(struct pci_dev *pdev, enum cxl_regloc_type type, if (!cxl_decode_regblock(pdev, reg_lo, reg_hi, map)) continue; - if (map->reg_type == type) - return 0; + if (map->reg_type == type) { + if (index == instance) + return 0; + instance++; + } } map->resource = CXL_RESOURCE_NONE; @@ -332,6 +337,29 @@ int cxl_find_regblock(struct pci_dev *pdev, enum cxl_regloc_type type, } EXPORT_SYMBOL_NS_GPL(cxl_find_regblock, CXL); +/** + * cxl_count_regblock() - Count instances of a given regblock type. + * @pdev: The CXL PCI device to enumerate. + * @type: Register Block Indicator id + * + * Some regblocks may be repeated. Count how many instances. + * + * Return: count of matching regblocks. + */ +int cxl_count_regblock(struct pci_dev *pdev, enum cxl_regloc_type type) +{ + struct cxl_register_map map; + int rc, count = 0; + + while (1) { + rc = cxl_find_regblock(pdev, type, &map, count); + if (rc) + return count; + count++; + } +} +EXPORT_SYMBOL_NS_GPL(cxl_count_regblock, CXL); + resource_size_t cxl_rcrb_to_component(struct device *dev, resource_size_t rcrb, enum cxl_rcrb which) diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h index aab87d74474d..0cc33e2a99d7 100644 --- a/drivers/cxl/cxl.h +++ b/drivers/cxl/cxl.h @@ -260,8 +260,9 @@ int cxl_map_device_regs(struct device *dev, struct cxl_device_regs *regs, struct cxl_register_map *map); enum cxl_regloc_type; +int cxl_count_regblock(struct pci_dev *pdev, enum cxl_regloc_type type); int cxl_find_regblock(struct pci_dev *pdev, enum cxl_regloc_type type, - struct cxl_register_map *map); + struct cxl_register_map *map, int index); enum cxl_rcrb { CXL_RCRB_DOWNSTREAM, diff --git a/drivers/cxl/pci.c b/drivers/cxl/pci.c index 60b23624d167..74443a5c3cc8 100644 --- a/drivers/cxl/pci.c +++ b/drivers/cxl/pci.c @@ -343,7 +343,7 @@ static int cxl_setup_regs(struct pci_dev *pdev, enum cxl_regloc_type type, { int rc; - rc = cxl_find_regblock(pdev, type, map); + rc = cxl_find_regblock(pdev, type, map, 0); if (rc) return rc; From patchwork Fri Mar 24 17:13:10 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jonathan Cameron X-Patchwork-Id: 74645 Return-Path: Delivered-To: ouuuleilei@gmail.com Received: by 2002:a59:b0ea:0:b0:3b6:4342:cba0 with SMTP id b10csp792075vqo; Fri, 24 Mar 2023 10:25:04 -0700 (PDT) X-Google-Smtp-Source: AKy350ZzmcX4WJDVGVhDGGk4vD/Bj38zhqmysw8UxhvEwvCMwKesqaronJZ3LU0iDAeySLvCdbMX X-Received: by 2002:a17:90b:38cd:b0:234:67ef:304b with SMTP id nn13-20020a17090b38cd00b0023467ef304bmr3829416pjb.37.1679678704632; Fri, 24 Mar 2023 10:25:04 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1679678704; cv=none; d=google.com; s=arc-20160816; b=Hnb8q4arr0xI7+slmA83i8cmCRTA4rtxPOnIDjk/i+gl713d0DDBvnLsDHh8UHX7M+ FxQEc7lwIKGcpppkFKGUILv1y4//TGh+pOZoX+3uWiEIxosJMIC3QNCVx/Lj5f9tZdYA r2J1hJvFLaqACokucig43r6vFbGXSKZykPEbHORZ6GjBuJJ+Vwd3bzzxqgYF3f+m9Hqb vuoosxoCYfDqzEOrglwOsA4NtS38jx7/114sE7Mefa7tLsqtAy6e3FX4AgJEY4k/TjRt tB39fe9047Sl2oyWY5diwWhZOUb/SucV9bTZTOXNBHyTgAZ/SkxWSxYLwWdgCp4gDYiY NRXA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:content-transfer-encoding:mime-version :references:in-reply-to:message-id:date:subject:cc:to:from; bh=Iy/RfxRzLkQ9aq41pZXqc351UxaPH+Nakq5sQQC+L4M=; b=yaQIV2wNTp4xUZZK7hKGMUi99jlwT9ckS8GiTVnf/IdG//VJ7haiP1o6xjQ9Cl9R5m 3lVHaFvr07S8GEUNiMeKvuHRrdGsdOxDvnw+XeMqhYWuApw/Hfgc/Ol/fTNpvF0io2LN JZehQvU3CX6m84KOjKBBfNs9zfoJdJSNGiNjKxjTEZ/5t3D5/5SG+tnjzv4YdPd9lQ9c 7INNU1F/hq0sB4T2q5WrRNVrX1wOMm2weUxvf2MK8XKlaYqYx5Pk1cmV5YnU3+jwR3dv dxsnkGox2L0In8KfwfUjGJFZSpVRVNZf8CmjO0PpmyclgbyaBAADpFPgQR+uuxIjrm5y PD0A== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: from out1.vger.email (out1.vger.email. [2620:137:e000::1:20]) by mx.google.com with ESMTP id 101-20020a17090a0fee00b0023676bd55d1si238917pjz.94.2023.03.24.10.24.52; Fri, 24 Mar 2023 10:25:04 -0700 (PDT) Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20; Authentication-Results: mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232149AbjCXROj (ORCPT + 99 others); Fri, 24 Mar 2023 13:14:39 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:47350 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232102AbjCXROb (ORCPT ); Fri, 24 Mar 2023 13:14:31 -0400 Received: from frasgout.his.huawei.com (frasgout.his.huawei.com [185.176.79.56]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id A573830CA; Fri, 24 Mar 2023 10:14:10 -0700 (PDT) Received: from lhrpeml500005.china.huawei.com (unknown [172.18.147.226]) by frasgout.his.huawei.com (SkyGuard) with ESMTP id 4Pjpcv6hgjz67PMg; Sat, 25 Mar 2023 01:10:39 +0800 (CST) Received: from SecurePC-101-06.china.huawei.com (10.122.247.231) by lhrpeml500005.china.huawei.com (7.191.163.240) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.21; Fri, 24 Mar 2023 17:14:08 +0000 From: Jonathan Cameron To: Liang Kan , , CC: , , , , , , , , Davidlohr Bueso , Dave Jiang Subject: [PATCH v2 2/5] perf: Allow a PMU to have a parent. Date: Fri, 24 Mar 2023 17:13:10 +0000 Message-ID: <20230324171313.18448-3-Jonathan.Cameron@huawei.com> X-Mailer: git-send-email 2.37.2 In-Reply-To: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> References: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> MIME-Version: 1.0 X-Originating-IP: [10.122.247.231] X-ClientProxiedBy: lhrpeml500002.china.huawei.com (7.191.160.78) To lhrpeml500005.china.huawei.com (7.191.163.240) X-CFilter-Loop: Reflected X-Spam-Status: No, score=-2.3 required=5.0 tests=RCVD_IN_DNSWL_MED, RCVD_IN_MSPIKE_H2,SPF_HELO_NONE,SPF_PASS autolearn=unavailable autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org X-getmail-retrieved-from-mailbox: =?utf-8?q?INBOX?= X-GMAIL-THRID: =?utf-8?q?1761270777287204324?= X-GMAIL-MSGID: =?utf-8?q?1761270777287204324?= Some PMUs have well defined parents such as PCI devices. As the device_initialize() and device_add() are all within pmu_dev_alloc() which is called from perf_pmu_register() there is no opportunity to set the parent from within a driver. Add a struct device *parent field to struct pmu and use that to set the parent. Signed-off-by: Jonathan Cameron --- include/linux/perf_event.h | 1 + kernel/events/core.c | 1 + 2 files changed, 2 insertions(+) diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index d5628a7b5eaa..b99db1eda72c 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -303,6 +303,7 @@ struct pmu { struct module *module; struct device *dev; + struct device *parent; const struct attribute_group **attr_groups; const struct attribute_group **attr_update; const char *name; diff --git a/kernel/events/core.c b/kernel/events/core.c index fb3e436bcd4a..a84c282221f2 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -11367,6 +11367,7 @@ static int pmu_dev_alloc(struct pmu *pmu) dev_set_drvdata(pmu->dev, pmu); pmu->dev->bus = &pmu_bus; + pmu->dev->parent = pmu->parent; pmu->dev->release = pmu_dev_release; ret = dev_set_name(pmu->dev, "%s", pmu->name); From patchwork Fri Mar 24 17:13:11 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jonathan Cameron X-Patchwork-Id: 74651 Return-Path: Delivered-To: ouuuleilei@gmail.com Received: by 2002:a59:b0ea:0:b0:3b6:4342:cba0 with SMTP id b10csp794621vqo; Fri, 24 Mar 2023 10:28:56 -0700 (PDT) X-Google-Smtp-Source: AK7set99yR0jTwykzYCXXeXYK+tXsjlnvV1k1wO6MzIcVrVJQYDgNhcVAffAvaVGNxKiFUeWJURj X-Received: by 2002:a05:6a20:9325:b0:d4:c1ba:9f4e with SMTP id r37-20020a056a20932500b000d4c1ba9f4emr3100050pzh.35.1679678936710; Fri, 24 Mar 2023 10:28:56 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1679678936; cv=none; d=google.com; s=arc-20160816; b=r/cZjOKYNC7dXzxB/Hooe6CEb0SsCbRYnKF94Muh0y6QnFZN3Zp8mjWvewxM7demVn SuWcj7b2DkuoA/sSxbeTqs+RVi32EnMT45csMASZ+j+km5X9t+/oB2zaUTSRjm0KoBt9 Kzy2Dw5A5XRYiQE/6SckRQikWs2susJcl6+WlcjO46h77uJUfFvPFvTkotW9it+XKLPk D0o8Dhykq4LeXpCxbjUhhKlPsHbzorG7pFJE8Dc/txnIJxJfs9Jas/QokTSsrZ7GSNs9 j/Sv3m6/NCQj0a0oEKEX1b+MSDxOBkmuGFnsehKRZcPchZAnLtytpEVNioU+1e69WXMm MqCg== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:content-transfer-encoding:mime-version :references:in-reply-to:message-id:date:subject:cc:to:from; bh=RcPih+GgNbC+Q3eKDPDBxR4DighcCns22h7JanQ/+1I=; b=Nh0Lj8SU5XtqVLvT+q246E5njBGzXheAZQbezBgFxwiHu9MRDBYv9ubFm92JzEVmc3 t5qMOYKZk2By4ylKYV05Sls8ap8wAyHiPbCfib1Q0RxBOD4aW3lmv8dBeV2iYf4yVvSL PEmWwL77+OoZdr/B7ebaf+p3IKOcofW8IspE2SFEERi5aqh2rD+5yKZR06m/GPNQZfnx iyKceWaBnaDw3HDfrhrpElcqqEwIWpU149Gw8lWVKFW90x7BfQz1O0zDV8AxL++yDe/n Xtw/cxbDeRamXc5CCc3XV6pj7VvzeV14QLnFloL0bSOkXWOecONfkSBB8XT/wtxeJKCE oAlA== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: from out1.vger.email (out1.vger.email. [2620:137:e000::1:20]) by mx.google.com with ESMTP id m24-20020a634c58000000b0050c0a00c1edsi21419827pgl.703.2023.03.24.10.28.43; Fri, 24 Mar 2023 10:28:56 -0700 (PDT) Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20; Authentication-Results: mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232187AbjCXRPp (ORCPT + 99 others); Fri, 24 Mar 2023 13:15:45 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:46916 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232102AbjCXRPc (ORCPT ); Fri, 24 Mar 2023 13:15:32 -0400 Received: from frasgout.his.huawei.com (frasgout.his.huawei.com [185.176.79.56]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 52D76CDCB; Fri, 24 Mar 2023 10:15:02 -0700 (PDT) Received: from lhrpeml500005.china.huawei.com (unknown [172.18.147.201]) by frasgout.his.huawei.com (SkyGuard) with ESMTP id 4Pjphx0KpFz6J6P3; Sat, 25 Mar 2023 01:14:09 +0800 (CST) Received: from SecurePC-101-06.china.huawei.com (10.122.247.231) by lhrpeml500005.china.huawei.com (7.191.163.240) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.21; Fri, 24 Mar 2023 17:14:39 +0000 From: Jonathan Cameron To: Liang Kan , , CC: , , , , , , , , Davidlohr Bueso , Dave Jiang Subject: [PATCH v2 3/5] cxl/pci: Find and register CXL PMU devices Date: Fri, 24 Mar 2023 17:13:11 +0000 Message-ID: <20230324171313.18448-4-Jonathan.Cameron@huawei.com> X-Mailer: git-send-email 2.37.2 In-Reply-To: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> References: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> MIME-Version: 1.0 X-Originating-IP: [10.122.247.231] X-ClientProxiedBy: lhrpeml100001.china.huawei.com (7.191.160.183) To lhrpeml500005.china.huawei.com (7.191.163.240) X-CFilter-Loop: Reflected X-Spam-Status: No, score=-2.3 required=5.0 tests=RCVD_IN_DNSWL_MED, RCVD_IN_MSPIKE_H2,SPF_HELO_NONE,SPF_PASS autolearn=unavailable autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org X-getmail-retrieved-from-mailbox: =?utf-8?q?INBOX?= X-GMAIL-THRID: =?utf-8?q?1761271020820321507?= X-GMAIL-MSGID: =?utf-8?q?1761271020820321507?= CXL PMU devices can be found from entries in the Register Locator DVSEC. In order to register the minimum number of IRQ vectors necessary to support all CPMUs found, separate the registration into two steps. First find the devices, and query the IRQs used and then register the devices. Between these two steps, request the IRQ vectors necessary and enable bus master support. Future IRQ users for CXL type 3 devices (e.g. DOEs) will need to follow a similar pattern the number of vectors necessary is known before any parts of the driver stack rely on their availability. Reviewed-by: Dave Jiang Signed-off-by: Jonathan Cameron --- v2: - Add missing cpmu.h - Fix free of unallocated ida by reordering that to be before device_initialize(). (Thanks to Davidlohr) --- drivers/cxl/core/Makefile | 1 + drivers/cxl/core/core.h | 3 ++ drivers/cxl/core/cpmu.c | 72 +++++++++++++++++++++++++++++++++++++++ drivers/cxl/core/port.c | 2 ++ drivers/cxl/core/regs.c | 16 +++++++++ drivers/cxl/cpmu.h | 56 ++++++++++++++++++++++++++++++ drivers/cxl/cxl.h | 14 ++++++++ drivers/cxl/cxlpci.h | 1 + drivers/cxl/pci.c | 25 +++++++++++++- 9 files changed, 189 insertions(+), 1 deletion(-) diff --git a/drivers/cxl/core/Makefile b/drivers/cxl/core/Makefile index ca4ae31d8f57..45e5543aff52 100644 --- a/drivers/cxl/core/Makefile +++ b/drivers/cxl/core/Makefile @@ -12,5 +12,6 @@ cxl_core-y += memdev.o cxl_core-y += mbox.o cxl_core-y += pci.o cxl_core-y += hdm.o +cxl_core-y += cpmu.o cxl_core-$(CONFIG_TRACING) += trace.o cxl_core-$(CONFIG_CXL_REGION) += region.o diff --git a/drivers/cxl/core/core.h b/drivers/cxl/core/core.h index cde475e13216..05e18fed3a75 100644 --- a/drivers/cxl/core/core.h +++ b/drivers/cxl/core/core.h @@ -17,12 +17,14 @@ extern struct device_attribute dev_attr_region; extern const struct device_type cxl_pmem_region_type; extern const struct device_type cxl_dax_region_type; extern const struct device_type cxl_region_type; +extern const struct device_type cxl_cpmu_type; void cxl_decoder_kill_region(struct cxl_endpoint_decoder *cxled); #define CXL_REGION_ATTR(x) (&dev_attr_##x.attr) #define CXL_REGION_TYPE(x) (&cxl_region_type) #define SET_CXL_REGION_ATTR(x) (&dev_attr_##x.attr), #define CXL_PMEM_REGION_TYPE(x) (&cxl_pmem_region_type) #define CXL_DAX_REGION_TYPE(x) (&cxl_dax_region_type) +#define CXL_CPMU_TYPE(x) (&cxl_cpmu_region_type) int cxl_region_init(void); void cxl_region_exit(void); #else @@ -41,6 +43,7 @@ static inline void cxl_region_exit(void) #define SET_CXL_REGION_ATTR(x) #define CXL_PMEM_REGION_TYPE(x) NULL #define CXL_DAX_REGION_TYPE(x) NULL +#define CXL_CPMU_TYPE(x) NULL #endif struct cxl_send_command; diff --git a/drivers/cxl/core/cpmu.c b/drivers/cxl/core/cpmu.c new file mode 100644 index 000000000000..8aff35055c05 --- /dev/null +++ b/drivers/cxl/core/cpmu.c @@ -0,0 +1,72 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2022 Huawei. All rights reserved. */ + +#include +#include +#include +#include +#include +#include +#include "core.h" + +static DEFINE_IDA(cpmu_ida); + +static void cxl_cpmu_release(struct device *dev) +{ + struct cxl_cpmu *cpmu = container_of(dev, struct cxl_cpmu, dev); + + ida_free(&cpmu_ida, cpmu->id); + kfree(cpmu); +} + +const struct device_type cxl_cpmu_type = { + .name = "cxl_cpmu", + .release = cxl_cpmu_release, +}; + +static void remove_dev(void *dev) +{ + device_del(dev); +} + +int devm_cxl_cpmu_add(struct device *parent, struct cxl_cpmu_regs *regs, int index) +{ + struct cxl_cpmu *cpmu; + struct device *dev; + int rc; + + cpmu = kzalloc(sizeof(*cpmu), GFP_KERNEL); + if (!cpmu) + return -ENOMEM; + + rc = ida_alloc(&cpmu_ida, GFP_KERNEL); + if (rc < 0) { + kfree(cpmu); + return rc; + } + cpmu->id = rc; + + cpmu->base = regs->cpmu; + dev = &cpmu->dev; + device_initialize(dev); + device_set_pm_not_required(dev); + dev->parent = parent; + dev->bus = &cxl_bus_type; + dev->type = &cxl_cpmu_type; + + rc = dev_set_name(dev, "cpmu%d", cpmu->id); + if (rc) + goto err; + + rc = device_add(dev); + if (rc) + goto err; + + return devm_add_action_or_reset(parent, remove_dev, dev); + +err: + put_device(&cpmu->dev); + return rc; +} +EXPORT_SYMBOL_NS_GPL(devm_cxl_cpmu_add, CXL); + diff --git a/drivers/cxl/core/port.c b/drivers/cxl/core/port.c index 97cc03dbceee..2154bf8d2aad 100644 --- a/drivers/cxl/core/port.c +++ b/drivers/cxl/core/port.c @@ -57,6 +57,8 @@ static int cxl_device_id(const struct device *dev) return CXL_DEVICE_MEMORY_EXPANDER; if (dev->type == CXL_REGION_TYPE()) return CXL_DEVICE_REGION; + if (dev->type == &cxl_cpmu_type) + return CXL_DEVICE_CPMU; return 0; } diff --git a/drivers/cxl/core/regs.c b/drivers/cxl/core/regs.c index 7389dd1af967..99d6ebe3aba9 100644 --- a/drivers/cxl/core/regs.c +++ b/drivers/cxl/core/regs.c @@ -6,6 +6,7 @@ #include #include #include +#include #include "core.h" @@ -360,6 +361,21 @@ int cxl_count_regblock(struct pci_dev *pdev, enum cxl_regloc_type type) } EXPORT_SYMBOL_NS_GPL(cxl_count_regblock, CXL); +int cxl_map_cpmu_regs(struct pci_dev *pdev, struct cxl_cpmu_regs *regs, + struct cxl_register_map *map) +{ + struct device *dev = &pdev->dev; + resource_size_t phys_addr; + + phys_addr = map->resource; + regs->cpmu = devm_cxl_iomap_block(dev, phys_addr, CPMU_REGMAP_SIZE); + if (!regs->cpmu) + return -ENOMEM; + + return 0; +} +EXPORT_SYMBOL_NS_GPL(cxl_map_cpmu_regs, CXL); + resource_size_t cxl_rcrb_to_component(struct device *dev, resource_size_t rcrb, enum cxl_rcrb which) diff --git a/drivers/cxl/cpmu.h b/drivers/cxl/cpmu.h new file mode 100644 index 000000000000..2cf78a6b9b13 --- /dev/null +++ b/drivers/cxl/cpmu.h @@ -0,0 +1,56 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * Copyright(c) 2022 Huawei + * CXL Specification rev 3.0 Setion 8.2.7 (CPMU Register Interface) + */ +#ifndef CXL_CPMU_H +#define CXL_CPMU_H + +#define CPMU_CAP_REG 0x0 +#define CPMU_CAP_NUM_COUNTERS_MSK GENMASK_ULL(4, 0) +#define CPMU_CAP_COUNTER_WIDTH_MSK GENMASK_ULL(15, 8) +#define CPMU_CAP_NUM_EVN_CAP_REG_SUP_MSK GENMASK_ULL(24, 20) +#define CPMU_CAP_FILTERS_SUP_MSK GENMASK_ULL(39, 32) +#define CPMU_FILTER_HDM BIT(0) +#define CPMU_FILTER_CHAN_RANK_BANK BIT(1) +#define CPMU_CAP_MSI_N_MSK GENMASK_ULL(47, 44) +#define CPMU_CAP_WRITEABLE_WHEN_FROZEN BIT_ULL(48) +#define CPMU_CAP_FREEZE BIT_ULL(49) +#define CPMU_CAP_INT BIT_ULL(50) +#define CPMU_CAP_VERSION_MSK GENMASK_ULL(63, 60) + +#define CPMU_OVERFLOW_REG 0x10 +#define CPMU_FREEZE_REG 0x18 +#define CPMU_EVENT_CAP_REG(n) (0x100 + 8 * (n)) +#define CPMU_EVENT_CAP_SUPPORTED_EVENTS_MSK GENMASK_ULL(31, 0) +#define CPMU_EVENT_CAP_GROUP_ID_MSK GENMASK_ULL(47, 32) +#define CPMU_EVENT_CAP_VENDOR_ID_MSK GENMASK_ULL(63, 48) + +#define CPMU_COUNTER_CFG_REG(n) (0x200 + 8 * (n)) +#define CPMU_COUNTER_CFG_TYPE_MSK GENMASK_ULL(1, 0) +#define CPMU_COUNTER_CFG_TYPE_FREE_RUN 0 +#define CPMU_COUNTER_CFG_TYPE_FIXED_FUN 1 +#define CPMU_COUNTER_CFG_TYPE_CONFIGURABLE 2 +#define CPMU_COUNTER_CFG_ENABLE BIT_ULL(8) +#define CPMU_COUNTER_CFG_INT_ON_OVRFLW BIT_ULL(9) +#define CPMU_COUNTER_CFG_FREEZE_ON_OVRFLW BIT_ULL(10) +#define CPMU_COUNTER_CFG_EDGE BIT_ULL(11) +#define CPMU_COUNTER_CFG_INVERT BIT_ULL(12) +#define CPMU_COUNTER_CFG_THRESHOLD_MSK GENMASK_ULL(23, 16) +#define CPMU_COUNTER_CFG_EVENTS_MSK GENMASK_ULL(55, 24) +#define CPMU_COUNTER_CFG_EVENT_GRP_ID_IDX_MSK GENMASK_ULL(63, 59) + +#define CPMU_FILTER_CFG_REG(n, f) (0x400 + 4 * ((f) + (n) * 8)) +#define CPMU_FILTER_CFG_VALUE_MSK GENMASK(15, 0) + +#define CPMU_COUNTER_REG(n) (0xc00 + 8 * (n)) + +#define CPMU_REGMAP_SIZE 0xe00 /* Table 8-32 CXL 3.0 specification */ +struct cxl_cpmu { + struct device dev; + void __iomem *base; + int id; +}; + +#define to_cxl_cpmu(dev) container_of(dev, struct cxl_cpmu, dev) +#endif diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h index 0cc33e2a99d7..a7b8a0c46715 100644 --- a/drivers/cxl/cxl.h +++ b/drivers/cxl/cxl.h @@ -209,6 +209,10 @@ struct cxl_regs { struct_group_tagged(cxl_device_regs, device_regs, void __iomem *status, *mbox, *memdev; ); + + struct_group_tagged(cxl_cpmu_regs, cpmu_regs, + void __iomem *cpmu; + ); }; struct cxl_reg_map { @@ -229,6 +233,10 @@ struct cxl_device_reg_map { struct cxl_reg_map memdev; }; +struct cxl_cpmu_reg_map { + struct cxl_reg_map cpmu; +}; + /** * struct cxl_register_map - DVSEC harvested register block mapping parameters * @base: virtual base of the register-block-BAR + @block_offset @@ -237,6 +245,7 @@ struct cxl_device_reg_map { * @reg_type: see enum cxl_regloc_type * @component_map: cxl_reg_map for component registers * @device_map: cxl_reg_maps for device registers + * @cpmu_map: cxl_reg_maps for CXL Performance Monitoring Units */ struct cxl_register_map { void __iomem *base; @@ -246,6 +255,7 @@ struct cxl_register_map { union { struct cxl_component_reg_map component_map; struct cxl_device_reg_map device_map; + struct cxl_cpmu_reg_map cpmu_map; }; }; @@ -258,6 +268,8 @@ int cxl_map_component_regs(struct device *dev, struct cxl_component_regs *regs, unsigned long map_mask); int cxl_map_device_regs(struct device *dev, struct cxl_device_regs *regs, struct cxl_register_map *map); +int cxl_map_cpmu_regs(struct pci_dev *pdev, struct cxl_cpmu_regs *regs, + struct cxl_register_map *map); enum cxl_regloc_type; int cxl_count_regblock(struct pci_dev *pdev, enum cxl_regloc_type type); @@ -750,6 +762,7 @@ void cxl_driver_unregister(struct cxl_driver *cxl_drv); #define CXL_DEVICE_REGION 6 #define CXL_DEVICE_PMEM_REGION 7 #define CXL_DEVICE_DAX_REGION 8 +#define CXL_DEVICE_CPMU 9 #define MODULE_ALIAS_CXL(type) MODULE_ALIAS("cxl:t" __stringify(type) "*") #define CXL_MODALIAS_FMT "cxl:t%d" @@ -789,6 +802,7 @@ static inline struct cxl_dax_region *to_cxl_dax_region(struct device *dev) } #endif +int devm_cxl_cpmu_add(struct device *parent, struct cxl_cpmu_regs *regs, int idx); /* * Unit test builds overrides this to __weak, find the 'strong' version * of these symbols in tools/testing/cxl/. diff --git a/drivers/cxl/cxlpci.h b/drivers/cxl/cxlpci.h index be6a2ef3cce3..2fd495ab3de1 100644 --- a/drivers/cxl/cxlpci.h +++ b/drivers/cxl/cxlpci.h @@ -65,6 +65,7 @@ enum cxl_regloc_type { CXL_REGLOC_RBI_COMPONENT, CXL_REGLOC_RBI_VIRT, CXL_REGLOC_RBI_MEMDEV, + CXL_REGLOC_RBI_CPMU, CXL_REGLOC_RBI_TYPES }; diff --git a/drivers/cxl/pci.c b/drivers/cxl/pci.c index 74443a5c3cc8..e8bc36cc2724 100644 --- a/drivers/cxl/pci.c +++ b/drivers/cxl/pci.c @@ -704,7 +704,7 @@ static int cxl_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id) struct cxl_register_map map; struct cxl_memdev *cxlmd; struct cxl_dev_state *cxlds; - int rc; + int i, rc, cpmu_count; /* * Double check the anonymous union trickery in struct cxl_regs @@ -781,6 +781,29 @@ static int cxl_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id) if (rc) return rc; + cpmu_count = cxl_count_regblock(pdev, CXL_REGLOC_RBI_CPMU); + for (i = 0; i < cpmu_count; i++) { + struct cxl_cpmu_regs cpmu_regs; + + rc = cxl_find_regblock(pdev, CXL_REGLOC_RBI_CPMU, &map, i); + if (rc) { + dev_dbg(&pdev->dev, "Could not find CPMU regblock\n"); + break; + } + + rc = cxl_map_cpmu_regs(pdev, &cpmu_regs, &map); + if (rc) { + dev_dbg(&pdev->dev, "Could not map CPMU regs\n"); + break; + } + + rc = devm_cxl_cpmu_add(cxlds->dev, &cpmu_regs, i); + if (rc) { + dev_dbg(&pdev->dev, "Could not add CPMU instance\n"); + break; + } + } + cxlmd = devm_cxl_add_memdev(cxlds); if (IS_ERR(cxlmd)) return PTR_ERR(cxlmd); From patchwork Fri Mar 24 17:13:12 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jonathan Cameron X-Patchwork-Id: 74650 Return-Path: Delivered-To: ouuuleilei@gmail.com Received: by 2002:a59:b0ea:0:b0:3b6:4342:cba0 with SMTP id b10csp792927vqo; Fri, 24 Mar 2023 10:26:24 -0700 (PDT) X-Google-Smtp-Source: AKy350YdCeH7mGEbjnfPkBEhJl+oaWd0PdOBAM2xkEbS7GXVWgfQDA8nuICdfXaOzCPvampiEgYX X-Received: by 2002:a17:903:1c5:b0:1a1:cf70:33d5 with SMTP id e5-20020a17090301c500b001a1cf7033d5mr3700394plh.56.1679678784506; Fri, 24 Mar 2023 10:26:24 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1679678784; cv=none; d=google.com; s=arc-20160816; b=rNW1ts9+VEn4Fo6L+QB/icGEkjv1wv3blK/ERsE5F7by6nQywDczLe4FPQptJaAw8a UNoalsz8MiN6zKO/E2uzqDsnxkXdR7bDQJVHZLm8nNE6+ogWcoNHkSsUbrEzCB6rwBql 559k4QxQcObd5S/LhvRnnRP4cdLKIdShs42Hs85+P3uXrwo0FaRR/jlvAL9Z+fk2jt8z auq8orlIcS6mb+9E80SbbMoq4heDqXDS0fgEtQorVilZZZA0z0JyLsRLgQLuT6HmqhZI +/gmFBghnMu1R6j9zI+zRkEjIqYNbRVV3ZhlTZeF7Y8G2dX0sjMukUUhe+NvljGJud1V Llig== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:content-transfer-encoding:mime-version :references:in-reply-to:message-id:date:subject:cc:to:from; bh=DsObcZ13R77onTxt1Lg/KMnxFXurGvEc3iMtSgm0aCw=; b=yNNOdxJxkEJ9+V4ifonh2OlR+X91jOo9lb4AX0rWqGV6jc7I0i8GSE+M8hMp56cd84 1XxQPYp6w6owGF8XKj6MMX3+S7PkChpkzqnYEsVeDpAD+QyuQ8kBfyRZWeJlVI49ajPv c/kaxa1rdcZ48+IA9QmU6drwIqNinBY4wbqL6glkfFf+/MGVRilyyZYEqhtmeHzMI7kG swWoKYr3WY6sshx1yLEIPY/0vZ9xl60PPB2Ij1SEBcuio21jCS7B/B/QkGWYwTjy2pNM G6LdJWyPPMkRdO2tpsGYZEGI2iKsPgJnhwI7ECSxnN8fqWLJcR59jCsP3rKfh9/n3MBx 3YVA== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: from out1.vger.email (out1.vger.email. [2620:137:e000::1:20]) by mx.google.com with ESMTP id g21-20020a170902d1d500b0019aa6450ff3si20017515plb.585.2023.03.24.10.26.11; Fri, 24 Mar 2023 10:26:24 -0700 (PDT) Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20; Authentication-Results: mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232307AbjCXRQ1 (ORCPT + 99 others); Fri, 24 Mar 2023 13:16:27 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:49192 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S231984AbjCXRQJ (ORCPT ); Fri, 24 Mar 2023 13:16:09 -0400 Received: from frasgout.his.huawei.com (frasgout.his.huawei.com [185.176.79.56]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 42EB87ED1; Fri, 24 Mar 2023 10:15:38 -0700 (PDT) Received: from lhrpeml500005.china.huawei.com (unknown [172.18.147.200]) by frasgout.his.huawei.com (SkyGuard) with ESMTP id 4PjphV72dKz6J7BP; Sat, 25 Mar 2023 01:13:46 +0800 (CST) Received: from SecurePC-101-06.china.huawei.com (10.122.247.231) by lhrpeml500005.china.huawei.com (7.191.163.240) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.21; Fri, 24 Mar 2023 17:15:09 +0000 From: Jonathan Cameron To: Liang Kan , , CC: , , , , , , , , Davidlohr Bueso , Dave Jiang Subject: [PATCH v2 4/5] cxl: CXL Performance Monitoring Unit driver Date: Fri, 24 Mar 2023 17:13:12 +0000 Message-ID: <20230324171313.18448-5-Jonathan.Cameron@huawei.com> X-Mailer: git-send-email 2.37.2 In-Reply-To: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> References: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> MIME-Version: 1.0 X-Originating-IP: [10.122.247.231] X-ClientProxiedBy: lhrpeml500002.china.huawei.com (7.191.160.78) To lhrpeml500005.china.huawei.com (7.191.163.240) X-CFilter-Loop: Reflected X-Spam-Status: No, score=-2.3 required=5.0 tests=RCVD_IN_DNSWL_MED, RCVD_IN_MSPIKE_H2,SPF_HELO_NONE,SPF_PASS autolearn=unavailable autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org X-getmail-retrieved-from-mailbox: =?utf-8?q?INBOX?= X-GMAIL-THRID: =?utf-8?q?1761270861135675217?= X-GMAIL-MSGID: =?utf-8?q?1761270861135675217?= CXL rev 3.0 introduces a standard performance monitoring hardware block to CXL. Instances are discovered using CXL Register Locator DVSEC entries. Each CXL component may have multiple PMUs. This initial driver supports on a subset of types of counter. It support counters that are either fixed or configurable, but requires that they support the ability to freeze and write value whilst frozen. Development done with QEMU model which will be posted shortly. Reviewed-by: Dave Jiang Signed-off-by: Jonathan Cameron --- Thanks to Kan Liang for detailed review and discussion of the configurability description to userspace. Note that complexity is left for a future patch set. v2: - Split the events list in to configurable and fixed cases. - Add utility functions to search each list. - Allow fallback to configurable counter if fixed counter for same thing is already in use. - Fix check of filter. - Add a macro for the CXL defined events and use the PCI_DVSEC_VENDOR_ID_CXL define. Sure this isn't DVSEC but the value was first used there and naming that define is complex for fun legal reasons. - Verify the event capability for what is requested is possible in cpmu_event_init() to error out earlier if not. - Drop an unnecessary check on enable in cpmu_pmu_enable() - Use bitmap manipulation to make it cheaper to find an unused counter. --- drivers/cxl/Kconfig | 12 + drivers/cxl/Makefile | 1 + drivers/cxl/cpmu.c | 937 +++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 950 insertions(+) diff --git a/drivers/cxl/Kconfig b/drivers/cxl/Kconfig index ff4e78117b31..d68fc5769c58 100644 --- a/drivers/cxl/Kconfig +++ b/drivers/cxl/Kconfig @@ -139,4 +139,16 @@ config CXL_REGION_INVALIDATION_TEST If unsure, or if this kernel is meant for production environments, say N. +config CXL_CPMU + tristate "CXL Performance Monitoring Unit" + default CXL_BUS + help + Support performance monitoring as defined in CXL rev 3.0 + section 13.2: Performance Monitoring. CXL components may have + one or more CXL Performance Monitoring Units (CPMUs). + + Say 'y/m' to enable a driver that will attach to performance + monitoring units and provide standard perf based interfaces. + + If unsure say 'm'. endif diff --git a/drivers/cxl/Makefile b/drivers/cxl/Makefile index db321f48ba52..024bb739554b 100644 --- a/drivers/cxl/Makefile +++ b/drivers/cxl/Makefile @@ -5,6 +5,7 @@ obj-$(CONFIG_CXL_MEM) += cxl_mem.o obj-$(CONFIG_CXL_ACPI) += cxl_acpi.o obj-$(CONFIG_CXL_PMEM) += cxl_pmem.o obj-$(CONFIG_CXL_PORT) += cxl_port.o +obj-$(CONFIG_CXL_CPMU) += cpmu.o cxl_mem-y := mem.o cxl_pci-y := pci.o diff --git a/drivers/cxl/cpmu.c b/drivers/cxl/cpmu.c new file mode 100644 index 000000000000..7a259f9de05e --- /dev/null +++ b/drivers/cxl/cpmu.c @@ -0,0 +1,937 @@ +// SPDX-License-Identifier: GPL-2.0-only + +/* + * Copyright(c) 2023 Huawei + * + * The CXL 3.0 specification includes a standard Performance Monitoring Unit, + * called the CXL PMU, or CPMU. In order to allow a high degree of + * implementation flexibility the specification provides a wide range of + * options all of which are self describing. + * + * Details in CXL rev 3.0 section 8.2.7 CPMU Register Interface + * + * TODO + * o Discoverability of counters. Allow perftool to provide summed counters + * and vendor defined counters. + * o Support free running counters - copy the Intel uncore PMU handling for these. + * o CPMUs which do not support freeze. + * o Add filter validation in cpmu_event_init() so problems are detected earlier. + * o Reject configurations that the hardware is ignoring + * (e.g. invert when not invertible) + * o Support CPMUs with no interrupts using an HRTIMER. + */ + +#include +#include +#include +#include +#include +#include +#include +#include "cpmu.h" +#include "cxlpci.h" +#include "cxl.h" + +/* CXL rev 3.0 Table 13-5 Events under CXL Vendor ID */ +#define CPMU_GID_CLOCK_TICKS 0x00 +#define CPMU_GID_D2H_REQ 0x0010 +#define CPMU_GID_D2H_RSP 0x0011 +#define CPMU_GID_H2D_REQ 0x0012 +#define CPMU_GID_H2D_RSP 0x0013 +#define CPMU_GID_CACHE_DATA 0x0014 +#define CPMU_GID_M2S_REQ 0x0020 +#define CPMU_GID_M2S_RWD 0x0021 +#define CPMU_GID_M2S_BIRSP 0x0022 +#define CPMU_GID_S2M_BISNP 0x0023 +#define CPMU_GID_S2M_NDR 0x0024 +#define CPMU_GID_S2M_DRS 0x0025 +#define CPMU_GID_DDR 0x8000 + +static int cpmu_cpuhp_state_num; + +struct cpmu_event { + u16 vid; + u16 gid; + u32 msk; + union { + int counter_idx; /* fixed counters */ + int event_idx; /* configurable counters */ + }; + struct list_head node; +}; + +#define CPMU_MAX_COUNTERS 64 +struct cpmu_info { + struct pmu pmu; + void __iomem *base; + struct perf_event **hw_events; + struct list_head events_configurable; + struct list_head events_fixed; + DECLARE_BITMAP(used_counter_bm, CPMU_MAX_COUNTERS); + DECLARE_BITMAP(conf_counter_bm, CPMU_MAX_COUNTERS); + u16 counter_width; + u8 num_counters; + u8 num_event_capabilities; + int on_cpu; + struct hlist_node node; + bool freeze_for_enable; + bool filter_hdm; + int irq; +}; + +#define pmu_to_cpmu_info(_pmu) container_of(_pmu, struct cpmu_info, pmu) + +/* + * All CPMU counters are discoverable via the Event Capabilities Registers. + * Each Event Capability register contains a a VID / GroupID. + * A counter may then count any combination (by summing) of events in + * that group which are in the Supported Events Bitmask. + * However, there are some complexities to the scheme. + * - Fixed function counters refer to an Event Capabilities register. + * That event capability register is not then used for Configurable + * counters. + */ +static int cpmu_parse_caps(struct device *dev, struct cpmu_info *info) +{ + DECLARE_BITMAP(fixed_counter_event_cap_bm, 32) = {0}; + void __iomem *base = info->base; + u64 val, eval; + int i; + + val = readq(base + CPMU_CAP_REG); + info->freeze_for_enable = FIELD_GET(CPMU_CAP_WRITEABLE_WHEN_FROZEN, val) & + FIELD_GET(CPMU_CAP_FREEZE, val); + if (!info->freeze_for_enable) { + dev_err(dev, "Driver does not support CPMUs that do not support freeze for enable\n"); + return -ENODEV; + } + + info->num_counters = FIELD_GET(CPMU_CAP_NUM_COUNTERS_MSK, val) + 1; + info->counter_width = FIELD_GET(CPMU_CAP_COUNTER_WIDTH_MSK, val); + info->num_event_capabilities = FIELD_GET(CPMU_CAP_NUM_EVN_CAP_REG_SUP_MSK, val) + 1; + + info->filter_hdm = FIELD_GET(CPMU_CAP_FILTERS_SUP_MSK, val) & CPMU_FILTER_HDM; + if (FIELD_GET(CPMU_CAP_INT, val)) + info->irq = FIELD_GET(CPMU_CAP_MSI_N_MSK, val); + else + info->irq = -1; + + /* First handle fixed function counters; note if configurable counters found */ + for (i = 0; i < info->num_counters; i++) { + struct cpmu_event *cpmu_ev; + u32 events_msk; + u8 group_idx; + + val = readq(base + CPMU_COUNTER_CFG_REG(i)); + + if (FIELD_GET(CPMU_COUNTER_CFG_TYPE_MSK, val) == + CPMU_COUNTER_CFG_TYPE_CONFIGURABLE) { + set_bit(i, info->conf_counter_bm); + } + + if (FIELD_GET(CPMU_COUNTER_CFG_TYPE_MSK, val) != + CPMU_COUNTER_CFG_TYPE_FIXED_FUN) + continue; + + /* In this case we know which fields are const */ + group_idx = FIELD_GET(CPMU_COUNTER_CFG_EVENT_GRP_ID_IDX_MSK, val); + events_msk = FIELD_GET(CPMU_COUNTER_CFG_EVENTS_MSK, val); + eval = readq(base + CPMU_EVENT_CAP_REG(group_idx)); + cpmu_ev = devm_kzalloc(dev, sizeof(*cpmu_ev), GFP_KERNEL); + if (!cpmu_ev) + return -ENOMEM; + + cpmu_ev->vid = FIELD_GET(CPMU_EVENT_CAP_VENDOR_ID_MSK, eval); + cpmu_ev->gid = FIELD_GET(CPMU_EVENT_CAP_GROUP_ID_MSK, eval); + /* For a fixed purpose counter use the events mask from the counter CFG */ + cpmu_ev->msk = events_msk; + cpmu_ev->counter_idx = i; + /* This list add is never unwound as all entries deleted on remove */ + list_add(&cpmu_ev->node, &info->events_fixed); + /* + * Configurable counters must not use an Event Capability registers that + * is in use for a Fixed counter + */ + set_bit(group_idx, fixed_counter_event_cap_bm); + } + + if (!bitmap_empty(info->conf_counter_bm, CPMU_MAX_COUNTERS)) { + struct cpmu_event *cpmu_ev; + int j; + /* Walk event capabilities unused by fixed counters */ + for_each_clear_bit(j, fixed_counter_event_cap_bm, + info->num_event_capabilities) { + cpmu_ev = devm_kzalloc(dev, sizeof(*cpmu_ev), GFP_KERNEL); + if (!cpmu_ev) + return -ENOMEM; + + eval = readq(base + CPMU_EVENT_CAP_REG(j)); + cpmu_ev->vid = FIELD_GET(CPMU_EVENT_CAP_VENDOR_ID_MSK, eval); + cpmu_ev->gid = FIELD_GET(CPMU_EVENT_CAP_GROUP_ID_MSK, eval); + cpmu_ev->msk = FIELD_GET(CPMU_EVENT_CAP_SUPPORTED_EVENTS_MSK, eval); + cpmu_ev->event_idx = j; + list_add(&cpmu_ev->node, &info->events_configurable); + } + } + + return 0; +} + +static ssize_t cpmu_event_sysfs_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct perf_pmu_events_attr *pmu_attr = + container_of(attr, struct perf_pmu_events_attr, attr); + + return sysfs_emit(buf, "config=%#llx\n", pmu_attr->id); +} + +#define CPMU_PMU_EVENT_ATTR(_name, _vid, _gid, _msk) \ + PMU_EVENT_ATTR_ID(_name, cpmu_event_sysfs_show, \ + ((u64)(_vid) << 48) | ((u64)(_gid) << 32) | (u64)(_msk)) + +/* For CXL spec defined events */ +#define CPMU_PMU_EVENT_CXL_ATTR(_name, _gid, _msk) \ + CPMU_PMU_EVENT_ATTR(_name, PCI_DVSEC_VENDOR_ID_CXL, _gid, _msk) + +static struct attribute *cpmu_event_attrs[] = { + CPMU_PMU_EVENT_CXL_ATTR(clock_ticks, CPMU_GID_CLOCK_TICKS, BIT(0)), + /* CXL rev 3.0 Table 3-17 - Device to Host Requests */ + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_rdcurr, CPMU_GID_D2H_REQ, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_rdown, CPMU_GID_D2H_REQ, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_rdshared, CPMU_GID_D2H_REQ, BIT(3)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_rdany, CPMU_GID_D2H_REQ, BIT(4)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_rdownnodata, CPMU_GID_D2H_REQ, BIT(5)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_itomwr, CPMU_GID_D2H_REQ, BIT(6)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_wrcurr, CPMU_GID_D2H_REQ, BIT(7)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_clflush, CPMU_GID_D2H_REQ, BIT(8)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_cleanevict, CPMU_GID_D2H_REQ, BIT(9)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_dirtyevict, CPMU_GID_D2H_REQ, BIT(10)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_cleanevictnodata, CPMU_GID_D2H_REQ, BIT(11)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_wowrinv, CPMU_GID_D2H_REQ, BIT(12)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_wowrinvf, CPMU_GID_D2H_REQ, BIT(13)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_wrinv, CPMU_GID_D2H_REQ, BIT(14)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_req_cacheflushed, CPMU_GID_D2H_REQ, BIT(16)), + /* CXL rev 3.0 Table 3-20 - D2H Repsonse Encodings */ + CPMU_PMU_EVENT_CXL_ATTR(d2h_rsp_rspihiti, CPMU_GID_D2H_RSP, BIT(4)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_rsp_rspvhitv, CPMU_GID_D2H_RSP, BIT(6)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_rsp_rspihitse, CPMU_GID_D2H_RSP, BIT(5)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_rsp_rspshitse, CPMU_GID_D2H_RSP, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_rsp_rspsfwdm, CPMU_GID_D2H_RSP, BIT(7)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_rsp_rspifwdm, CPMU_GID_D2H_RSP, BIT(15)), + CPMU_PMU_EVENT_CXL_ATTR(d2h_rsp_rspvfwdv, CPMU_GID_D2H_RSP, BIT(22)), + /* CXL rev 3.0 Table 3-21 - CXL.cache - Mapping of H2D Requests to D2H Responses */ + CPMU_PMU_EVENT_CXL_ATTR(h2d_req_snpdata, CPMU_GID_H2D_REQ, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_req_snpinv, CPMU_GID_H2D_REQ, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_req_snpcur, CPMU_GID_H2D_REQ, BIT(3)), + /* CXL rev 3.0 Table 3-22 - H2D Response Opcode Encodings */ + CPMU_PMU_EVENT_CXL_ATTR(h2d_rsp_writepull, CPMU_GID_H2D_RSP, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_rsp_go, CPMU_GID_H2D_RSP, BIT(4)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_rsp_gowritepull, CPMU_GID_H2D_RSP, BIT(5)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_rsp_extcmp, CPMU_GID_H2D_RSP, BIT(6)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_rsp_gowritepulldrop, CPMU_GID_H2D_RSP, BIT(8)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_rsp_fastgowritepull, CPMU_GID_H2D_RSP, BIT(13)), + CPMU_PMU_EVENT_CXL_ATTR(h2d_rsp_goerrwritepull, CPMU_GID_H2D_RSP, BIT(15)), + /* CXL rev 3.0 Table 13-5 directly lists these */ + CPMU_PMU_EVENT_CXL_ATTR(cachedata_d2h_data, CPMU_GID_CACHE_DATA, BIT(0)), + CPMU_PMU_EVENT_CXL_ATTR(cachedata_h2d_data, CPMU_GID_CACHE_DATA, BIT(1)), + /* CXL rev 3.0 Table 3-29 M2S Req Memory Opcodes */ + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_meminv, CPMU_GID_M2S_REQ, BIT(0)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_memrd, CPMU_GID_M2S_REQ, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_memrddata, CPMU_GID_M2S_REQ, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_memrdfwd, CPMU_GID_M2S_REQ, BIT(3)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_memwrfwd, CPMU_GID_M2S_REQ, BIT(4)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_memspecrd, CPMU_GID_M2S_REQ, BIT(8)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_meminvnt, CPMU_GID_M2S_REQ, BIT(9)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_req_memcleanevict, CPMU_GID_M2S_REQ, BIT(10)), + /* CXL rev 3.0 Table 3-35 M2S RwD Memory Opcodes */ + CPMU_PMU_EVENT_CXL_ATTR(m2s_rwd_memwr, CPMU_GID_M2S_RWD, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_rwd_memwrptl, CPMU_GID_M2S_RWD, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_rwd_biconflict, CPMU_GID_M2S_RWD, BIT(4)), + /* CXL rev 3.0 Table 3-38 M2S BIRsp Memory Opcodes */ + CPMU_PMU_EVENT_CXL_ATTR(m2s_birsp_i, CPMU_GID_M2S_BIRSP, BIT(0)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_birsp_s, CPMU_GID_M2S_BIRSP, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_birsp_e, CPMU_GID_M2S_BIRSP, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_birsp_iblk, CPMU_GID_M2S_BIRSP, BIT(4)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_birsp_sblk, CPMU_GID_M2S_BIRSP, BIT(5)), + CPMU_PMU_EVENT_CXL_ATTR(m2s_birsp_eblk, CPMU_GID_M2S_BIRSP, BIT(6)), + /* CXL rev 3.0 Table 3-40 S2M BISnp Opcodes */ + CPMU_PMU_EVENT_CXL_ATTR(s2m_bisnp_cur, CPMU_GID_S2M_BISNP, BIT(0)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_bisnp_data, CPMU_GID_S2M_BISNP, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_bisnp_inv, CPMU_GID_S2M_BISNP, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_bisnp_curblk, CPMU_GID_S2M_BISNP, BIT(4)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_bisnp_datblk, CPMU_GID_S2M_BISNP, BIT(5)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_bisnp_invblk, CPMU_GID_S2M_BISNP, BIT(6)), + /* CXL rev 3.0 Table 3-43 S2M NDR Opcopdes */ + CPMU_PMU_EVENT_CXL_ATTR(s2m_ndr_cmp, CPMU_GID_S2M_NDR, BIT(0)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_ndr_cmps, CPMU_GID_S2M_NDR, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_ndr_cmpe, CPMU_GID_S2M_NDR, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_ndr_biconflictack, CPMU_GID_S2M_NDR, BIT(3)), + /* CXL rev 3.0 Table 3-46 S2M DRS opcodes */ + CPMU_PMU_EVENT_CXL_ATTR(s2m_drs_memdata, CPMU_GID_S2M_DRS, BIT(0)), + CPMU_PMU_EVENT_CXL_ATTR(s2m_drs_memdatanxm, CPMU_GID_S2M_DRS, BIT(1)), + /* CXL rev 3.0 Table 13-5 directly lists these */ + CPMU_PMU_EVENT_CXL_ATTR(ddr_act, CPMU_GID_DDR, BIT(0)), + CPMU_PMU_EVENT_CXL_ATTR(ddr_pre, CPMU_GID_DDR, BIT(1)), + CPMU_PMU_EVENT_CXL_ATTR(ddr_casrd, CPMU_GID_DDR, BIT(2)), + CPMU_PMU_EVENT_CXL_ATTR(ddr_caswr, CPMU_GID_DDR, BIT(3)), + CPMU_PMU_EVENT_CXL_ATTR(ddr_refresh, CPMU_GID_DDR, BIT(4)), + CPMU_PMU_EVENT_CXL_ATTR(ddr_selfrefreshent, CPMU_GID_DDR, BIT(5)), + CPMU_PMU_EVENT_CXL_ATTR(ddr_rfm, CPMU_GID_DDR, BIT(6)), + NULL +}; + +static struct cpmu_event *cpmu_find_fixed_counter_event(struct cpmu_info *info, + int vid, int gid, int msk) +{ + struct cpmu_event *cpmu_ev; + + list_for_each_entry(cpmu_ev, &info->events_fixed, node) { + if (vid != cpmu_ev->vid || gid != cpmu_ev->gid) + continue; + + /* Precise match for fixed counter */ + if (msk == cpmu_ev->msk) + return cpmu_ev; + } + + return ERR_PTR(-EINVAL); +} + +static struct cpmu_event *cpmu_find_config_counter_event(struct cpmu_info *info, + int vid, int gid, int msk) +{ + struct cpmu_event *cpmu_ev; + + list_for_each_entry(cpmu_ev, &info->events_configurable, node) { + if (vid != cpmu_ev->vid || gid != cpmu_ev->gid) + continue; + + /* Request mask must be subset of supported */ + if (msk & ~cpmu_ev->msk) + continue; + + return cpmu_ev; + } + + return ERR_PTR(-EINVAL); +} + +static umode_t cpmu_event_is_visible(struct kobject *kobj, struct attribute *attr, int a) +{ + struct device_attribute *dev_attr = container_of(attr, struct device_attribute, attr); + struct perf_pmu_events_attr *pmu_attr = + container_of(dev_attr, struct perf_pmu_events_attr, attr); + struct device *dev = kobj_to_dev(kobj); + struct cpmu_info *info = dev_get_drvdata(dev); + int vid = FIELD_GET(GENMASK_ULL(63, 48), pmu_attr->id); + int gid = FIELD_GET(GENMASK_ULL(47, 32), pmu_attr->id); + int msk = FIELD_GET(GENMASK_ULL(31, 0), pmu_attr->id); + + if (!IS_ERR(cpmu_find_fixed_counter_event(info, vid, gid, msk))) + return attr->mode; + + if (!IS_ERR(cpmu_find_config_counter_event(info, vid, gid, msk))) + return attr->mode; + + return 0; +} + +static const struct attribute_group cpmu_events = { + .name = "events", + .attrs = cpmu_event_attrs, + .is_visible = cpmu_event_is_visible, +}; + +static ssize_t cpmu_format_sysfs_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct dev_ext_attribute *eattr; + + eattr = container_of(attr, struct dev_ext_attribute, attr); + + return sysfs_emit(buf, "%s\n", (char *)eattr->var); +} + +#define CPMU_FORMAT_ATTR(_name, _format)\ + (&((struct dev_ext_attribute[]) { \ + { \ + .attr = __ATTR(_name, 0444, \ + cpmu_format_sysfs_show, NULL), \ + .var = (void *)_format \ + } \ + })[0].attr.attr) + +enum { + cpmu_mask_attr, + cpmu_gid_attr, + cpmu_vid_attr, + cpmu_threshold_attr, + cpmu_invert_attr, + cpmu_edge_attr, + cpmu_hdm_filter_en_attr, + cpmu_hdm_attr, +}; + +static struct attribute *cpmu_format_attr[] = { + [cpmu_mask_attr] = CPMU_FORMAT_ATTR(mask, "config:0-31"), + [cpmu_gid_attr] = CPMU_FORMAT_ATTR(gid, "config:32-47"), + [cpmu_vid_attr] = CPMU_FORMAT_ATTR(vid, "config:48-63"), + [cpmu_threshold_attr] = CPMU_FORMAT_ATTR(threshold, "config1:0-15"), + [cpmu_invert_attr] = CPMU_FORMAT_ATTR(invert, "config1:16"), + [cpmu_edge_attr] = CPMU_FORMAT_ATTR(edge, "config1:17"), + [cpmu_hdm_filter_en_attr] = CPMU_FORMAT_ATTR(hdm_filter_en, "config1:18"), + [cpmu_hdm_attr] = CPMU_FORMAT_ATTR(hdm, "config2:0-15"), + NULL +}; + +static umode_t cpmu_format_is_visible(struct kobject *kobj, struct attribute *attr, int a) +{ + struct device *dev = kobj_to_dev(kobj); + struct cpmu_info *info = dev_get_drvdata(dev); + + /* + * Filter capability at the CPMU level, so hide the attributes if the particular + * filter is not supported. + */ + if (attr == cpmu_format_attr[cpmu_hdm_filter_en_attr] || + attr == cpmu_format_attr[cpmu_hdm_attr]) { + if (info->filter_hdm) + return 0444; + else + return 0; + } else { + return 0444; + } +} + +static const struct attribute_group cpmu_format_group = { + .name = "format", + .attrs = cpmu_format_attr, + .is_visible = cpmu_format_is_visible, +}; + +static u32 cpmu_config_get_mask(struct perf_event *event) +{ + return FIELD_GET(GENMASK_ULL(31, 0), event->attr.config); +} + +static u16 cpmu_config_get_gid(struct perf_event *event) +{ + return FIELD_GET(GENMASK_ULL(47, 32), event->attr.config); +} + +static u16 cpmu_config_get_vid(struct perf_event *event) +{ + return FIELD_GET(GENMASK_ULL(63, 48), event->attr.config); +} + +static u8 cpmu_config1_get_threshold(struct perf_event *event) +{ + return FIELD_GET(GENMASK_ULL(15, 0), event->attr.config1); +} + +static bool cpmu_config1_get_invert(struct perf_event *event) +{ + return FIELD_GET(BIT(16), event->attr.config1); +} + +static bool cpmu_config1_get_edge(struct perf_event *event) +{ + return FIELD_GET(BIT(17), event->attr.config1); +} + +/* + * CPMU specification allows for 8 filters, each with a 16 bit value... + * So we need to find 8x16bits to store it in. + * As the value used for disable is 0xffff, a separate enable switch + * is needed. + */ + +static bool cpmu_config1_hdm_filter_en(struct perf_event *event) +{ + return FIELD_GET(BIT(14), event->attr.config1); +} + +static u16 cpmu_config2_get_hdm_decoder(struct perf_event *event) +{ + return FIELD_GET(GENMASK(15, 0), event->attr.config2); +} + +static ssize_t cpumask_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct cpmu_info *info = dev_get_drvdata(dev); + + return cpumap_print_to_pagebuf(true, buf, cpumask_of(info->on_cpu)); +} +static DEVICE_ATTR_RO(cpumask); + +static struct attribute *cpmu_cpumask_attrs[] = { + &dev_attr_cpumask.attr, + NULL +}; + +static const struct attribute_group cpmu_cpumask_group = { + .attrs = cpmu_cpumask_attrs, +}; + +static const struct attribute_group *cpmu_attr_groups[] = { + &cpmu_events, + &cpmu_format_group, + &cpmu_cpumask_group, + NULL +}; + +/* If counter_idx == NULL, don't try to allocate a counter. */ +static int cpmu_get_event_idx(struct perf_event *event, int *counter_idx, int *event_idx) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + DECLARE_BITMAP(configurable_and_free, CPMU_MAX_COUNTERS); + struct cpmu_event *cpmu_ev; + u32 mask; + u16 gid, vid; + int i; + + vid = cpmu_config_get_vid(event); + gid = cpmu_config_get_gid(event); + mask = cpmu_config_get_mask(event); + + cpmu_ev = cpmu_find_fixed_counter_event(info, vid, gid, mask); + if (!IS_ERR(cpmu_ev)) { + if (!counter_idx) + return 0; + if (!info->hw_events[cpmu_ev->counter_idx]) { + *counter_idx = cpmu_ev->counter_idx; + return 0; + } + /* Fixed counter is in use, but maybe a configurable one? */ + } + + cpmu_ev = cpmu_find_config_counter_event(info, vid, gid, mask); + if (!IS_ERR(cpmu_ev)) { + if (!counter_idx) + return 0; + + bitmap_andnot(configurable_and_free, info->conf_counter_bm, + info->used_counter_bm, CPMU_MAX_COUNTERS); + + i = find_first_bit(configurable_and_free, CPMU_MAX_COUNTERS); + if (i == CPMU_MAX_COUNTERS) + return -EINVAL; + + *counter_idx = i; + return 0; + } + + return -EINVAL; +} + +static int cpmu_event_init(struct perf_event *event) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + + event->cpu = info->on_cpu; + /* Top level type sanity check - is this a Hardware Event being requested */ + if (event->attr.type != event->pmu->type) + return -ENOENT; + + if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK) + return -EOPNOTSUPP; + /* TODO: Validation of any filter */ + + /* + * Verify that it is possible to count what was requested. Either must + * be a fixed counter that is a precise match or a configurable counter + * where this is a subset. + */ + return cpmu_get_event_idx(event, NULL, NULL); +} + +static void cpmu_pmu_enable(struct pmu *pmu) +{ + struct cpmu_info *info = pmu_to_cpmu_info(pmu); + void __iomem *base = info->base; + + /* We don't have a global enable, but we 'might' have a global freeze which we can use */ + if (info->freeze_for_enable) { + /* Can assume frozen at this stage */ + writeq(0, base + CPMU_FREEZE_REG); + + return; + } +} + +static void cpmu_pmu_disable(struct pmu *pmu) +{ + struct cpmu_info *info = pmu_to_cpmu_info(pmu); + void __iomem *base = info->base; + + if (info->freeze_for_enable) { + /* + * Whilst bits above number of counters are RsvdZ + * they are unlikely to be repurposed given + * number of counters is allowed to be 64 leaving + * no reserved bits. Hence this is only slightly + * naughty. + */ + writeq(GENMASK(63, 0), base + CPMU_FREEZE_REG); + return; + } +} + +static void cpmu_event_start(struct perf_event *event, int flags) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + struct hw_perf_event *hwc = &event->hw; + void __iomem *base = info->base; + u64 cfg, prev_cnt; + + if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED))) + return; + + WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE)); + hwc->state = 0; + + /* + * Currently only hdm filter control is implemnted, this code will + * want generalizing when more filters are added. + */ + if (info->filter_hdm) { + if (cpmu_config1_hdm_filter_en(event)) + cfg = cpmu_config2_get_hdm_decoder(event); + else + cfg = GENMASK(15, 0); + writeq(cfg, base + CPMU_FILTER_CFG_REG(hwc->idx, 0)); + } + + cfg = readq(base + CPMU_COUNTER_CFG_REG(hwc->idx)); + cfg |= FIELD_PREP(CPMU_COUNTER_CFG_INT_ON_OVRFLW, 1); + cfg |= FIELD_PREP(CPMU_COUNTER_CFG_ENABLE, 1); + cfg |= FIELD_PREP(CPMU_COUNTER_CFG_EDGE, cpmu_config1_get_edge(event) ? 1 : 0); + cfg |= FIELD_PREP(CPMU_COUNTER_CFG_INVERT, cpmu_config1_get_invert(event) ? 1 : 0); + + /* Fixed purpose counters have next two fields RO */ + if (test_bit(hwc->idx, info->conf_counter_bm)) { + cfg |= FIELD_PREP(CPMU_COUNTER_CFG_EVENT_GRP_ID_IDX_MSK, hwc->event_base); + cfg |= FIELD_PREP(CPMU_COUNTER_CFG_EVENTS_MSK, cpmu_config_get_mask(event)); + } + cfg &= ~CPMU_COUNTER_CFG_THRESHOLD_MSK; + /* + * For events that generate only 1 count per clock the CXL 3.0 spec + * states the threshold shall be set to 1 but if set to 0 it will + * count the raw value anwyay? + * There is no definition of what events will count multiple per cycle + * and hence to which non 1 values of threshold can apply. + * (CXL 3.0 8.2.7.2.1 Counter Configuration - threshold field definition) + */ + cfg |= FIELD_PREP(CPMU_COUNTER_CFG_THRESHOLD_MSK, + cpmu_config1_get_threshold(event)); + writeq(cfg, base + CPMU_COUNTER_CFG_REG(hwc->idx)); + + local64_set(&hwc->prev_count, 0); + writeq(0, base + CPMU_COUNTER_REG(hwc->idx)); + + if (flags & PERF_EF_RELOAD) { + prev_cnt = local64_read(&hwc->prev_count); + writeq(prev_cnt, base + CPMU_COUNTER_REG(hwc->idx)); + } + + perf_event_update_userpage(event); +} + +static u64 cpmu_read_counter(struct perf_event *event) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + void __iomem *base = info->base; + + return readq(base + CPMU_COUNTER_REG(event->hw.idx)); +} + +static void __cpmu_read(struct perf_event *event, bool overflow) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + struct hw_perf_event *hwc = &event->hw; + u64 new_cnt, prev_cnt, delta; + + do { + prev_cnt = local64_read(&hwc->prev_count); + new_cnt = cpmu_read_counter(event); + } while (local64_cmpxchg(&hwc->prev_count, prev_cnt, new_cnt) != prev_cnt); + + /* + * If we know an overflow occur then take that into account. + * Note counter is not reset as that would lose events + */ + delta = (new_cnt - prev_cnt) & GENMASK(info->counter_width - 1, 0); + if (overflow && delta < GENMASK(info->counter_width - 1, 0)) + delta += (1UL << info->counter_width); + + local64_add(delta, &event->count); +} + +static void cpmu_read(struct perf_event *event) +{ + __cpmu_read(event, false); +} + +static void cpmu_event_stop(struct perf_event *event, int flags) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + void __iomem *base = info->base; + struct hw_perf_event *hwc = &event->hw; + u64 cfg; + + cpmu_read(event); + WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED); + hwc->state |= PERF_HES_STOPPED; + + cfg = readq(base + CPMU_COUNTER_CFG_REG(hwc->idx)); + cfg &= ~(FIELD_PREP(CPMU_COUNTER_CFG_INT_ON_OVRFLW, 1) | + FIELD_PREP(CPMU_COUNTER_CFG_ENABLE, 1)); + writeq(cfg, base + CPMU_COUNTER_CFG_REG(hwc->idx)); + + if (hwc->state & PERF_HES_UPTODATE) + return; + + hwc->state |= PERF_HES_UPTODATE; +} + +/* + * Reset ensures no possibility of any information leaking to wrong + * counter. Note that all fields written during start() + */ +static void cpmu_reset_counter(struct cpmu_info *info, int idx) +{ + void __iomem *base = info->base; + + /* Much of this register is read only */ + writeq(0, base + CPMU_EVENT_CAP_REG(idx)); + /* Filters are not per counter, so no reset here */ + writeq(0, base + CPMU_COUNTER_REG(idx)); +} + +static int cpmu_event_add(struct perf_event *event, int flags) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + struct hw_perf_event *hwc = &event->hw; + int idx, rc; + int event_idx = 0; + + hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE; + + rc = cpmu_get_event_idx(event, &idx, &event_idx); + if (rc < 0) + return rc; + + hwc->idx = idx; + + /* Only set for configurable counters */ + hwc->event_base = event_idx; + info->hw_events[idx] = event; + set_bit(idx, info->used_counter_bm); + + cpmu_reset_counter(info, idx); + + if (flags & PERF_EF_START) + cpmu_event_start(event, PERF_EF_RELOAD); + + return 0; +} + +static void cpmu_event_del(struct perf_event *event, int flags) +{ + struct cpmu_info *info = pmu_to_cpmu_info(event->pmu); + struct hw_perf_event *hwc = &event->hw; + + cpmu_event_stop(event, PERF_EF_UPDATE); + clear_bit(hwc->idx, info->used_counter_bm); + info->hw_events[hwc->idx] = NULL; + perf_event_update_userpage(event); +} + +static irqreturn_t cpmu_irq(int irq, void *data) +{ + struct cpmu_info *info = data; + void __iomem *base = info->base; + u64 overflowed; + DECLARE_BITMAP(overflowedbm, 64); + int i; + + overflowed = readq(base + CPMU_OVERFLOW_REG); + + /* Interrupt may be shared, so maybe it isn't ours */ + if (!overflowed) + return IRQ_NONE; + + bitmap_from_arr64(overflowedbm, &overflowed, 64); + for_each_set_bit(i, overflowedbm, info->num_counters) { + struct perf_event *event = info->hw_events[i]; + + if (!event) { + dev_dbg(info->pmu.dev, + "overflow but on non enabled counter %d\n", i); + continue; + } + + __cpmu_read(event, true); + } + + writeq(overflowed, base + CPMU_OVERFLOW_REG); + + return IRQ_HANDLED; +} + +static int cxl_cpmu_probe(struct device *dev) +{ + struct cxl_cpmu *cpmu = to_cxl_cpmu(dev); + struct pci_dev *pdev = to_pci_dev(dev->parent); + struct cpmu_info *info; + char *irq_name; + int rc, irq; + + info = devm_kzalloc(dev, sizeof(*info), GFP_KERNEL); + if (!info) + return -ENOMEM; + + INIT_LIST_HEAD(&info->events_fixed); + INIT_LIST_HEAD(&info->events_configurable); + + info->base = cpmu->base; + + info->on_cpu = -1; + rc = cpmu_parse_caps(dev, info); + if (rc) + return rc; + + info->hw_events = devm_kcalloc(dev, sizeof(*info->hw_events), + info->num_counters, GFP_KERNEL); + if (!info->hw_events) + return -ENOMEM; + + info->pmu = (struct pmu) { + .name = dev_name(dev), + .parent = dev, + .module = THIS_MODULE, + .event_init = cpmu_event_init, + .pmu_enable = cpmu_pmu_enable, + .pmu_disable = cpmu_pmu_disable, + .add = cpmu_event_add, + .del = cpmu_event_del, + .start = cpmu_event_start, + .stop = cpmu_event_stop, + .read = cpmu_read, + .task_ctx_nr = perf_invalid_context, + .attr_groups = cpmu_attr_groups, + .capabilities = PERF_PMU_CAP_NO_EXCLUDE, + }; + + if (info->irq <= 0) + return -EINVAL; + + rc = pci_irq_vector(pdev, info->irq); + if (rc < 0) + return rc; + irq = rc; + + irq_name = devm_kasprintf(dev, GFP_KERNEL, "%s_overflow\n", dev_name(dev)); + if (!irq_name) + return -ENOMEM; + + rc = devm_request_irq(dev, irq, cpmu_irq, IRQF_SHARED, irq_name, info); + if (rc) + return rc; + info->irq = irq; + + rc = cpuhp_state_add_instance(cpmu_cpuhp_state_num, &info->node); + if (rc) + return rc; + + rc = perf_pmu_register(&info->pmu, info->pmu.name, -1); + if (rc) + return rc; + + dev_set_drvdata(dev, info); + + return 0; +} + +static void cxl_cpmu_remove(struct device *dev) +{ + struct cpmu_info *info = dev_get_drvdata(dev); + + perf_pmu_unregister(&info->pmu); + cpuhp_state_remove_instance_nocalls(cpmu_cpuhp_state_num, &info->node); +} + +static struct cxl_driver cxl_cpmu_driver = { + .name = "cxl_cpmu", + .probe = cxl_cpmu_probe, + .remove = cxl_cpmu_remove, + .id = CXL_DEVICE_CPMU, +}; + +static int cpmu_online_cpu(unsigned int cpu, struct hlist_node *node) +{ + struct cpmu_info *info = hlist_entry_safe(node, struct cpmu_info, node); + + if (info->on_cpu != -1) + return 0; + + info->on_cpu = cpu; + WARN_ON(irq_set_affinity(info->irq, cpumask_of(cpu))); + + return 0; +} + +static int cpmu_offline_cpu(unsigned int cpu, struct hlist_node *node) +{ + struct cpmu_info *info = hlist_entry_safe(node, struct cpmu_info, node); + unsigned int target; + + if (info->on_cpu != cpu) + return 0; + + info->on_cpu = -1; + target = cpumask_first(cpu_online_mask); + if (target >= nr_cpu_ids) { + dev_err(info->pmu.dev, "Unable to find a suitable CPU\n"); + return 0; + } + + perf_pmu_migrate_context(&info->pmu, cpu, target); + info->on_cpu = target; + WARN_ON(irq_set_affinity(info->irq, cpumask_of(target))); + + return 0; +} + +static __init int cxl_cpmu_init(void) +{ + int rc; + + rc = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, + "AP_PERF_CPMU_ONLINE", + cpmu_online_cpu, cpmu_offline_cpu); + if (rc < 0) + return rc; + cpmu_cpuhp_state_num = rc; + + rc = cxl_driver_register(&cxl_cpmu_driver); + if (rc) + cpuhp_remove_multi_state(cpmu_cpuhp_state_num); + + return rc; +} + +static __exit void cxl_cpmu_exit(void) +{ + cxl_driver_unregister(&cxl_cpmu_driver); + cpuhp_remove_multi_state(cpmu_cpuhp_state_num); +} + +MODULE_LICENSE("GPL"); +MODULE_IMPORT_NS(CXL); +module_init(cxl_cpmu_init); +module_exit(cxl_cpmu_exit); +MODULE_ALIAS_CXL(CXL_DEVICE_CPMU); From patchwork Fri Mar 24 17:13:13 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jonathan Cameron X-Patchwork-Id: 74649 Return-Path: Delivered-To: ouuuleilei@gmail.com Received: by 2002:a59:b0ea:0:b0:3b6:4342:cba0 with SMTP id b10csp792640vqo; Fri, 24 Mar 2023 10:25:57 -0700 (PDT) X-Google-Smtp-Source: AKy350YzZp/ZIGiHk5/0LG+tNvy+FDLR48xEjORIZ9QIruqGNkUCoPvuFZdOwL4PMA7gyDlL6BFd X-Received: by 2002:a62:1c0a:0:b0:627:eece:5ca5 with SMTP id c10-20020a621c0a000000b00627eece5ca5mr3732939pfc.18.1679678756993; Fri, 24 Mar 2023 10:25:56 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1679678756; cv=none; d=google.com; s=arc-20160816; b=xZuezMCjraAMeVqDwEroYksz7PBvaTreG6F2ktKP/htdcMBMR5vsr0vWViUB9bKALE rtx2pvKvHiOUELCIUaizEwk1O9ESqI7PujvsBz6hR2ot0MjV1MFxyfCeyCYts1fg6qTI DKLwr49jY8oHG2pT3G9kp+dAhEmBoN/AcbyhATOtGWwmroXXcYdBs4ufnoaYRW44Jjny /rV5Ec8dzraNRabDjuAXKZSJ5iNIkrrTujIFubFnIQnhCYck2HDHWO8l0OGkFPI2wz7s iVwAila6z7pQ5ScSrtase55Z2Q3X92V+FSGaVHsVZrctzX4dbr1xqTOHwwB8qB8IryVz 8Jug== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:content-transfer-encoding:mime-version :references:in-reply-to:message-id:date:subject:cc:to:from; bh=D3Ssi1QFP6wbDkdRlH6j2rw9x14u+y8fwZ6vuNRVuKI=; b=0xmGorw+q/BS4AAU4wOFAYBYQDNfqTeiGi1P2B6B2Bvnd9ghdkCebSOBeWhQLJWuAX wNf0OWJkaSoOohgZ+qVcFTGFPheLLQNx+EPHWxLxZVX5SgCTzbfJ9YKkOIMU3Yi55es6 AmJBsNoeYIr6zk5tJ61lRduiGYV9KzLeCnoYHwcvmZDyRdZ4/qgG6zmV2ZAD0CjBL46T /A8G8FskcALa0b2aJiyE27iVMF1u92QPIfgER7Mr46cqTchZOqQIVzDq1uUUSTvJ0PUz 1nedyFpr0UlbqrFZRYQd/Xs8VZTb99amowTNiMr1NZhI9Xh5lJU3Sq5TG03U+Q+6LihG +PMg== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: from out1.vger.email (out1.vger.email. [2620:137:e000::1:20]) by mx.google.com with ESMTP id r17-20020a056a00217100b0060102d22785si20120113pff.283.2023.03.24.10.25.44; Fri, 24 Mar 2023 10:25:56 -0700 (PDT) Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20; Authentication-Results: mx.google.com; spf=pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=QUARANTINE sp=QUARANTINE dis=NONE) header.from=huawei.com Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232280AbjCXRQw (ORCPT + 99 others); Fri, 24 Mar 2023 13:16:52 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:54678 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232283AbjCXRQY (ORCPT ); Fri, 24 Mar 2023 13:16:24 -0400 Received: from frasgout.his.huawei.com (frasgout.his.huawei.com [185.176.79.56]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id E101D22C82; Fri, 24 Mar 2023 10:15:56 -0700 (PDT) Received: from lhrpeml500005.china.huawei.com (unknown [172.18.147.200]) by frasgout.his.huawei.com (SkyGuard) with ESMTP id 4Pjpj60V2Pz6J6vj; Sat, 25 Mar 2023 01:14:18 +0800 (CST) Received: from SecurePC-101-06.china.huawei.com (10.122.247.231) by lhrpeml500005.china.huawei.com (7.191.163.240) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.21; Fri, 24 Mar 2023 17:15:40 +0000 From: Jonathan Cameron To: Liang Kan , , CC: , , , , , , , , Davidlohr Bueso , Dave Jiang Subject: [PATCH v2 5/5] docs: perf: Minimal introduction the the CXL PMU device and driver. Date: Fri, 24 Mar 2023 17:13:13 +0000 Message-ID: <20230324171313.18448-6-Jonathan.Cameron@huawei.com> X-Mailer: git-send-email 2.37.2 In-Reply-To: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> References: <20230324171313.18448-1-Jonathan.Cameron@huawei.com> MIME-Version: 1.0 X-Originating-IP: [10.122.247.231] X-ClientProxiedBy: lhrpeml500002.china.huawei.com (7.191.160.78) To lhrpeml500005.china.huawei.com (7.191.163.240) X-CFilter-Loop: Reflected X-Spam-Status: No, score=-2.3 required=5.0 tests=RCVD_IN_DNSWL_MED, RCVD_IN_MSPIKE_H2,SPF_HELO_NONE,SPF_PASS autolearn=unavailable autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on lindbergh.monkeyblade.net Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org X-getmail-retrieved-from-mailbox: =?utf-8?q?INBOX?= X-GMAIL-THRID: =?utf-8?q?1761270832576242104?= X-GMAIL-MSGID: =?utf-8?q?1761270832576242104?= Very basic introduction to the device and the current driver support provided. I expect to expand on this in future versions of this patch set. Reviewed-by: Dave Jiang Signed-off-by: Jonathan Cameron --- v2: - Not renamed the device as now an earlier patch allows the pmu->device to have a parent thus making the association explicit. v1: - Add docs for how to use a Vendor Defined Counter. --- Documentation/admin-guide/perf/cxl.rst | 65 ++++++++++++++++++++++++ Documentation/admin-guide/perf/index.rst | 1 + 2 files changed, 66 insertions(+) diff --git a/Documentation/admin-guide/perf/cxl.rst b/Documentation/admin-guide/perf/cxl.rst new file mode 100644 index 000000000000..46235dff4b21 --- /dev/null +++ b/Documentation/admin-guide/perf/cxl.rst @@ -0,0 +1,65 @@ +.. SPDX-License-Identifier: GPL-2.0 + +====================================== +CXL Performance Monitoring Unit (CPMU) +====================================== + +The CXL rev 3.0 specification provides a definition of CXL Performance +Monitoring Unit in section 13.2: Performance Monitoring. + +CXL components (e.g. Root Port, Switch Upstream Port, End Point) may have +any number of CPMU instances. CPMU capabilities are fully discoverable from +the devices. The specification provides event definitions for all CXL protocol +message types and a set of additional events for things commonly counted on +CXL devices (e.g. DRAM events). + +CPMU driver +=========== + +The CPMU driver register a perf PMU with the name cpmu on the CXL bus. + + /sys/bus/cxl/device/cpmu + +The associated PMU is registered as + + /sys/bus/event_sources/devices/cpmu + +In common with other CXL bus devices, the id has no specific meaning and the +relationship to specific CXL device should be established via the device parent +of the device on the CXL bus. + +PMU driver provides description of available events and filter options in sysfs. + +The "format" directory describes all formats of the config (event vendor id, +group id and mask) config1 (threshold, filter enables) and config2 (filter +parameters) fields of the perf_event_attr structure. The "events" directory +describes all documented events show in perf list. + +The events shown in perf list are the most fine grained events with a single +bit of the event mask set. More general events may be enable by setting +multiple mask bits in config. For example, all Device to Host Read Requests +may be captured on a single counter by setting the bits for all of + +* d2h_req_rdcurr +* d2h_req_rdown +* d2h_req_rdshared +* d2h_req_rdany +* d2h_req_rdownnodata + +Example of usage:: + + $#perf list + cpmu0/clock_ticks/ [Kernel PMU event] + cpmu0/d2h_req_itomwr/ [Kernel PMU event] + cpmu0/d2h_req_rdany/ [Kernel PMU event] + cpmu0/d2h_req_rdcurr/ [Kernel PMU event] + ----------------------------------------------------------- + + $# perf stat -e cpmu0/clock_ticks/ -e cpmu0/d2h_req_itowrm/ + +Vendor specific events may also be available and if so can be used via + + $# perf stat -e cpmu0/vid=VID,gid=GID,mask=MASK/ + +The driver does not support sampling. So "perf record" and attaching to +a task are unsupported. diff --git a/Documentation/admin-guide/perf/index.rst b/Documentation/admin-guide/perf/index.rst index 9de64a40adab..f60be04e4e33 100644 --- a/Documentation/admin-guide/perf/index.rst +++ b/Documentation/admin-guide/perf/index.rst @@ -21,3 +21,4 @@ Performance monitor support alibaba_pmu nvidia-pmu meson-ddr-pmu + cxl