Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752192Ab2FKU1c (ORCPT ); Mon, 11 Jun 2012 16:27:32 -0400 Received: from acsinet15.oracle.com ([141.146.126.227]:27583 "EHLO acsinet15.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752040Ab2FKU1a (ORCPT ); Mon, 11 Jun 2012 16:27:30 -0400 Date: Mon, 11 Jun 2012 16:20:08 -0400 From: Konrad Rzeszutek Wilk To: "Liu, Jinsong" Cc: "'xen-devel@lists.xensource.com'" , "'linux-kernel@vger.kernel.org'" Subject: Re: [Xen-devel] [PATCH] Xen physical cpus online/offline sys interface Message-ID: <20120611202008.GA14435@phenom.dumpdata.com> References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.21 (2010-09-15) X-Source-IP: ucsinet22.oracle.com [156.151.31.94] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 14642 Lines: 505 On Mon, Jun 11, 2012 at 05:56:29AM +0000, Liu, Jinsong wrote: > Konrad, > > Our old patch of cpu online/offline is based on old kernel tree. > I rebase it to your latest xen.git > git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git branch devel/mce.v3 applied. > > Thanks, > Jinsong > > ===================== > >From 29152bb054d87334ef970ad037888b7bb032f8ef Mon Sep 17 00:00:00 2001 > From: Liu, Jinsong > Date: Mon, 11 Jun 2012 20:38:08 +0800 > Subject: [PATCH] Xen physical cpus online/offline sys interface > > This patch provide Xen physical cpus online/offline sys interface. > User can use it for their own purpose, like power saving: > by offlining some cpus when light workload it save power greatly. > > Its basic workflow is, user online/offline cpu via sys interface, > then hypercall xen to implement, after done xen inject virq back to dom0, > and then dom0 sync cpu status. > > Signed-off-by: Jiang, Yunhong > Signed-off-by: Liu, Jinsong > --- > .../ABI/testing/sysfs-devices-system-xen_cpu | 20 + > drivers/xen/Makefile | 1 + > drivers/xen/pcpu.c | 371 ++++++++++++++++++++ > include/xen/interface/platform.h | 8 + > include/xen/interface/xen.h | 1 + > 5 files changed, 401 insertions(+), 0 deletions(-) > create mode 100644 Documentation/ABI/testing/sysfs-devices-system-xen_cpu > create mode 100644 drivers/xen/pcpu.c > > diff --git a/Documentation/ABI/testing/sysfs-devices-system-xen_cpu b/Documentation/ABI/testing/sysfs-devices-system-xen_cpu > new file mode 100644 > index 0000000..9ca02fb > --- /dev/null > +++ b/Documentation/ABI/testing/sysfs-devices-system-xen_cpu > @@ -0,0 +1,20 @@ > +What: /sys/devices/system/xen_cpu/ > +Date: May 2012 > +Contact: Liu, Jinsong > +Description: > + A collection of global/individual Xen physical cpu attributes > + > + Individual physical cpu attributes are contained in > + subdirectories named by the Xen's logical cpu number, e.g.: > + /sys/devices/system/xen_cpu/xen_cpu#/ > + > + > +What: /sys/devices/system/xen_cpu/xen_cpu#/online > +Date: May 2012 > +Contact: Liu, Jinsong > +Description: > + Interface to online/offline Xen physical cpus > + > + When running under Xen platform, it provide user interface > + to online/offline physical cpus, except cpu0 due to several > + logic restrictions and assumptions. > diff --git a/drivers/xen/Makefile b/drivers/xen/Makefile > index a787029..d80bea5 100644 > --- a/drivers/xen/Makefile > +++ b/drivers/xen/Makefile > @@ -17,6 +17,7 @@ obj-$(CONFIG_XEN_SYS_HYPERVISOR) += sys-hypervisor.o > obj-$(CONFIG_XEN_PVHVM) += platform-pci.o > obj-$(CONFIG_XEN_TMEM) += tmem.o > obj-$(CONFIG_SWIOTLB_XEN) += swiotlb-xen.o > +obj-$(CONFIG_XEN_DOM0) += pcpu.o > obj-$(CONFIG_XEN_DOM0) += pci.o acpi.o > obj-$(CONFIG_XEN_MCE_LOG) += mcelog.o > obj-$(CONFIG_XEN_PCIDEV_BACKEND) += xen-pciback/ > diff --git a/drivers/xen/pcpu.c b/drivers/xen/pcpu.c > new file mode 100644 > index 0000000..067fcfa > --- /dev/null > +++ b/drivers/xen/pcpu.c > @@ -0,0 +1,371 @@ > +/****************************************************************************** > + * pcpu.c > + * Management physical cpu in dom0, get pcpu info and provide sys interface > + * > + * Copyright (c) 2012 Intel Corporation > + * Author: Liu, Jinsong > + * Author: Jiang, Yunhong > + * > + * This program is free software; you can redistribute it and/or > + * modify it under the terms of the GNU General Public License version 2 > + * as published by the Free Software Foundation; or, when distributed > + * separately from the Linux kernel or incorporated into other > + * software packages, subject to the following license: > + * > + * Permission is hereby granted, free of charge, to any person obtaining a copy > + * of this source file (the "Software"), to deal in the Software without > + * restriction, including without limitation the rights to use, copy, modify, > + * merge, publish, distribute, sublicense, and/or sell copies of the Software, > + * and to permit persons to whom the Software is furnished to do so, subject to > + * the following conditions: > + * > + * The above copyright notice and this permission notice shall be included in > + * all copies or substantial portions of the Software. > + * > + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR > + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, > + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE > + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER > + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING > + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS > + * IN THE SOFTWARE. > + */ > + > +#include > +#include > +#include > +#include > +#include > + > +#include > +#include > +#include > +#include > +#include > +#include > + > +#define XEN_PCPU "xen_cpu: " > + > +/* > + * @cpu_id: Xen physical cpu logic number > + * @flags: Xen physical cpu status flag > + * - XEN_PCPU_FLAGS_ONLINE: cpu is online > + * - XEN_PCPU_FLAGS_INVALID: cpu is not present > + */ > +struct pcpu { > + struct list_head list; > + struct device dev; > + uint32_t cpu_id; > + uint32_t flags; > +}; > + > +static struct bus_type xen_pcpu_subsys = { > + .name = "xen_cpu", > + .dev_name = "xen_cpu", > +}; > + > +static DEFINE_MUTEX(xen_pcpu_lock); > + > +static LIST_HEAD(xen_pcpus); > + > +static int xen_pcpu_down(uint32_t cpu_id) > +{ > + struct xen_platform_op op = { > + .cmd = XENPF_cpu_offline, > + .interface_version = XENPF_INTERFACE_VERSION, > + .u.cpu_ol.cpuid = cpu_id, > + }; > + > + return HYPERVISOR_dom0_op(&op); > +} > + > +static int xen_pcpu_up(uint32_t cpu_id) > +{ > + struct xen_platform_op op = { > + .cmd = XENPF_cpu_online, > + .interface_version = XENPF_INTERFACE_VERSION, > + .u.cpu_ol.cpuid = cpu_id, > + }; > + > + return HYPERVISOR_dom0_op(&op); > +} > + > +static ssize_t show_online(struct device *dev, > + struct device_attribute *attr, > + char *buf) > +{ > + struct pcpu *cpu = container_of(dev, struct pcpu, dev); > + > + return sprintf(buf, "%u\n", !!(cpu->flags & XEN_PCPU_FLAGS_ONLINE)); > +} > + > +static ssize_t __ref store_online(struct device *dev, > + struct device_attribute *attr, > + const char *buf, size_t count) > +{ > + struct pcpu *pcpu = container_of(dev, struct pcpu, dev); > + unsigned long long val; > + ssize_t ret; > + > + if (!capable(CAP_SYS_ADMIN)) > + return -EPERM; > + > + if (kstrtoull(buf, 0, &val) < 0) > + return -EINVAL; > + > + switch (val) { > + case 0: > + ret = xen_pcpu_down(pcpu->cpu_id); > + break; > + case 1: > + ret = xen_pcpu_up(pcpu->cpu_id); > + break; > + default: > + ret = -EINVAL; > + } > + > + if (ret >= 0) > + ret = count; > + return ret; > +} > +static DEVICE_ATTR(online, S_IRUGO | S_IWUSR, show_online, store_online); > + > +static bool xen_pcpu_online(uint32_t flags) > +{ > + return !!(flags & XEN_PCPU_FLAGS_ONLINE); > +} > + > +static void pcpu_online_status(struct xenpf_pcpuinfo *info, > + struct pcpu *pcpu) > +{ > + if (xen_pcpu_online(info->flags) && > + !xen_pcpu_online(pcpu->flags)) { > + /* the pcpu is onlined */ > + pcpu->flags |= XEN_PCPU_FLAGS_ONLINE; > + kobject_uevent(&pcpu->dev.kobj, KOBJ_ONLINE); > + } else if (!xen_pcpu_online(info->flags) && > + xen_pcpu_online(pcpu->flags)) { > + /* The pcpu is offlined */ > + pcpu->flags &= ~XEN_PCPU_FLAGS_ONLINE; > + kobject_uevent(&pcpu->dev.kobj, KOBJ_OFFLINE); > + } > +} > + > +static struct pcpu *get_pcpu(uint32_t cpu_id) > +{ > + struct pcpu *pcpu; > + > + list_for_each_entry(pcpu, &xen_pcpus, list) { > + if (pcpu->cpu_id == cpu_id) > + return pcpu; > + } > + > + return NULL; > +} > + > +static void pcpu_release(struct device *dev) > +{ > + struct pcpu *pcpu = container_of(dev, struct pcpu, dev); > + > + list_del(&pcpu->list); > + kfree(pcpu); > +} > + > +static void unregister_and_remove_pcpu(struct pcpu *pcpu) > +{ > + struct device *dev; > + > + if (!pcpu) > + return; > + > + dev = &pcpu->dev; > + if (dev->id) > + device_remove_file(dev, &dev_attr_online); > + > + /* pcpu remove would be implicitly done */ > + device_unregister(dev); > +} > + > +static int register_pcpu(struct pcpu *pcpu) > +{ > + struct device *dev; > + int err = -EINVAL; > + > + if (!pcpu) > + return err; > + > + dev = &pcpu->dev; > + dev->bus = &xen_pcpu_subsys; > + dev->id = pcpu->cpu_id; > + dev->release = pcpu_release; > + > + err = device_register(dev); > + if (err) { > + pcpu_release(dev); > + return err; > + } > + > + /* > + * Xen never offline cpu0 due to several restrictions > + * and assumptions. This basically doesn't add a sys control > + * to user, one cannot attempt to offline BSP. > + */ > + if (dev->id) { > + err = device_create_file(dev, &dev_attr_online); > + if (err) { > + device_unregister(dev); > + return err; > + } > + } > + > + return 0; > +} > + > +static struct pcpu *create_and_register_pcpu(struct xenpf_pcpuinfo *info) > +{ > + struct pcpu *pcpu; > + int err; > + > + if (info->flags & XEN_PCPU_FLAGS_INVALID) > + return ERR_PTR(-ENODEV); > + > + pcpu = kzalloc(sizeof(struct pcpu), GFP_KERNEL); > + if (!pcpu) > + return ERR_PTR(-ENOMEM); > + > + INIT_LIST_HEAD(&pcpu->list); > + pcpu->cpu_id = info->xen_cpuid; > + pcpu->flags = info->flags; > + > + /* Need hold on xen_pcpu_lock before pcpu list manipulations */ > + list_add_tail(&pcpu->list, &xen_pcpus); > + > + err = register_pcpu(pcpu); > + if (err) { > + pr_warning(XEN_PCPU "Failed to register pcpu%u\n", > + info->xen_cpuid); > + return ERR_PTR(-ENOENT); > + } > + > + return pcpu; > +} > + > +/* > + * Caller should hold the xen_pcpu_lock > + */ > +static int sync_pcpu(uint32_t cpu, uint32_t *max_cpu) > +{ > + int ret; > + struct pcpu *pcpu = NULL; > + struct xenpf_pcpuinfo *info; > + struct xen_platform_op op = { > + .cmd = XENPF_get_cpuinfo, > + .interface_version = XENPF_INTERFACE_VERSION, > + .u.pcpu_info.xen_cpuid = cpu, > + }; > + > + ret = HYPERVISOR_dom0_op(&op); > + if (ret) > + return ret; > + > + info = &op.u.pcpu_info; > + if (max_cpu) > + *max_cpu = info->max_present; > + > + pcpu = get_pcpu(cpu); > + > + /* > + * Only those at cpu present map has its sys interface. > + */ > + if (info->flags & XEN_PCPU_FLAGS_INVALID) { > + if (pcpu) > + unregister_and_remove_pcpu(pcpu); > + return 0; > + } > + > + if (!pcpu) { > + pcpu = create_and_register_pcpu(info); > + if (IS_ERR_OR_NULL(pcpu)) > + return -ENODEV; > + } else > + pcpu_online_status(info, pcpu); > + > + return 0; > +} > + > +/* > + * Sync dom0's pcpu information with xen hypervisor's > + */ > +static int xen_sync_pcpus(void) > +{ > + /* > + * Boot cpu always have cpu_id 0 in xen > + */ > + uint32_t cpu = 0, max_cpu = 0; > + int err = 0; > + struct pcpu *pcpu, *tmp; > + > + mutex_lock(&xen_pcpu_lock); > + > + while (!err && (cpu <= max_cpu)) { > + err = sync_pcpu(cpu, &max_cpu); > + cpu++; > + } > + > + if (err) > + list_for_each_entry_safe(pcpu, tmp, &xen_pcpus, list) > + unregister_and_remove_pcpu(pcpu); > + > + mutex_unlock(&xen_pcpu_lock); > + > + return err; > +} > + > +static void xen_pcpu_work_fn(struct work_struct *work) > +{ > + xen_sync_pcpus(); > +} > +static DECLARE_WORK(xen_pcpu_work, xen_pcpu_work_fn); > + > +static irqreturn_t xen_pcpu_interrupt(int irq, void *dev_id) > +{ > + schedule_work(&xen_pcpu_work); > + return IRQ_HANDLED; > +} > + > +static int __init xen_pcpu_init(void) > +{ > + int irq, ret; > + > + if (!xen_initial_domain()) > + return -ENODEV; > + > + irq = bind_virq_to_irqhandler(VIRQ_PCPU_STATE, 0, > + xen_pcpu_interrupt, 0, > + "xen-pcpu", NULL); > + if (irq < 0) { > + pr_warning(XEN_PCPU "Failed to bind pcpu virq\n"); > + return irq; > + } > + > + ret = subsys_system_register(&xen_pcpu_subsys, NULL); > + if (ret) { > + pr_warning(XEN_PCPU "Failed to register pcpu subsys\n"); > + goto err1; > + } > + > + ret = xen_sync_pcpus(); > + if (ret) { > + pr_warning(XEN_PCPU "Failed to sync pcpu info\n"); > + goto err2; > + } > + > + return 0; > + > +err2: > + bus_unregister(&xen_pcpu_subsys); > +err1: > + unbind_from_irqhandler(irq, NULL); > + return ret; > +} > +arch_initcall(xen_pcpu_init); > diff --git a/include/xen/interface/platform.h b/include/xen/interface/platform.h > index 486653f..61fa661 100644 > --- a/include/xen/interface/platform.h > +++ b/include/xen/interface/platform.h > @@ -314,6 +314,13 @@ struct xenpf_pcpuinfo { > }; > DEFINE_GUEST_HANDLE_STRUCT(xenpf_pcpuinfo); > > +#define XENPF_cpu_online 56 > +#define XENPF_cpu_offline 57 > +struct xenpf_cpu_ol { > + uint32_t cpuid; > +}; > +DEFINE_GUEST_HANDLE_STRUCT(xenpf_cpu_ol); > + > struct xen_platform_op { > uint32_t cmd; > uint32_t interface_version; /* XENPF_INTERFACE_VERSION */ > @@ -330,6 +337,7 @@ struct xen_platform_op { > struct xenpf_getidletime getidletime; > struct xenpf_set_processor_pminfo set_pminfo; > struct xenpf_pcpuinfo pcpu_info; > + struct xenpf_cpu_ol cpu_ol; > uint8_t pad[128]; > } u; > }; > diff --git a/include/xen/interface/xen.h b/include/xen/interface/xen.h > index a890804..0801468 100644 > --- a/include/xen/interface/xen.h > +++ b/include/xen/interface/xen.h > @@ -80,6 +80,7 @@ > #define VIRQ_CONSOLE 2 /* (DOM0) Bytes received on emergency console. */ > #define VIRQ_DOM_EXC 3 /* (DOM0) Exceptional event for some domain. */ > #define VIRQ_DEBUGGER 6 /* (DOM0) A domain has paused for debugging. */ > +#define VIRQ_PCPU_STATE 9 /* (DOM0) PCPU state changed */ > > /* Architecture-specific VIRQ definitions. */ > #define VIRQ_ARCH_0 16 > -- > 1.7.1 > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > http://lists.xen.org/xen-devel -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/