Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S936201AbZDIQkH (ORCPT ); Thu, 9 Apr 2009 12:40:07 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S935603AbZDIQal (ORCPT ); Thu, 9 Apr 2009 12:30:41 -0400 Received: from victor.provo.novell.com ([137.65.250.26]:41271 "EHLO victor.provo.novell.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S935568AbZDIQaN (ORCPT ); Thu, 9 Apr 2009 12:30:13 -0400 From: Gregory Haskins Subject: [RFC PATCH v2 17/19] kvm: Add guest-side support for VBUS To: linux-kernel@vger.kernel.org Cc: agraf@suse.de, pmullaney@novell.com, pmorreale@novell.com, anthony@codemonkey.ws, rusty@rustcorp.com.au, netdev@vger.kernel.org, kvm@vger.kernel.org, avi@redhat.com, bhutchings@solarflare.com, andi@firstfloor.org, gregkh@suse.de, herber@gondor.apana.org.au, chrisw@sous-sol.org, shemminger@vyatta.com Date: Thu, 09 Apr 2009 12:32:10 -0400 Message-ID: <20090409163210.32740.10975.stgit@dev.haskins.net> In-Reply-To: <20090409155200.32740.19358.stgit@dev.haskins.net> References: <20090409155200.32740.19358.stgit@dev.haskins.net> User-Agent: StGIT/0.14.3 MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 18399 Lines: 789 This adds a driver to interface between the host VBUS support, and the guest-vbus bus model. Signed-off-by: Gregory Haskins --- arch/x86/Kconfig | 9 + drivers/Makefile | 1 drivers/vbus/proxy/Makefile | 2 drivers/vbus/proxy/kvm.c | 726 +++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 738 insertions(+), 0 deletions(-) create mode 100644 drivers/vbus/proxy/Makefile create mode 100644 drivers/vbus/proxy/kvm.c diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 91fefd5..8661495 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -451,6 +451,15 @@ config KVM_GUEST_DYNIRQ depends on KVM_GUEST default y +config KVM_GUEST_VBUS + tristate "KVM virtual-bus (VBUS) guest-side support" + depends on KVM_GUEST + select VBUS_DRIVERS + default y + ---help--- + This option enables guest-side support for accessing virtual-bus + devices. + source "arch/x86/lguest/Kconfig" config PARAVIRT diff --git a/drivers/Makefile b/drivers/Makefile index 98fab51..4f2cb93 100644 --- a/drivers/Makefile +++ b/drivers/Makefile @@ -107,3 +107,4 @@ obj-$(CONFIG_VIRTIO) += virtio/ obj-$(CONFIG_STAGING) += staging/ obj-y += platform/ obj-$(CONFIG_VBUS_DEVICES) += vbus/devices/ +obj-$(CONFIG_VBUS_DRIVERS) += vbus/proxy/ diff --git a/drivers/vbus/proxy/Makefile b/drivers/vbus/proxy/Makefile new file mode 100644 index 0000000..c18d58d --- /dev/null +++ b/drivers/vbus/proxy/Makefile @@ -0,0 +1,2 @@ +kvm-guest-vbus-objs += kvm.o +obj-$(CONFIG_KVM_GUEST_VBUS) += kvm-guest-vbus.o diff --git a/drivers/vbus/proxy/kvm.c b/drivers/vbus/proxy/kvm.c new file mode 100644 index 0000000..82e28b4 --- /dev/null +++ b/drivers/vbus/proxy/kvm.c @@ -0,0 +1,726 @@ +/* + * Copyright (C) 2009 Novell. All Rights Reserved. + * + * Author: + * Gregory Haskins + * + * This file is free software; you can redistribute it and/or modify + * it under the terms of version 2 of the GNU General Public License + * as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +MODULE_AUTHOR("Gregory Haskins"); +MODULE_LICENSE("GPL"); +MODULE_VERSION("1"); + +static int kvm_vbus_hypercall(unsigned long nr, void *data, unsigned long len) +{ + return kvm_hypercall3(KVM_HC_VBUS, nr, __pa(data), len); +} + +struct kvm_vbus { + spinlock_t lock; + struct ioq eventq; + struct kvm_vbus_event *ring; + int irq; +}; + +static struct kvm_vbus kvm_vbus; + +struct kvm_vbus_device { + char type[VBUS_MAX_DEVTYPE_LEN]; + u64 handle; + struct list_head shms; + struct vbus_device_proxy vdev; +}; + +/* + * ------------------- + * common routines + * ------------------- + */ + +struct kvm_vbus_device * +to_dev(struct vbus_device_proxy *vdev) +{ + return container_of(vdev, struct kvm_vbus_device, vdev); +} + +static void +_signal_init(struct shm_signal *signal, struct shm_signal_desc *desc, + struct shm_signal_ops *ops) +{ + desc->magic = SHM_SIGNAL_MAGIC; + desc->ver = SHM_SIGNAL_VER; + + shm_signal_init(signal); + + signal->locale = shm_locality_north; + signal->ops = ops; + signal->desc = desc; +} + +/* + * ------------------- + * _signal + * ------------------- + */ + +struct _signal { + struct kvm_vbus *kvbus; + struct shm_signal signal; + u64 handle; + struct rb_node node; + struct list_head list; +}; + +static struct _signal * +to_signal(struct shm_signal *signal) +{ + return container_of(signal, struct _signal, signal); +} + +static struct _signal * +node_to_signal(struct rb_node *node) +{ + return container_of(node, struct _signal, node); +} + +static int +_signal_inject(struct shm_signal *signal) +{ + struct _signal *_signal = to_signal(signal); + + kvm_vbus_hypercall(KVM_VBUS_OP_SHMSIGNAL, + &_signal->handle, sizeof(_signal->handle)); + + return 0; +} + +static void +_signal_release(struct shm_signal *signal) +{ + struct _signal *_signal = to_signal(signal); + + kfree(_signal); +} + +static struct shm_signal_ops _signal_ops = { + .inject = _signal_inject, + .release = _signal_release, +}; + +/* + * ------------------- + * vbus_device_proxy routines + * ------------------- + */ + +static int +kvm_vbus_device_open(struct vbus_device_proxy *vdev, int version, int flags) +{ + struct kvm_vbus_device *dev = to_dev(vdev); + struct vbus_deviceopen params; + int ret; + + if (dev->handle) + return -EINVAL; + + params.devid = vdev->id; + params.version = version; + + ret = kvm_vbus_hypercall(KVM_VBUS_OP_DEVOPEN, + ¶ms, sizeof(params)); + if (ret < 0) + return ret; + + dev->handle = params.handle; + + return 0; +} + +static int +kvm_vbus_device_close(struct vbus_device_proxy *vdev, int flags) +{ + struct kvm_vbus_device *dev = to_dev(vdev); + unsigned long iflags; + int ret; + + if (!dev->handle) + return -EINVAL; + + spin_lock_irqsave(&kvm_vbus.lock, iflags); + + while (!list_empty(&dev->shms)) { + struct _signal *_signal; + + _signal = list_first_entry(&dev->shms, struct _signal, list); + + list_del(&_signal->list); + + spin_unlock_irqrestore(&kvm_vbus.lock, iflags); + shm_signal_put(&_signal->signal); + spin_lock_irqsave(&kvm_vbus.lock, iflags); + } + + spin_unlock_irqrestore(&kvm_vbus.lock, iflags); + + /* + * The DEVICECLOSE will implicitly close all of the shm on the + * host-side, so there is no need to do an explicit per-shm + * hypercall + */ + ret = kvm_vbus_hypercall(KVM_VBUS_OP_DEVCLOSE, + &dev->handle, sizeof(dev->handle)); + + if (ret < 0) + printk(KERN_ERR "KVM-VBUS: Error closing device %s/%lld: %d\n", + vdev->type, vdev->id, ret); + + dev->handle = 0; + + return 0; +} + +static int +kvm_vbus_device_shm(struct vbus_device_proxy *vdev, int id, int prio, + void *ptr, size_t len, + struct shm_signal_desc *sdesc, struct shm_signal **signal, + int flags) +{ + struct kvm_vbus_device *dev = to_dev(vdev); + struct _signal *_signal = NULL; + struct vbus_deviceshm params; + unsigned long iflags; + int ret; + + if (!dev->handle) + return -EINVAL; + + params.devh = dev->handle; + params.id = id; + params.flags = flags; + params.datap = (u64)__pa(ptr); + params.len = len; + + if (signal) { + /* + * The signal descriptor must be embedded within the + * provided ptr + */ + if (!sdesc + || (len < sizeof(*sdesc)) + || ((void *)sdesc < ptr) + || ((void *)sdesc > (ptr + len - sizeof(*sdesc)))) + return -EINVAL; + + _signal = kzalloc(sizeof(*_signal), GFP_KERNEL); + if (!_signal) + return -ENOMEM; + + _signal_init(&_signal->signal, sdesc, &_signal_ops); + + /* + * take another reference for the host. This is dropped + * by a SHMCLOSE event + */ + shm_signal_get(&_signal->signal); + + params.signal.offset = (u64)sdesc - (u64)ptr; + params.signal.prio = prio; + params.signal.cookie = (u64)_signal; + + } else + params.signal.offset = -1; /* yes, this is a u32, but its ok */ + + ret = kvm_vbus_hypercall(KVM_VBUS_OP_DEVSHM, + ¶ms, sizeof(params)); + if (ret < 0) { + if (_signal) { + /* + * We held two references above, so we need to drop + * both of them + */ + shm_signal_put(&_signal->signal); + shm_signal_put(&_signal->signal); + } + + return ret; + } + + if (signal) { + _signal->handle = params.handle; + + spin_lock_irqsave(&kvm_vbus.lock, iflags); + + list_add_tail(&_signal->list, &dev->shms); + + spin_unlock_irqrestore(&kvm_vbus.lock, iflags); + + shm_signal_get(&_signal->signal); + *signal = &_signal->signal; + } + + return 0; +} + +static int +kvm_vbus_device_call(struct vbus_device_proxy *vdev, u32 func, void *data, + size_t len, int flags) +{ + struct kvm_vbus_device *dev = to_dev(vdev); + struct vbus_devicecall params = { + .devh = dev->handle, + .func = func, + .datap = (u64)__pa(data), + .len = len, + .flags = flags, + }; + + if (!dev->handle) + return -EINVAL; + + return kvm_vbus_hypercall(KVM_VBUS_OP_DEVCALL, ¶ms, sizeof(params)); +} + +static void +kvm_vbus_device_release(struct vbus_device_proxy *vdev) +{ + struct kvm_vbus_device *_dev = to_dev(vdev); + + kvm_vbus_device_close(vdev, 0); + + kfree(_dev); +} + +struct vbus_device_proxy_ops kvm_vbus_device_ops = { + .open = kvm_vbus_device_open, + .close = kvm_vbus_device_close, + .shm = kvm_vbus_device_shm, + .call = kvm_vbus_device_call, + .release = kvm_vbus_device_release, +}; + +/* + * ------------------- + * vbus events + * ------------------- + */ + +static void +event_devadd(struct kvm_vbus_add_event *event) +{ + int ret; + struct kvm_vbus_device *new = kzalloc(sizeof(*new), GFP_KERNEL); + if (!new) { + printk(KERN_ERR "KVM_VBUS: Out of memory on add_event\n"); + return; + } + + INIT_LIST_HEAD(&new->shms); + + memcpy(new->type, event->type, VBUS_MAX_DEVTYPE_LEN); + new->vdev.type = new->type; + new->vdev.id = event->id; + new->vdev.ops = &kvm_vbus_device_ops; + + sprintf(new->vdev.dev.bus_id, "%lld", event->id); + + ret = vbus_device_proxy_register(&new->vdev); + if (ret < 0) + panic("failed to register device %lld(%s): %d\n", + event->id, event->type, ret); +} + +static void +event_devdrop(struct kvm_vbus_handle_event *event) +{ + struct vbus_device_proxy *dev = vbus_device_proxy_find(event->handle); + + if (!dev) { + printk(KERN_WARNING "KVM-VBUS: devdrop failed: %lld\n", + event->handle); + return; + } + + vbus_device_proxy_unregister(dev); +} + +static void +event_shmsignal(struct kvm_vbus_handle_event *event) +{ + struct _signal *_signal = (struct _signal *)event->handle; + + _shm_signal_wakeup(&_signal->signal); +} + +static void +event_shmclose(struct kvm_vbus_handle_event *event) +{ + struct _signal *_signal = (struct _signal *)event->handle; + + /* + * This reference was taken during the DEVICESHM call + */ + shm_signal_put(&_signal->signal); +} + +/* + * ------------------- + * eventq routines + * ------------------- + */ + +static struct ioq_notifier eventq_notifier; + +static int __init +eventq_init(int qlen) +{ + struct ioq_iterator iter; + int ret; + int i; + + kvm_vbus.ring = kzalloc(sizeof(struct kvm_vbus_event) * qlen, + GFP_KERNEL); + if (!kvm_vbus.ring) + return -ENOMEM; + + /* + * We want to iterate on the "valid" index. By default the iterator + * will not "autoupdate" which means it will not hypercall the host + * with our changes. This is good, because we are really just + * initializing stuff here anyway. Note that you can always manually + * signal the host with ioq_signal() if the autoupdate feature is not + * used. + */ + ret = ioq_iter_init(&kvm_vbus.eventq, &iter, ioq_idxtype_valid, 0); + BUG_ON(ret < 0); + + /* + * Seek to the tail of the valid index (which should be our first + * item since the queue is brand-new) + */ + ret = ioq_iter_seek(&iter, ioq_seek_tail, 0, 0); + BUG_ON(ret < 0); + + /* + * Now populate each descriptor with an empty vbus_event and mark it + * valid + */ + for (i = 0; i < qlen; i++) { + struct kvm_vbus_event *event = &kvm_vbus.ring[i]; + size_t len = sizeof(*event); + struct ioq_ring_desc *desc = iter.desc; + + BUG_ON(iter.desc->valid); + + desc->cookie = (u64)event; + desc->ptr = (u64)__pa(event); + desc->len = len; /* total length */ + desc->valid = 1; + + /* + * This push operation will simultaneously advance the + * valid-tail index and increment our position in the queue + * by one. + */ + ret = ioq_iter_push(&iter, 0); + BUG_ON(ret < 0); + } + + kvm_vbus.eventq.notifier = &eventq_notifier; + + /* + * And finally, ensure that we can receive notification + */ + ioq_notify_enable(&kvm_vbus.eventq, 0); + + return 0; +} + +/* Invoked whenever the hypervisor ioq_signal()s our eventq */ +static void +eventq_wakeup(struct ioq_notifier *notifier) +{ + struct ioq_iterator iter; + int ret; + + /* We want to iterate on the head of the in-use index */ + ret = ioq_iter_init(&kvm_vbus.eventq, &iter, ioq_idxtype_inuse, 0); + BUG_ON(ret < 0); + + ret = ioq_iter_seek(&iter, ioq_seek_head, 0, 0); + BUG_ON(ret < 0); + + /* + * The EOM is indicated by finding a packet that is still owned by + * the south side. + * + * FIXME: This in theory could run indefinitely if the host keeps + * feeding us events since there is nothing like a NAPI budget. We + * might need to address that + */ + while (!iter.desc->sown) { + struct ioq_ring_desc *desc = iter.desc; + struct kvm_vbus_event *event; + + event = (struct kvm_vbus_event *)desc->cookie; + + switch (event->eventid) { + case KVM_VBUS_EVENT_DEVADD: + event_devadd(&event->data.add); + break; + case KVM_VBUS_EVENT_DEVDROP: + event_devdrop(&event->data.handle); + break; + case KVM_VBUS_EVENT_SHMSIGNAL: + event_shmsignal(&event->data.handle); + break; + case KVM_VBUS_EVENT_SHMCLOSE: + event_shmclose(&event->data.handle); + break; + default: + printk(KERN_WARNING "KVM_VBUS: Unexpected event %d\n", + event->eventid); + break; + }; + + memset(event, 0, sizeof(*event)); + + /* Advance the in-use head */ + ret = ioq_iter_pop(&iter, 0); + BUG_ON(ret < 0); + } + + /* And let the south side know that we changed the queue */ + ioq_signal(&kvm_vbus.eventq, 0); +} + +static struct ioq_notifier eventq_notifier = { + .signal = &eventq_wakeup, +}; + +/* Injected whenever the host issues an ioq_signal() on the eventq */ +irqreturn_t +eventq_intr(int irq, void *dev) +{ + _shm_signal_wakeup(kvm_vbus.eventq.signal); + + return IRQ_HANDLED; +} + +/* + * ------------------- + */ + +static int +eventq_signal_inject(struct shm_signal *signal) +{ + u64 handle = 0; /* The eventq uses the special-case handle=0 */ + + kvm_vbus_hypercall(KVM_VBUS_OP_SHMSIGNAL, &handle, sizeof(handle)); + + return 0; +} + +static void +eventq_signal_release(struct shm_signal *signal) +{ + kfree(signal); +} + +static struct shm_signal_ops eventq_signal_ops = { + .inject = eventq_signal_inject, + .release = eventq_signal_release, +}; + +/* + * ------------------- + */ + +static void +eventq_ioq_release(struct ioq *ioq) +{ + /* released as part of the kvm_vbus object */ +} + +static struct ioq_ops eventq_ioq_ops = { + .release = eventq_ioq_release, +}; + +/* + * ------------------- + */ + +static void +kvm_vbus_release(void) +{ + if (kvm_vbus.irq > 0) { + free_irq(kvm_vbus.irq, NULL); + destroy_kvm_dynirq(kvm_vbus.irq); + } + + kfree(kvm_vbus.eventq.head_desc); + kfree(kvm_vbus.ring); +} + +static int __init +kvm_vbus_open(void) +{ + struct kvm_vbus_busopen params = { + .magic = KVM_VBUS_MAGIC, + .version = KVM_VBUS_VERSION, + .capabilities = 0, + }; + + return kvm_vbus_hypercall(KVM_VBUS_OP_BUSOPEN, ¶ms, sizeof(params)); +} + +#define QLEN 1024 + +static int __init +kvm_vbus_register(void) +{ + struct kvm_vbus_busreg params = { + .count = 1, + .eventq = { + { + .irq = kvm_vbus.irq, + .count = QLEN, + .ring = (u64)__pa(kvm_vbus.eventq.head_desc), + .data = (u64)__pa(kvm_vbus.ring), + }, + }, + }; + + return kvm_vbus_hypercall(KVM_VBUS_OP_BUSREG, ¶ms, sizeof(params)); +} + +static int __init +_ioq_init(size_t ringsize, struct ioq *ioq, struct ioq_ops *ops) +{ + struct shm_signal *signal = NULL; + struct ioq_ring_head *head = NULL; + size_t len = IOQ_HEAD_DESC_SIZE(ringsize); + + head = kzalloc(len, GFP_KERNEL | GFP_DMA); + if (!head) + return -ENOMEM; + + signal = kzalloc(sizeof(*signal), GFP_KERNEL); + if (!signal) { + kfree(head); + return -ENOMEM; + } + + head->magic = IOQ_RING_MAGIC; + head->ver = IOQ_RING_VER; + head->count = ringsize; + + _signal_init(signal, &head->signal, &eventq_signal_ops); + + ioq_init(ioq, ops, ioq_locality_north, head, signal, ringsize); + + return 0; +} + +int __init +kvm_vbus_init(void) +{ + int ret; + + memset(&kvm_vbus, 0, sizeof(kvm_vbus)); + + ret = kvm_para_has_feature(KVM_FEATURE_VBUS); + if (!ret) + return -ENOENT; + + ret = kvm_vbus_open(); + if (ret < 0) { + printk(KERN_ERR "KVM_VBUS: Could not register with host: %d\n", + ret); + goto out_fail; + } + + spin_lock_init(&kvm_vbus.lock); + + /* + * Allocate an IOQ to use for host-2-guest event notification + */ + ret = _ioq_init(QLEN, &kvm_vbus.eventq, &eventq_ioq_ops); + if (ret < 0) { + printk(KERN_ERR "KVM_VBUS: Cound not init eventq\n"); + goto out_fail; + } + + ret = eventq_init(QLEN); + if (ret < 0) { + printk(KERN_ERR "KVM_VBUS: Cound not setup ring\n"); + goto out_fail; + } + + /* + * Dynamically assign a free IRQ to this resource + */ + kvm_vbus.irq = create_kvm_dynirq(0); + if (kvm_vbus.irq < 0) { + printk(KERN_ERR "KVM_VBUS: Failed to create IRQ\n"); + goto out_fail; + } + + ret = request_irq(kvm_vbus.irq, eventq_intr, 0, "vbus", NULL); + if (ret < 0) { + printk(KERN_ERR "KVM_VBUS: Failed to register IRQ %d\n: %d", + kvm_vbus.irq, ret); + goto out_fail; + } + + /* + * Finally register our queue on the host to start receiving events + */ + ret = kvm_vbus_register(); + if (ret < 0) { + printk(KERN_ERR "KVM_VBUS: Could not register with host: %d\n", + ret); + goto out_fail; + } + + return 0; + + out_fail: + kvm_vbus_release(); + + return ret; + +} + +static void __exit +kvm_vbus_exit(void) +{ + kvm_vbus_release(); +} + +module_init(kvm_vbus_init); +module_exit(kvm_vbus_exit); + -- 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/