From patchwork Thu Sep 1 19:50:43 2011 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Alex Williamson X-Patchwork-Id: 1120422 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by demeter2.kernel.org (8.14.4/8.14.4) with ESMTP id p81Jprsa028417 for ; Thu, 1 Sep 2011 19:51:53 GMT Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757907Ab1IATvu (ORCPT ); Thu, 1 Sep 2011 15:51:50 -0400 Received: from mx1.redhat.com ([209.132.183.28]:11804 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757676Ab1IATvt (ORCPT ); Thu, 1 Sep 2011 15:51:49 -0400 Received: from int-mx09.intmail.prod.int.phx2.redhat.com (int-mx09.intmail.prod.int.phx2.redhat.com [10.5.11.22]) by mx1.redhat.com (8.14.4/8.14.4) with ESMTP id p81JojnZ021251 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Thu, 1 Sep 2011 15:50:45 -0400 Received: from s20.home (ovpn01.gateway.prod.ext.phx2.redhat.com [10.5.9.1]) by int-mx09.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id p81JohEN029720; Thu, 1 Sep 2011 15:50:44 -0400 From: Alex Williamson Subject: [RFC PATCH 3/5] VFIO: Base framework for new VFIO driver To: chrisw@sous-sol.org, aik@au1.ibm.com, pmac@au1.ibm.com, dwg@au1.ibm.com, joerg.roedel@amd.com, agraf@suse.de, benve@cisco.com, aafabbri@cisco.com, B08248@freescale.com, B07421@freescale.com, avi@redhat.com, kvm@vger.kernel.org, qemu-devel@nongnu.org, iommu@lists.linux-foundation.org, linux-pci@vger.kernel.org Cc: alex.williamson@redhat.com Date: Thu, 01 Sep 2011 13:50:43 -0600 Message-ID: <20110901195043.2391.31843.stgit@s20.home> In-Reply-To: <20110901194915.2391.97400.stgit@s20.home> References: <20110901194915.2391.97400.stgit@s20.home> User-Agent: StGIT/0.14.3 MIME-Version: 1.0 X-Scanned-By: MIMEDefang 2.68 on 10.5.11.22 Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org X-Greylist: IP, sender and recipient auto-whitelisted, not delayed by milter-greylist-4.2.6 (demeter2.kernel.org [140.211.167.43]); Thu, 01 Sep 2011 19:51:53 +0000 (UTC) Signed-off-by: Alex Williamson --- drivers/Kconfig | 2 drivers/Makefile | 1 drivers/vfio/Kconfig | 5 drivers/vfio/Makefile | 3 drivers/vfio/vfio_device.c | 109 +++++ drivers/vfio/vfio_iommu.c | 81 ++++ drivers/vfio/vfio_main.c | 879 +++++++++++++++++++++++++++++++++++++++++++ drivers/vfio/vfio_private.h | 82 ++++ 8 files changed, 1162 insertions(+), 0 deletions(-) create mode 100644 drivers/vfio/Kconfig create mode 100644 drivers/vfio/Makefile create mode 100644 drivers/vfio/vfio_device.c create mode 100644 drivers/vfio/vfio_iommu.c create mode 100644 drivers/vfio/vfio_main.c create mode 100644 drivers/vfio/vfio_private.h -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/drivers/Kconfig b/drivers/Kconfig index 3bb154d..5b5fffc 100644 --- a/drivers/Kconfig +++ b/drivers/Kconfig @@ -112,6 +112,8 @@ source "drivers/auxdisplay/Kconfig" source "drivers/uio/Kconfig" +source "drivers/vfio/Kconfig" + source "drivers/vlynq/Kconfig" source "drivers/xen/Kconfig" diff --git a/drivers/Makefile b/drivers/Makefile index 09f3232..6b17848 100644 --- a/drivers/Makefile +++ b/drivers/Makefile @@ -56,6 +56,7 @@ obj-$(CONFIG_ATM) += atm/ obj-$(CONFIG_FUSION) += message/ obj-y += firewire/ obj-$(CONFIG_UIO) += uio/ +obj-$(CONFIG_VFIO) += vfio/ obj-y += cdrom/ obj-y += auxdisplay/ obj-$(CONFIG_PCCARD) += pcmcia/ diff --git a/drivers/vfio/Kconfig b/drivers/vfio/Kconfig new file mode 100644 index 0000000..a150521 --- /dev/null +++ b/drivers/vfio/Kconfig @@ -0,0 +1,5 @@ +menuconfig VFIO + tristate "Non-Privileged User Space driver" + depends on IOMMU_API + help + If you don't know what to do here, say N. diff --git a/drivers/vfio/Makefile b/drivers/vfio/Makefile new file mode 100644 index 0000000..5eaa074 --- /dev/null +++ b/drivers/vfio/Makefile @@ -0,0 +1,3 @@ +obj-$(CONFIG_VFIO) := vfio.o + +vfio-y := vfio_main.o vfio_iommu.o vfio_device.o diff --git a/drivers/vfio/vfio_device.c b/drivers/vfio/vfio_device.c new file mode 100644 index 0000000..101cbbf --- /dev/null +++ b/drivers/vfio/vfio_device.c @@ -0,0 +1,109 @@ +/* + * Copyright (C) 2011 Red Hat, Inc. All rights reserved. + * Author: Alex Williamson + * + * 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. + * + * Derived from original vfio: + * Copyright 2010 Cisco Systems, Inc. All rights reserved. + * Author: Tom Lyon, pugs@cisco.com + */ + +/* + * VFIO device module: Common device handling and callouts to other drivers + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "vfio_private.h" + +static int vfio_device_release(struct inode *inode, struct file *filep) +{ + struct vfio_device *vdev = filep->private_data; + + mutex_lock(&vdev->vfio->group_lock); + vdev->refcnt--; + vdev->iommu->refcnt--; + mutex_unlock(&vdev->vfio->group_lock); + + return 0; +} + +static long vfio_device_unl_ioctl(struct file *filep, + unsigned int cmd, unsigned long arg) +{ + struct vfio_device *vdev = filep->private_data; + int ret = -EINVAL; + + switch (cmd) { + // TBD - what can we handle as common device ioctls? + default: + if (vdev->ops->fops.unlocked_ioctl) + ret = vdev->ops->fops.unlocked_ioctl(filep, cmd, arg); + } + return ret; +} + +static ssize_t vfio_device_read(struct file *filep, char __user *buf, + size_t count, loff_t *ppos) +{ + struct vfio_device *vdev = filep->private_data; + + if (vdev->ops->fops.read) + return vdev->ops->fops.read(filep, buf, count, ppos); + + return -EINVAL; +} + +static ssize_t vfio_device_write(struct file *filep, const char __user *buf, + size_t count, loff_t *ppos) +{ + struct vfio_device *vdev = filep->private_data; + + if (vdev->ops->fops.write) + return vdev->ops->fops.write(filep, buf, count, ppos); + + return -EINVAL; +} + +static int vfio_device_mmap(struct file *filep, struct vm_area_struct *vma) +{ + struct vfio_device *vdev = filep->private_data; + + if (vdev->ops->fops.mmap) + return vdev->ops->fops.mmap(filep, vma); + + return -EINVAL; +} + +#ifdef CONFIG_COMPAT +static long vfio_device_compat_ioctl(struct file *filep, + unsigned int cmd, unsigned long arg) +{ + arg = (unsigned long)compat_ptr(arg); + return vfio_device_unl_ioctl(filep, cmd, arg); +} +#endif /* CONFIG_COMPAT */ + +const struct file_operations vfio_device_fops = { + .owner = THIS_MODULE, + .release = vfio_device_release, + .read = vfio_device_read, + .write = vfio_device_write, + .unlocked_ioctl = vfio_device_unl_ioctl, +#ifdef CONFIG_COMPAT + .compat_ioctl = vfio_device_compat_ioctl, +#endif + .mmap = vfio_device_mmap, +}; diff --git a/drivers/vfio/vfio_iommu.c b/drivers/vfio/vfio_iommu.c new file mode 100644 index 0000000..1a6f321 --- /dev/null +++ b/drivers/vfio/vfio_iommu.c @@ -0,0 +1,81 @@ +/* + * Copyright (C) 2011 Red Hat, Inc. All rights reserved. + * Author: Alex Williamson + * + * 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. + * + * Derived from original vfio: + * Copyright 2010 Cisco Systems, Inc. All rights reserved. + * Author: Tom Lyon, pugs@cisco.com + */ + +/* + * VFIO iomm module: iommu fd callbacks + */ + +#include +#include +#include +#include +#include +#include +#include + +#include "vfio_private.h" + +static int vfio_iommu_release(struct inode *inode, struct file *filep) +{ + struct vfio_iommu *viommu = filep->private_data; + + mutex_lock(&viommu->vfio->group_lock); + viommu->refcnt--; + mutex_unlock(&viommu->vfio->group_lock); + return 0; +} + +static long vfio_iommu_unl_ioctl(struct file *filep, + unsigned int cmd, unsigned long arg) +{ + struct vfio_iommu *viommu = filep->private_data; + struct vfio_dma_map dm; + int ret = -ENOSYS; + + switch (cmd) { + case VFIO_IOMMU_MAP_DMA: + if (copy_from_user(&dm, (void __user *)arg, sizeof dm)) + return -EFAULT; + ret = 0; // XXX - Do something + if (!ret && copy_to_user((void __user *)arg, &dm, sizeof dm)) + ret = -EFAULT; + break; + + case VFIO_IOMMU_UNMAP_DMA: + if (copy_from_user(&dm, (void __user *)arg, sizeof dm)) + return -EFAULT; + ret = 0; // XXX - Do something + if (!ret && copy_to_user((void __user *)arg, &dm, sizeof dm)) + ret = -EFAULT; + break; + } + return ret; +} + +#ifdef CONFIG_COMPAT +static long vfio_iommu_compat_ioctl(struct file *filep, + unsigned int cmd, unsigned long arg) +{ + arg = (unsigned long)compat_ptr(arg); + return vfio_iommu_unl_ioctl(filep, cmd, arg); +} +#endif /* CONFIG_COMPAT */ + +const struct file_operations vfio_iommu_fops = { + .owner = THIS_MODULE, + .release = vfio_iommu_release, + .unlocked_ioctl = vfio_iommu_unl_ioctl, +#ifdef CONFIG_COMPAT + .compat_ioctl = vfio_iommu_compat_ioctl, +#endif +}; diff --git a/drivers/vfio/vfio_main.c b/drivers/vfio/vfio_main.c new file mode 100644 index 0000000..7f05692 --- /dev/null +++ b/drivers/vfio/vfio_main.c @@ -0,0 +1,879 @@ +/* + * Copyright (C) 2011 Red Hat, Inc. All rights reserved. + * Author: Alex Williamson + * + * 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. + * + * Derived from original vfio: + * Copyright 2010 Cisco Systems, Inc. All rights reserved. + * Author: Tom Lyon, pugs@cisco.com + */ + +/* + * VFIO main module: IOMMU group framework + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "vfio_private.h" + +#define DRIVER_VERSION "0.2" +#define DRIVER_AUTHOR "Alex Williamson " +#define DRIVER_DESC "VFIO - User Level meta-driver" + +#define MAX_PATH 256 + +static int allow_unsafe_intrs; +module_param(allow_unsafe_intrs, int, 0); +MODULE_PARM_DESC(allow_unsafe_intrs, + "Allow use of IOMMUs which do not support interrupt remapping"); + +static struct vfio vfio; +static const struct file_operations vfio_group_fops; + +static inline void vfio_container_reset_read(struct vfio_container *vcontainer) +{ + kfree(vcontainer->read_buf); + vcontainer->read_buf = NULL; +} + +int vfio_group_add_dev(struct device *dev, void *data) +{ + struct vfio_device_ops *ops = data; + struct list_head *pos; + struct vfio_group *vgroup = NULL; + struct vfio_device *vdev = NULL; + unsigned int group; + int ret = 0, new_group = 0; + + if (iommu_device_group(dev, &group)) + return 0; + + mutex_lock(&vfio.group_lock); + + list_for_each(pos, &vfio.group_list) { + vgroup = list_entry(pos, struct vfio_group, next); + if (vgroup->group == group) + break; + vgroup = NULL; + } + + if (!vgroup) { + int id; + + if (unlikely(idr_pre_get(&vfio.idr, GFP_KERNEL) == 0)) { + ret = -ENOMEM; + goto out; + } + vgroup = kzalloc(sizeof(*vgroup), GFP_KERNEL); + if (!vgroup) { + ret = -ENOMEM; + goto out; + } + + vgroup->group = group; + INIT_LIST_HEAD(&vgroup->device_list); + + ret = idr_get_new(&vfio.idr, vgroup, &id); + if (ret == 0 && id > MINORMASK) { + idr_remove(&vfio.idr, id); + kfree(vgroup); + ret = -ENOSPC; + goto out; + } + + vgroup->devt = MKDEV(MAJOR(vfio.devt), id); + list_add(&vgroup->next, &vfio.group_list); + device_create(vfio.class, NULL, vgroup->devt, + vgroup, "%u", group); + + new_group = 1; + } else { + list_for_each(pos, &vgroup->device_list) { + vdev = list_entry(pos, struct vfio_device, next); + if (vdev->dev == dev) + break; + vdev = NULL; + } + } + + if (!vdev) { + /* Adding a device for a group that's already in use? */ + /* Maybe we should attach to the domain so others can't */ + BUG_ON(vgroup->container && + vgroup->container->iommu && + vgroup->container->iommu->refcnt); + + vdev = ops->new(dev); + if (IS_ERR(vdev)) { + /* If we just created this vgroup, tear it down */ + if (new_group) { + device_destroy(vfio.class, vgroup->devt); + idr_remove(&vfio.idr, MINOR(vgroup->devt)); + list_del(&vgroup->next); + kfree(vgroup); + } + ret = PTR_ERR(vdev); + goto out; + } + list_add(&vdev->next, &vgroup->device_list); + vdev->dev = dev; + vdev->ops = ops; + vdev->vfio = &vfio; + } +out: + mutex_unlock(&vfio.group_lock); + return ret; +} + +void vfio_group_del_dev(struct device *dev) +{ + struct list_head *pos; + struct vfio_container *vcontainer; + struct vfio_group *vgroup = NULL; + struct vfio_device *vdev = NULL; + unsigned int group; + + if (iommu_device_group(dev, &group)) + return; + + mutex_lock(&vfio.group_lock); + + list_for_each(pos, &vfio.group_list) { + vgroup = list_entry(pos, struct vfio_group, next); + if (vgroup->group == group) + break; + vgroup = NULL; + } + + if (!vgroup) + goto out; + + vcontainer = vgroup->container; + + list_for_each(pos, &vgroup->device_list) { + vdev = list_entry(pos, struct vfio_device, next); + if (vdev->dev == dev) + break; + vdev = NULL; + } + + if (!vdev) + goto out; + + /* XXX Did a device we're using go away? */ + BUG_ON(vdev->refcnt); + + if (vcontainer && vcontainer->iommu) { + iommu_detach_device(vcontainer->iommu->domain, vdev->dev); + vfio_container_reset_read(vcontainer); + } + + list_del(&vdev->next); + vdev->ops->free(vdev); + + if (list_empty(&vgroup->device_list) && vgroup->refcnt == 0) { + device_destroy(vfio.class, vgroup->devt); + idr_remove(&vfio.idr, MINOR(vgroup->devt)); + list_del(&vgroup->next); + kfree(vgroup); + } +out: + mutex_unlock(&vfio.group_lock); +} + +static int __vfio_group_viable(struct vfio_container *vcontainer) +{ + struct list_head *gpos, *dpos; + + list_for_each(gpos, &vfio.group_list) { + struct vfio_group *vgroup; + vgroup = list_entry(gpos, struct vfio_group, next); + if (vgroup->container != vcontainer) + continue; + + list_for_each(dpos, &vgroup->device_list) { + struct vfio_device *vdev; + vdev = list_entry(dpos, struct vfio_device, next); + + if (!vdev->dev->driver || + vdev->dev->driver->owner != THIS_MODULE) + return 0; + } + } + return 1; +} + +static int __vfio_close_iommu(struct vfio_container *vcontainer) +{ + struct list_head *gpos, *dpos; + struct vfio_iommu *viommu = vcontainer->iommu; + struct vfio_group *vgroup; + struct vfio_device *vdev; + + if (!viommu) + return 0; + + if (viommu->refcnt) + return -EBUSY; + + list_for_each(gpos, &vfio.group_list) { + vgroup = list_entry(gpos, struct vfio_group, next); + if (vgroup->container != vcontainer) + continue; + + list_for_each(dpos, &vgroup->device_list) { + vdev = list_entry(dpos, struct vfio_device, next); + iommu_detach_device(viommu->domain, vdev->dev); + vdev->iommu = NULL; + } + } + iommu_domain_free(viommu->domain); + kfree(viommu); + vcontainer->iommu = NULL; + return 0; +} + +static int __vfio_open_iommu(struct vfio_container *vcontainer) +{ + struct list_head *gpos, *dpos; + struct vfio_iommu *viommu; + struct vfio_group *vgroup; + struct vfio_device *vdev; + + if (!__vfio_group_viable(vcontainer)) + return -EBUSY; + + viommu = kzalloc(sizeof(*viommu), GFP_KERNEL); + if (!viommu) + return -ENOMEM; + + viommu->domain = iommu_domain_alloc(); + if (!viommu->domain) { + kfree(viommu); + return -EFAULT; + } + + viommu->vfio = &vfio; + vcontainer->iommu = viommu; + + list_for_each(gpos, &vfio.group_list) { + vgroup = list_entry(gpos, struct vfio_group, next); + if (vgroup->container != vcontainer) + continue; + + list_for_each(dpos, &vgroup->device_list) { + int ret; + + vdev = list_entry(dpos, struct vfio_device, next); + + ret = iommu_attach_device(viommu->domain, vdev->dev); + if (ret) { + __vfio_close_iommu(vcontainer); + return ret; + } + vdev->iommu = viommu; + } + } + + if (!allow_unsafe_intrs && + !iommu_domain_has_cap(viommu->domain, IOMMU_CAP_INTR_REMAP)) { + __vfio_close_iommu(vcontainer); + return -EFAULT; + } + + return 0; +} + +static int vfio_group_merge(struct vfio_group *vgroup, int fd) +{ + struct vfio_group *vgroup2; + struct iommu_domain *domain; + struct list_head *pos; + struct file *file; + int ret = 0; + + mutex_lock(&vfio.group_lock); + + file = fget(fd); + if (!file) { + ret = -EBADF; + goto out_noput; + } + if (file->f_op != &vfio_group_fops) { + ret = -EINVAL; + goto out; + } + + vgroup2 = file->private_data; + if (!vgroup2 || vgroup2 == vgroup || vgroup2->mm != vgroup->mm || + (vgroup2->container->iommu && vgroup2->container->iommu->refcnt)) { + ret = -EINVAL; + goto out; + } + + if (!vgroup->container->iommu) { + ret = __vfio_open_iommu(vgroup->container); + if (ret) + goto out; + } + + if (!vgroup2->container->iommu) { + ret = __vfio_open_iommu(vgroup2->container); + if (ret) + goto out; + } + + if (iommu_domain_has_cap(vgroup->container->iommu->domain, + IOMMU_CAP_CACHE_COHERENCY) != + iommu_domain_has_cap(vgroup2->container->iommu->domain, + IOMMU_CAP_CACHE_COHERENCY)) { + ret = -EINVAL; + goto out; + } + + ret = __vfio_close_iommu(vgroup2->container); + if (ret) + goto out; + + domain = vgroup->container->iommu->domain; + + list_for_each(pos, &vgroup2->device_list) { + struct vfio_device *vdev; + + vdev = list_entry(pos, struct vfio_device, next); + + ret = iommu_attach_device(domain, vdev->dev); + if (ret) { + list_for_each(pos, &vgroup2->device_list) { + struct vfio_device *vdev2; + + vdev2 = list_entry(pos, + struct vfio_device, next); + if (vdev2 == vdev) + break; + + iommu_detach_device(domain, vdev2->dev); + vdev2->iommu = NULL; + } + goto out; + } + vdev->iommu = vgroup->container->iommu; + } + + kfree(vgroup2->container->read_buf); + kfree(vgroup2->container); + + vgroup2->container = vgroup->container; + vgroup->container->refcnt++; + vfio_container_reset_read(vgroup->container); + +out: + fput(file); +out_noput: + mutex_unlock(&vfio.group_lock); + return ret; +} + +static int vfio_group_unmerge(struct vfio_group *vgroup, int fd) +{ + struct vfio_group *vgroup2; + struct vfio_container *vcontainer2; + struct vfio_device *vdev; + struct list_head *pos; + struct file *file; + int ret = 0; + + vcontainer2 = kzalloc(sizeof(*vcontainer2), GFP_KERNEL); + if (!vcontainer2) + return -ENOMEM; + + mutex_lock(&vfio.group_lock); + + file = fget(fd); + if (!file) { + ret = -EBADF; + goto out_noput; + } + if (file->f_op != &vfio_group_fops) { + ret = -EINVAL; + goto out; + } + + vgroup2 = file->private_data; + if (!vgroup2 || vgroup2 == vgroup || + vgroup2->container != vgroup->container) { + ret = -EINVAL; + goto out; + } + + list_for_each(pos, &vgroup2->device_list) { + vdev = list_entry(pos, struct vfio_device, next); + if (vdev->refcnt) { + ret = -EBUSY; + goto out; + } + } + + list_for_each(pos, &vgroup2->device_list) { + vdev = list_entry(pos, struct vfio_device, next); + iommu_detach_device(vgroup->container->iommu->domain, + vdev->dev); + vdev->iommu = NULL; + } + + vgroup2->container = vcontainer2; + vcontainer2->refcnt++; + vgroup->container->refcnt--; + vfio_container_reset_read(vgroup->container); +out: + fput(file); +out_noput: + if (ret) + kfree(vcontainer2); + mutex_unlock(&vfio.group_lock); + return ret; +} + +static int vfio_group_get_iommu_fd(struct vfio_group *vgroup) +{ + int ret = 0; + struct vfio_iommu *viommu; + + mutex_lock(&vfio.group_lock); + + if (!vgroup->container->iommu) { + ret = __vfio_open_iommu(vgroup->container); + if (ret) + goto out; + } + + viommu = vgroup->container->iommu; + + if (!viommu->file) { + viommu->file = anon_inode_getfile("vfio-iommu", + &vfio_iommu_fops, + viommu, O_RDWR); + if (IS_ERR(viommu->file)) { + ret = PTR_ERR(viommu->file); + viommu->file = NULL; + goto out; + } + } + ret = get_unused_fd(); + if (ret < 0) + goto out; + + fd_install(ret, viommu->file); + + vgroup->container->iommu->refcnt++; +out: + mutex_unlock(&vfio.group_lock); + return ret; +} + +static int vfio_group_get_device_fd(struct vfio_group *vgroup, char *buf) +{ + struct vfio_container *vcontainer = vgroup->container; + struct list_head *gpos, *dpos; + int ret = -ENODEV; + + mutex_lock(&vfio.group_lock); + + if (!vcontainer->iommu) { + ret = __vfio_open_iommu(vcontainer); + if (ret) + goto out; + } + + list_for_each(gpos, &vfio.group_list) { + vgroup = list_entry(gpos, struct vfio_group, next); + if (vgroup->container != vcontainer) + continue; + + list_for_each(dpos, &vgroup->device_list) { + struct vfio_device *vdev; + char buf2[MAX_PATH]; + + vdev = list_entry(dpos, struct vfio_device, next); + + snprintf(buf2, MAX_PATH, "%s", dev_name(vdev->dev)); + + if (!strncmp(buf, buf2, MAX_PATH)) { + if (!vdev->file) { + vdev->file = anon_inode_getfile( + "vfio-device", + &vfio_device_fops, + vdev, O_RDWR); + if (IS_ERR(vdev->file)) { + ret = PTR_ERR(vdev->file); + vdev->file = NULL; + goto out; + } + } + ret = get_unused_fd(); + if (ret < 0) + goto out; + + fd_install(ret, vdev->file); + + vdev->refcnt++; + vcontainer->iommu->refcnt++; + goto out; + } + } + } +out: + mutex_unlock(&vfio.group_lock); + return ret; +} + +static long vfio_group_unl_ioctl(struct file *filep, + unsigned int cmd, unsigned long arg) +{ + struct vfio_group *vgroup = filep->private_data; + + if (vgroup->mm != current->mm) + return -EIO; + + switch (cmd) { + case VFIO_GROUP_MERGE: + case VFIO_GROUP_UNMERGE: + { + int fd; + + if (get_user(fd, (int __user *)arg)) + return -EFAULT; + if (fd < 0) + return -EINVAL; + + if (cmd == VFIO_GROUP_MERGE) + return vfio_group_merge(vgroup, fd); + else + return vfio_group_unmerge(vgroup, fd); + } + case VFIO_GROUP_GET_IOMMU_FD: + return vfio_group_get_iommu_fd(vgroup); + case VFIO_GROUP_GET_DEVICE_FD: + { + char *buf; + int ret; + + buf = strndup_user((const char __user *)arg, MAX_PATH); + if (IS_ERR(buf)) + return PTR_ERR(buf); + + ret = vfio_group_get_device_fd(vgroup, buf); + kfree(buf); + return ret; + } + } + return -ENOSYS; +} + + +#ifdef CONFIG_COMPAT +static long vfio_group_compat_ioctl(struct file *filep, + unsigned int cmd, unsigned long arg) +{ + arg = (unsigned long)compat_ptr(arg); + return vfio_group_unl_ioctl(filep, cmd, arg); +} +#endif /* CONFIG_COMPAT */ + +static int vfio_group_open(struct inode *inode, struct file *filep) +{ + struct vfio_group *vgroup; + int ret = 0; + + mutex_lock(&vfio.group_lock); + + vgroup = idr_find(&vfio.idr, iminor(inode)); + + if (!vgroup) { + ret = -ENODEV; + goto out; + } + + if (!vgroup->refcnt) { + struct vfio_container *vcontainer; + vcontainer = kzalloc(sizeof(*vcontainer), GFP_KERNEL); + if (!vcontainer) { + ret = -ENOMEM; + goto out; + } + vgroup->container = vcontainer; + vgroup->mm = current->mm; + } else if (current->mm != vgroup->mm) { + ret = -EBUSY; + goto out; + } + filep->private_data = vgroup; + vgroup->refcnt++; + vgroup->container->refcnt++; +out: + mutex_unlock(&vfio.group_lock); + + return ret; +} + +static int vfio_group_release(struct inode *inode, struct file *filep) +{ + struct vfio_group *vgroup = filep->private_data; + struct vfio_container *vcontainer = vgroup->container; + struct list_head *pos; + int ret = 0; + + mutex_lock(&vfio.group_lock); + + if (vgroup->refcnt > 1) { + vgroup->refcnt--; + vcontainer->refcnt--; + goto out; + } + + list_for_each(pos, &vgroup->device_list) { + struct vfio_device *vdev; + vdev = list_entry(pos, struct vfio_device, next); + if (vdev->refcnt) { + ret = -EBUSY; + goto out; + } + } + + /* Merged group? */ + if (vcontainer->refcnt > 1) { + if (vcontainer->iommu) { + list_for_each(pos, &vgroup->device_list) { + struct vfio_device *vdev; + vdev = list_entry(pos, + struct vfio_device, next); + iommu_detach_device(vcontainer->iommu->domain, + vdev->dev); + vdev->iommu = NULL; + } + } + vcontainer->refcnt--; + vfio_container_reset_read(vcontainer); + } else { + if (vcontainer->iommu && vcontainer->iommu->refcnt) { + ret = -EBUSY; + goto out; + } + + ret = __vfio_close_iommu(vcontainer); + if (ret) + goto out; + + kfree(vcontainer->read_buf); + kfree(vcontainer); + } + + vgroup->refcnt--; + vgroup->mm = NULL; + vgroup->container = NULL; + + /* Possible we had the group open while device members were removed */ + if (list_empty(&vgroup->device_list)) { + device_destroy(vfio.class, vgroup->devt); + idr_remove(&vfio.idr, MINOR(vgroup->devt)); + list_del(&vgroup->next); + kfree(vgroup); + } +out: + mutex_unlock(&vfio.group_lock); + return 0; +} + +static int __vfio_container_create_read_buf(struct vfio_container *vcontainer) +{ + struct list_head *gpos, *dpos; + struct vfio_group *vgroup; + struct vfio_device *vdev; + int off = 0; + char *buf; + + buf = kzalloc(MAX_PATH, GFP_KERNEL); + if (!buf) + return -ENOMEM; + + list_for_each(gpos, &vfio.group_list) { + vgroup = list_entry(gpos, struct vfio_group, next); + if (vgroup->container != vcontainer) + continue; + + off += snprintf(buf + off, MAX_PATH, + "group: %u\n", vgroup->group); + buf = krealloc(buf, off + MAX_PATH, GFP_KERNEL); + if (!buf) + return -ENOMEM; + memset(buf + off, 0, MAX_PATH); + + list_for_each(dpos, &vgroup->device_list) { + vdev = list_entry(dpos, struct vfio_device, next); + + off += snprintf(buf + off, MAX_PATH, + "device: %s\n", dev_name(vdev->dev)); + buf = krealloc(buf, off + MAX_PATH, GFP_KERNEL); + if (!buf) + return -ENOMEM; + memset(buf + off, 0, MAX_PATH); + } + } + buf = krealloc(buf, off + 1, GFP_KERNEL); + if (!buf) + return -ENOMEM; + + vcontainer->read_buf = buf; + return 0; +} + +static ssize_t vfio_group_read(struct file *filep, char __user *buf, + size_t count, loff_t *ppos) +{ + struct vfio_group *vgroup = filep->private_data; + struct vfio_container *vcontainer; + ssize_t ret = 0; + + mutex_lock(&vfio.group_lock); + + vcontainer = vgroup->container; + + if (!vcontainer) { + ret = -EINVAL; + goto out; + } + + if (!vcontainer->read_buf) { + ret = __vfio_container_create_read_buf(vcontainer); + if (ret) + goto out; + } + + if (*ppos >= strlen(vcontainer->read_buf) + 1) { + ret = 0; + goto out; + } + + if (*ppos + count > strlen(vcontainer->read_buf) + 1) + count = strlen(vcontainer->read_buf) + 1 - *ppos; + + if (copy_to_user(buf, vcontainer->read_buf + *ppos, count)) { + ret = -EFAULT; + goto out; + } + + *ppos += count; + ret = count; +out: + mutex_unlock(&vfio.group_lock); + return ret; +} + +static const struct file_operations vfio_group_fops = { + .owner = THIS_MODULE, + .open = vfio_group_open, + .release = vfio_group_release, + .read = vfio_group_read, + .unlocked_ioctl = vfio_group_unl_ioctl, +#ifdef CONFIG_COMPAT + .compat_ioctl = vfio_group_compat_ioctl, +#endif +}; + +static void vfio_class_release(struct kref *kref) +{ + class_destroy(vfio.class); + vfio.class = NULL; +} + +static char *vfio_devnode(struct device *dev, mode_t *mode) +{ + return kasprintf(GFP_KERNEL, "vfio/%s", dev_name(dev)); +} + +static int __init vfio_init(void) +{ + int ret; + + idr_init(&vfio.idr); + mutex_init(&vfio.group_lock); + INIT_LIST_HEAD(&vfio.group_list); + + kref_init(&vfio.kref); + vfio.class = class_create(THIS_MODULE, "vfio"); + if (IS_ERR(vfio.class)) { + ret = PTR_ERR(vfio.class); + goto err_class; + } + + vfio.class->devnode = vfio_devnode; + + /* FIXME - how many minors to allocate... all of them! */ + ret = alloc_chrdev_region(&vfio.devt, 0, MINORMASK, "vfio"); + if (ret) + goto err_chrdev; + + cdev_init(&vfio.cdev, &vfio_group_fops); + ret = cdev_add(&vfio.cdev, vfio.devt, MINORMASK); + if (ret) + goto err_cdev; + + pr_info(DRIVER_DESC " version: " DRIVER_VERSION "\n"); + + return 0; + +err_cdev: + unregister_chrdev_region(vfio.devt, MINORMASK); +err_chrdev: + kref_put(&vfio.kref, vfio_class_release); +err_class: + return ret; +} + +static void __exit vfio_cleanup(void) +{ + struct list_head *gpos, *gppos; + + list_for_each_safe(gpos, gppos, &vfio.group_list) { + struct vfio_group *vgroup; + struct list_head *dpos, *dppos; + + vgroup = list_entry(gpos, struct vfio_group, next); + + list_for_each_safe(dpos, dppos, &vgroup->device_list) { + struct vfio_device *vdev; + + vdev = list_entry(dpos, struct vfio_device, next); + vfio_group_del_dev(vdev->dev); + } + } + + idr_destroy(&vfio.idr); + cdev_del(&vfio.cdev); + unregister_chrdev_region(vfio.devt, MINORMASK); + kref_put(&vfio.kref, vfio_class_release); +} + +module_init(vfio_init); +module_exit(vfio_cleanup); + +MODULE_VERSION(DRIVER_VERSION); +MODULE_LICENSE("GPL v2"); +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); diff --git a/drivers/vfio/vfio_private.h b/drivers/vfio/vfio_private.h new file mode 100644 index 0000000..2cc300c --- /dev/null +++ b/drivers/vfio/vfio_private.h @@ -0,0 +1,82 @@ +/* + * Copyright (C) 2011 Red Hat, Inc. All rights reserved. + * Author: Alex Williamson + * + * 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. + * + * Derived from original vfio: + * Copyright 2010 Cisco Systems, Inc. All rights reserved. + * Author: Tom Lyon, pugs@cisco.com + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#ifndef VFIO_PRIVATE_H +#define VFIO_PRIVATE_H + +extern const struct file_operations vfio_iommu_fops; +extern const struct file_operations vfio_device_fops; + +struct vfio { + dev_t devt; + struct cdev cdev; + struct list_head group_list; + struct mutex group_lock; + struct kref kref; + struct class *class; + struct idr idr; +}; + +struct vfio_device_ops { + struct vfio_device *(* new)(struct device *); + void (* free)(struct vfio_device *); + struct file_operations fops; +}; + +struct vfio_iommu { + struct iommu_domain *domain; + struct vfio *vfio; + int refcnt; + struct file *file; +}; + +struct vfio_device { + struct device *dev; + struct list_head next; + struct file *file; + struct vfio_device_ops *ops; + struct vfio *vfio; + struct vfio_iommu *iommu; + int refcnt; +}; + +struct vfio_container { + struct vfio_iommu *iommu; + char *read_buf; + int refcnt; +}; + +struct vfio_group { + dev_t devt; + unsigned int group; + int refcnt; + struct mm_struct *mm; + struct vfio_container *container; + struct list_head device_list; + struct list_head next; +}; + +extern int vfio_group_add_dev(struct device *dev, void *data); +extern void vfio_group_del_dev(struct device *dev); + +#endif /* VFIO_PRIVATE_H */