Add two new ioctl for VFIO containers. VFIO_DEVICE_BIND_PROCESS creates a
bond between a container and a process address space, identified by a
device-specific ID named PASID. This allows the device to target DMA
transactions at the process virtual addresses without a need for mapping
and unmapping buffers explicitly in the IOMMU. The process page tables are
shared with the IOMMU, and mechanisms such as PCI ATS/PRI may be used to
handle faults. VFIO_DEVICE_UNBIND_PROCESS removed a bond identified by a
PASID.
Signed-off-by: Jean-Philippe Brucker <jean-philippe.brucker@xxxxxxx>
---
drivers/vfio/vfio_iommu_type1.c | 243 +++++++++++++++++++++++++++++++++++++++-
include/uapi/linux/vfio.h | 69 ++++++++++++
2 files changed, 311 insertions(+), 1 deletion(-)
diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c
index 92155cce926d..4bfb92273cb5 100644
--- a/drivers/vfio/vfio_iommu_type1.c
+++ b/drivers/vfio/vfio_iommu_type1.c
@@ -30,6 +30,7 @@
#include <linux/iommu.h>
#include <linux/module.h>
#include <linux/mm.h>
+#include <linux/ptrace.h>
#include <linux/rbtree.h>
#include <linux/sched/signal.h>
#include <linux/sched/mm.h>
@@ -60,6 +61,7 @@ MODULE_PARM_DESC(disable_hugepages,
struct vfio_iommu {
struct list_head domain_list;
+ struct list_head process_list;
struct vfio_domain *external_domain; /* domain for external user */
struct mutex lock;
struct rb_root dma_list;
@@ -92,6 +94,12 @@ struct vfio_group {
struct list_head next;
};
+struct vfio_process {
+ int pasid;
+ struct pid *pid;
+ struct list_head next;
+};
+
/*
* Guest RAM pinning working set or DMA target
*/
@@ -1114,6 +1122,25 @@ static int vfio_iommu_replay(struct vfio_iommu *iommu,
return 0;
}
+static int vfio_iommu_replay_bind(struct vfio_iommu *iommu, struct vfio_group *group)
+{
+ int ret;
+ u32 pasid;
+ struct vfio_process *vfio_process;
+
+ list_for_each_entry(vfio_process, &iommu->process_list, next) {
+ struct task_struct *task = get_pid_task(vfio_process->pid,
+ PIDTYPE_PID);
+
+ ret = iommu_process_bind_group(group->iommu_group, task, &pasid, 0);
+ put_task_struct(task);
+ if (ret)
+ return ret;
+ }
+
+ return 0;
+}
+
/*
* We change our unmap behavior slightly depending on whether the IOMMU
* supports fine-grained superpages. IOMMUs like AMD-Vi will use a superpage
@@ -1301,8 +1328,9 @@ static int vfio_iommu_type1_attach_group(void *iommu_data,
list_add(&group->next, &d->group_list);
iommu_domain_free(domain->domain);
kfree(domain);
+ ret = vfio_iommu_replay_bind(iommu, group);
mutex_unlock(&iommu->lock);
- return 0;
+ return ret;
}
ret = iommu_attach_group(domain->domain, iommu_group);
@@ -1318,6 +1346,10 @@ static int vfio_iommu_type1_attach_group(void *iommu_data,
if (ret)
goto out_detach;
+ ret = vfio_iommu_replay_bind(iommu, group);
+ if (ret)
+ goto out_detach;
+
if (resv_msi) {
ret = iommu_get_msi_cookie(domain->domain, resv_msi_base);
if (ret)
@@ -1349,6 +1381,21 @@ static void vfio_iommu_unmap_unpin_all(struct vfio_iommu *iommu)
vfio_remove_dma(iommu, rb_entry(node, struct vfio_dma, node));
}
+static void vfio_iommu_unbind_all(struct vfio_iommu *iommu)
+{
+ struct vfio_process *process, *process_tmp;
+
+ list_for_each_entry_safe(process, process_tmp, &iommu->process_list, next) {
+ /*
+ * No need to unbind manually, iommu_detach_group should
+ * do it for us.
+ */
+ put_pid(process->pid);
+ kfree(process);
+ }
+ INIT_LIST_HEAD(&iommu->process_list);
+}
+
static void vfio_iommu_unmap_unpin_reaccount(struct vfio_iommu *iommu)
{
struct rb_node *n, *p;
@@ -1438,6 +1485,7 @@ static void vfio_iommu_type1_detach_group(void *iommu_data,
vfio_iommu_unmap_unpin_all(iommu);
else
vfio_iommu_unmap_unpin_reaccount(iommu);
+ vfio_iommu_unbind_all(iommu);
}
iommu_domain_free(domain->domain);
list_del(&domain->next);
@@ -1472,6 +1520,7 @@ static void *vfio_iommu_type1_open(unsigned long arg)
}
INIT_LIST_HEAD(&iommu->domain_list);
+ INIT_LIST_HEAD(&iommu->process_list);
iommu->dma_list = RB_ROOT;
mutex_init(&iommu->lock);
BLOCKING_INIT_NOTIFIER_HEAD(&iommu->notifier);
@@ -1506,6 +1555,7 @@ static void vfio_iommu_type1_release(void *iommu_data)
kfree(iommu->external_domain);
}
+ vfio_iommu_unbind_all(iommu);
vfio_iommu_unmap_unpin_all(iommu);
list_for_each_entry_safe(domain, domain_tmp,
@@ -1534,6 +1584,159 @@ static int vfio_domains_have_iommu_cache(struct vfio_iommu *iommu)
return ret;
}
+static long vfio_iommu_type1_bind_process(struct vfio_iommu *iommu,
+ void __user *arg,
+ struct vfio_iommu_type1_bind *bind)
+{
+ struct vfio_iommu_type1_bind_process params;
+ struct vfio_process *vfio_process;
+ struct vfio_domain *domain;
+ struct task_struct *task;
+ struct vfio_group *group;
+ struct mm_struct *mm;
+ unsigned long minsz;
+ struct pid *pid;
+ int ret;
+
+ minsz = sizeof(*bind) + sizeof(params);
+ if (bind->argsz < minsz)
+ return -EINVAL;
+
+ arg += sizeof(*bind);
+ ret = copy_from_user(¶ms, arg, sizeof(params));
+ if (ret)
+ return -EFAULT;
+
+ if (params.flags & ~VFIO_IOMMU_BIND_PID)
+ return -EINVAL;
+
+ if (params.flags & VFIO_IOMMU_BIND_PID) {
+ pid_t vpid;
+
+ minsz += sizeof(pid_t);
+ if (bind->argsz < minsz)
+ return -EINVAL;
+
+ ret = copy_from_user(&vpid, arg + sizeof(params), sizeof(pid_t));
+ if (ret)
+ return -EFAULT;
+
+ rcu_read_lock();
+ task = find_task_by_vpid(vpid);
+ if (task)
+ get_task_struct(task);
+ rcu_read_unlock();
+ if (!task)
+ return -ESRCH;
+
+ /* Ensure current has RW access on the mm */
+ mm = mm_access(task, PTRACE_MODE_ATTACH_REALCREDS);
+ if (!mm || IS_ERR(mm)) {
+ put_task_struct(task);
+ return IS_ERR(mm) ? PTR_ERR(mm) : -ESRCH;
+ }
+ mmput(mm);
+ } else {
+ get_task_struct(current);
+ task = current;
+ }
+
+ pid = get_task_pid(task, PIDTYPE_PID);
+ mutex_lock(&iommu->lock);
+ list_for_each_entry(vfio_process, &iommu->process_list, next) {
+ if (vfio_process->pid != pid)
+ continue;
+
+ params.pasid = vfio_process->pasid;
+
+ mutex_unlock(&iommu->lock);
+ put_pid(pid);
+ put_task_struct(task);
+ return copy_to_user(arg, ¶ms, sizeof(params)) ?
+ -EFAULT : 0;
+ }
+
+ vfio_process = kzalloc(sizeof(*vfio_process), GFP_KERNEL);
+ if (!vfio_process) {
+ mutex_unlock(&iommu->lock);
+ put_pid(pid);
+ put_task_struct(task);
+ return -ENOMEM;
+ }
+
+ list_for_each_entry(domain, &iommu->domain_list, next) {
+ list_for_each_entry(group, &domain->group_list, next) {
+ ret = iommu_process_bind_group(group->iommu_group, task,
+ ¶ms.pasid, 0);
+ if (ret)
+ break;
+ }
+ if (ret)
+ break;
+ }
+
+ if (!ret) {
+ vfio_process->pid = pid;
+ vfio_process->pasid = params.pasid;
+ list_add(&vfio_process->next, &iommu->process_list);
+ }
+
+ mutex_unlock(&iommu->lock);
+
+ put_task_struct(task);
+
+ if (ret)
+ kfree(vfio_process);
+ else
+ ret = copy_to_user(arg, ¶ms, sizeof(params)) ?
+ -EFAULT : 0;
+
+ return ret;
+}
+
+static long vfio_iommu_type1_unbind_process(struct vfio_iommu *iommu,
+ void __user *arg,
+ struct vfio_iommu_type1_bind *bind)
+{
+ int ret = -EINVAL;
+ unsigned long minsz;
+ struct vfio_process *process;
+ struct vfio_group *group;
+ struct vfio_domain *domain;
+ struct vfio_iommu_type1_bind_process params;
+
+ minsz = sizeof(*bind) + sizeof(params);
+ if (bind->argsz < minsz)
+ return -EINVAL;
+
+ arg += sizeof(*bind);
+ ret = copy_from_user(¶ms, arg, sizeof(params));
+ if (ret)
+ return -EFAULT;
+
+ if (params.flags)
+ return -EINVAL;
+
+ mutex_lock(&iommu->lock);
+ list_for_each_entry(process, &iommu->process_list, next) {
+ if (process->pasid != params.pasid)
+ continue;
+
+ list_for_each_entry(domain, &iommu->domain_list, next)
+ list_for_each_entry(group, &domain->group_list, next)
+ iommu_process_unbind_group(group->iommu_group,
+ process->pasid);
+
+ put_pid(process->pid);
+ list_del(&process->next);
+ kfree(process);
+ break;
+ }
+ mutex_unlock(&iommu->lock);
+
+ return ret;
+}
+
static long vfio_iommu_type1_ioctl(void *iommu_data,
unsigned int cmd, unsigned long arg)
{
@@ -1604,6 +1807,44 @@ static long vfio_iommu_type1_ioctl(void *iommu_data,
return copy_to_user((void __user *)arg, &unmap, minsz) ?
-EFAULT : 0;
+
+ } else if (cmd == VFIO_IOMMU_BIND) {
+ struct vfio_iommu_type1_bind bind;
+
+ minsz = offsetofend(struct vfio_iommu_type1_bind, mode);
+
+ if (copy_from_user(&bind, (void __user *)arg, minsz))
+ return -EFAULT;
+
+ if (bind.argsz < minsz)
+ return -EINVAL;
+
+ switch (bind.mode) {
+ case VFIO_IOMMU_BIND_PROCESS:
+ return vfio_iommu_type1_bind_process(iommu, (void *)arg,
+ &bind);
+ default:
+ return -EINVAL;
+ }
+
+ } else if (cmd == VFIO_IOMMU_UNBIND) {
+ struct vfio_iommu_type1_bind bind;
+
+ minsz = offsetofend(struct vfio_iommu_type1_bind, mode);
+
+ if (copy_from_user(&bind, (void __user *)arg, minsz))
+ return -EFAULT;
+
+ if (bind.argsz < minsz)
+ return -EINVAL;
+
+ switch (bind.mode) {
+ case VFIO_IOMMU_BIND_PROCESS:
+ return vfio_iommu_type1_unbind_process(iommu, (void *)arg,
+ &bind);
+ default:
+ return -EINVAL;
+ }
}
return -ENOTTY;
diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
index ae461050661a..6da8321c33dc 100644
--- a/include/uapi/linux/vfio.h
+++ b/include/uapi/linux/vfio.h
@@ -565,6 +565,75 @@ struct vfio_iommu_type1_dma_unmap {
#define VFIO_IOMMU_ENABLE _IO(VFIO_TYPE, VFIO_BASE + 15)
#define VFIO_IOMMU_DISABLE _IO(VFIO_TYPE, VFIO_BASE + 16)
+/*
+ * Allocate a PASID for a local process, and use it to attach this process to
+ * devices in the container. Devices can then tag their DMA traffic with the
+ * returned @pasid to perform transactions on the associated virtual address
+ * space. Mapping and unmapping of buffers is performed by standard functions
+ * such as mmap and malloc.
+ *
+ * If flag is VFIO_IOMMU_BIND_PID, bind to a process different from the calling
+ * one. data contains the pid of that process, a s32. Given that the caller owns
+ * the device, setting this flag grants the caller read and write permissions on
+ * the entire address space of foreign process described by @pid. Therefore,
+ * permission to perform the bind operation on a foreign process is governed by
+ * the ptrace access mode PTRACE_MODE_ATTACH_REALCREDS check. See man ptrace(2)
+ * for more information.
+ *
+ * On success, VFIO writes a Process Address Space ID (PASID) into @pasid. This
+ * ID is unique to a process and can be used on all devices in the container.
+ *
+ * On fork, the child inherits the device fd and can use the bonds setup by its
+ * parent. Consequently, the child has R/W access on the address spaces bound by
+ * its parent. After an execv, the device fd is closed and the child doesn't
+ * have access to the address space anymore.
+ */
+struct vfio_iommu_type1_bind_process {
+ __u32 flags;
+#define VFIO_IOMMU_BIND_PID (1 << 0)
+ __u32 pasid;
+ __u8 data[];
+};
+
+/*
+ * Only mode supported at the moment is VFIO_IOMMU_BIND_PROCESS, which takes
+ * vfio_iommu_type1_bind_process in data.
+ */
+struct vfio_iommu_type1_bind {
+ __u32 argsz;
+ __u32 mode;
+#define VFIO_IOMMU_BIND_PROCESS (1 << 0)
+ __u8 data[];
+};
+
+/*
+ * VFIO_IOMMU_BIND - _IOWR(VFIO_TYPE, VFIO_BASE + 22, struct vfio_iommu_bind)
+ *
+ * Manage address spaces of devices in this container. Initially a TYPE1
+ * container can only have one address space, managed with
+ * VFIO_IOMMU_MAP/UNMAP_DMA.
+ *
+ * An IOMMU of type VFIO_TYPE1_NESTING_IOMMU can be managed by both MAP/UNMAP
+ * and BIND ioctls at the same time. MAP/UNMAP acts on the stage-2 (host) page
+ * tables, and BIND manages the stage-1 (guest) page tables. Other types of
+ * IOMMU may allow MAP/UNMAP and BIND to coexist, where MAP/UNMAP controls
+ * non-PASID traffic and BIND controls PASID traffic. But this depends on the
+ * underlying IOMMU architecture and isn't guaranteed.
+ *
+ * Availability of this feature depends on the device, its bus, the underlying
+ * IOMMU and the CPU architecture.
+ *
+ * returns: 0 on success, -errno on failure.
+ */
+#define VFIO_IOMMU_BIND _IO(VFIO_TYPE, VFIO_BASE + 22)
+
+/*
+ * VFIO_IOMMU_UNBIND - _IOWR(VFIO_TYPE, VFIO_BASE + 23, struct vfio_iommu_bind)
+ *
+ * Undo what was done by the corresponding VFIO_IOMMU_BIND ioctl.
+ */
+#define VFIO_IOMMU_UNBIND _IO(VFIO_TYPE, VFIO_BASE + 23)
+
/* -------- Additional API for SPAPR TCE (Server POWERPC) IOMMU -------- */
/*
--
2.13.3
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
