Merge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
Pull vhost infrastructure updates from Michael S. Tsirkin:
"This reworks vhost core dropping unnecessary RCU uses in favor of VQ
mutexes which are used on fast path anyway. This fixes worst-case
latency for users which change the memory mappings a lot. Memory
allocation for vhost-net now supports fallback on vmalloc (same as for
vhost-scsi) this makes it possible to create the device on systems
where memory is very fragmented, with slightly lower performance"
* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
vhost: move memory pointer to VQs
vhost: move acked_features to VQs
vhost: replace rcu with mutex
vhost-net: extend device allocation to vmalloc
diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index be414d2..971a760 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c
@@ -17,6 +17,7 @@
#include <linux/workqueue.h>
#include <linux/file.h>
#include <linux/slab.h>
+#include <linux/vmalloc.h>
#include <linux/net.h>
#include <linux/if_packet.h>
@@ -373,7 +374,7 @@
% UIO_MAXIOV == nvq->done_idx))
break;
- head = vhost_get_vq_desc(&net->dev, vq, vq->iov,
+ head = vhost_get_vq_desc(vq, vq->iov,
ARRAY_SIZE(vq->iov),
&out, &in,
NULL, NULL);
@@ -505,7 +506,7 @@
r = -ENOBUFS;
goto err;
}
- r = vhost_get_vq_desc(vq->dev, vq, vq->iov + seg,
+ r = vhost_get_vq_desc(vq, vq->iov + seg,
ARRAY_SIZE(vq->iov) - seg, &out,
&in, log, log_num);
if (unlikely(r < 0))
@@ -584,9 +585,9 @@
vhost_hlen = nvq->vhost_hlen;
sock_hlen = nvq->sock_hlen;
- vq_log = unlikely(vhost_has_feature(&net->dev, VHOST_F_LOG_ALL)) ?
+ vq_log = unlikely(vhost_has_feature(vq, VHOST_F_LOG_ALL)) ?
vq->log : NULL;
- mergeable = vhost_has_feature(&net->dev, VIRTIO_NET_F_MRG_RXBUF);
+ mergeable = vhost_has_feature(vq, VIRTIO_NET_F_MRG_RXBUF);
while ((sock_len = peek_head_len(sock->sk))) {
sock_len += sock_hlen;
@@ -699,18 +700,30 @@
handle_rx(net);
}
+static void vhost_net_free(void *addr)
+{
+ if (is_vmalloc_addr(addr))
+ vfree(addr);
+ else
+ kfree(addr);
+}
+
static int vhost_net_open(struct inode *inode, struct file *f)
{
- struct vhost_net *n = kmalloc(sizeof *n, GFP_KERNEL);
+ struct vhost_net *n;
struct vhost_dev *dev;
struct vhost_virtqueue **vqs;
int i;
- if (!n)
- return -ENOMEM;
+ n = kmalloc(sizeof *n, GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT);
+ if (!n) {
+ n = vmalloc(sizeof *n);
+ if (!n)
+ return -ENOMEM;
+ }
vqs = kmalloc(VHOST_NET_VQ_MAX * sizeof(*vqs), GFP_KERNEL);
if (!vqs) {
- kfree(n);
+ vhost_net_free(n);
return -ENOMEM;
}
@@ -827,7 +840,7 @@
* since jobs can re-queue themselves. */
vhost_net_flush(n);
kfree(n->dev.vqs);
- kfree(n);
+ vhost_net_free(n);
return 0;
}
@@ -1038,15 +1051,13 @@
mutex_unlock(&n->dev.mutex);
return -EFAULT;
}
- n->dev.acked_features = features;
- smp_wmb();
for (i = 0; i < VHOST_NET_VQ_MAX; ++i) {
mutex_lock(&n->vqs[i].vq.mutex);
+ n->vqs[i].vq.acked_features = features;
n->vqs[i].vhost_hlen = vhost_hlen;
n->vqs[i].sock_hlen = sock_hlen;
mutex_unlock(&n->vqs[i].vq.mutex);
}
- vhost_net_flush(n);
mutex_unlock(&n->dev.mutex);
return 0;
}
diff --git a/drivers/vhost/scsi.c b/drivers/vhost/scsi.c
index aeb5131..e9c280f 100644
--- a/drivers/vhost/scsi.c
+++ b/drivers/vhost/scsi.c
@@ -606,7 +606,7 @@
again:
vhost_disable_notify(&vs->dev, vq);
- head = vhost_get_vq_desc(&vs->dev, vq, vq->iov,
+ head = vhost_get_vq_desc(vq, vq->iov,
ARRAY_SIZE(vq->iov), &out, &in,
NULL, NULL);
if (head < 0) {
@@ -945,7 +945,7 @@
vhost_disable_notify(&vs->dev, vq);
for (;;) {
- head = vhost_get_vq_desc(&vs->dev, vq, vq->iov,
+ head = vhost_get_vq_desc(vq, vq->iov,
ARRAY_SIZE(vq->iov), &out, &in,
NULL, NULL);
pr_debug("vhost_get_vq_desc: head: %d, out: %u in: %u\n",
@@ -1373,6 +1373,9 @@
static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)
{
+ struct vhost_virtqueue *vq;
+ int i;
+
if (features & ~VHOST_SCSI_FEATURES)
return -EOPNOTSUPP;
@@ -1382,9 +1385,13 @@
mutex_unlock(&vs->dev.mutex);
return -EFAULT;
}
- vs->dev.acked_features = features;
- smp_wmb();
- vhost_scsi_flush(vs);
+
+ for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
+ vq = &vs->vqs[i].vq;
+ mutex_lock(&vq->mutex);
+ vq->acked_features = features;
+ mutex_unlock(&vq->mutex);
+ }
mutex_unlock(&vs->dev.mutex);
return 0;
}
@@ -1591,10 +1598,6 @@
return;
mutex_lock(&vs->dev.mutex);
- if (!vhost_has_feature(&vs->dev, VIRTIO_SCSI_F_HOTPLUG)) {
- mutex_unlock(&vs->dev.mutex);
- return;
- }
if (plug)
reason = VIRTIO_SCSI_EVT_RESET_RESCAN;
@@ -1603,8 +1606,9 @@
vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
mutex_lock(&vq->mutex);
- tcm_vhost_send_evt(vs, tpg, lun,
- VIRTIO_SCSI_T_TRANSPORT_RESET, reason);
+ if (vhost_has_feature(vq, VIRTIO_SCSI_F_HOTPLUG))
+ tcm_vhost_send_evt(vs, tpg, lun,
+ VIRTIO_SCSI_T_TRANSPORT_RESET, reason);
mutex_unlock(&vq->mutex);
mutex_unlock(&vs->dev.mutex);
}
diff --git a/drivers/vhost/test.c b/drivers/vhost/test.c
index c2a54fb..d9c501e 100644
--- a/drivers/vhost/test.c
+++ b/drivers/vhost/test.c
@@ -53,7 +53,7 @@
vhost_disable_notify(&n->dev, vq);
for (;;) {
- head = vhost_get_vq_desc(&n->dev, vq, vq->iov,
+ head = vhost_get_vq_desc(vq, vq->iov,
ARRAY_SIZE(vq->iov),
&out, &in,
NULL, NULL);
@@ -241,15 +241,18 @@
static int vhost_test_set_features(struct vhost_test *n, u64 features)
{
+ struct vhost_virtqueue *vq;
+
mutex_lock(&n->dev.mutex);
if ((features & (1 << VHOST_F_LOG_ALL)) &&
!vhost_log_access_ok(&n->dev)) {
mutex_unlock(&n->dev.mutex);
return -EFAULT;
}
- n->dev.acked_features = features;
- smp_wmb();
- vhost_test_flush(n);
+ vq = &n->vqs[VHOST_TEST_VQ];
+ mutex_lock(&vq->mutex);
+ vq->acked_features = features;
+ mutex_unlock(&vq->mutex);
mutex_unlock(&n->dev.mutex);
return 0;
}
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 78987e4..c90f437 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -18,7 +18,6 @@
#include <linux/mmu_context.h>
#include <linux/miscdevice.h>
#include <linux/mutex.h>
-#include <linux/rcupdate.h>
#include <linux/poll.h>
#include <linux/file.h>
#include <linux/highmem.h>
@@ -191,6 +190,7 @@
vq->log_used = false;
vq->log_addr = -1ull;
vq->private_data = NULL;
+ vq->acked_features = 0;
vq->log_base = NULL;
vq->error_ctx = NULL;
vq->error = NULL;
@@ -198,6 +198,7 @@
vq->call_ctx = NULL;
vq->call = NULL;
vq->log_ctx = NULL;
+ vq->memory = NULL;
}
static int vhost_worker(void *data)
@@ -415,11 +416,18 @@
/* Caller should have device mutex */
void vhost_dev_reset_owner(struct vhost_dev *dev, struct vhost_memory *memory)
{
+ int i;
+
vhost_dev_cleanup(dev, true);
/* Restore memory to default empty mapping. */
memory->nregions = 0;
- RCU_INIT_POINTER(dev->memory, memory);
+ dev->memory = memory;
+ /* We don't need VQ locks below since vhost_dev_cleanup makes sure
+ * VQs aren't running.
+ */
+ for (i = 0; i < dev->nvqs; ++i)
+ dev->vqs[i]->memory = memory;
}
EXPORT_SYMBOL_GPL(vhost_dev_reset_owner);
@@ -462,10 +470,8 @@
fput(dev->log_file);
dev->log_file = NULL;
/* No one will access memory at this point */
- kfree(rcu_dereference_protected(dev->memory,
- locked ==
- lockdep_is_held(&dev->mutex)));
- RCU_INIT_POINTER(dev->memory, NULL);
+ kfree(dev->memory);
+ dev->memory = NULL;
WARN_ON(!list_empty(&dev->work_list));
if (dev->worker) {
kthread_stop(dev->worker);
@@ -524,11 +530,13 @@
for (i = 0; i < d->nvqs; ++i) {
int ok;
+ bool log;
+
mutex_lock(&d->vqs[i]->mutex);
+ log = log_all || vhost_has_feature(d->vqs[i], VHOST_F_LOG_ALL);
/* If ring is inactive, will check when it's enabled. */
if (d->vqs[i]->private_data)
- ok = vq_memory_access_ok(d->vqs[i]->log_base, mem,
- log_all);
+ ok = vq_memory_access_ok(d->vqs[i]->log_base, mem, log);
else
ok = 1;
mutex_unlock(&d->vqs[i]->mutex);
@@ -538,12 +546,12 @@
return 1;
}
-static int vq_access_ok(struct vhost_dev *d, unsigned int num,
+static int vq_access_ok(struct vhost_virtqueue *vq, unsigned int num,
struct vring_desc __user *desc,
struct vring_avail __user *avail,
struct vring_used __user *used)
{
- size_t s = vhost_has_feature(d, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
+ size_t s = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
return access_ok(VERIFY_READ, desc, num * sizeof *desc) &&
access_ok(VERIFY_READ, avail,
sizeof *avail + num * sizeof *avail->ring + s) &&
@@ -555,26 +563,19 @@
/* Caller should have device mutex but not vq mutex */
int vhost_log_access_ok(struct vhost_dev *dev)
{
- struct vhost_memory *mp;
-
- mp = rcu_dereference_protected(dev->memory,
- lockdep_is_held(&dev->mutex));
- return memory_access_ok(dev, mp, 1);
+ return memory_access_ok(dev, dev->memory, 1);
}
EXPORT_SYMBOL_GPL(vhost_log_access_ok);
/* Verify access for write logging. */
/* Caller should have vq mutex and device mutex */
-static int vq_log_access_ok(struct vhost_dev *d, struct vhost_virtqueue *vq,
+static int vq_log_access_ok(struct vhost_virtqueue *vq,
void __user *log_base)
{
- struct vhost_memory *mp;
- size_t s = vhost_has_feature(d, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
+ size_t s = vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
- mp = rcu_dereference_protected(vq->dev->memory,
- lockdep_is_held(&vq->mutex));
- return vq_memory_access_ok(log_base, mp,
- vhost_has_feature(vq->dev, VHOST_F_LOG_ALL)) &&
+ return vq_memory_access_ok(log_base, vq->memory,
+ vhost_has_feature(vq, VHOST_F_LOG_ALL)) &&
(!vq->log_used || log_access_ok(log_base, vq->log_addr,
sizeof *vq->used +
vq->num * sizeof *vq->used->ring + s));
@@ -584,8 +585,8 @@
/* Caller should have vq mutex and device mutex */
int vhost_vq_access_ok(struct vhost_virtqueue *vq)
{
- return vq_access_ok(vq->dev, vq->num, vq->desc, vq->avail, vq->used) &&
- vq_log_access_ok(vq->dev, vq, vq->log_base);
+ return vq_access_ok(vq, vq->num, vq->desc, vq->avail, vq->used) &&
+ vq_log_access_ok(vq, vq->log_base);
}
EXPORT_SYMBOL_GPL(vhost_vq_access_ok);
@@ -593,6 +594,7 @@
{
struct vhost_memory mem, *newmem, *oldmem;
unsigned long size = offsetof(struct vhost_memory, regions);
+ int i;
if (copy_from_user(&mem, m, size))
return -EFAULT;
@@ -611,15 +613,19 @@
return -EFAULT;
}
- if (!memory_access_ok(d, newmem,
- vhost_has_feature(d, VHOST_F_LOG_ALL))) {
+ if (!memory_access_ok(d, newmem, 0)) {
kfree(newmem);
return -EFAULT;
}
- oldmem = rcu_dereference_protected(d->memory,
- lockdep_is_held(&d->mutex));
- rcu_assign_pointer(d->memory, newmem);
- synchronize_rcu();
+ oldmem = d->memory;
+ d->memory = newmem;
+
+ /* All memory accesses are done under some VQ mutex. */
+ for (i = 0; i < d->nvqs; ++i) {
+ mutex_lock(&d->vqs[i]->mutex);
+ d->vqs[i]->memory = newmem;
+ mutex_unlock(&d->vqs[i]->mutex);
+ }
kfree(oldmem);
return 0;
}
@@ -718,7 +724,7 @@
* If it is not, we don't as size might not have been setup.
* We will verify when backend is configured. */
if (vq->private_data) {
- if (!vq_access_ok(d, vq->num,
+ if (!vq_access_ok(vq, vq->num,
(void __user *)(unsigned long)a.desc_user_addr,
(void __user *)(unsigned long)a.avail_user_addr,
(void __user *)(unsigned long)a.used_user_addr)) {
@@ -858,7 +864,7 @@
vq = d->vqs[i];
mutex_lock(&vq->mutex);
/* If ring is inactive, will check when it's enabled. */
- if (vq->private_data && !vq_log_access_ok(d, vq, base))
+ if (vq->private_data && !vq_log_access_ok(vq, base))
r = -EFAULT;
else
vq->log_base = base;
@@ -1044,7 +1050,7 @@
}
EXPORT_SYMBOL_GPL(vhost_init_used);
-static int translate_desc(struct vhost_dev *dev, u64 addr, u32 len,
+static int translate_desc(struct vhost_virtqueue *vq, u64 addr, u32 len,
struct iovec iov[], int iov_size)
{
const struct vhost_memory_region *reg;
@@ -1053,9 +1059,7 @@
u64 s = 0;
int ret = 0;
- rcu_read_lock();
-
- mem = rcu_dereference(dev->memory);
+ mem = vq->memory;
while ((u64)len > s) {
u64 size;
if (unlikely(ret >= iov_size)) {
@@ -1077,7 +1081,6 @@
++ret;
}
- rcu_read_unlock();
return ret;
}
@@ -1102,7 +1105,7 @@
return next;
}
-static int get_indirect(struct vhost_dev *dev, struct vhost_virtqueue *vq,
+static int get_indirect(struct vhost_virtqueue *vq,
struct iovec iov[], unsigned int iov_size,
unsigned int *out_num, unsigned int *in_num,
struct vhost_log *log, unsigned int *log_num,
@@ -1121,7 +1124,7 @@
return -EINVAL;
}
- ret = translate_desc(dev, indirect->addr, indirect->len, vq->indirect,
+ ret = translate_desc(vq, indirect->addr, indirect->len, vq->indirect,
UIO_MAXIOV);
if (unlikely(ret < 0)) {
vq_err(vq, "Translation failure %d in indirect.\n", ret);
@@ -1161,7 +1164,7 @@
return -EINVAL;
}
- ret = translate_desc(dev, desc.addr, desc.len, iov + iov_count,
+ ret = translate_desc(vq, desc.addr, desc.len, iov + iov_count,
iov_size - iov_count);
if (unlikely(ret < 0)) {
vq_err(vq, "Translation failure %d indirect idx %d\n",
@@ -1198,7 +1201,7 @@
* This function returns the descriptor number found, or vq->num (which is
* never a valid descriptor number) if none was found. A negative code is
* returned on error. */
-int vhost_get_vq_desc(struct vhost_dev *dev, struct vhost_virtqueue *vq,
+int vhost_get_vq_desc(struct vhost_virtqueue *vq,
struct iovec iov[], unsigned int iov_size,
unsigned int *out_num, unsigned int *in_num,
struct vhost_log *log, unsigned int *log_num)
@@ -1272,7 +1275,7 @@
return -EFAULT;
}
if (desc.flags & VRING_DESC_F_INDIRECT) {
- ret = get_indirect(dev, vq, iov, iov_size,
+ ret = get_indirect(vq, iov, iov_size,
out_num, in_num,
log, log_num, &desc);
if (unlikely(ret < 0)) {
@@ -1283,7 +1286,7 @@
continue;
}
- ret = translate_desc(dev, desc.addr, desc.len, iov + iov_count,
+ ret = translate_desc(vq, desc.addr, desc.len, iov + iov_count,
iov_size - iov_count);
if (unlikely(ret < 0)) {
vq_err(vq, "Translation failure %d descriptor idx %d\n",
@@ -1426,11 +1429,11 @@
* interrupts. */
smp_mb();
- if (vhost_has_feature(dev, VIRTIO_F_NOTIFY_ON_EMPTY) &&
+ if (vhost_has_feature(vq, VIRTIO_F_NOTIFY_ON_EMPTY) &&
unlikely(vq->avail_idx == vq->last_avail_idx))
return true;
- if (!vhost_has_feature(dev, VIRTIO_RING_F_EVENT_IDX)) {
+ if (!vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX)) {
__u16 flags;
if (__get_user(flags, &vq->avail->flags)) {
vq_err(vq, "Failed to get flags");
@@ -1491,7 +1494,7 @@
if (!(vq->used_flags & VRING_USED_F_NO_NOTIFY))
return false;
vq->used_flags &= ~VRING_USED_F_NO_NOTIFY;
- if (!vhost_has_feature(dev, VIRTIO_RING_F_EVENT_IDX)) {
+ if (!vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX)) {
r = vhost_update_used_flags(vq);
if (r) {
vq_err(vq, "Failed to enable notification at %p: %d\n",
@@ -1528,7 +1531,7 @@
if (vq->used_flags & VRING_USED_F_NO_NOTIFY)
return;
vq->used_flags |= VRING_USED_F_NO_NOTIFY;
- if (!vhost_has_feature(dev, VIRTIO_RING_F_EVENT_IDX)) {
+ if (!vhost_has_feature(vq, VIRTIO_RING_F_EVENT_IDX)) {
r = vhost_update_used_flags(vq);
if (r)
vq_err(vq, "Failed to enable notification at %p: %d\n",
diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index 35eeb2a..3eda654 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -104,20 +104,18 @@
struct iovec *indirect;
struct vring_used_elem *heads;
/* Protected by virtqueue mutex. */
+ struct vhost_memory *memory;
void *private_data;
+ unsigned acked_features;
/* Log write descriptors */
void __user *log_base;
struct vhost_log *log;
};
struct vhost_dev {
- /* Readers use RCU to access memory table pointer
- * log base pointer and features.
- * Writers use mutex below.*/
- struct vhost_memory __rcu *memory;
+ struct vhost_memory *memory;
struct mm_struct *mm;
struct mutex mutex;
- unsigned acked_features;
struct vhost_virtqueue **vqs;
int nvqs;
struct file *log_file;
@@ -140,7 +138,7 @@
int vhost_vq_access_ok(struct vhost_virtqueue *vq);
int vhost_log_access_ok(struct vhost_dev *);
-int vhost_get_vq_desc(struct vhost_dev *, struct vhost_virtqueue *,
+int vhost_get_vq_desc(struct vhost_virtqueue *,
struct iovec iov[], unsigned int iov_count,
unsigned int *out_num, unsigned int *in_num,
struct vhost_log *log, unsigned int *log_num);
@@ -174,13 +172,8 @@
(1ULL << VHOST_F_LOG_ALL),
};
-static inline int vhost_has_feature(struct vhost_dev *dev, int bit)
+static inline int vhost_has_feature(struct vhost_virtqueue *vq, int bit)
{
- unsigned acked_features;
-
- /* TODO: check that we are running from vhost_worker or dev mutex is
- * held? */
- acked_features = rcu_dereference_index_check(dev->acked_features, 1);
- return acked_features & (1 << bit);
+ return vq->acked_features & (1 << bit);
}
#endif