Re: [Qemu-devel] [RFC PATCH 1/8] virtio: convert to use DMA api

["Michael S. Tsirkin" <mst@redhat.com>]

On Fri, Mar 25, 2016 at 10:13:22AM +0800, Jason Wang wrote:
> Currently, all virtio devices bypass IOMMU completely. This is because
> address_space_memory is assumed and used during DMA emulation. This
> patch converts the virtio core API to use DMA API. This idea is
> 
> - introducing a new transport specific helper to query the dma address
>   space. (only pci version is implemented).
> - query and use this address space during virtio device guest memory
>   accessing
> 
> With this virtio devices will not bypass IOMMU anymore. Tested with
> intel_iommu=on/strict with:
> 
> - virtio guest DMA series posted in https://lkml.org/lkml/2015/10/28/64.
> - vfio (unsafe interrupt mode) dpdk l2fwd in guest
> 
> TODO:
> - Feature bit for this
> - Implement this for all transports

Nice. The only thing that worries me here is that ring
addresses are only translated once at DRIVER_OK.

In theory, rings could be non-contigious and
mapped to contigious ranges of bus addresses by
the IOMMU. Might be useful for very large rings
or memory-constrained guests.

Thoughts? Is this worth worrying about?

> Cc: Michael S. Tsirkin <mst@redhat.com>
> Cc: Stefan Hajnoczi <stefanha@redhat.com>
> Cc: Kevin Wolf <kwolf@redhat.com>
> Cc: Amit Shah <amit.shah@redhat.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: qemu-block@nongnu.org
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  hw/block/virtio-blk.c             |  2 +-
>  hw/char/virtio-serial-bus.c       |  3 +-
>  hw/scsi/virtio-scsi.c             |  4 ++-
>  hw/virtio/virtio-pci.c            |  9 ++++++
>  hw/virtio/virtio.c                | 58 +++++++++++++++++++++++----------------
>  include/hw/virtio/virtio-access.h | 42 +++++++++++++++++++++-------
>  include/hw/virtio/virtio-bus.h    |  1 +
>  include/hw/virtio/virtio.h        |  4 +--
>  8 files changed, 85 insertions(+), 38 deletions(-)
> 
> diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
> index cb710f1..9411f99 100644
> --- a/hw/block/virtio-blk.c
> +++ b/hw/block/virtio-blk.c
> @@ -829,7 +829,7 @@ static int virtio_blk_load_device(VirtIODevice *vdev, QEMUFile *f,
>  
>      while (qemu_get_sbyte(f)) {
>          VirtIOBlockReq *req;
> -        req = qemu_get_virtqueue_element(f, sizeof(VirtIOBlockReq));
> +        req = qemu_get_virtqueue_element(vdev, f, sizeof(VirtIOBlockReq));
>          virtio_blk_init_request(s, req);
>          req->next = s->rq;
>          s->rq = req;
> diff --git a/hw/char/virtio-serial-bus.c b/hw/char/virtio-serial-bus.c
> index 99cb683..bdc5393 100644
> --- a/hw/char/virtio-serial-bus.c
> +++ b/hw/char/virtio-serial-bus.c
> @@ -687,6 +687,7 @@ static void virtio_serial_post_load_timer_cb(void *opaque)
>  static int fetch_active_ports_list(QEMUFile *f, int version_id,
>                                     VirtIOSerial *s, uint32_t nr_active_ports)
>  {
> +    VirtIODevice *vdev = VIRTIO_DEVICE(s);
>      uint32_t i;
>  
>      s->post_load = g_malloc0(sizeof(*s->post_load));
> @@ -722,7 +723,7 @@ static int fetch_active_ports_list(QEMUFile *f, int version_id,
>                  qemu_get_be64s(f, &port->iov_offset);
>  
>                  port->elem =
> -                    qemu_get_virtqueue_element(f, sizeof(VirtQueueElement));
> +                    qemu_get_virtqueue_element(vdev, f, sizeof(VirtQueueElement));
>  
>                  /*
>                   *  Port was throttled on source machine.  Let's
> diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
> index 0c30d2e..26ce701 100644
> --- a/hw/scsi/virtio-scsi.c
> +++ b/hw/scsi/virtio-scsi.c
> @@ -196,12 +196,14 @@ static void *virtio_scsi_load_request(QEMUFile *f, SCSIRequest *sreq)
>      SCSIBus *bus = sreq->bus;
>      VirtIOSCSI *s = container_of(bus, VirtIOSCSI, bus);
>      VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(s);
> +    VirtIODevice *vdev = VIRTIO_DEVICE(s);
>      VirtIOSCSIReq *req;
>      uint32_t n;
>  
>      qemu_get_be32s(f, &n);
>      assert(n < vs->conf.num_queues);
> -    req = qemu_get_virtqueue_element(f, sizeof(VirtIOSCSIReq) + vs->cdb_size);
> +    req = qemu_get_virtqueue_element(vdev, f,
> +                                     sizeof(VirtIOSCSIReq) + vs->cdb_size);
>      virtio_scsi_init_req(s, vs->cmd_vqs[n], req);
>  
>      if (virtio_scsi_parse_req(req, sizeof(VirtIOSCSICmdReq) + vs->cdb_size,
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index 0dadb66..5508b1c 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -1211,6 +1211,14 @@ static int virtio_pci_query_nvectors(DeviceState *d)
>      return proxy->nvectors;
>  }
>  
> +static AddressSpace *virtio_pci_get_dma_as(DeviceState *d)
> +{
> +    VirtIOPCIProxy *proxy = VIRTIO_PCI(d);
> +    PCIDevice *dev = &proxy->pci_dev;
> +
> +    return pci_get_address_space(dev);
> +}
> +
>  static int virtio_pci_add_mem_cap(VirtIOPCIProxy *proxy,
>                                     struct virtio_pci_cap *cap)
>  {
> @@ -2495,6 +2503,7 @@ static void virtio_pci_bus_class_init(ObjectClass *klass, void *data)
>      k->device_plugged = virtio_pci_device_plugged;
>      k->device_unplugged = virtio_pci_device_unplugged;
>      k->query_nvectors = virtio_pci_query_nvectors;
> +    k->get_dma_as = virtio_pci_get_dma_as;
>  }
>  
>  static const TypeInfo virtio_pci_bus_info = {
> diff --git a/hw/virtio/virtio.c b/hw/virtio/virtio.c
> index 08275a9..37c9951 100644
> --- a/hw/virtio/virtio.c
> +++ b/hw/virtio/virtio.c
> @@ -21,6 +21,7 @@
>  #include "hw/virtio/virtio-bus.h"
>  #include "migration/migration.h"
>  #include "hw/virtio/virtio-access.h"
> +#include "sysemu/dma.h"
>  
>  /*
>   * The alignment to use between consumer and producer parts of vring.
> @@ -118,7 +119,7 @@ void virtio_queue_update_rings(VirtIODevice *vdev, int n)
>  static void vring_desc_read(VirtIODevice *vdev, VRingDesc *desc,
>                              hwaddr desc_pa, int i)
>  {
> -    address_space_read(&address_space_memory, desc_pa + i * sizeof(VRingDesc),
> +    address_space_read(virtio_get_dma_as(vdev), desc_pa + i * sizeof(VRingDesc),
>                         MEMTXATTRS_UNSPECIFIED, (void *)desc, sizeof(VRingDesc));
>      virtio_tswap64s(vdev, &desc->addr);
>      virtio_tswap32s(vdev, &desc->len);
> @@ -160,7 +161,7 @@ static inline void vring_used_write(VirtQueue *vq, VRingUsedElem *uelem,
>      virtio_tswap32s(vq->vdev, &uelem->id);
>      virtio_tswap32s(vq->vdev, &uelem->len);
>      pa = vq->vring.used + offsetof(VRingUsed, ring[i]);
> -    address_space_write(&address_space_memory, pa, MEMTXATTRS_UNSPECIFIED,
> +    address_space_write(virtio_get_dma_as(vq->vdev), pa, MEMTXATTRS_UNSPECIFIED,
>                         (void *)uelem, sizeof(VRingUsedElem));
>  }
>  
> @@ -240,6 +241,7 @@ int virtio_queue_empty(VirtQueue *vq)
>  static void virtqueue_unmap_sg(VirtQueue *vq, const VirtQueueElement *elem,
>                                 unsigned int len)
>  {
> +    AddressSpace *dma_as = virtio_get_dma_as(vq->vdev);
>      unsigned int offset;
>      int i;
>  
> @@ -247,17 +249,17 @@ static void virtqueue_unmap_sg(VirtQueue *vq, const VirtQueueElement *elem,
>      for (i = 0; i < elem->in_num; i++) {
>          size_t size = MIN(len - offset, elem->in_sg[i].iov_len);
>  
> -        cpu_physical_memory_unmap(elem->in_sg[i].iov_base,
> -                                  elem->in_sg[i].iov_len,
> -                                  1, size);
> +        dma_memory_unmap(dma_as, elem->in_sg[i].iov_base, elem->in_sg[i].iov_len,
> +                         DMA_DIRECTION_FROM_DEVICE, size);
>  
>          offset += size;
>      }
>  
>      for (i = 0; i < elem->out_num; i++)
> -        cpu_physical_memory_unmap(elem->out_sg[i].iov_base,
> -                                  elem->out_sg[i].iov_len,
> -                                  0, elem->out_sg[i].iov_len);
> +        dma_memory_unmap(dma_as, elem->out_sg[i].iov_base,
> +                         elem->out_sg[i].iov_len,
> +                         DMA_DIRECTION_TO_DEVICE,
> +                         elem->out_sg[i].iov_len);
>  }
>  
>  void virtqueue_discard(VirtQueue *vq, const VirtQueueElement *elem,
> @@ -447,7 +449,8 @@ int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes,
>      return in_bytes <= in_total && out_bytes <= out_total;
>  }
>  
> -static void virtqueue_map_desc(unsigned int *p_num_sg, hwaddr *addr, struct iovec *iov,
> +static void virtqueue_map_desc(VirtIODevice *vdev,
> +                               unsigned int *p_num_sg, hwaddr *addr, struct iovec *iov,
>                                 unsigned int max_num_sg, bool is_write,
>                                 hwaddr pa, size_t sz)
>  {
> @@ -462,7 +465,10 @@ static void virtqueue_map_desc(unsigned int *p_num_sg, hwaddr *addr, struct iove
>              exit(1);
>          }
>  
> -        iov[num_sg].iov_base = cpu_physical_memory_map(pa, &len, is_write);
> +        iov[num_sg].iov_base = dma_memory_map(virtio_get_dma_as(vdev), pa, &len,
> +                                              is_write ?
> +                                              DMA_DIRECTION_FROM_DEVICE:
> +                                              DMA_DIRECTION_TO_DEVICE);
>          iov[num_sg].iov_len = len;
>          addr[num_sg] = pa;
>  
> @@ -473,9 +479,9 @@ static void virtqueue_map_desc(unsigned int *p_num_sg, hwaddr *addr, struct iove
>      *p_num_sg = num_sg;
>  }
>  
> -static void virtqueue_map_iovec(struct iovec *sg, hwaddr *addr,
> -                                unsigned int *num_sg, unsigned int max_size,
> -                                int is_write)
> +static void virtqueue_map_iovec(VirtIODevice *vdev, struct iovec *sg,
> +                                hwaddr *addr, unsigned int *num_sg,
> +                                unsigned int max_size, int is_write)
>  {
>      unsigned int i;
>      hwaddr len;
> @@ -494,7 +500,10 @@ static void virtqueue_map_iovec(struct iovec *sg, hwaddr *addr,
>  
>      for (i = 0; i < *num_sg; i++) {
>          len = sg[i].iov_len;
> -        sg[i].iov_base = cpu_physical_memory_map(addr[i], &len, is_write);
> +        sg[i].iov_base = dma_memory_map(virtio_get_dma_as(vdev),
> +                                        addr[i], &len, is_write ?
> +                                        DMA_DIRECTION_FROM_DEVICE :
> +                                        DMA_DIRECTION_TO_DEVICE);
>          if (!sg[i].iov_base) {
>              error_report("virtio: error trying to map MMIO memory");
>              exit(1);
> @@ -506,12 +515,15 @@ static void virtqueue_map_iovec(struct iovec *sg, hwaddr *addr,
>      }
>  }
>  
> -void virtqueue_map(VirtQueueElement *elem)
> +void virtqueue_map(VirtIODevice *vdev, VirtQueueElement *elem)
>  {
> -    virtqueue_map_iovec(elem->in_sg, elem->in_addr, &elem->in_num,
> -                        VIRTQUEUE_MAX_SIZE, 1);
> -    virtqueue_map_iovec(elem->out_sg, elem->out_addr, &elem->out_num,
> -                        VIRTQUEUE_MAX_SIZE, 0);
> +    virtqueue_map_iovec(vdev, elem->in_sg, elem->in_addr, &elem->in_num,
> +                        MIN(ARRAY_SIZE(elem->in_sg), ARRAY_SIZE(elem->in_addr)),
> +                        1);
> +    virtqueue_map_iovec(vdev, elem->out_sg, elem->out_addr, &elem->out_num,
> +                        MIN(ARRAY_SIZE(elem->out_sg),
> +                        ARRAY_SIZE(elem->out_addr)),
> +                        0);
>  }
>  
>  void *virtqueue_alloc_element(size_t sz, unsigned out_num, unsigned in_num)
> @@ -580,14 +592,14 @@ void *virtqueue_pop(VirtQueue *vq, size_t sz)
>      /* Collect all the descriptors */
>      do {
>          if (desc.flags & VRING_DESC_F_WRITE) {
> -            virtqueue_map_desc(&in_num, addr + out_num, iov + out_num,
> +            virtqueue_map_desc(vdev, &in_num, addr + out_num, iov + out_num,
>                                 VIRTQUEUE_MAX_SIZE - out_num, true, desc.addr, desc.len);
>          } else {
>              if (in_num) {
>                  error_report("Incorrect order for descriptors");
>                  exit(1);
>              }
> -            virtqueue_map_desc(&out_num, addr, iov,
> +            virtqueue_map_desc(vdev, &out_num, addr, iov,
>                                 VIRTQUEUE_MAX_SIZE, false, desc.addr, desc.len);
>          }
>  
> @@ -633,7 +645,7 @@ typedef struct VirtQueueElementOld {
>      struct iovec out_sg[VIRTQUEUE_MAX_SIZE];
>  } VirtQueueElementOld;
>  
> -void *qemu_get_virtqueue_element(QEMUFile *f, size_t sz)
> +void *qemu_get_virtqueue_element(VirtIODevice *vdev, QEMUFile *f, size_t sz)
>  {
>      VirtQueueElement *elem;
>      VirtQueueElementOld data;
> @@ -664,7 +676,7 @@ void *qemu_get_virtqueue_element(QEMUFile *f, size_t sz)
>          elem->out_sg[i].iov_len = data.out_sg[i].iov_len;
>      }
>  
> -    virtqueue_map(elem);
> +    virtqueue_map(vdev, elem);
>      return elem;
>  }
>  
> diff --git a/include/hw/virtio/virtio-access.h b/include/hw/virtio/virtio-access.h
> index 8dc84f5..967cc75 100644
> --- a/include/hw/virtio/virtio-access.h
> +++ b/include/hw/virtio/virtio-access.h
> @@ -15,8 +15,20 @@
>  #ifndef _QEMU_VIRTIO_ACCESS_H
>  #define _QEMU_VIRTIO_ACCESS_H
>  #include "hw/virtio/virtio.h"
> +#include "hw/virtio/virtio-bus.h"
>  #include "exec/address-spaces.h"
>  
> +static inline AddressSpace *virtio_get_dma_as(VirtIODevice *vdev)
> +{
> +    BusState *qbus = qdev_get_parent_bus(DEVICE(vdev));
> +    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
> +
> +    if (k->get_dma_as) {
> +        return k->get_dma_as(qbus->parent);
> +    }
> +    return &address_space_memory;
> +}
> +
>  static inline bool virtio_access_is_big_endian(VirtIODevice *vdev)
>  {
>  #if defined(TARGET_IS_BIENDIAN)
> @@ -34,45 +46,55 @@ static inline bool virtio_access_is_big_endian(VirtIODevice *vdev)
>  
>  static inline uint16_t virtio_lduw_phys(VirtIODevice *vdev, hwaddr pa)
>  {
> +    AddressSpace *dma_as = virtio_get_dma_as(vdev);
> +
>      if (virtio_access_is_big_endian(vdev)) {
> -        return lduw_be_phys(&address_space_memory, pa);
> +        return lduw_be_phys(dma_as, pa);
>      }
> -    return lduw_le_phys(&address_space_memory, pa);
> +    return lduw_le_phys(dma_as, pa);
>  }
>  
>  static inline uint32_t virtio_ldl_phys(VirtIODevice *vdev, hwaddr pa)
>  {
> +    AddressSpace *dma_as = virtio_get_dma_as(vdev);
> +
>      if (virtio_access_is_big_endian(vdev)) {
> -        return ldl_be_phys(&address_space_memory, pa);
> +        return ldl_be_phys(dma_as, pa);
>      }
> -    return ldl_le_phys(&address_space_memory, pa);
> +    return ldl_le_phys(dma_as, pa);
>  }
>  
>  static inline uint64_t virtio_ldq_phys(VirtIODevice *vdev, hwaddr pa)
>  {
> +    AddressSpace *dma_as = virtio_get_dma_as(vdev);
> +
>      if (virtio_access_is_big_endian(vdev)) {
> -        return ldq_be_phys(&address_space_memory, pa);
> +        return ldq_be_phys(dma_as, pa);
>      }
> -    return ldq_le_phys(&address_space_memory, pa);
> +    return ldq_le_phys(dma_as, pa);
>  }
>  
>  static inline void virtio_stw_phys(VirtIODevice *vdev, hwaddr pa,
>                                     uint16_t value)
>  {
> +    AddressSpace *dma_as = virtio_get_dma_as(vdev);
> +
>      if (virtio_access_is_big_endian(vdev)) {
> -        stw_be_phys(&address_space_memory, pa, value);
> +        stw_be_phys(dma_as, pa, value);
>      } else {
> -        stw_le_phys(&address_space_memory, pa, value);
> +        stw_le_phys(dma_as, pa, value);
>      }
>  }
>  
>  static inline void virtio_stl_phys(VirtIODevice *vdev, hwaddr pa,
>                                     uint32_t value)
>  {
> +    AddressSpace *dma_as = virtio_get_dma_as(vdev);
> +
>      if (virtio_access_is_big_endian(vdev)) {
> -        stl_be_phys(&address_space_memory, pa, value);
> +        stl_be_phys(dma_as, pa, value);
>      } else {
> -        stl_le_phys(&address_space_memory, pa, value);
> +        stl_le_phys(dma_as, pa, value);
>      }
>  }
>  
> diff --git a/include/hw/virtio/virtio-bus.h b/include/hw/virtio/virtio-bus.h
> index 3f2c136..17c07af 100644
> --- a/include/hw/virtio/virtio-bus.h
> +++ b/include/hw/virtio/virtio-bus.h
> @@ -76,6 +76,7 @@ typedef struct VirtioBusClass {
>       * Note that changing this will break migration for this transport.
>       */
>      bool has_variable_vring_alignment;
> +    AddressSpace *(*get_dma_as)(DeviceState *d);
>  } VirtioBusClass;
>  
>  struct VirtioBusState {
> diff --git a/include/hw/virtio/virtio.h b/include/hw/virtio/virtio.h
> index 2b5b248..0908bf6 100644
> --- a/include/hw/virtio/virtio.h
> +++ b/include/hw/virtio/virtio.h
> @@ -153,9 +153,9 @@ void virtqueue_discard(VirtQueue *vq, const VirtQueueElement *elem,
>  void virtqueue_fill(VirtQueue *vq, const VirtQueueElement *elem,
>                      unsigned int len, unsigned int idx);
>  
> -void virtqueue_map(VirtQueueElement *elem);
> +void virtqueue_map(VirtIODevice *vdev, VirtQueueElement *elem);
>  void *virtqueue_pop(VirtQueue *vq, size_t sz);
> -void *qemu_get_virtqueue_element(QEMUFile *f, size_t sz);
> +void *qemu_get_virtqueue_element(VirtIODevice *vdev, QEMUFile *f, size_t sz);
>  void qemu_put_virtqueue_element(QEMUFile *f, VirtQueueElement *elem);
>  int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes,
>                            unsigned int out_bytes);
> -- 
> 2.5.0