[Qemu-devel] [PATCH v3 0/4] 9pfs: handle transport errors

[Qemu-devel] [PATCH v3 0/4] 9pfs: handle transport errors

[Greg Kurz]

The 9p protocol relies on a reliable transport, but the current code
treats transport errors (ie, failure to marshal or unmarshal) as if
they were coming from the backend. This doesn't make sense: if the
transport failed, we should notify the guest that the transport is
broken and needs to be reset, using transport specific means.

This series only modifies the virtio transport so that it can notify
the guest about transport failures.

Changes since v2:
- complete rework

--
Greg

---

Greg Kurz (4):
      virtio-9p: record element after sanity checks
      virtio-9p: message header is 7-byte long
      virtio-9p: break device if buffers are misconfigured
      9pfs: handle transport errors in pdu_complete()


 hw/9pfs/9p.c               |   25 +++++++++++++-------
 hw/9pfs/9p.h               |    7 +++++-
 hw/9pfs/virtio-9p-device.c |   56 ++++++++++++++++++++++++++++++++++++--------
 hw/9pfs/xen-9p-backend.c   |    3 ++
 4 files changed, 70 insertions(+), 21 deletions(-)

[Qemu-devel] [PATCH v3 1/4] virtio-9p: record element after sanity checks

[Greg Kurz]

If the guest sends a malformed request, we end up with a dangling pointer
in V9fsVirtioState. This doesn't seem to cause any bug, but let's remove
this side effect anyway.

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/9pfs/virtio-9p-device.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
index 245abd8aaef1..3380bfc0c551 100644
--- a/hw/9pfs/virtio-9p-device.c
+++ b/hw/9pfs/virtio-9p-device.c
@@ -61,7 +61,6 @@ static void handle_9p_output(VirtIODevice *vdev, VirtQueue *vq)
         }
         QEMU_BUILD_BUG_ON(sizeof(out) != 7);
 
-        v->elems[pdu->idx] = elem;
         len = iov_to_buf(elem->out_sg, elem->out_num, 0,
                          &out, sizeof(out));
         if (len != sizeof(out)) {
@@ -70,6 +69,8 @@ static void handle_9p_output(VirtIODevice *vdev, VirtQueue *vq)
             goto out_free_req;
         }
 
+        v->elems[pdu->idx] = elem;
+
         pdu_submit(pdu, &out);
     }
 

[Qemu-devel] [PATCH v3 2/4] virtio-9p: message header is 7-byte long

[Greg Kurz]

The 9p spec at http://man.cat-v.org/plan_9/5/intro reads:

 "Each 9P message begins with a four-byte size field specify-
  ing the length in bytes of the complete message including
  the four bytes of the size field itself.  The next byte is
  the message type, one of the constants in the enumeration in
  the include file <fcall.h>.  The next two bytes are an iden-
  tifying tag, described below."

ie, each message starts with a 7-byte long header.

The core 9P code already assumes this pretty much everywhere. This patch
does the following:
- makes the assumption explicit in the common 9p.h header, since it isn't
  related to the transport
- open codes the header size in handle_9p_output() and hardens the sanity
  check on the space needed for the reply message

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/9pfs/9p.h               |    5 +++++
 hw/9pfs/virtio-9p-device.c |    8 +++-----
 2 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/hw/9pfs/9p.h b/hw/9pfs/9p.h
index c886ba78d2ee..aac1b0b2ce3d 100644
--- a/hw/9pfs/9p.h
+++ b/hw/9pfs/9p.h
@@ -124,6 +124,11 @@ typedef struct {
     uint8_t id;
     uint16_t tag_le;
 } QEMU_PACKED P9MsgHeader;
+/* According to the specification, 9p messages start with a 7-byte header.
+ * Since most of the code uses this header size in literal form, we must be
+ * sure this is indeed the case.
+ */
+QEMU_BUILD_BUG_ON(sizeof(P9MsgHeader) != 7);
 
 struct V9fsPDU
 {
diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
index 3380bfc0c551..1a68c1622d3a 100644
--- a/hw/9pfs/virtio-9p-device.c
+++ b/hw/9pfs/virtio-9p-device.c
@@ -53,17 +53,15 @@ static void handle_9p_output(VirtIODevice *vdev, VirtQueue *vq)
             goto out_free_pdu;
         }
 
-        if (elem->in_num == 0) {
+        if (iov_size(elem->in_sg, elem->in_num) < 7) {
             virtio_error(vdev,
                          "The guest sent a VirtFS request without space for "
                          "the reply");
             goto out_free_req;
         }
-        QEMU_BUILD_BUG_ON(sizeof(out) != 7);
 
-        len = iov_to_buf(elem->out_sg, elem->out_num, 0,
-                         &out, sizeof(out));
-        if (len != sizeof(out)) {
+        len = iov_to_buf(elem->out_sg, elem->out_num, 0, &out, 7);
+        if (len != 7) {
             virtio_error(vdev, "The guest sent a malformed VirtFS request: "
                          "header size is %zd, should be 7", len);
             goto out_free_req;

[Qemu-devel] [PATCH v3 3/4] virtio-9p: break device if buffers are misconfigured

[Greg Kurz]

The 9P protocol is transport agnostic: if the guest misconfigured the
buffers, the best we can do is to set the broken flag on the device.

We check if the transport isn't broken already to avoid printing extra
error messages.

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/9pfs/9p.c               |    2 +-
 hw/9pfs/9p.h               |    2 +-
 hw/9pfs/virtio-9p-device.c |   45 ++++++++++++++++++++++++++++++++++++++++----
 hw/9pfs/xen-9p-backend.c   |    3 ++-
 4 files changed, 45 insertions(+), 7 deletions(-)

diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
index 96d268334865..da0d6da65b45 100644
--- a/hw/9pfs/9p.c
+++ b/hw/9pfs/9p.c
@@ -1664,7 +1664,7 @@ static void v9fs_init_qiov_from_pdu(QEMUIOVector *qiov, V9fsPDU *pdu,
     unsigned int niov;
 
     if (is_write) {
-        pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov);
+        pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov, size + skip);
     } else {
         pdu->s->transport->init_in_iov_from_pdu(pdu, &iov, &niov, size + skip);
     }
diff --git a/hw/9pfs/9p.h b/hw/9pfs/9p.h
index aac1b0b2ce3d..d1cfeaf10e4f 100644
--- a/hw/9pfs/9p.h
+++ b/hw/9pfs/9p.h
@@ -363,7 +363,7 @@ struct V9fsTransport {
     void        (*init_in_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
                                         unsigned int *pniov, size_t size);
     void        (*init_out_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
-                                         unsigned int *pniov);
+                                         unsigned int *pniov, size_t size);
     void        (*push_and_notify)(V9fsPDU *pdu);
 };
 
diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
index 1a68c1622d3a..873f42e24818 100644
--- a/hw/9pfs/virtio-9p-device.c
+++ b/hw/9pfs/virtio-9p-device.c
@@ -146,8 +146,18 @@ static ssize_t virtio_pdu_vmarshal(V9fsPDU *pdu, size_t offset,
     V9fsState *s = pdu->s;
     V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
     VirtQueueElement *elem = v->elems[pdu->idx];
+    int ret;
 
-    return v9fs_iov_vmarshal(elem->in_sg, elem->in_num, offset, 1, fmt, ap);
+    ret = v9fs_iov_vmarshal(elem->in_sg, elem->in_num, offset, 1, fmt, ap);
+    if (ret < 0) {
+        VirtIODevice *vdev = VIRTIO_DEVICE(v);
+
+        if (!vdev->broken) {
+            virtio_error(vdev, "Failed to encode VirtFS reply type %d",
+                         pdu->id + 1);
+        }
+    }
+    return ret;
 }
 
 static ssize_t virtio_pdu_vunmarshal(V9fsPDU *pdu, size_t offset,
@@ -156,28 +166,55 @@ static ssize_t virtio_pdu_vunmarshal(V9fsPDU *pdu, size_t offset,
     V9fsState *s = pdu->s;
     V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
     VirtQueueElement *elem = v->elems[pdu->idx];
+    int ret;
 
-    return v9fs_iov_vunmarshal(elem->out_sg, elem->out_num, offset, 1, fmt, ap);
+    ret = v9fs_iov_vunmarshal(elem->out_sg, elem->out_num, offset, 1, fmt, ap);
+    if (ret < 0) {
+        VirtIODevice *vdev = VIRTIO_DEVICE(v);
+
+        if (!vdev->broken) {
+            virtio_error(vdev, "Failed to decode VirtFS request type %d",
+                         pdu->id);
+        }
+    }
+    return ret;
 }
 
-/* The size parameter is used by other transports. Do not drop it. */
 static void virtio_init_in_iov_from_pdu(V9fsPDU *pdu, struct iovec **piov,
                                         unsigned int *pniov, size_t size)
 {
     V9fsState *s = pdu->s;
     V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
     VirtQueueElement *elem = v->elems[pdu->idx];
+    size_t buf_size = iov_size(elem->in_sg, elem->in_num);
+
+    if (buf_size < size) {
+        VirtIODevice *vdev = VIRTIO_DEVICE(v);
+
+        virtio_error(vdev,
+                     "VirtFS reply type %d needs %zu bytes, buffer has %zu",
+                     pdu->id + 1, size, buf_size);
+    }
 
     *piov = elem->in_sg;
     *pniov = elem->in_num;
 }
 
 static void virtio_init_out_iov_from_pdu(V9fsPDU *pdu, struct iovec **piov,
-                                         unsigned int *pniov)
+                                         unsigned int *pniov, size_t size)
 {
     V9fsState *s = pdu->s;
     V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
     VirtQueueElement *elem = v->elems[pdu->idx];
+    size_t buf_size = iov_size(elem->out_sg, elem->out_num);
+
+    if (buf_size < size) {
+        VirtIODevice *vdev = VIRTIO_DEVICE(v);
+
+        virtio_error(vdev,
+                     "VirtFS request type %d needs %zu bytes, buffer has %zu",
+                     pdu->id, size, buf_size);
+    }
 
     *piov = elem->out_sg;
     *pniov = elem->out_num;
diff --git a/hw/9pfs/xen-9p-backend.c b/hw/9pfs/xen-9p-backend.c
index 922cc967be63..a82cf817fe45 100644
--- a/hw/9pfs/xen-9p-backend.c
+++ b/hw/9pfs/xen-9p-backend.c
@@ -147,7 +147,8 @@ static ssize_t xen_9pfs_pdu_vunmarshal(V9fsPDU *pdu,
 
 static void xen_9pfs_init_out_iov_from_pdu(V9fsPDU *pdu,
                                            struct iovec **piov,
-                                           unsigned int *pniov)
+                                           unsigned int *pniov,
+                                           size_t size)
 {
     Xen9pfsDev *xen_9pfs = container_of(pdu->s, Xen9pfsDev, state);
     Xen9pfsRing *ring = &xen_9pfs->rings[pdu->tag % xen_9pfs->num_rings];

[Qemu-devel] [PATCH v3 4/4] 9pfs: handle transport errors in pdu_complete()

[Greg Kurz]

Contrary to what is written in the comment, a buggy guest can misconfigure
the transport buffers and pdu_marshal() may return an error.  If this ever
happens, it is up to the transport layer to handle the situation (9P is
transport agnostic).

This fixes Coverity issue CID1348518.

Signed-off-by: Greg Kurz <groug@kaod.org>
---
 hw/9pfs/9p.c |   23 +++++++++++++++--------
 1 file changed, 15 insertions(+), 8 deletions(-)

diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
index da0d6da65b45..f7871e6f3e5a 100644
--- a/hw/9pfs/9p.c
+++ b/hw/9pfs/9p.c
@@ -624,15 +624,11 @@ void pdu_free(V9fsPDU *pdu)
     QLIST_INSERT_HEAD(&s->free_list, pdu, next);
 }
 
-/*
- * We don't do error checking for pdu_marshal/unmarshal here
- * because we always expect to have enough space to encode
- * error details
- */
 static void coroutine_fn pdu_complete(V9fsPDU *pdu, ssize_t len)
 {
     int8_t id = pdu->id + 1; /* Response */
     V9fsState *s = pdu->s;
+    int ret;
 
     if (len < 0) {
         int err = -len;
@@ -644,11 +640,19 @@ static void coroutine_fn pdu_complete(V9fsPDU *pdu, ssize_t len)
             str.data = strerror(err);
             str.size = strlen(str.data);
 
-            len += pdu_marshal(pdu, len, "s", &str);
+            ret = pdu_marshal(pdu, len, "s", &str);
+            if (ret < 0) {
+                goto out_notify;
+            }
+            len += ret;
             id = P9_RERROR;
         }
 
-        len += pdu_marshal(pdu, len, "d", err);
+        ret = pdu_marshal(pdu, len, "d", err);
+        if (ret < 0) {
+            goto out_notify;
+        }
+        len += ret;
 
         if (s->proto_version == V9FS_PROTO_2000L) {
             id = P9_RLERROR;
@@ -657,12 +661,15 @@ static void coroutine_fn pdu_complete(V9fsPDU *pdu, ssize_t len)
     }
 
     /* fill out the header */
-    pdu_marshal(pdu, 0, "dbw", (int32_t)len, id, pdu->tag);
+    if (pdu_marshal(pdu, 0, "dbw", (int32_t)len, id, pdu->tag) < 0) {
+        goto out_notify;
+    }
 
     /* keep these in sync */
     pdu->size = len;
     pdu->id = id;
 
+out_notify:
     pdu->s->transport->push_and_notify(pdu);
 
     /* Now wakeup anybody waiting in flush for this request */

Re: [Qemu-devel] [PATCH v3 1/4] virtio-9p: record element after sanity checks

[Michael S. Tsirkin]

On Wed, Jun 21, 2017 at 07:41:58PM +0200, Greg Kurz wrote:
> If the guest sends a malformed request, we end up with a dangling pointer
> in V9fsVirtioState. This doesn't seem to cause any bug, but let's remove
> this side effect anyway.
> 
> Signed-off-by: Greg Kurz <groug@kaod.org>

Reviewed-by: Michael S. Tsirkin <mst@redhat.com>

> ---
>  hw/9pfs/virtio-9p-device.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
> index 245abd8aaef1..3380bfc0c551 100644
> --- a/hw/9pfs/virtio-9p-device.c
> +++ b/hw/9pfs/virtio-9p-device.c
> @@ -61,7 +61,6 @@ static void handle_9p_output(VirtIODevice *vdev, VirtQueue *vq)
>          }
>          QEMU_BUILD_BUG_ON(sizeof(out) != 7);
>  
> -        v->elems[pdu->idx] = elem;
>          len = iov_to_buf(elem->out_sg, elem->out_num, 0,
>                           &out, sizeof(out));
>          if (len != sizeof(out)) {
> @@ -70,6 +69,8 @@ static void handle_9p_output(VirtIODevice *vdev, VirtQueue *vq)
>              goto out_free_req;
>          }
>  
> +        v->elems[pdu->idx] = elem;
> +
>          pdu_submit(pdu, &out);
>      }
>  

Re: [Qemu-devel] [PATCH v3 3/4] virtio-9p: break device if buffers are misconfigured

[Michael S. Tsirkin]

On Wed, Jun 21, 2017 at 07:42:15PM +0200, Greg Kurz wrote:
> The 9P protocol is transport agnostic: if the guest misconfigured the
> buffers, the best we can do is to set the broken flag on the device.
> 
> We check if the transport isn't broken already to avoid printing extra
> error messages.
> 
> Signed-off-by: Greg Kurz <groug@kaod.org>
> ---
>  hw/9pfs/9p.c               |    2 +-
>  hw/9pfs/9p.h               |    2 +-
>  hw/9pfs/virtio-9p-device.c |   45 ++++++++++++++++++++++++++++++++++++++++----
>  hw/9pfs/xen-9p-backend.c   |    3 ++-
>  4 files changed, 45 insertions(+), 7 deletions(-)
> 
> diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
> index 96d268334865..da0d6da65b45 100644
> --- a/hw/9pfs/9p.c
> +++ b/hw/9pfs/9p.c
> @@ -1664,7 +1664,7 @@ static void v9fs_init_qiov_from_pdu(QEMUIOVector *qiov, V9fsPDU *pdu,
>      unsigned int niov;
>  
>      if (is_write) {
> -        pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov);
> +        pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov, size + skip);
>      } else {
>          pdu->s->transport->init_in_iov_from_pdu(pdu, &iov, &niov, size + skip);
>      }
> diff --git a/hw/9pfs/9p.h b/hw/9pfs/9p.h
> index aac1b0b2ce3d..d1cfeaf10e4f 100644
> --- a/hw/9pfs/9p.h
> +++ b/hw/9pfs/9p.h
> @@ -363,7 +363,7 @@ struct V9fsTransport {
>      void        (*init_in_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
>                                          unsigned int *pniov, size_t size);
>      void        (*init_out_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
> -                                         unsigned int *pniov);
> +                                         unsigned int *pniov, size_t size);
>      void        (*push_and_notify)(V9fsPDU *pdu);
>  };
>  
> diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
> index 1a68c1622d3a..873f42e24818 100644
> --- a/hw/9pfs/virtio-9p-device.c
> +++ b/hw/9pfs/virtio-9p-device.c
> @@ -146,8 +146,18 @@ static ssize_t virtio_pdu_vmarshal(V9fsPDU *pdu, size_t offset,
>      V9fsState *s = pdu->s;
>      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
>      VirtQueueElement *elem = v->elems[pdu->idx];
> +    int ret;
>  
> -    return v9fs_iov_vmarshal(elem->in_sg, elem->in_num, offset, 1, fmt, ap);
> +    ret = v9fs_iov_vmarshal(elem->in_sg, elem->in_num, offset, 1, fmt, ap);
> +    if (ret < 0) {
> +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> +
> +        if (!vdev->broken) {
> +            virtio_error(vdev, "Failed to encode VirtFS reply type %d",
> +                         pdu->id + 1);
> +        }
> +    }
> +    return ret;
>  }
>  
>  static ssize_t virtio_pdu_vunmarshal(V9fsPDU *pdu, size_t offset,
> @@ -156,28 +166,55 @@ static ssize_t virtio_pdu_vunmarshal(V9fsPDU *pdu, size_t offset,
>      V9fsState *s = pdu->s;
>      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
>      VirtQueueElement *elem = v->elems[pdu->idx];
> +    int ret;
>  
> -    return v9fs_iov_vunmarshal(elem->out_sg, elem->out_num, offset, 1, fmt, ap);
> +    ret = v9fs_iov_vunmarshal(elem->out_sg, elem->out_num, offset, 1, fmt, ap);
> +    if (ret < 0) {
> +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> +
> +        if (!vdev->broken) {

Do we really need this test? And why?

> +            virtio_error(vdev, "Failed to decode VirtFS request type %d",
> +                         pdu->id);
> +        }
> +    }
> +    return ret;
>  }
>  
> -/* The size parameter is used by other transports. Do not drop it. */
>  static void virtio_init_in_iov_from_pdu(V9fsPDU *pdu, struct iovec **piov,
>                                          unsigned int *pniov, size_t size)
>  {
>      V9fsState *s = pdu->s;
>      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
>      VirtQueueElement *elem = v->elems[pdu->idx];
> +    size_t buf_size = iov_size(elem->in_sg, elem->in_num);
> +
> +    if (buf_size < size) {
> +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> +
> +        virtio_error(vdev,
> +                     "VirtFS reply type %d needs %zu bytes, buffer has %zu",
> +                     pdu->id + 1, size, buf_size);
> +    }
>  
>      *piov = elem->in_sg;
>      *pniov = elem->in_num;
>  }
>  
>  static void virtio_init_out_iov_from_pdu(V9fsPDU *pdu, struct iovec **piov,
> -                                         unsigned int *pniov)
> +                                         unsigned int *pniov, size_t size)
>  {
>      V9fsState *s = pdu->s;
>      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
>      VirtQueueElement *elem = v->elems[pdu->idx];
> +    size_t buf_size = iov_size(elem->out_sg, elem->out_num);
> +
> +    if (buf_size < size) {
> +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> +
> +        virtio_error(vdev,
> +                     "VirtFS request type %d needs %zu bytes, buffer has %zu",
> +                     pdu->id, size, buf_size);
> +    }
>  
>      *piov = elem->out_sg;
>      *pniov = elem->out_num;
> diff --git a/hw/9pfs/xen-9p-backend.c b/hw/9pfs/xen-9p-backend.c
> index 922cc967be63..a82cf817fe45 100644
> --- a/hw/9pfs/xen-9p-backend.c
> +++ b/hw/9pfs/xen-9p-backend.c
> @@ -147,7 +147,8 @@ static ssize_t xen_9pfs_pdu_vunmarshal(V9fsPDU *pdu,
>  
>  static void xen_9pfs_init_out_iov_from_pdu(V9fsPDU *pdu,
>                                             struct iovec **piov,
> -                                           unsigned int *pniov)
> +                                           unsigned int *pniov,
> +                                           size_t size)
>  {
>      Xen9pfsDev *xen_9pfs = container_of(pdu->s, Xen9pfsDev, state);
>      Xen9pfsRing *ring = &xen_9pfs->rings[pdu->tag % xen_9pfs->num_rings];

Re: [Qemu-devel] [PATCH v3 3/4] virtio-9p: break device if buffers are misconfigured

[Greg Kurz]

[-- Attachment #1: Type: text/plain, Size: 6454 bytes --]

On Thu, 22 Jun 2017 01:09:24 +0300
"Michael S. Tsirkin" <mst@redhat.com> wrote:

> On Wed, Jun 21, 2017 at 07:42:15PM +0200, Greg Kurz wrote:
> > The 9P protocol is transport agnostic: if the guest misconfigured the
> > buffers, the best we can do is to set the broken flag on the device.
> > 
> > We check if the transport isn't broken already to avoid printing extra
> > error messages.
> > 
> > Signed-off-by: Greg Kurz <groug@kaod.org>
> > ---
> >  hw/9pfs/9p.c               |    2 +-
> >  hw/9pfs/9p.h               |    2 +-
> >  hw/9pfs/virtio-9p-device.c |   45 ++++++++++++++++++++++++++++++++++++++++----
> >  hw/9pfs/xen-9p-backend.c   |    3 ++-
> >  4 files changed, 45 insertions(+), 7 deletions(-)
> > 
> > diff --git a/hw/9pfs/9p.c b/hw/9pfs/9p.c
> > index 96d268334865..da0d6da65b45 100644
> > --- a/hw/9pfs/9p.c
> > +++ b/hw/9pfs/9p.c
> > @@ -1664,7 +1664,7 @@ static void v9fs_init_qiov_from_pdu(QEMUIOVector *qiov, V9fsPDU *pdu,
> >      unsigned int niov;
> >  
> >      if (is_write) {
> > -        pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov);
> > +        pdu->s->transport->init_out_iov_from_pdu(pdu, &iov, &niov, size + skip);
> >      } else {
> >          pdu->s->transport->init_in_iov_from_pdu(pdu, &iov, &niov, size + skip);
> >      }
> > diff --git a/hw/9pfs/9p.h b/hw/9pfs/9p.h
> > index aac1b0b2ce3d..d1cfeaf10e4f 100644
> > --- a/hw/9pfs/9p.h
> > +++ b/hw/9pfs/9p.h
> > @@ -363,7 +363,7 @@ struct V9fsTransport {
> >      void        (*init_in_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
> >                                          unsigned int *pniov, size_t size);
> >      void        (*init_out_iov_from_pdu)(V9fsPDU *pdu, struct iovec **piov,
> > -                                         unsigned int *pniov);
> > +                                         unsigned int *pniov, size_t size);
> >      void        (*push_and_notify)(V9fsPDU *pdu);
> >  };
> >  
> > diff --git a/hw/9pfs/virtio-9p-device.c b/hw/9pfs/virtio-9p-device.c
> > index 1a68c1622d3a..873f42e24818 100644
> > --- a/hw/9pfs/virtio-9p-device.c
> > +++ b/hw/9pfs/virtio-9p-device.c
> > @@ -146,8 +146,18 @@ static ssize_t virtio_pdu_vmarshal(V9fsPDU *pdu, size_t offset,
> >      V9fsState *s = pdu->s;
> >      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
> >      VirtQueueElement *elem = v->elems[pdu->idx];
> > +    int ret;
> >  
> > -    return v9fs_iov_vmarshal(elem->in_sg, elem->in_num, offset, 1, fmt, ap);
> > +    ret = v9fs_iov_vmarshal(elem->in_sg, elem->in_num, offset, 1, fmt, ap);
> > +    if (ret < 0) {
> > +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> > +
> > +        if (!vdev->broken) {
> > +            virtio_error(vdev, "Failed to encode VirtFS reply type %d",
> > +                         pdu->id + 1);
> > +        }
> > +    }
> > +    return ret;
> >  }
> >  
> >  static ssize_t virtio_pdu_vunmarshal(V9fsPDU *pdu, size_t offset,
> > @@ -156,28 +166,55 @@ static ssize_t virtio_pdu_vunmarshal(V9fsPDU *pdu, size_t offset,
> >      V9fsState *s = pdu->s;
> >      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
> >      VirtQueueElement *elem = v->elems[pdu->idx];
> > +    int ret;
> >  
> > -    return v9fs_iov_vunmarshal(elem->out_sg, elem->out_num, offset, 1, fmt, ap);
> > +    ret = v9fs_iov_vunmarshal(elem->out_sg, elem->out_num, offset, 1, fmt, ap);
> > +    if (ret < 0) {
> > +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> > +
> > +        if (!vdev->broken) {  
> 
> Do we really need this test? And why?
> 

Indeed, you're right. We need this for marshal since several active PDUs
may try to send data back to the client. But we definitely don't need it
for unmarshal since we only submit one PDU at a time.

I'll remove it.

> > +            virtio_error(vdev, "Failed to decode VirtFS request type %d",
> > +                         pdu->id);
> > +        }
> > +    }
> > +    return ret;
> >  }
> >  
> > -/* The size parameter is used by other transports. Do not drop it. */
> >  static void virtio_init_in_iov_from_pdu(V9fsPDU *pdu, struct iovec **piov,
> >                                          unsigned int *pniov, size_t size)
> >  {
> >      V9fsState *s = pdu->s;
> >      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
> >      VirtQueueElement *elem = v->elems[pdu->idx];
> > +    size_t buf_size = iov_size(elem->in_sg, elem->in_num);
> > +
> > +    if (buf_size < size) {
> > +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> > +
> > +        virtio_error(vdev,
> > +                     "VirtFS reply type %d needs %zu bytes, buffer has %zu",
> > +                     pdu->id + 1, size, buf_size);
> > +    }
> >  
> >      *piov = elem->in_sg;
> >      *pniov = elem->in_num;
> >  }
> >  
> >  static void virtio_init_out_iov_from_pdu(V9fsPDU *pdu, struct iovec **piov,
> > -                                         unsigned int *pniov)
> > +                                         unsigned int *pniov, size_t size)
> >  {
> >      V9fsState *s = pdu->s;
> >      V9fsVirtioState *v = container_of(s, V9fsVirtioState, state);
> >      VirtQueueElement *elem = v->elems[pdu->idx];
> > +    size_t buf_size = iov_size(elem->out_sg, elem->out_num);
> > +
> > +    if (buf_size < size) {
> > +        VirtIODevice *vdev = VIRTIO_DEVICE(v);
> > +
> > +        virtio_error(vdev,
> > +                     "VirtFS request type %d needs %zu bytes, buffer has %zu",
> > +                     pdu->id, size, buf_size);
> > +    }
> >  
> >      *piov = elem->out_sg;
> >      *pniov = elem->out_num;
> > diff --git a/hw/9pfs/xen-9p-backend.c b/hw/9pfs/xen-9p-backend.c
> > index 922cc967be63..a82cf817fe45 100644
> > --- a/hw/9pfs/xen-9p-backend.c
> > +++ b/hw/9pfs/xen-9p-backend.c
> > @@ -147,7 +147,8 @@ static ssize_t xen_9pfs_pdu_vunmarshal(V9fsPDU *pdu,
> >  
> >  static void xen_9pfs_init_out_iov_from_pdu(V9fsPDU *pdu,
> >                                             struct iovec **piov,
> > -                                           unsigned int *pniov)
> > +                                           unsigned int *pniov,
> > +                                           size_t size)
> >  {
> >      Xen9pfsDev *xen_9pfs = container_of(pdu->s, Xen9pfsDev, state);
> >      Xen9pfsRing *ring = &xen_9pfs->rings[pdu->tag % xen_9pfs->num_rings];  


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]