Commit 7e607460 authored by Johannes Berg's avatar Johannes Berg Committed by Richard Weinberger

um: virtio: Keep reading on -EAGAIN

When we get an interrupt from the socket getting readable,
and start reading, there's a possibility for a race. This
depends on the implementation of the device, but e.g. with
qemu's libvhost-user, we can see:

 device                 virtio_uml
---------------------------------------
  write header
                         get interrupt
                         read header
                         read body -> returns -EAGAIN
  write body

The -EAGAIN return is because the socket is non-blocking,
and then this leads us to abandon this message.

In fact, we've already read the header, so when the get
another signal/interrupt for the body, we again read it
as though it's a new message header, and also abandon it
for the same reason (wrong size etc.)

This essentially breaks things, and if that message was
one that required a response, it leads to a deadlock as
the device is waiting for the response but we'll never
reply.

Fix this by spinning on -EAGAIN as well when we read the
message body. We need to handle -EAGAIN as "no message"
while reading the header, since we share an interrupt.

Note that this situation is highly unlikely to occur in
normal usage, since there will be very few messages and
only in the startup phase. With the inband call feature
this does tend to happen (eventually) though.
Signed-off-by: default avatarJohannes Berg <johannes.berg@intel.com>
Signed-off-by: default avatarRichard Weinberger <richard@nod.at>
parent 04e5b1fb
...@@ -91,7 +91,7 @@ static int full_sendmsg_fds(int fd, const void *buf, unsigned int len, ...@@ -91,7 +91,7 @@ static int full_sendmsg_fds(int fd, const void *buf, unsigned int len,
return 0; return 0;
} }
static int full_read(int fd, void *buf, int len) static int full_read(int fd, void *buf, int len, bool abortable)
{ {
int rc; int rc;
...@@ -101,7 +101,7 @@ static int full_read(int fd, void *buf, int len) ...@@ -101,7 +101,7 @@ static int full_read(int fd, void *buf, int len)
buf += rc; buf += rc;
len -= rc; len -= rc;
} }
} while (len && (rc > 0 || rc == -EINTR)); } while (len && (rc > 0 || rc == -EINTR || (!abortable && rc == -EAGAIN)));
if (rc < 0) if (rc < 0)
return rc; return rc;
...@@ -112,7 +112,7 @@ static int full_read(int fd, void *buf, int len) ...@@ -112,7 +112,7 @@ static int full_read(int fd, void *buf, int len)
static int vhost_user_recv_header(int fd, struct vhost_user_msg *msg) static int vhost_user_recv_header(int fd, struct vhost_user_msg *msg)
{ {
return full_read(fd, msg, sizeof(msg->header)); return full_read(fd, msg, sizeof(msg->header), true);
} }
static int vhost_user_recv(struct virtio_uml_device *vu_dev, static int vhost_user_recv(struct virtio_uml_device *vu_dev,
...@@ -135,7 +135,7 @@ static int vhost_user_recv(struct virtio_uml_device *vu_dev, ...@@ -135,7 +135,7 @@ static int vhost_user_recv(struct virtio_uml_device *vu_dev,
size = msg->header.size; size = msg->header.size;
if (size > max_payload_size) if (size > max_payload_size)
return -EPROTO; return -EPROTO;
return full_read(fd, &msg->payload, size); return full_read(fd, &msg->payload, size, false);
} }
static int vhost_user_recv_resp(struct virtio_uml_device *vu_dev, static int vhost_user_recv_resp(struct virtio_uml_device *vu_dev,
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment