[ovs-dev] [PATCH] netlink-socket: Work around kernel Netlink dump thread races.

[Alex Wang]

Thx for the explanation, looks good to me,

And confirm that it solve the pkt duplication issue,

Acked-by: Alex Wang <alexw at nicira.com>


On Thu, Jul 10, 2014 at 4:03 PM, Ben Pfaff <blp at nicira.com> wrote:

> The Linux kernel Netlink implementation has two races that cause problems
> for processes that attempt to dump a table in a multithreaded manner.
>
> The first race is in the structure of the kernel netlink_recv() function.
> This function pulls a message from the socket queue and, if there is none,
> reports EAGAIN:
>         skb = skb_recv_datagram(sk, flags, noblock, &err);
>         if (skb == NULL)
>                 goto out;
> Only if a message is successfully read from the socket queue does the
> function, toward the end, try to queue up a new message to be dumped:
>         if (nlk->cb && atomic_read(&sk->sk_rmem_alloc) <= sk->sk_rcvbuf /
> 2) {
>                 ret = netlink_dump(sk);
>                 if (ret) {
>                         sk->sk_err = ret;
>                         sk->sk_error_report(sk);
>                 }
>         }
> This means that if thread A reads a message from a dump, then thread B
> attempts to read one before A queues up the next, B will get EAGAIN.  This
> means that, following EAGAIN, B needs to wait until A returns to userspace
> before it tries to read the socket again.  nl_dump_next() already does
> this, using 'dump->status_seq' (although the need for it has never been
> explained clearly, to my knowledge).
>
> The second race is more serious.  Suppose thread X and thread Y both
> simultaneously attempt to queue up a new message to be dumped, using the
> call to netlink_dump() quoted above.  netlink_dump() begins with:
>         mutex_lock(nlk->cb_mutex);
>
>         cb = nlk->cb;
>         if (cb == NULL) {
>                 err = -EINVAL;
>                 goto errout_skb;
>         }
> Suppose that X gets cb_mutex first and finds that the dump is complete.  It
> will therefore, toward the end of netlink_dump(), clear nlk->cb to NULL to
> indicate that no dump is in progress and release the mutex:
>         nlk->cb = NULL;
>         mutex_unlock(nlk->cb_mutex);
> When Y grabs cb_mutex afterward, it will see that nlk->cb is NULL and
> return -EINVAL as quoted above.  netlink_recv() stuffs -EINVAL in sk_err,
> but that error is not reported immediately; instead, it is saved for the
> next read from the socket.  Since Open vSwitch maintains a pool of Netlink
> sockets, that next failure can crop up pretty much anywhere.  One of the
> worst places for it to crop up is in the execution of a later transaction
> (e.g. in nl_sock_transact_multiple__()), because userspace treats Netlink
> transactions as idempotent and will re-execute them when socket errors
> occur.  For a transaction that sends a packet, this causes packet
> duplication, which we actually observed in practice.  (ENOBUFS should
> actually cause transactions to be re-executed in many cases, but EINVAL
> should not; this is a separate bug in the userspace netlink code.)
>
> VMware-BZ: #1283188
> Reported-and-tested-by: Alex Wang <alexw at nicira.com>
> Signed-off-by: Ben Pfaff <blp at nicira.com>
> ---
>  lib/netlink-socket.c |   10 ++++++++++
>  lib/netlink-socket.h |    2 ++
>  2 files changed, 12 insertions(+)
>
> diff --git a/lib/netlink-socket.c b/lib/netlink-socket.c
> index 42ba7e1..8f8d603 100644
> --- a/lib/netlink-socket.c
> +++ b/lib/netlink-socket.c
> @@ -714,6 +714,7 @@ nl_dump_start(struct nl_dump *dump, int protocol,
> const struct ofpbuf *request)
>      atomic_init(&dump->status, status << 1);
>      dump->nl_seq = nl_msg_nlmsghdr(request)->nlmsg_seq;
>      dump->status_seq = seq_create();
> +    ovs_mutex_init(&dump->mutex);
>  }
>
>  /* Attempts to retrieve another reply from 'dump' into 'buffer'. 'dump'
> must
> @@ -756,7 +757,15 @@ nl_dump_next(struct nl_dump *dump, struct ofpbuf
> *reply, struct ofpbuf *buffer)
>              return false;
>          }
>
> +        /* Take the mutex here to avoid a in-kernel race.  If two threads
> try
> +         * to read from a Netlink dump socket at once, then the socket
> error
> +         * can be set to EINVAL, which will be encountered on the next
> recv on
> +         * that socket, which could be anywhere due to the way that we
> pool
> +         * Netlink sockets.  Serializing the recv calls avoids the issue.
> */
> +        ovs_mutex_lock(&dump->mutex);
>          retval = nl_sock_recv__(dump->sock, buffer, false);
> +        ovs_mutex_unlock(&dump->mutex);
> +
>          if (retval) {
>              ofpbuf_clear(buffer);
>              if (retval == EAGAIN) {
> @@ -835,6 +844,7 @@ nl_dump_done(struct nl_dump *dump)
>      }
>      nl_pool_release(dump->sock);
>      seq_destroy(dump->status_seq);
> +    ovs_mutex_destroy(&dump->mutex);
>      return status >> 1;
>  }
>
> diff --git a/lib/netlink-socket.h b/lib/netlink-socket.h
> index dd32409..8ac201a 100644
> --- a/lib/netlink-socket.h
> +++ b/lib/netlink-socket.h
> @@ -52,6 +52,7 @@
>  #include <stdint.h>
>  #include "ofpbuf.h"
>  #include "ovs-atomic.h"
> +#include "ovs-thread.h"
>
>  struct nl_sock;
>
> @@ -114,6 +115,7 @@ struct nl_dump {
>      atomic_uint status;         /* Low bit set if we read final message.
>                                   * Other bits hold an errno (0 for
> success). */
>      struct seq *status_seq;     /* Tracks changes to the above 'status'.
> */
> +    struct ovs_mutex mutex;
>  };
>
>  void nl_dump_start(struct nl_dump *, int protocol,
> --
> 1.7.10.4
>
>