qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
To: "Daniel P. Berrange" <berrange@redhat.com>
Cc: Amit Shah <amit.shah@redhat.com>,
	qemu-devel@nongnu.org, Juan Quintela <quintela@redhat.com>
Subject: Re: [Qemu-devel] [PATCH v1 10/22] migration: convert tcp socket protocol to use QIOChannel
Date: Wed, 3 Feb 2016 10:33:10 +0000	[thread overview]
Message-ID: <20160203103309.GG2376@work-vm> (raw)
In-Reply-To: <20160203100217.GA30222@redhat.com>

* Daniel P. Berrange (berrange@redhat.com) wrote:
> On Tue, Feb 02, 2016 at 06:19:03PM +0000, Dr. David Alan Gilbert wrote:
> > * Daniel P. Berrange (berrange@redhat.com) wrote:
> > > Convert the tcp socket migration protocol driver to use
> > > QIOChannel and QEMUFileChannel, instead of plain sockets
> > > APIs.
> > > 
> > > While this now looks pretty similar to the migration/unix.c
> > > file from the previous patch, it was decided not to merge
> > > the two, because when TLS is added to the TCP impl later,
> > > this file diverge from unix.c once again.
> > 
> > Hmm OK, although I'd kind of like to see merging, but lets
> > see the TLS code later in the series....
> > 
> > > Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
> > > ---
> > >  migration/tcp.c | 119 ++++++++++++++++++++++++++++++++++++++------------------
> > >  1 file changed, 82 insertions(+), 37 deletions(-)
> > > 
> > > diff --git a/migration/tcp.c b/migration/tcp.c
> > > index ae89172..ac73977 100644
> > > --- a/migration/tcp.c
> > > +++ b/migration/tcp.c
> > > @@ -2,9 +2,11 @@
> > >   * QEMU live migration
> > >   *
> > >   * Copyright IBM, Corp. 2008
> > > + * Copyright Red Hat, Inc. 2015
> > >   *
> > >   * Authors:
> > >   *  Anthony Liguori   <aliguori@us.ibm.com>
> > > + *  Daniel P. Berrange <berrange@redhat.com>
> > >   *
> > >   * This work is licensed under the terms of the GNU GPL, version 2.  See
> > >   * the COPYING file in the top-level directory.
> > > @@ -17,11 +19,9 @@
> > >  
> > >  #include "qemu-common.h"
> > >  #include "qemu/error-report.h"
> > > -#include "qemu/sockets.h"
> > >  #include "migration/migration.h"
> > >  #include "migration/qemu-file.h"
> > > -#include "block/block.h"
> > > -#include "qemu/main-loop.h"
> > > +#include "io/channel-socket.h"
> > >  
> > >  //#define DEBUG_MIGRATION_TCP
> > >  
> > > @@ -33,71 +33,116 @@
> > >      do { } while (0)
> > >  #endif
> > >  
> > > -static void tcp_wait_for_connect(int fd, Error *err, void *opaque)
> > > +
> > > +static SocketAddress *tcp_build_address(const char *host_port, Error **errp)
> > > +{
> > > +    InetSocketAddress *iaddr = inet_parse(host_port, errp);
> > > +    SocketAddress *saddr;
> > > +
> > > +    if (!iaddr) {
> > > +        return NULL;
> > > +    }
> > > +
> > > +    saddr = g_new0(SocketAddress, 1);
> > > +    saddr->type = SOCKET_ADDRESS_KIND_INET;
> > > +    saddr->u.inet = iaddr;
> > > +
> > > +    return saddr;
> > > +}
> > > +
> > > +
> > > +static void tcp_outgoing_migration(Object *src,
> > > +                                   Error *err,
> > > +                                   gpointer opaque)
> > >  {
> > >      MigrationState *s = opaque;
> > > +    QIOChannel *sioc = QIO_CHANNEL(src);
> > >  
> > > -    if (fd < 0) {
> > > +    if (err) {
> > >          DPRINTF("migrate connect error: %s\n", error_get_pretty(err));
> > >          s->file = NULL;
> > >          migrate_fd_error(s);
> > >      } else {
> > >          DPRINTF("migrate connect success\n");
> > > -        s->file = qemu_fopen_socket(fd, "wb");
> > > +        s->file = qemu_fopen_channel_output(sioc);
> > >          migrate_fd_connect(s);
> > >      }
> > > +    object_unref(src);
> > >  }
> > >  
> > > -void tcp_start_outgoing_migration(MigrationState *s, const char *host_port, Error **errp)
> > > +
> > > +void tcp_start_outgoing_migration(MigrationState *s,
> > > +                                  const char *host_port,
> > > +                                  Error **errp)
> > >  {
> > > -    inet_nonblocking_connect(host_port, tcp_wait_for_connect, s, errp);
> > > +    SocketAddress *saddr = tcp_build_address(host_port, errp);
> > > +    QIOChannelSocket *sioc;
> > > +
> > > +    if (!saddr) {
> > > +        return;
> > > +    }
> > > +
> > > +    sioc = qio_channel_socket_new();
> > > +    qio_channel_socket_connect_async(sioc,
> > > +                                     saddr,
> > > +                                     tcp_outgoing_migration,
> > > +                                     s,
> > > +                                     NULL);
> > > +    qapi_free_SocketAddress(saddr);
> > >  }
> > >  
> > > -static void tcp_accept_incoming_migration(void *opaque)
> > > +
> > > +static gboolean tcp_accept_incoming_migration(QIOChannel *ioc,
> > > +                                              GIOCondition condition,
> > > +                                              gpointer opaque)
> > >  {
> > > -    struct sockaddr_in addr;
> > > -    socklen_t addrlen = sizeof(addr);
> > > -    int s = (intptr_t)opaque;
> > >      QEMUFile *f;
> > > -    int c, err;
> > > +    QIOChannelSocket *cioc;
> > > +    Error *err = NULL;
> > >  
> > > -    do {
> > > -        c = qemu_accept(s, (struct sockaddr *)&addr, &addrlen);
> > > -        err = socket_error();
> > > -    } while (c < 0 && err == EINTR);
> > > -    qemu_set_fd_handler(s, NULL, NULL, NULL);
> > > -    closesocket(s);
> > > -
> > > -    DPRINTF("accepted migration\n");
> > > -
> > > -    if (c < 0) {
> > > +    cioc = qio_channel_socket_accept(QIO_CHANNEL_SOCKET(ioc),
> > > +                                     &err);
> > > +    if (!cioc) {
> > >          error_report("could not accept migration connection (%s)",
> > > -                     strerror(err));
> > > -        return;
> > > -    }
> > > -
> > > -    f = qemu_fopen_socket(c, "rb");
> > > -    if (f == NULL) {
> > > -        error_report("could not qemu_fopen socket");
> > > +                     error_get_pretty(err));
> > >          goto out;
> > >      }
> > >  
> > > +    DPRINTF("accepted migration\n");
> > > +
> > > +    f = qemu_fopen_channel_input(QIO_CHANNEL(cioc));
> > > +    object_unref(OBJECT(cioc));
> > > +
> > >      process_incoming_migration(f);
> > > -    return;
> > >  
> > >  out:
> > > -    closesocket(c);
> > > +    /* Close listening socket as its no longer needed */
> > > +    qio_channel_close(ioc, NULL);
> > > +    return FALSE;
> > >  }
> > >  
> > > +
> > >  void tcp_start_incoming_migration(const char *host_port, Error **errp)
> > >  {
> > > -    int s;
> > > +    SocketAddress *saddr = tcp_build_address(host_port, errp);
> > > +    QIOChannelSocket *listen_ioc;
> > >  
> > > -    s = inet_listen(host_port, NULL, 256, SOCK_STREAM, 0, errp);
> > > -    if (s < 0) {
> > > +    if (!saddr) {
> > >          return;
> > >      }
> > >  
> > > -    qemu_set_fd_handler(s, tcp_accept_incoming_migration, NULL,
> > > -                        (void *)(intptr_t)s);
> > > +    listen_ioc = qio_channel_socket_new();
> > > +    if (qio_channel_socket_listen_sync(listen_ioc, saddr, errp) < 0) {
> > 
> > In this case, although weird, that could block couldn't it?
> > The case I'm thinking of is using the migrate_incoming command with a hostname
> > (I've used it with hostnames before as aliases for networks to listen on,
> > admittedly it be rare for it to be DNS that would block).
> 
> Yes it certainly could block and this is undesirable in general. The problem
> is that the only way we can provide errors back for the monitor reponse is
> if we do this synchronously from this method.  If I ran async then we'd have
> no ability to report errors about non-existant host names, already in use tcp
> port, etc.
> 
> I'm not really sure what todo about this long term, but the only way would
> be if the monitor response was able to be sent back asychronously from
> this method call. Then we can start an async listen call, and respond to
> the monitor when done.
> 
> Since the current code being replaced is also blocking, I felt solving
> that problem is best left to a future patch series. At least with this
> explicit method naming with _sync() suffix we can easily identify the
> problems in future

Yeh that's fair enough.  Failed incoming migrations generally just quit anyway.

Dave

> 
> Regards,
> Daniel
> -- 
> |: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
> |: http://libvirt.org              -o-             http://virt-manager.org :|
> |: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
> |: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

  reply	other threads:[~2016-02-03 10:33 UTC|newest]

Thread overview: 66+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-01-12 11:43 [Qemu-devel] [PATCH v1 00/22] Convert migration to QIOChannel & support TLS Daniel P. Berrange
2016-01-12 11:43 ` [Qemu-devel] [PATCH v1 01/22] s390: use FILE instead of QEMUFile for creating text file Daniel P. Berrange
2016-01-12 11:58   ` Cornelia Huck
2016-01-12 12:01     ` Daniel P. Berrange
2016-01-12 12:05       ` Cornelia Huck
2016-02-12 17:19   ` Dr. David Alan Gilbert
2016-01-12 11:43 ` [Qemu-devel] [PATCH v1 02/22] migration: remove use of qemu_bufopen from vmstate tests Daniel P. Berrange
2016-01-28 17:45   ` Dr. David Alan Gilbert
2016-01-12 11:43 ` [Qemu-devel] [PATCH v1 03/22] migration: ensure qemu_fflush() always writes full data amount Daniel P. Berrange
2016-01-28 17:53   ` Dr. David Alan Gilbert
2016-02-03 13:31     ` Daniel P. Berrange
2016-01-12 11:43 ` [Qemu-devel] [PATCH v1 04/22] migration: split migration hooks out of QEMUFileOps Daniel P. Berrange
2016-01-28 17:57   ` Dr. David Alan Gilbert
2016-01-12 11:43 ` [Qemu-devel] [PATCH v1 05/22] migration: introduce set_blocking function in QEMUFileOps Daniel P. Berrange
2016-01-28 18:00   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 06/22] migration: force QEMUFile to blocking mode for outgoing migration Daniel P. Berrange
2016-01-28 18:17   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 07/22] migration: introduce a new QEMUFile impl based on QIOChannel Daniel P. Berrange
2016-02-02 17:06   ` Dr. David Alan Gilbert
2016-02-03 13:37     ` Daniel P. Berrange
2016-02-12 17:16       ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 08/22] migration: convert post-copy to use QIOChannelBuffer Daniel P. Berrange
2016-01-25 19:38   ` Dr. David Alan Gilbert
2016-01-25 22:15     ` Daniel P. Berrange
2016-01-26 18:59       ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 09/22] migration: convert unix socket protocol to use QIOChannel Daniel P. Berrange
2016-02-02 18:02   ` Dr. David Alan Gilbert
2016-02-03 11:25     ` Daniel P. Berrange
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 10/22] migration: convert tcp " Daniel P. Berrange
2016-02-02 18:19   ` Dr. David Alan Gilbert
2016-02-03 10:02     ` Daniel P. Berrange
2016-02-03 10:33       ` Dr. David Alan Gilbert [this message]
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 11/22] migration: convert fd " Daniel P. Berrange
2016-02-02 18:46   ` Dr. David Alan Gilbert
2016-02-03 10:05     ` Daniel P. Berrange
2016-02-03 10:29       ` Dr. David Alan Gilbert
2016-02-03 10:39         ` Daniel P. Berrange
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 12/22] migration: convert exec " Daniel P. Berrange
2016-02-02 18:53   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 13/22] migration: convert RDMA to use QIOChannel interface Daniel P. Berrange
2016-02-02 20:01   ` Dr. David Alan Gilbert
2016-02-03 11:37     ` Daniel P. Berrange
2016-02-03 13:23       ` Dr. David Alan Gilbert
2016-02-03 13:25         ` Daniel P. Berrange
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 14/22] migration: convert savevm to use QIOChannel for writing to files Daniel P. Berrange
2016-02-03  9:52   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 15/22] migration: delete QEMUFile buffer implementation Daniel P. Berrange
2016-02-03  9:54   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 16/22] migration: delete QEMUSizedBuffer struct Daniel P. Berrange
2016-02-03  9:55   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 17/22] migration: delete QEMUFile sockets implementation Daniel P. Berrange
2016-02-03  9:56   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 18/22] migration: delete QEMUFile stdio implementation Daniel P. Berrange
2016-02-03  9:58   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 19/22] migration: move definition of struct QEMUFile back into qemu-file.c Daniel P. Berrange
2016-02-05 18:32   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 20/22] migration: support TLS encryption with TCP migration backend Daniel P. Berrange
2016-02-12 17:09   ` Dr. David Alan Gilbert
2016-02-12 17:25     ` Daniel P. Berrange
2016-02-15 11:00       ` Daniel P. Berrange
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 21/22] migration: remove support for non-iovec based write handlers Daniel P. Berrange
2016-02-12 15:48   ` Dr. David Alan Gilbert
2016-01-12 11:44 ` [Qemu-devel] [PATCH v1 22/22] migration: remove qemu_get_fd method from QEMUFile Daniel P. Berrange
2016-02-12 15:51   ` Dr. David Alan Gilbert
2016-01-12 11:59 ` [Qemu-devel] [PATCH v1 00/22] Convert migration to QIOChannel & support TLS Daniel P. Berrange
2016-01-20 18:01 ` Daniel P. Berrange

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160203103309.GG2376@work-vm \
    --to=dgilbert@redhat.com \
    --cc=amit.shah@redhat.com \
    --cc=berrange@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=quintela@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).