From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:41890) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aQuK8-0007da-Sk for qemu-devel@nongnu.org; Wed, 03 Feb 2016 05:06:12 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1aQuK2-0000YQ-Ql for qemu-devel@nongnu.org; Wed, 03 Feb 2016 05:06:08 -0500 Received: from mx1.redhat.com ([209.132.183.28]:41064) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aQuK2-0000YA-J8 for qemu-devel@nongnu.org; Wed, 03 Feb 2016 05:06:02 -0500 Received: from int-mx09.intmail.prod.int.phx2.redhat.com (int-mx09.intmail.prod.int.phx2.redhat.com [10.5.11.22]) by mx1.redhat.com (Postfix) with ESMTPS id 26B837AEA7 for ; Wed, 3 Feb 2016 10:06:02 +0000 (UTC) Date: Wed, 3 Feb 2016 10:05:58 +0000 From: "Daniel P. Berrange" Message-ID: <20160203100558.GB30222@redhat.com> References: <1452599056-27357-1-git-send-email-berrange@redhat.com> <1452599056-27357-12-git-send-email-berrange@redhat.com> <20160202184601.GD4498@work-vm> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20160202184601.GD4498@work-vm> Subject: Re: [Qemu-devel] [PATCH v1 11/22] migration: convert fd socket protocol to use QIOChannel Reply-To: "Daniel P. Berrange" List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Dr. David Alan Gilbert" Cc: Amit Shah , qemu-devel@nongnu.org, Juan Quintela On Tue, Feb 02, 2016 at 06:46:01PM +0000, Dr. David Alan Gilbert wrote: > * Daniel P. Berrange (berrange@redhat.com) wrote: > > Convert the fd socket migration protocol driver to use > > QIOChannel and QEMUFileChannel, instead of plain sockets > > APIs. It can be unconditionally built because the > > QIOChannel APIs it uses will take care to report suitable > > error messages if needed. > > > > Signed-off-by: Daniel P. Berrange > > --- > > migration/Makefile.objs | 4 ++-- > > migration/fd.c | 57 ++++++++++++++++++++++++++++++++----------------- > > migration/migration.c | 4 ---- > > 3 files changed, 39 insertions(+), 26 deletions(-) > > > > diff --git a/migration/Makefile.objs b/migration/Makefile.objs > > index a5f8a03..64f95cd 100644 > > --- a/migration/Makefile.objs > > +++ b/migration/Makefile.objs > > @@ -1,11 +1,11 @@ > > -common-obj-y += migration.o tcp.o unix.o > > +common-obj-y += migration.o tcp.o unix.o fd.o > > common-obj-y += vmstate.o > > common-obj-y += qemu-file.o qemu-file-buf.o qemu-file-unix.o qemu-file-stdio.o > > common-obj-y += qemu-file-channel.o > > common-obj-y += xbzrle.o postcopy-ram.o > > > > common-obj-$(CONFIG_RDMA) += rdma.o > > -common-obj-$(CONFIG_POSIX) += exec.o fd.o > > +common-obj-$(CONFIG_POSIX) += exec.o > > > > common-obj-y += block.o > > > > diff --git a/migration/fd.c b/migration/fd.c > > index 3e4bed0..8d48e0d 100644 > > --- a/migration/fd.c > > +++ b/migration/fd.c > > @@ -20,6 +20,8 @@ > > #include "monitor/monitor.h" > > #include "migration/qemu-file.h" > > #include "block/block.h" > > +#include "io/channel-file.h" > > +#include "io/channel-socket.h" > > > > //#define DEBUG_MIGRATION_FD > > > > @@ -33,56 +35,71 @@ > > > > static bool fd_is_socket(int fd) > > { > > - struct stat stat; > > - int ret = fstat(fd, &stat); > > - if (ret == -1) { > > - /* When in doubt say no */ > > - return false; > > - } > > - return S_ISSOCK(stat.st_mode); > > + int optval; > > + socklen_t optlen; > > + optlen = sizeof(optval); > > + return getsockopt(fd, > > + SOL_SOCKET, > > + SO_TYPE, > > + (char *)&optval, > > + &optlen) == 0; > > Should that be qemu_getsockopt ? Yes. > > void fd_start_outgoing_migration(MigrationState *s, const char *fdname, Error **errp) > > { > > + QIOChannel *ioc; > > int fd = monitor_get_fd(cur_mon, fdname, errp); > > if (fd == -1) { > > return; > > } > > > > if (fd_is_socket(fd)) { > > - s->file = qemu_fopen_socket(fd, "wb"); > > + ioc = QIO_CHANNEL(qio_channel_socket_new_fd(fd, errp)); > > + if (!ioc) { > > + close(fd); > > + return; > > + } > > Have you considered moving this fd_is_socket detection > glue down into channel-file.c? It's here so that a socket > passed via an fd will be able to use a shutdown() method. > It would be nice to do the same thing to sockets in the same > situation in other places, so it would make sense if it worked > on any fd that went through channels. > The tricky bit is at that level you return a QIOChannelFile rather > than a generic QIOChannel pointer. > (One place I'd like to be able to a shutdown is on an nbd channel > for example). Yeah, the complexity is that we have to return two different class instances depending on the type of FD in use. We could introduce some helper method todo this though. > > -static void fd_accept_incoming_migration(void *opaque) > > +static gboolean fd_accept_incoming_migration(QIOChannel *ioc, > > + GIOCondition condition, > > + gpointer opaque) > > { > > QEMUFile *f = opaque; > > - > > - qemu_set_fd_handler(qemu_get_fd(f), NULL, NULL, NULL); > > process_incoming_migration(f); > > + return FALSE; > > Please comment the magical FALSE. FALSE is the standard value any glib event loop handled need to return to cause glib to unregister it. I'm not sure we really want to comment every event handle in this way. > > void fd_start_incoming_migration(const char *infd, Error **errp) > > { > > - int fd; > > QEMUFile *f; > > + QIOChannel *ioc; > > + int fd; > > > > DPRINTF("Attempting to start an incoming migration via fd\n"); > > > > fd = strtol(infd, NULL, 0); > > if (fd_is_socket(fd)) { > > - f = qemu_fopen_socket(fd, "rb"); > > + ioc = QIO_CHANNEL(qio_channel_socket_new_fd(fd, errp)); > > + if (!ioc) { > > + close(fd); > > + return; > > + } > > Wouldn't it be better to move this check outside of this if, so that > you test the output of both the socket_new_fd and the file_new_fd ? qio_channel_file_new_fd() can never fail - only the socket_new_fd can fail (when trying to query the sockaddr_t data). > > } else { > > - f = qemu_fdopen(fd, "rb"); > > - } > > - if(f == NULL) { > > - error_setg_errno(errp, errno, "failed to open the source descriptor"); > > - return; > > + ioc = QIO_CHANNEL(qio_channel_file_new_fd(fd)); > > } > > + f = qemu_fopen_channel_input(ioc); > > + object_unref(OBJECT(ioc)); > > > > - qemu_set_fd_handler(fd, fd_accept_incoming_migration, NULL, f); > > + qio_channel_add_watch(ioc, > > + G_IO_IN, > > + fd_accept_incoming_migration, > > + f, > > + NULL); > > } 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 :|