* [Qemu-devel] [PATCH 1/9] ide/macio: Fix macio DMA initialisation.
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 2/9] virtio-blk: fix unplug + virsh reboot Stefan Hajnoczi
` (7 subsequent siblings)
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel
Cc: Anthony Liguori, Mark Cave-Ayland, Andreas Färber, Stefan Hajnoczi
From: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Commit 07a7484e5d713f1eb7c1c37b18a8ab0d56d88875 accidentally introduced a bug
in the initialisation of the second macio DMA device which could cause some
DMA operations to segfault QEMU.
CC: Andreas Färber <afaerber@suse.de>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
hw/macio.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/hw/macio.c b/hw/macio.c
index 74bdcd1..0c6a6b8 100644
--- a/hw/macio.c
+++ b/hw/macio.c
@@ -188,7 +188,7 @@ static int macio_newworld_initfn(PCIDevice *d)
sysbus_dev = SYS_BUS_DEVICE(&ns->ide[1]);
sysbus_connect_irq(sysbus_dev, 0, ns->irqs[3]);
sysbus_connect_irq(sysbus_dev, 1, ns->irqs[4]);
- macio_ide_register_dma(&ns->ide[0], s->dbdma, 0x1a);
+ macio_ide_register_dma(&ns->ide[1], s->dbdma, 0x1a);
ret = qdev_init(DEVICE(&ns->ide[1]));
if (ret < 0) {
return ret;
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* [Qemu-devel] [PATCH 2/9] virtio-blk: fix unplug + virsh reboot
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 1/9] ide/macio: Fix macio DMA initialisation Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext Stefan Hajnoczi
` (6 subsequent siblings)
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Christian Borntraeger, Anthony Liguori, Stefan Hajnoczi
From: Christian Borntraeger <borntraeger@de.ibm.com>
virtio-blk registers a vmstate change handler. Unfortunately this
handler is not unregistered on unplug, leading to some random
crashes if the system is restarted, e.g. via virsh reboot.
Lets unregister the vmstate change handler if the device is removed.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
hw/virtio-blk.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/hw/virtio-blk.c b/hw/virtio-blk.c
index 34913ee..f5e6ee9 100644
--- a/hw/virtio-blk.c
+++ b/hw/virtio-blk.c
@@ -36,6 +36,7 @@ typedef struct VirtIOBlock
VirtIOBlkConf *blk;
unsigned short sector_mask;
DeviceState *qdev;
+ VMChangeStateEntry *change;
#ifdef CONFIG_VIRTIO_BLK_DATA_PLANE
VirtIOBlockDataPlane *dataplane;
#endif
@@ -681,7 +682,7 @@ VirtIODevice *virtio_blk_init(DeviceState *dev, VirtIOBlkConf *blk)
}
#endif
- qemu_add_vm_change_state_handler(virtio_blk_dma_restart_cb, s);
+ s->change = qemu_add_vm_change_state_handler(virtio_blk_dma_restart_cb, s);
s->qdev = dev;
register_savevm(dev, "virtio-blk", virtio_blk_id++, 2,
virtio_blk_save, virtio_blk_load, s);
@@ -702,6 +703,7 @@ void virtio_blk_exit(VirtIODevice *vdev)
virtio_blk_data_plane_destroy(s->dataplane);
s->dataplane = NULL;
#endif
+ qemu_del_vm_change_state_handler(s->change);
unregister_savevm(s->qdev, "virtio-blk", s);
blockdev_mark_auto_del(s->bs);
virtio_cleanup(vdev);
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 1/9] ide/macio: Fix macio DMA initialisation Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 2/9] virtio-blk: fix unplug + virsh reboot Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-08 8:37 ` Christian Borntraeger
2013-03-04 9:15 ` [Qemu-devel] [PATCH 4/9] slirp/tcp_subr.c: fix coding style in tcp_connect Stefan Hajnoczi
` (5 subsequent siblings)
8 siblings, 1 reply; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Paolo Bonzini, Anthony Liguori, Stefan Hajnoczi
From: Paolo Bonzini <pbonzini@redhat.com>
During the review of the dataplane code, the EventPoll API morphed itself
(not concidentially) into something very very similar to an AioContext.
Thus, it is trivial to convert virtio-blk-dataplane to use AioContext,
and a first baby step towards letting dataplane talk directly to the
QEMU block layer.
The only interesting note is the value-copy of EventNotifiers. At least
in my opinion this is part of the EventNotifier API and is even portable
to Windows. Of course, in this case you should not close the notifier's
underlying file descriptors or handle with event_notifier_cleanup.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
hw/dataplane/Makefile.objs | 2 +-
hw/dataplane/event-poll.c | 100 ---------------------------------------------
hw/dataplane/event-poll.h | 40 ------------------
hw/dataplane/virtio-blk.c | 48 +++++++++++++---------
4 files changed, 29 insertions(+), 161 deletions(-)
delete mode 100644 hw/dataplane/event-poll.c
delete mode 100644 hw/dataplane/event-poll.h
diff --git a/hw/dataplane/Makefile.objs b/hw/dataplane/Makefile.objs
index 3e47d05..701111c 100644
--- a/hw/dataplane/Makefile.objs
+++ b/hw/dataplane/Makefile.objs
@@ -1 +1 @@
-obj-$(CONFIG_VIRTIO_BLK_DATA_PLANE) += hostmem.o vring.o event-poll.o ioq.o virtio-blk.o
+obj-$(CONFIG_VIRTIO_BLK_DATA_PLANE) += hostmem.o vring.o ioq.o virtio-blk.o
diff --git a/hw/dataplane/event-poll.c b/hw/dataplane/event-poll.c
deleted file mode 100644
index 2b55c6e..0000000
--- a/hw/dataplane/event-poll.c
+++ /dev/null
@@ -1,100 +0,0 @@
-/*
- * Event loop with file descriptor polling
- *
- * Copyright 2012 IBM, Corp.
- * Copyright 2012 Red Hat, Inc. and/or its affiliates
- *
- * Authors:
- * Stefan Hajnoczi <stefanha@redhat.com>
- *
- * This work is licensed under the terms of the GNU GPL, version 2 or later.
- * See the COPYING file in the top-level directory.
- *
- */
-
-#include <sys/epoll.h>
-#include "hw/dataplane/event-poll.h"
-
-/* Add an event notifier and its callback for polling */
-void event_poll_add(EventPoll *poll, EventHandler *handler,
- EventNotifier *notifier, EventCallback *callback)
-{
- struct epoll_event event = {
- .events = EPOLLIN,
- .data.ptr = handler,
- };
- handler->notifier = notifier;
- handler->callback = callback;
- if (epoll_ctl(poll->epoll_fd, EPOLL_CTL_ADD,
- event_notifier_get_fd(notifier), &event) != 0) {
- fprintf(stderr, "failed to add event handler to epoll: %m\n");
- exit(1);
- }
-}
-
-/* Event callback for stopping event_poll() */
-static void handle_stop(EventHandler *handler)
-{
- /* Do nothing */
-}
-
-void event_poll_init(EventPoll *poll)
-{
- /* Create epoll file descriptor */
- poll->epoll_fd = epoll_create1(EPOLL_CLOEXEC);
- if (poll->epoll_fd < 0) {
- fprintf(stderr, "epoll_create1 failed: %m\n");
- exit(1);
- }
-
- /* Set up stop notifier */
- if (event_notifier_init(&poll->stop_notifier, 0) < 0) {
- fprintf(stderr, "failed to init stop notifier\n");
- exit(1);
- }
- event_poll_add(poll, &poll->stop_handler,
- &poll->stop_notifier, handle_stop);
-}
-
-void event_poll_cleanup(EventPoll *poll)
-{
- event_notifier_cleanup(&poll->stop_notifier);
- close(poll->epoll_fd);
- poll->epoll_fd = -1;
-}
-
-/* Block until the next event and invoke its callback */
-void event_poll(EventPoll *poll)
-{
- EventHandler *handler;
- struct epoll_event event;
- int nevents;
-
- /* Wait for the next event. Only do one event per call to keep the
- * function simple, this could be changed later. */
- do {
- nevents = epoll_wait(poll->epoll_fd, &event, 1, -1);
- } while (nevents < 0 && errno == EINTR);
- if (unlikely(nevents != 1)) {
- fprintf(stderr, "epoll_wait failed: %m\n");
- exit(1); /* should never happen */
- }
-
- /* Find out which event handler has become active */
- handler = event.data.ptr;
-
- /* Clear the eventfd */
- event_notifier_test_and_clear(handler->notifier);
-
- /* Handle the event */
- handler->callback(handler);
-}
-
-/* Stop event_poll()
- *
- * This function can be used from another thread.
- */
-void event_poll_notify(EventPoll *poll)
-{
- event_notifier_set(&poll->stop_notifier);
-}
diff --git a/hw/dataplane/event-poll.h b/hw/dataplane/event-poll.h
deleted file mode 100644
index 3e8d3ec..0000000
--- a/hw/dataplane/event-poll.h
+++ /dev/null
@@ -1,40 +0,0 @@
-/*
- * Event loop with file descriptor polling
- *
- * Copyright 2012 IBM, Corp.
- * Copyright 2012 Red Hat, Inc. and/or its affiliates
- *
- * Authors:
- * Stefan Hajnoczi <stefanha@redhat.com>
- *
- * This work is licensed under the terms of the GNU GPL, version 2 or later.
- * See the COPYING file in the top-level directory.
- *
- */
-
-#ifndef EVENT_POLL_H
-#define EVENT_POLL_H
-
-#include "qemu/event_notifier.h"
-
-typedef struct EventHandler EventHandler;
-typedef void EventCallback(EventHandler *handler);
-struct EventHandler {
- EventNotifier *notifier; /* eventfd */
- EventCallback *callback; /* callback function */
-};
-
-typedef struct {
- int epoll_fd; /* epoll(2) file descriptor */
- EventNotifier stop_notifier; /* stop poll notifier */
- EventHandler stop_handler; /* stop poll handler */
-} EventPoll;
-
-void event_poll_add(EventPoll *poll, EventHandler *handler,
- EventNotifier *notifier, EventCallback *callback);
-void event_poll_init(EventPoll *poll);
-void event_poll_cleanup(EventPoll *poll);
-void event_poll(EventPoll *poll);
-void event_poll_notify(EventPoll *poll);
-
-#endif /* EVENT_POLL_H */
diff --git a/hw/dataplane/virtio-blk.c b/hw/dataplane/virtio-blk.c
index 3f2da22..aa9b040 100644
--- a/hw/dataplane/virtio-blk.c
+++ b/hw/dataplane/virtio-blk.c
@@ -14,13 +14,13 @@
#include "trace.h"
#include "qemu/iov.h"
-#include "event-poll.h"
#include "qemu/thread.h"
#include "vring.h"
#include "ioq.h"
#include "migration/migration.h"
#include "hw/virtio-blk.h"
#include "hw/dataplane/virtio-blk.h"
+#include "block/aio.h"
enum {
SEG_MAX = 126, /* maximum number of I/O segments */
@@ -51,9 +51,14 @@ struct VirtIOBlockDataPlane {
Vring vring; /* virtqueue vring */
EventNotifier *guest_notifier; /* irq */
- EventPoll event_poll; /* event poller */
- EventHandler io_handler; /* Linux AIO completion handler */
- EventHandler notify_handler; /* virtqueue notify handler */
+ /* Note that these EventNotifiers are assigned by value. This is
+ * fine as long as you do not call event_notifier_cleanup on them
+ * (because you don't own the file descriptor or handle; you just
+ * use it).
+ */
+ AioContext *ctx;
+ EventNotifier io_notifier; /* Linux AIO completion */
+ EventNotifier host_notifier; /* doorbell */
IOQueue ioqueue; /* Linux AIO queue (should really be per
dataplane thread) */
@@ -256,10 +261,10 @@ static int process_request(IOQueue *ioq, struct iovec iov[],
}
}
-static void handle_notify(EventHandler *handler)
+static void handle_notify(EventNotifier *e)
{
- VirtIOBlockDataPlane *s = container_of(handler, VirtIOBlockDataPlane,
- notify_handler);
+ VirtIOBlockDataPlane *s = container_of(e, VirtIOBlockDataPlane,
+ host_notifier);
/* There is one array of iovecs into which all new requests are extracted
* from the vring. Requests are read from the vring and the translated
@@ -286,6 +291,7 @@ static void handle_notify(EventHandler *handler)
unsigned int out_num = 0, in_num = 0;
unsigned int num_queued;
+ event_notifier_test_and_clear(&s->host_notifier);
for (;;) {
/* Disable guest->host notifies to avoid unnecessary vmexits */
vring_disable_notification(s->vdev, &s->vring);
@@ -334,11 +340,12 @@ static void handle_notify(EventHandler *handler)
}
}
-static void handle_io(EventHandler *handler)
+static void handle_io(EventNotifier *e)
{
- VirtIOBlockDataPlane *s = container_of(handler, VirtIOBlockDataPlane,
- io_handler);
+ VirtIOBlockDataPlane *s = container_of(e, VirtIOBlockDataPlane,
+ io_notifier);
+ event_notifier_test_and_clear(&s->io_notifier);
if (ioq_run_completion(&s->ioqueue, complete_request, s) > 0) {
notify_guest(s);
}
@@ -348,7 +355,7 @@ static void handle_io(EventHandler *handler)
* requests.
*/
if (unlikely(vring_more_avail(&s->vring))) {
- handle_notify(&s->notify_handler);
+ handle_notify(&s->host_notifier);
}
}
@@ -357,7 +364,7 @@ static void *data_plane_thread(void *opaque)
VirtIOBlockDataPlane *s = opaque;
do {
- event_poll(&s->event_poll);
+ aio_poll(s->ctx, true);
} while (!s->stopping || s->num_reqs > 0);
return NULL;
}
@@ -445,7 +452,7 @@ void virtio_blk_data_plane_start(VirtIOBlockDataPlane *s)
return;
}
- event_poll_init(&s->event_poll);
+ s->ctx = aio_context_new();
/* Set up guest notifier (irq) */
if (s->vdev->binding->set_guest_notifiers(s->vdev->binding_opaque, 1,
@@ -462,17 +469,16 @@ void virtio_blk_data_plane_start(VirtIOBlockDataPlane *s)
fprintf(stderr, "virtio-blk failed to set host notifier\n");
exit(1);
}
- event_poll_add(&s->event_poll, &s->notify_handler,
- virtio_queue_get_host_notifier(vq),
- handle_notify);
+ s->host_notifier = *virtio_queue_get_host_notifier(vq);
+ aio_set_event_notifier(s->ctx, &s->host_notifier, handle_notify, NULL);
/* Set up ioqueue */
ioq_init(&s->ioqueue, s->fd, REQ_MAX);
for (i = 0; i < ARRAY_SIZE(s->requests); i++) {
ioq_put_iocb(&s->ioqueue, &s->requests[i].iocb);
}
- event_poll_add(&s->event_poll, &s->io_handler,
- ioq_get_notifier(&s->ioqueue), handle_io);
+ s->io_notifier = *ioq_get_notifier(&s->ioqueue);
+ aio_set_event_notifier(s->ctx, &s->io_notifier, handle_io, NULL);
s->started = true;
trace_virtio_blk_data_plane_start(s);
@@ -498,15 +504,17 @@ void virtio_blk_data_plane_stop(VirtIOBlockDataPlane *s)
qemu_bh_delete(s->start_bh);
s->start_bh = NULL;
} else {
- event_poll_notify(&s->event_poll);
+ aio_notify(s->ctx);
qemu_thread_join(&s->thread);
}
+ aio_set_event_notifier(s->ctx, &s->io_notifier, NULL, NULL);
ioq_cleanup(&s->ioqueue);
+ aio_set_event_notifier(s->ctx, &s->host_notifier, NULL, NULL);
s->vdev->binding->set_host_notifier(s->vdev->binding_opaque, 0, false);
- event_poll_cleanup(&s->event_poll);
+ aio_context_unref(s->ctx);
/* Clean up guest notifier (irq) */
s->vdev->binding->set_guest_notifiers(s->vdev->binding_opaque, 1, false);
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* Re: [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext
2013-03-04 9:15 ` [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext Stefan Hajnoczi
@ 2013-03-08 8:37 ` Christian Borntraeger
2013-03-08 9:22 ` Paolo Bonzini
0 siblings, 1 reply; 23+ messages in thread
From: Christian Borntraeger @ 2013-03-08 8:37 UTC (permalink / raw)
To: Stefan Hajnoczi
Cc: Anthony Liguori, qemu-devel, Dominik Dingel, Jens Freimann,
Cornelia Huck, Paolo Bonzini
[-- Attachment #1: Type: text/plain, Size: 1584 bytes --]
On 04/03/13 10:15, Stefan Hajnoczi wrote:
> From: Paolo Bonzini <pbonzini@redhat.com>
>
> During the review of the dataplane code, the EventPoll API morphed itself
> (not concidentially) into something very very similar to an AioContext.
> Thus, it is trivial to convert virtio-blk-dataplane to use AioContext,
> and a first baby step towards letting dataplane talk directly to the
> QEMU block layer.
>
> The only interesting note is the value-copy of EventNotifiers. At least
> in my opinion this is part of the EventNotifier API and is even portable
> to Windows. Of course, in this case you should not close the notifier's
> underlying file descriptors or handle with event_notifier_cleanup.
>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Hmm, this broke data plane on our internal notifier prototype code on virtio-ccw
(attached for reference)
[...]
> + /* Note that these EventNotifiers are assigned by value. This is
> + * fine as long as you do not call event_notifier_cleanup on them
> + * (because you don't own the file descriptor or handle; you just
> + * use it).
> + */
And this might be the reason. Currently we only have eventfd to wire up
guest_to_host notifiers. The host_to_guest notification is not done
via vectors/irqfd, instead we let our qemu transport listen to the irq
eventfd. Worked fine so far with vhost and dataplane without this patch.
Any ideas how to properly enable a transport that has full host notifiers
but only poor mans guest notifiers?
Christian
[-- Attachment #2: guest-host-notifiers.patch --]
[-- Type: text/x-patch, Size: 5251 bytes --]
>From 76ceaec73c44f71b2e703accb157c09fef94ccd1 Mon Sep 17 00:00:00 2001
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Date: Tue, 19 Feb 2013 13:48:17 +0100
Subject: [PATCH 23/28] Re-add guest/host notifiers.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
---
hw/s390x/virtio-ccw.c | 90 +++++++++++++++++++++++++++++++++++++++++++++++++++
hw/s390x/virtio-ccw.h | 1 +
2 files changed, 91 insertions(+)
diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c
index 63c851f..d4fa42a 100644
--- a/hw/s390x/virtio-ccw.c
+++ b/hw/s390x/virtio-ccw.c
@@ -95,6 +95,7 @@ static void virtio_ccw_start_ioeventfd(VirtioCcwDevice *dev)
int n, r;
if (!(dev->flags & VIRTIO_CCW_FLAG_USE_IOEVENTFD) ||
+ dev->ioeventfd_disabled ||
dev->ioeventfd_started) {
return;
}
@@ -793,6 +794,89 @@ static void virtio_ccw_vmstate_change(DeviceState *d, bool running)
}
}
+static bool virtio_ccw_query_guest_notifiers(DeviceState *d)
+{
+ VirtioCcwDevice *dev = VIRTIO_CCW_DEVICE(d);
+
+ return !!(dev->sch->curr_status.pmcw.flags & PMCW_FLAGS_MASK_ENA);
+}
+
+static int virtio_ccw_set_host_notifier(DeviceState *d, int n, bool assign)
+{
+ VirtioCcwDevice *dev = VIRTIO_CCW_DEVICE(d);
+
+ /* Stop using the generic ioeventfd, we are doing eventfd handling
+ * ourselves below */
+ dev->ioeventfd_disabled = assign;
+ if (assign) {
+ virtio_ccw_stop_ioeventfd(dev);
+ }
+ return virtio_ccw_set_guest2host_notifier(dev, n, assign, false);
+}
+
+static int virtio_ccw_set_guest_notifier(VirtioCcwDevice *dev, int n,
+ bool assign, bool with_irqfd)
+{
+ VirtQueue *vq = virtio_get_queue(dev->vdev, n);
+ EventNotifier *notifier = virtio_queue_get_guest_notifier(vq);
+
+ if (assign) {
+ int r = event_notifier_init(notifier, 0);
+
+ if (r < 0) {
+ return r;
+ }
+ virtio_queue_set_guest_notifier_fd_handler(vq, true, with_irqfd);
+ /* We do not support irqfd for classic I/O interrupts, because the
+ * classic interrupts are intermixed with the subchannel status, that
+ * is queried with test subchannel. We want to use vhost, though.
+ * Lets make sure to have vhost running and wire up the irq fd to
+ * land in qemu (and only the irq fd) in this code.
+ */
+ if (dev->vdev->guest_notifier_mask) {
+ dev->vdev->guest_notifier_mask(dev->vdev, n, false);
+ }
+ /* get lost events and re-inject */
+ if (dev->vdev->guest_notifier_pending &&
+ dev->vdev->guest_notifier_pending(dev->vdev, n)) {
+ event_notifier_set(notifier);
+ }
+ } else {
+ if (dev->vdev->guest_notifier_mask) {
+ dev->vdev->guest_notifier_mask(dev->vdev, n, true);
+ }
+ virtio_queue_set_guest_notifier_fd_handler(vq, false, with_irqfd);
+ event_notifier_cleanup(notifier);
+ }
+ return 0;
+}
+
+static int virtio_ccw_set_guest_notifiers(DeviceState *d, int nvqs,
+ bool assigned)
+{
+ VirtioCcwDevice *dev = VIRTIO_CCW_DEVICE(d);
+ VirtIODevice *vdev = dev->vdev;
+ int r, n;
+
+ for (n = 0; n < nvqs; n++) {
+ if (!virtio_queue_get_num(vdev, n)) {
+ break;
+ }
+ /* false -> true, as soon as irqfd works */
+ r = virtio_ccw_set_guest_notifier(dev, n, assigned, false);
+ if (r < 0) {
+ goto assign_error;
+ }
+ }
+ return 0;
+
+assign_error:
+ while (--n >= 0) {
+ virtio_ccw_set_guest_notifier(dev, n, !assigned, false);
+ }
+ return r;
+}
+
static void virtio_ccw_save_queue(DeviceState *d, int n, QEMUFile *f)
{
VirtioCcwDevice *dev = VIRTIO_CCW_DEVICE(d);
@@ -840,6 +924,9 @@ static const VirtIOBindings virtio_ccw_bindings = {
.notify = virtio_ccw_notify,
.get_features = virtio_ccw_get_features,
.vmstate_change = virtio_ccw_vmstate_change,
+ .query_guest_notifiers = virtio_ccw_query_guest_notifiers,
+ .set_host_notifier = virtio_ccw_set_host_notifier,
+ .set_guest_notifiers = virtio_ccw_set_guest_notifiers,
.save_queue = virtio_ccw_save_queue,
.load_queue = virtio_ccw_load_queue,
.save_config = virtio_ccw_save_config,
@@ -1093,6 +1180,9 @@ static void virtio_ccw_bus_class_init(ObjectClass *klass, void *data)
k->notify = virtio_ccw_notify;
k->get_features = virtio_ccw_get_features;
k->vmstate_change = virtio_ccw_vmstate_change;
+ k->query_guest_notifiers = virtio_ccw_query_guest_notifiers;
+ k->set_host_notifier = virtio_ccw_set_host_notifier;
+ k->set_guest_notifiers = virtio_ccw_set_guest_notifiers;
k->save_queue = virtio_ccw_save_queue;
k->load_queue = virtio_ccw_load_queue;
k->save_config = virtio_ccw_save_config;
diff --git a/hw/s390x/virtio-ccw.h b/hw/s390x/virtio-ccw.h
index db83e0f..a2e066f 100644
--- a/hw/s390x/virtio-ccw.h
+++ b/hw/s390x/virtio-ccw.h
@@ -84,6 +84,7 @@ struct VirtioCcwDevice {
VirtIOSCSIConf scsi;
VirtioBusState bus;
bool ioeventfd_started;
+ bool ioeventfd_disabled;
uint32_t flags;
/* Guest provided values: */
hwaddr indicators;
--
1.8.0.1
^ permalink raw reply related [flat|nested] 23+ messages in thread* Re: [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext
2013-03-08 8:37 ` Christian Borntraeger
@ 2013-03-08 9:22 ` Paolo Bonzini
2013-03-08 12:44 ` Cornelia Huck
0 siblings, 1 reply; 23+ messages in thread
From: Paolo Bonzini @ 2013-03-08 9:22 UTC (permalink / raw)
To: Christian Borntraeger
Cc: Anthony Liguori, qemu-devel, Dominik Dingel, Jens Freimann,
Stefan Hajnoczi, Cornelia Huck
Il 08/03/2013 09:37, Christian Borntraeger ha scritto:
> + if (assign) {
> + int r = event_notifier_init(notifier, 0);
> +
> + if (r < 0) {
> + return r;
> + }
> + virtio_queue_set_guest_notifier_fd_handler(vq, true, with_irqfd);
Perhaps you can simply hard-code with_irqfd to false in this call to
virtio_queue_set_guest_notifier_fd_handler and the one below? Then the
guest notifier will be emulated in userspace and processed via
vdev->binding->notify.
You will not need to overwrite the EventNotifier which is IMO a pretty
ufly violation of encapsulation. :)
Paolo
> + /* We do not support irqfd for classic I/O interrupts, because the
> + * classic interrupts are intermixed with the subchannel status, that
> + * is queried with test subchannel. We want to use vhost, though.
> + * Lets make sure to have vhost running and wire up the irq fd to
> + * land in qemu (and only the irq fd) in this code.
> + */
> + if (dev->vdev->guest_notifier_mask) {
> + dev->vdev->guest_notifier_mask(dev->vdev, n, false);
> + }
> + /* get lost events and re-inject */
> + if (dev->vdev->guest_notifier_pending &&
> + dev->vdev->guest_notifier_pending(dev->vdev, n)) {
> + event_notifier_set(notifier);
> + }
> + } else {
> + if (dev->vdev->guest_notifier_mask) {
> + dev->vdev->guest_notifier_mask(dev->vdev, n, true);
> + }
> + virtio_queue_set_guest_notifier_fd_handler(vq, false, with_irqfd);
> + event_notifier_cleanup(notifier);
^ permalink raw reply [flat|nested] 23+ messages in thread* Re: [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext
2013-03-08 9:22 ` Paolo Bonzini
@ 2013-03-08 12:44 ` Cornelia Huck
2013-03-08 13:51 ` Paolo Bonzini
0 siblings, 1 reply; 23+ messages in thread
From: Cornelia Huck @ 2013-03-08 12:44 UTC (permalink / raw)
To: Paolo Bonzini
Cc: Anthony Liguori, qemu-devel, Dominik Dingel,
Christian Borntraeger, Jens Freimann, Stefan Hajnoczi
On Fri, 08 Mar 2013 10:22:36 +0100
Paolo Bonzini <pbonzini@redhat.com> wrote:
> Il 08/03/2013 09:37, Christian Borntraeger ha scritto:
> > + if (assign) {
> > + int r = event_notifier_init(notifier, 0);
> > +
> > + if (r < 0) {
> > + return r;
> > + }
> > + virtio_queue_set_guest_notifier_fd_handler(vq, true, with_irqfd);
>
> Perhaps you can simply hard-code with_irqfd to false in this call to
> virtio_queue_set_guest_notifier_fd_handler and the one below? Then the
> guest notifier will be emulated in userspace and processed via
> vdev->binding->notify.
Well, effectively with_irqfd is already hardcoded to false (as
virtio_ccw_set_guest_notifiers() always calls this function with
with_irqfd=false), so that doesn't seem to be the problem here.
>
> You will not need to overwrite the EventNotifier which is IMO a pretty
> ufly violation of encapsulation. :)
>
> Paolo
>
> > + /* We do not support irqfd for classic I/O interrupts, because the
> > + * classic interrupts are intermixed with the subchannel status, that
> > + * is queried with test subchannel. We want to use vhost, though.
> > + * Lets make sure to have vhost running and wire up the irq fd to
> > + * land in qemu (and only the irq fd) in this code.
> > + */
> > + if (dev->vdev->guest_notifier_mask) {
> > + dev->vdev->guest_notifier_mask(dev->vdev, n, false);
> > + }
> > + /* get lost events and re-inject */
> > + if (dev->vdev->guest_notifier_pending &&
> > + dev->vdev->guest_notifier_pending(dev->vdev, n)) {
> > + event_notifier_set(notifier);
> > + }
> > + } else {
> > + if (dev->vdev->guest_notifier_mask) {
> > + dev->vdev->guest_notifier_mask(dev->vdev, n, true);
> > + }
> > + virtio_queue_set_guest_notifier_fd_handler(vq, false, with_irqfd);
> > + event_notifier_cleanup(notifier);
>
^ permalink raw reply [flat|nested] 23+ messages in thread* Re: [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext
2013-03-08 12:44 ` Cornelia Huck
@ 2013-03-08 13:51 ` Paolo Bonzini
0 siblings, 0 replies; 23+ messages in thread
From: Paolo Bonzini @ 2013-03-08 13:51 UTC (permalink / raw)
To: Cornelia Huck
Cc: Anthony Liguori, qemu-devel, Dominik Dingel,
Christian Borntraeger, Jens Freimann, Stefan Hajnoczi
Il 08/03/2013 13:44, Cornelia Huck ha scritto:
>> > Perhaps you can simply hard-code with_irqfd to false in this call to
>> > virtio_queue_set_guest_notifier_fd_handler and the one below? Then the
>> > guest notifier will be emulated in userspace and processed via
>> > vdev->binding->notify.
>
> Well, effectively with_irqfd is already hardcoded to false (as
> virtio_ccw_set_guest_notifiers() always calls this function with
> with_irqfd=false), so that doesn't seem to be the problem here.
Actually, the guest->host notifier is not touched by either the old or
the new code. The code I modified only worries about the host->guest
notifier.
How did you track the problem to the assignment by value of EventNotifiers?
BTW:
>> You will not need to overwrite the EventNotifier which is IMO a pretty
>> ufly violation of encapsulation.
This was nonsense. :)
Paolo
^ permalink raw reply [flat|nested] 23+ messages in thread
* [Qemu-devel] [PATCH 4/9] slirp/tcp_subr.c: fix coding style in tcp_connect
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
` (2 preceding siblings ...)
2013-03-04 9:15 ` [Qemu-devel] [PATCH 3/9] dataplane: remove EventPoll in favor of AioContext Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 5/9] move socket_set_nodelay to osdep.c Stefan Hajnoczi
` (4 subsequent siblings)
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Anthony Liguori, Stefan Hajnoczi, MORITA Kazutaka
From: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Fix coding style in tcp_connect before the next patch.
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
slirp/tcp_subr.c | 140 ++++++++++++++++++++++++++++---------------------------
1 file changed, 72 insertions(+), 68 deletions(-)
diff --git a/slirp/tcp_subr.c b/slirp/tcp_subr.c
index 1542e43..317dc07 100644
--- a/slirp/tcp_subr.c
+++ b/slirp/tcp_subr.c
@@ -384,83 +384,87 @@ int tcp_fconnect(struct socket *so)
* the time it gets to accept(), so... We simply accept
* here and SYN the local-host.
*/
-void
-tcp_connect(struct socket *inso)
+void tcp_connect(struct socket *inso)
{
- Slirp *slirp = inso->slirp;
- struct socket *so;
- struct sockaddr_in addr;
- socklen_t addrlen = sizeof(struct sockaddr_in);
- struct tcpcb *tp;
- int s, opt;
+ Slirp *slirp = inso->slirp;
+ struct socket *so;
+ struct sockaddr_in addr;
+ socklen_t addrlen = sizeof(struct sockaddr_in);
+ struct tcpcb *tp;
+ int s, opt;
- DEBUG_CALL("tcp_connect");
- DEBUG_ARG("inso = %lx", (long)inso);
+ DEBUG_CALL("tcp_connect");
+ DEBUG_ARG("inso = %lx", (long)inso);
- /*
- * If it's an SS_ACCEPTONCE socket, no need to socreate()
- * another socket, just use the accept() socket.
- */
- if (inso->so_state & SS_FACCEPTONCE) {
- /* FACCEPTONCE already have a tcpcb */
- so = inso;
- } else {
- if ((so = socreate(slirp)) == NULL) {
- /* If it failed, get rid of the pending connection */
- closesocket(accept(inso->s,(struct sockaddr *)&addr,&addrlen));
- return;
- }
- if (tcp_attach(so) < 0) {
- free(so); /* NOT sofree */
- return;
- }
- so->so_laddr = inso->so_laddr;
- so->so_lport = inso->so_lport;
- }
+ /*
+ * If it's an SS_ACCEPTONCE socket, no need to socreate()
+ * another socket, just use the accept() socket.
+ */
+ if (inso->so_state & SS_FACCEPTONCE) {
+ /* FACCEPTONCE already have a tcpcb */
+ so = inso;
+ } else {
+ so = socreate(slirp);
+ if (so == NULL) {
+ /* If it failed, get rid of the pending connection */
+ closesocket(accept(inso->s, (struct sockaddr *)&addr, &addrlen));
+ return;
+ }
+ if (tcp_attach(so) < 0) {
+ free(so); /* NOT sofree */
+ return;
+ }
+ so->so_laddr = inso->so_laddr;
+ so->so_lport = inso->so_lport;
+ }
- (void) tcp_mss(sototcpcb(so), 0);
+ tcp_mss(sototcpcb(so), 0);
- if ((s = accept(inso->s,(struct sockaddr *)&addr,&addrlen)) < 0) {
- tcp_close(sototcpcb(so)); /* This will sofree() as well */
- return;
- }
- socket_set_nonblock(s);
- opt = 1;
- setsockopt(s,SOL_SOCKET,SO_REUSEADDR,(char *)&opt,sizeof(int));
- opt = 1;
- setsockopt(s,SOL_SOCKET,SO_OOBINLINE,(char *)&opt,sizeof(int));
- opt = 1;
- setsockopt(s,IPPROTO_TCP,TCP_NODELAY,(char *)&opt,sizeof(int));
-
- so->so_fport = addr.sin_port;
- so->so_faddr = addr.sin_addr;
- /* Translate connections from localhost to the real hostname */
- if (so->so_faddr.s_addr == 0 ||
- (so->so_faddr.s_addr & loopback_mask) ==
- (loopback_addr.s_addr & loopback_mask)) {
- so->so_faddr = slirp->vhost_addr;
- }
+ s = accept(inso->s, (struct sockaddr *)&addr, &addrlen);
+ if (s < 0) {
+ tcp_close(sototcpcb(so)); /* This will sofree() as well */
+ return;
+ }
+ socket_set_nonblock(s);
+ opt = 1;
+ setsockopt(s, SOL_SOCKET, SO_REUSEADDR, (char *)&opt, sizeof(int));
+ opt = 1;
+ setsockopt(s, SOL_SOCKET, SO_OOBINLINE, (char *)&opt, sizeof(int));
+ opt = 1;
+ setsockopt(s, IPPROTO_TCP, TCP_NODELAY, (char *)&opt, sizeof(int));
+
+ so->so_fport = addr.sin_port;
+ so->so_faddr = addr.sin_addr;
+ /* Translate connections from localhost to the real hostname */
+ if (so->so_faddr.s_addr == 0 ||
+ (so->so_faddr.s_addr & loopback_mask) ==
+ (loopback_addr.s_addr & loopback_mask)) {
+ so->so_faddr = slirp->vhost_addr;
+ }
- /* Close the accept() socket, set right state */
- if (inso->so_state & SS_FACCEPTONCE) {
- closesocket(so->s); /* If we only accept once, close the accept() socket */
- so->so_state = SS_NOFDREF; /* Don't select it yet, even though we have an FD */
- /* if it's not FACCEPTONCE, it's already NOFDREF */
- }
- so->s = s;
- so->so_state |= SS_INCOMING;
+ /* Close the accept() socket, set right state */
+ if (inso->so_state & SS_FACCEPTONCE) {
+ /* If we only accept once, close the accept() socket */
+ closesocket(so->s);
+
+ /* Don't select it yet, even though we have an FD */
+ /* if it's not FACCEPTONCE, it's already NOFDREF */
+ so->so_state = SS_NOFDREF;
+ }
+ so->s = s;
+ so->so_state |= SS_INCOMING;
- so->so_iptos = tcp_tos(so);
- tp = sototcpcb(so);
+ so->so_iptos = tcp_tos(so);
+ tp = sototcpcb(so);
- tcp_template(tp);
+ tcp_template(tp);
- tp->t_state = TCPS_SYN_SENT;
- tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
- tp->iss = slirp->tcp_iss;
- slirp->tcp_iss += TCP_ISSINCR/2;
- tcp_sendseqinit(tp);
- tcp_output(tp);
+ tp->t_state = TCPS_SYN_SENT;
+ tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
+ tp->iss = slirp->tcp_iss;
+ slirp->tcp_iss += TCP_ISSINCR/2;
+ tcp_sendseqinit(tp);
+ tcp_output(tp);
}
/*
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* [Qemu-devel] [PATCH 5/9] move socket_set_nodelay to osdep.c
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
` (3 preceding siblings ...)
2013-03-04 9:15 ` [Qemu-devel] [PATCH 4/9] slirp/tcp_subr.c: fix coding style in tcp_connect Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 6/9] sheepdog: accept URIs Stefan Hajnoczi
` (3 subsequent siblings)
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Anthony Liguori, Stefan Hajnoczi, MORITA Kazutaka
From: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
block/sheepdog.c | 11 +----------
gdbstub.c | 5 ++---
include/qemu/sockets.h | 1 +
qemu-char.c | 6 ------
slirp/tcp_subr.c | 3 +--
util/osdep.c | 6 ++++++
6 files changed, 11 insertions(+), 21 deletions(-)
diff --git a/block/sheepdog.c b/block/sheepdog.c
index d466b23..51e75ad 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -787,15 +787,6 @@ static int aio_flush_request(void *opaque)
!QLIST_EMPTY(&s->pending_aio_head);
}
-static int set_nodelay(int fd)
-{
- int ret, opt;
-
- opt = 1;
- ret = setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, (char *)&opt, sizeof(opt));
- return ret;
-}
-
/*
* Return a socket discriptor to read/write objects.
*
@@ -814,7 +805,7 @@ static int get_sheep_fd(BDRVSheepdogState *s)
socket_set_nonblock(fd);
- ret = set_nodelay(fd);
+ ret = socket_set_nodelay(fd);
if (ret) {
error_report("%s", strerror(errno));
closesocket(fd);
diff --git a/gdbstub.c b/gdbstub.c
index 32dfea9..e414ad9 100644
--- a/gdbstub.c
+++ b/gdbstub.c
@@ -2841,7 +2841,7 @@ static void gdb_accept(void)
GDBState *s;
struct sockaddr_in sockaddr;
socklen_t len;
- int val, fd;
+ int fd;
for(;;) {
len = sizeof(sockaddr);
@@ -2858,8 +2858,7 @@ static void gdb_accept(void)
}
/* set short latency */
- val = 1;
- setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, (char *)&val, sizeof(val));
+ socket_set_nodelay(fd);
s = g_malloc0(sizeof(GDBState));
s->c_cpu = first_cpu;
diff --git a/include/qemu/sockets.h b/include/qemu/sockets.h
index 803ae17..6125bf7 100644
--- a/include/qemu/sockets.h
+++ b/include/qemu/sockets.h
@@ -34,6 +34,7 @@ int inet_aton(const char *cp, struct in_addr *ia);
int qemu_socket(int domain, int type, int protocol);
int qemu_accept(int s, struct sockaddr *addr, socklen_t *addrlen);
int socket_set_cork(int fd, int v);
+int socket_set_nodelay(int fd);
void socket_set_block(int fd);
void socket_set_nonblock(int fd);
int send_all(int fd, const void *buf, int len1);
diff --git a/qemu-char.c b/qemu-char.c
index 160decc..36295b1 100644
--- a/qemu-char.c
+++ b/qemu-char.c
@@ -2365,12 +2365,6 @@ static void tcp_chr_telnet_init(int fd)
send(fd, (char *)buf, 3, 0);
}
-static void socket_set_nodelay(int fd)
-{
- int val = 1;
- setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, (char *)&val, sizeof(val));
-}
-
static int tcp_chr_add_client(CharDriverState *chr, int fd)
{
TCPCharDriver *s = chr->opaque;
diff --git a/slirp/tcp_subr.c b/slirp/tcp_subr.c
index 317dc07..7b7ad60 100644
--- a/slirp/tcp_subr.c
+++ b/slirp/tcp_subr.c
@@ -430,8 +430,7 @@ void tcp_connect(struct socket *inso)
setsockopt(s, SOL_SOCKET, SO_REUSEADDR, (char *)&opt, sizeof(int));
opt = 1;
setsockopt(s, SOL_SOCKET, SO_OOBINLINE, (char *)&opt, sizeof(int));
- opt = 1;
- setsockopt(s, IPPROTO_TCP, TCP_NODELAY, (char *)&opt, sizeof(int));
+ socket_set_nodelay(s);
so->so_fport = addr.sin_port;
so->so_faddr = addr.sin_addr;
diff --git a/util/osdep.c b/util/osdep.c
index 5b51a03..c408261 100644
--- a/util/osdep.c
+++ b/util/osdep.c
@@ -63,6 +63,12 @@ int socket_set_cork(int fd, int v)
#endif
}
+int socket_set_nodelay(int fd)
+{
+ int v = 1;
+ return setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, &v, sizeof(v));
+}
+
int qemu_madvise(void *addr, size_t len, int advice)
{
if (advice == QEMU_MADV_INVALID) {
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* [Qemu-devel] [PATCH 6/9] sheepdog: accept URIs
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
` (4 preceding siblings ...)
2013-03-04 9:15 ` [Qemu-devel] [PATCH 5/9] move socket_set_nodelay to osdep.c Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 7/9] sheepdog: use inet_connect to simplify connect code Stefan Hajnoczi
` (2 subsequent siblings)
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Anthony Liguori, Stefan Hajnoczi, MORITA Kazutaka
From: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
The URI syntax is consistent with the NBD and Gluster syntax. The
syntax is
sheepdog[+tcp]://[host:port]/vdiname[#snapid|#tag]
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
block/sheepdog.c | 139 +++++++++++++++++++++++++++++++++++++++++--------------
qemu-doc.texi | 16 +++----
qemu-options.hx | 18 ++-----
3 files changed, 117 insertions(+), 56 deletions(-)
diff --git a/block/sheepdog.c b/block/sheepdog.c
index 51e75ad..bfa8a00 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -13,6 +13,7 @@
*/
#include "qemu-common.h"
+#include "qemu/uri.h"
#include "qemu/error-report.h"
#include "qemu/sockets.h"
#include "block/block_int.h"
@@ -816,8 +817,52 @@ static int get_sheep_fd(BDRVSheepdogState *s)
return fd;
}
+static int sd_parse_uri(BDRVSheepdogState *s, const char *filename,
+ char *vdi, uint32_t *snapid, char *tag)
+{
+ URI *uri;
+ QueryParams *qp = NULL;
+ int ret = 0;
+
+ uri = uri_parse(filename);
+ if (!uri) {
+ return -EINVAL;
+ }
+
+ if (uri->path == NULL || !strcmp(uri->path, "/")) {
+ ret = -EINVAL;
+ goto out;
+ }
+ pstrcpy(vdi, SD_MAX_VDI_LEN, uri->path + 1);
+
+ /* sheepdog[+tcp]://[host:port]/vdiname */
+ s->addr = g_strdup(uri->server ?: SD_DEFAULT_ADDR);
+ if (uri->port) {
+ s->port = g_strdup_printf("%d", uri->port);
+ } else {
+ s->port = g_strdup(SD_DEFAULT_PORT);
+ }
+
+ /* snapshot tag */
+ if (uri->fragment) {
+ *snapid = strtoul(uri->fragment, NULL, 10);
+ if (*snapid == 0) {
+ pstrcpy(tag, SD_MAX_VDI_TAG_LEN, uri->fragment);
+ }
+ } else {
+ *snapid = CURRENT_VDI_ID; /* search current vdi */
+ }
+
+out:
+ if (qp) {
+ query_params_free(qp);
+ }
+ uri_free(uri);
+ return ret;
+}
+
/*
- * Parse a filename
+ * Parse a filename (old syntax)
*
* filename must be one of the following formats:
* 1. [vdiname]
@@ -836,9 +881,11 @@ static int get_sheep_fd(BDRVSheepdogState *s)
static int parse_vdiname(BDRVSheepdogState *s, const char *filename,
char *vdi, uint32_t *snapid, char *tag)
{
- char *p, *q;
- int nr_sep;
+ char *p, *q, *uri;
+ const char *host_spec, *vdi_spec;
+ int nr_sep, ret;
+ strstart(filename, "sheepdog:", (const char **)&filename);
p = q = g_strdup(filename);
/* count the number of separators */
@@ -851,38 +898,32 @@ static int parse_vdiname(BDRVSheepdogState *s, const char *filename,
}
p = q;
- /* use the first two tokens as hostname and port number. */
+ /* use the first two tokens as host_spec. */
if (nr_sep >= 2) {
- s->addr = p;
+ host_spec = p;
p = strchr(p, ':');
- *p++ = '\0';
-
- s->port = p;
+ p++;
p = strchr(p, ':');
*p++ = '\0';
} else {
- s->addr = NULL;
- s->port = 0;
+ host_spec = "";
}
- pstrcpy(vdi, SD_MAX_VDI_LEN, p);
+ vdi_spec = p;
- p = strchr(vdi, ':');
+ p = strchr(vdi_spec, ':');
if (p) {
- *p++ = '\0';
- *snapid = strtoul(p, NULL, 10);
- if (*snapid == 0) {
- pstrcpy(tag, SD_MAX_VDI_TAG_LEN, p);
- }
- } else {
- *snapid = CURRENT_VDI_ID; /* search current vdi */
+ *p++ = '#';
}
- if (s->addr == NULL) {
- g_free(q);
- }
+ uri = g_strdup_printf("sheepdog://%s/%s", host_spec, vdi_spec);
- return 0;
+ ret = sd_parse_uri(s, uri, vdi, snapid, tag);
+
+ g_free(q);
+ g_free(uri);
+
+ return ret;
}
static int find_vdi_name(BDRVSheepdogState *s, char *filename, uint32_t snapid,
@@ -1097,16 +1138,19 @@ static int sd_open(BlockDriverState *bs, const char *filename, int flags)
uint32_t snapid;
char *buf = NULL;
- strstart(filename, "sheepdog:", (const char **)&filename);
-
QLIST_INIT(&s->inflight_aio_head);
QLIST_INIT(&s->pending_aio_head);
s->fd = -1;
memset(vdi, 0, sizeof(vdi));
memset(tag, 0, sizeof(tag));
- if (parse_vdiname(s, filename, vdi, &snapid, tag) < 0) {
- ret = -EINVAL;
+
+ if (strstr(filename, "://")) {
+ ret = sd_parse_uri(s, filename, vdi, &snapid, tag);
+ } else {
+ ret = parse_vdiname(s, filename, vdi, &snapid, tag);
+ }
+ if (ret < 0) {
goto out;
}
s->fd = get_sheep_fd(s);
@@ -1275,17 +1319,17 @@ static int sd_create(const char *filename, QEMUOptionParameter *options)
char vdi[SD_MAX_VDI_LEN], tag[SD_MAX_VDI_TAG_LEN];
uint32_t snapid;
bool prealloc = false;
- const char *vdiname;
s = g_malloc0(sizeof(BDRVSheepdogState));
- strstart(filename, "sheepdog:", &vdiname);
-
memset(vdi, 0, sizeof(vdi));
memset(tag, 0, sizeof(tag));
- if (parse_vdiname(s, vdiname, vdi, &snapid, tag) < 0) {
- error_report("invalid filename");
- ret = -EINVAL;
+ if (strstr(filename, "://")) {
+ ret = sd_parse_uri(s, filename, vdi, &snapid, tag);
+ } else {
+ ret = parse_vdiname(s, filename, vdi, &snapid, tag);
+ }
+ if (ret < 0) {
goto out;
}
@@ -1392,6 +1436,7 @@ static void sd_close(BlockDriverState *bs)
qemu_aio_set_fd_handler(s->fd, NULL, NULL, NULL, NULL);
closesocket(s->fd);
g_free(s->addr);
+ g_free(s->port);
}
static int64_t sd_getlength(BlockDriverState *bs)
@@ -2054,7 +2099,7 @@ static QEMUOptionParameter sd_create_options[] = {
{ NULL }
};
-BlockDriver bdrv_sheepdog = {
+static BlockDriver bdrv_sheepdog = {
.format_name = "sheepdog",
.protocol_name = "sheepdog",
.instance_size = sizeof(BDRVSheepdogState),
@@ -2079,8 +2124,34 @@ BlockDriver bdrv_sheepdog = {
.create_options = sd_create_options,
};
+static BlockDriver bdrv_sheepdog_tcp = {
+ .format_name = "sheepdog",
+ .protocol_name = "sheepdog+tcp",
+ .instance_size = sizeof(BDRVSheepdogState),
+ .bdrv_file_open = sd_open,
+ .bdrv_close = sd_close,
+ .bdrv_create = sd_create,
+ .bdrv_getlength = sd_getlength,
+ .bdrv_truncate = sd_truncate,
+
+ .bdrv_co_readv = sd_co_readv,
+ .bdrv_co_writev = sd_co_writev,
+ .bdrv_co_flush_to_disk = sd_co_flush_to_disk,
+
+ .bdrv_snapshot_create = sd_snapshot_create,
+ .bdrv_snapshot_goto = sd_snapshot_goto,
+ .bdrv_snapshot_delete = sd_snapshot_delete,
+ .bdrv_snapshot_list = sd_snapshot_list,
+
+ .bdrv_save_vmstate = sd_save_vmstate,
+ .bdrv_load_vmstate = sd_load_vmstate,
+
+ .create_options = sd_create_options,
+};
+
static void bdrv_sheepdog_init(void)
{
bdrv_register(&bdrv_sheepdog);
+ bdrv_register(&bdrv_sheepdog_tcp);
}
block_init(bdrv_sheepdog_init);
diff --git a/qemu-doc.texi b/qemu-doc.texi
index 747e052..2083e29 100644
--- a/qemu-doc.texi
+++ b/qemu-doc.texi
@@ -830,7 +830,7 @@ QEMU-based virtual machines.
You can create a Sheepdog disk image with the command:
@example
-qemu-img create sheepdog:@var{image} @var{size}
+qemu-img create sheepdog:///@var{image} @var{size}
@end example
where @var{image} is the Sheepdog image name and @var{size} is its
size.
@@ -838,29 +838,29 @@ size.
To import the existing @var{filename} to Sheepdog, you can use a
convert command.
@example
-qemu-img convert @var{filename} sheepdog:@var{image}
+qemu-img convert @var{filename} sheepdog:///@var{image}
@end example
You can boot from the Sheepdog disk image with the command:
@example
-qemu-system-i386 sheepdog:@var{image}
+qemu-system-i386 sheepdog:///@var{image}
@end example
You can also create a snapshot of the Sheepdog image like qcow2.
@example
-qemu-img snapshot -c @var{tag} sheepdog:@var{image}
+qemu-img snapshot -c @var{tag} sheepdog:///@var{image}
@end example
where @var{tag} is a tag name of the newly created snapshot.
To boot from the Sheepdog snapshot, specify the tag name of the
snapshot.
@example
-qemu-system-i386 sheepdog:@var{image}:@var{tag}
+qemu-system-i386 sheepdog:///@var{image}#@var{tag}
@end example
You can create a cloned image from the existing snapshot.
@example
-qemu-img create -b sheepdog:@var{base}:@var{tag} sheepdog:@var{image}
+qemu-img create -b sheepdog:///@var{base}#@var{tag} sheepdog:///@var{image}
@end example
where @var{base} is a image name of the source snapshot and @var{tag}
is its tag name.
@@ -868,8 +868,8 @@ is its tag name.
If the Sheepdog daemon doesn't run on the local host, you need to
specify one of the Sheepdog servers to connect to.
@example
-qemu-img create sheepdog:@var{hostname}:@var{port}:@var{image} @var{size}
-qemu-system-i386 sheepdog:@var{hostname}:@var{port}:@var{image}
+qemu-img create sheepdog://@var{hostname}:@var{port}/@var{image} @var{size}
+qemu-system-i386 sheepdog://@var{hostname}:@var{port}/@var{image}
@end example
@node disk_images_iscsi
diff --git a/qemu-options.hx b/qemu-options.hx
index 797d992..e8fb78c 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -2099,23 +2099,13 @@ QEMU supports using either local sheepdog devices or remote networked
devices.
Syntax for specifying a sheepdog device
-@table @list
-``sheepdog:<vdiname>''
-
-``sheepdog:<vdiname>:<snapid>''
-
-``sheepdog:<vdiname>:<tag>''
-
-``sheepdog:<host>:<port>:<vdiname>''
-
-``sheepdog:<host>:<port>:<vdiname>:<snapid>''
-
-``sheepdog:<host>:<port>:<vdiname>:<tag>''
-@end table
+@example
+sheepdog[+tcp]://[host:port]/vdiname[#snapid|#tag]
+@end example
Example
@example
-qemu-system-i386 --drive file=sheepdog:192.0.2.1:30000:MyVirtualMachine
+qemu-system-i386 --drive file=sheepdog://192.0.2.1:30000/MyVirtualMachine
@end example
See also @url{http://http://www.osrg.net/sheepdog/}.
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* [Qemu-devel] [PATCH 7/9] sheepdog: use inet_connect to simplify connect code
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
` (5 preceding siblings ...)
2013-03-04 9:15 ` [Qemu-devel] [PATCH 6/9] sheepdog: accept URIs Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 8/9] sheepdog: add support for connecting to unix domain socket Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 9/9] block: for HMP commit() operations on 'all', skip non-COW drives Stefan Hajnoczi
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Anthony Liguori, Stefan Hajnoczi, MORITA Kazutaka
From: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
This uses the form "<host>:<port>" for the representation of the
sheepdog server to use inet_connect.
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
block/sheepdog.c | 111 +++++++++++++++----------------------------------------
1 file changed, 30 insertions(+), 81 deletions(-)
diff --git a/block/sheepdog.c b/block/sheepdog.c
index bfa8a00..b5cbdfe 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -22,7 +22,7 @@
#define SD_PROTO_VER 0x01
#define SD_DEFAULT_ADDR "localhost"
-#define SD_DEFAULT_PORT "7000"
+#define SD_DEFAULT_PORT 7000
#define SD_OP_CREATE_AND_WRITE_OBJ 0x01
#define SD_OP_READ_OBJ 0x02
@@ -298,8 +298,7 @@ typedef struct BDRVSheepdogState {
bool is_snapshot;
uint32_t cache_flags;
- char *addr;
- char *port;
+ char *host_spec;
int fd;
CoMutex lock;
@@ -447,56 +446,18 @@ static SheepdogAIOCB *sd_aio_setup(BlockDriverState *bs, QEMUIOVector *qiov,
return acb;
}
-static int connect_to_sdog(const char *addr, const char *port)
+static int connect_to_sdog(BDRVSheepdogState *s)
{
- char hbuf[NI_MAXHOST], sbuf[NI_MAXSERV];
- int fd, ret;
- struct addrinfo hints, *res, *res0;
-
- if (!addr) {
- addr = SD_DEFAULT_ADDR;
- port = SD_DEFAULT_PORT;
- }
+ int fd;
+ Error *err = NULL;
- memset(&hints, 0, sizeof(hints));
- hints.ai_socktype = SOCK_STREAM;
+ fd = inet_connect(s->host_spec, &err);
- ret = getaddrinfo(addr, port, &hints, &res0);
- if (ret) {
- error_report("unable to get address info %s, %s",
- addr, strerror(errno));
- return -errno;
+ if (err != NULL) {
+ qerror_report_err(err);
+ error_free(err);
}
- for (res = res0; res; res = res->ai_next) {
- ret = getnameinfo(res->ai_addr, res->ai_addrlen, hbuf, sizeof(hbuf),
- sbuf, sizeof(sbuf), NI_NUMERICHOST | NI_NUMERICSERV);
- if (ret) {
- continue;
- }
-
- fd = socket(res->ai_family, res->ai_socktype, res->ai_protocol);
- if (fd < 0) {
- continue;
- }
-
- reconnect:
- ret = connect(fd, res->ai_addr, res->ai_addrlen);
- if (ret < 0) {
- if (errno == EINTR) {
- goto reconnect;
- }
- close(fd);
- break;
- }
-
- dprintf("connected to %s:%s\n", addr, port);
- goto success;
- }
- fd = -errno;
- error_report("failed connect to %s:%s", addr, port);
-success:
- freeaddrinfo(res0);
return fd;
}
@@ -798,9 +759,8 @@ static int get_sheep_fd(BDRVSheepdogState *s)
{
int ret, fd;
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
- error_report("%s", strerror(errno));
return fd;
}
@@ -836,12 +796,8 @@ static int sd_parse_uri(BDRVSheepdogState *s, const char *filename,
pstrcpy(vdi, SD_MAX_VDI_LEN, uri->path + 1);
/* sheepdog[+tcp]://[host:port]/vdiname */
- s->addr = g_strdup(uri->server ?: SD_DEFAULT_ADDR);
- if (uri->port) {
- s->port = g_strdup_printf("%d", uri->port);
- } else {
- s->port = g_strdup(SD_DEFAULT_PORT);
- }
+ s->host_spec = g_strdup_printf("%s:%d", uri->server ?: SD_DEFAULT_ADDR,
+ uri->port ?: SD_DEFAULT_PORT);
/* snapshot tag */
if (uri->fragment) {
@@ -935,7 +891,7 @@ static int find_vdi_name(BDRVSheepdogState *s, char *filename, uint32_t snapid,
unsigned int wlen, rlen = 0;
char buf[SD_MAX_VDI_LEN + SD_MAX_VDI_TAG_LEN];
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
return fd;
}
@@ -1178,9 +1134,8 @@ static int sd_open(BlockDriverState *bs, const char *filename, int flags)
s->is_snapshot = true;
}
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
- error_report("failed to connect");
ret = fd;
goto out;
}
@@ -1213,9 +1168,8 @@ out:
return ret;
}
-static int do_sd_create(char *filename, int64_t vdi_size,
- uint32_t base_vid, uint32_t *vdi_id, int snapshot,
- const char *addr, const char *port)
+static int do_sd_create(BDRVSheepdogState *s, char *filename, int64_t vdi_size,
+ uint32_t base_vid, uint32_t *vdi_id, int snapshot)
{
SheepdogVdiReq hdr;
SheepdogVdiRsp *rsp = (SheepdogVdiRsp *)&hdr;
@@ -1223,7 +1177,7 @@ static int do_sd_create(char *filename, int64_t vdi_size,
unsigned int wlen, rlen = 0;
char buf[SD_MAX_VDI_LEN];
- fd = connect_to_sdog(addr, port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
return fd;
}
@@ -1390,7 +1344,7 @@ static int sd_create(const char *filename, QEMUOptionParameter *options)
bdrv_delete(bs);
}
- ret = do_sd_create(vdi, vdi_size, base_vid, &vid, 0, s->addr, s->port);
+ ret = do_sd_create(s, vdi, vdi_size, base_vid, &vid, 0);
if (!prealloc || ret) {
goto out;
}
@@ -1411,7 +1365,7 @@ static void sd_close(BlockDriverState *bs)
dprintf("%s\n", s->name);
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
return;
}
@@ -1435,8 +1389,7 @@ static void sd_close(BlockDriverState *bs)
qemu_aio_set_fd_handler(s->fd, NULL, NULL, NULL, NULL);
closesocket(s->fd);
- g_free(s->addr);
- g_free(s->port);
+ g_free(s->host_spec);
}
static int64_t sd_getlength(BlockDriverState *bs)
@@ -1460,7 +1413,7 @@ static int sd_truncate(BlockDriverState *bs, int64_t offset)
return -EINVAL;
}
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
return fd;
}
@@ -1536,17 +1489,15 @@ static int sd_create_branch(BDRVSheepdogState *s)
buf = g_malloc(SD_INODE_SIZE);
- ret = do_sd_create(s->name, s->inode.vdi_size, s->inode.vdi_id, &vid, 1,
- s->addr, s->port);
+ ret = do_sd_create(s, s->name, s->inode.vdi_size, s->inode.vdi_id, &vid, 1);
if (ret) {
goto out;
}
dprintf("%" PRIx32 " is created.\n", vid);
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
- error_report("failed to connect");
ret = fd;
goto out;
}
@@ -1805,7 +1756,7 @@ static int sd_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info)
datalen = SD_INODE_SIZE - sizeof(s->inode.data_vdi_id);
/* refresh inode. */
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
ret = fd;
goto cleanup;
@@ -1818,8 +1769,8 @@ static int sd_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info)
goto cleanup;
}
- ret = do_sd_create(s->name, s->inode.vdi_size, s->inode.vdi_id, &new_vid, 1,
- s->addr, s->port);
+ ret = do_sd_create(s, s->name, s->inode.vdi_size, s->inode.vdi_id, &new_vid,
+ 1);
if (ret < 0) {
error_report("failed to create inode for snapshot. %s",
strerror(errno));
@@ -1874,9 +1825,8 @@ static int sd_snapshot_goto(BlockDriverState *bs, const char *snapshot_id)
goto out;
}
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
- error_report("failed to connect");
ret = fd;
goto out;
}
@@ -1938,7 +1888,7 @@ static int sd_snapshot_list(BlockDriverState *bs, QEMUSnapshotInfo **psn_tab)
vdi_inuse = g_malloc(max);
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
ret = fd;
goto out;
@@ -1965,9 +1915,8 @@ static int sd_snapshot_list(BlockDriverState *bs, QEMUSnapshotInfo **psn_tab)
hval = fnv_64a_buf(s->name, strlen(s->name), FNV1A_64_INIT);
start_nr = hval & (SD_NR_VDIS - 1);
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
- error_report("failed to connect");
ret = fd;
goto out;
}
@@ -2024,7 +1973,7 @@ static int do_load_save_vmstate(BDRVSheepdogState *s, uint8_t *data,
uint32_t vdi_index;
uint64_t offset;
- fd = connect_to_sdog(s->addr, s->port);
+ fd = connect_to_sdog(s);
if (fd < 0) {
return fd;
}
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* [Qemu-devel] [PATCH 8/9] sheepdog: add support for connecting to unix domain socket
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
` (6 preceding siblings ...)
2013-03-04 9:15 ` [Qemu-devel] [PATCH 7/9] sheepdog: use inet_connect to simplify connect code Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
2013-03-04 9:15 ` [Qemu-devel] [PATCH 9/9] block: for HMP commit() operations on 'all', skip non-COW drives Stefan Hajnoczi
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Anthony Liguori, Stefan Hajnoczi, MORITA Kazutaka
From: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
This patch adds support for a unix domain socket for a connection
between qemu and local sheepdog server. You can use the unix domain
socket with the following syntax:
$ qemu sheepdog+unix:///<vdiname>?socket=<socket path>[#snapid]
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
block/sheepdog.c | 82 +++++++++++++++++++++++++++++++++++++++++++++++---------
qemu-doc.texi | 6 +++++
qemu-options.hx | 2 +-
3 files changed, 77 insertions(+), 13 deletions(-)
diff --git a/block/sheepdog.c b/block/sheepdog.c
index b5cbdfe..c711c28 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -299,6 +299,7 @@ typedef struct BDRVSheepdogState {
uint32_t cache_flags;
char *host_spec;
+ bool is_unix;
int fd;
CoMutex lock;
@@ -451,7 +452,18 @@ static int connect_to_sdog(BDRVSheepdogState *s)
int fd;
Error *err = NULL;
- fd = inet_connect(s->host_spec, &err);
+ if (s->is_unix) {
+ fd = unix_connect(s->host_spec, &err);
+ } else {
+ fd = inet_connect(s->host_spec, &err);
+
+ if (err == NULL) {
+ int ret = socket_set_nodelay(fd);
+ if (ret < 0) {
+ error_report("%s", strerror(errno));
+ }
+ }
+ }
if (err != NULL) {
qerror_report_err(err);
@@ -757,7 +769,7 @@ static int aio_flush_request(void *opaque)
*/
static int get_sheep_fd(BDRVSheepdogState *s)
{
- int ret, fd;
+ int fd;
fd = connect_to_sdog(s);
if (fd < 0) {
@@ -766,13 +778,6 @@ static int get_sheep_fd(BDRVSheepdogState *s)
socket_set_nonblock(fd);
- ret = socket_set_nodelay(fd);
- if (ret) {
- error_report("%s", strerror(errno));
- closesocket(fd);
- return -errno;
- }
-
qemu_aio_set_fd_handler(fd, co_read_response, NULL, aio_flush_request, s);
return fd;
}
@@ -789,15 +794,42 @@ static int sd_parse_uri(BDRVSheepdogState *s, const char *filename,
return -EINVAL;
}
+ /* transport */
+ if (!strcmp(uri->scheme, "sheepdog")) {
+ s->is_unix = false;
+ } else if (!strcmp(uri->scheme, "sheepdog+tcp")) {
+ s->is_unix = false;
+ } else if (!strcmp(uri->scheme, "sheepdog+unix")) {
+ s->is_unix = true;
+ } else {
+ ret = -EINVAL;
+ goto out;
+ }
+
if (uri->path == NULL || !strcmp(uri->path, "/")) {
ret = -EINVAL;
goto out;
}
pstrcpy(vdi, SD_MAX_VDI_LEN, uri->path + 1);
- /* sheepdog[+tcp]://[host:port]/vdiname */
- s->host_spec = g_strdup_printf("%s:%d", uri->server ?: SD_DEFAULT_ADDR,
- uri->port ?: SD_DEFAULT_PORT);
+ qp = query_params_parse(uri->query);
+ if (qp->n > 1 || (s->is_unix && !qp->n) || (!s->is_unix && qp->n)) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ if (s->is_unix) {
+ /* sheepdog+unix:///vdiname?socket=path */
+ if (uri->server || uri->port || strcmp(qp->p[0].name, "socket")) {
+ ret = -EINVAL;
+ goto out;
+ }
+ s->host_spec = g_strdup(qp->p[0].value);
+ } else {
+ /* sheepdog[+tcp]://[host:port]/vdiname */
+ s->host_spec = g_strdup_printf("%s:%d", uri->server ?: SD_DEFAULT_ADDR,
+ uri->port ?: SD_DEFAULT_PORT);
+ }
/* snapshot tag */
if (uri->fragment) {
@@ -2098,9 +2130,35 @@ static BlockDriver bdrv_sheepdog_tcp = {
.create_options = sd_create_options,
};
+static BlockDriver bdrv_sheepdog_unix = {
+ .format_name = "sheepdog",
+ .protocol_name = "sheepdog+unix",
+ .instance_size = sizeof(BDRVSheepdogState),
+ .bdrv_file_open = sd_open,
+ .bdrv_close = sd_close,
+ .bdrv_create = sd_create,
+ .bdrv_getlength = sd_getlength,
+ .bdrv_truncate = sd_truncate,
+
+ .bdrv_co_readv = sd_co_readv,
+ .bdrv_co_writev = sd_co_writev,
+ .bdrv_co_flush_to_disk = sd_co_flush_to_disk,
+
+ .bdrv_snapshot_create = sd_snapshot_create,
+ .bdrv_snapshot_goto = sd_snapshot_goto,
+ .bdrv_snapshot_delete = sd_snapshot_delete,
+ .bdrv_snapshot_list = sd_snapshot_list,
+
+ .bdrv_save_vmstate = sd_save_vmstate,
+ .bdrv_load_vmstate = sd_load_vmstate,
+
+ .create_options = sd_create_options,
+};
+
static void bdrv_sheepdog_init(void)
{
bdrv_register(&bdrv_sheepdog);
bdrv_register(&bdrv_sheepdog_tcp);
+ bdrv_register(&bdrv_sheepdog_unix);
}
block_init(bdrv_sheepdog_init);
diff --git a/qemu-doc.texi b/qemu-doc.texi
index 2083e29..af84bef 100644
--- a/qemu-doc.texi
+++ b/qemu-doc.texi
@@ -865,6 +865,12 @@ qemu-img create -b sheepdog:///@var{base}#@var{tag} sheepdog:///@var{image}
where @var{base} is a image name of the source snapshot and @var{tag}
is its tag name.
+You can use an unix socket instead of an inet socket:
+
+@example
+qemu-system-i386 sheepdog+unix:///@var{image}?socket=@var{path}
+@end example
+
If the Sheepdog daemon doesn't run on the local host, you need to
specify one of the Sheepdog servers to connect to.
@example
diff --git a/qemu-options.hx b/qemu-options.hx
index e8fb78c..f598d7a 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -2100,7 +2100,7 @@ devices.
Syntax for specifying a sheepdog device
@example
-sheepdog[+tcp]://[host:port]/vdiname[#snapid|#tag]
+sheepdog[+tcp|+unix]://[host:port]/vdiname[?socket=path][#snapid|#tag]
@end example
Example
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread* [Qemu-devel] [PATCH 9/9] block: for HMP commit() operations on 'all', skip non-COW drives
2013-03-04 9:15 [Qemu-devel] [PULL 0/9] Block patches Stefan Hajnoczi
` (7 preceding siblings ...)
2013-03-04 9:15 ` [Qemu-devel] [PATCH 8/9] sheepdog: add support for connecting to unix domain socket Stefan Hajnoczi
@ 2013-03-04 9:15 ` Stefan Hajnoczi
8 siblings, 0 replies; 23+ messages in thread
From: Stefan Hajnoczi @ 2013-03-04 9:15 UTC (permalink / raw)
To: qemu-devel; +Cc: Anthony Liguori, Jeff Cody, Stefan Hajnoczi
From: Jeff Cody <jcody@redhat.com>
During a commit of 'all' using the HMP non-live commit, the operation
is aborted and returns error on the first error enountered. When
non-COW drives are in use (e.g. ejected floppy, cdrom, or drives without
a backing parent), that means a commit all will return an error of either
-ENOMEDIUM or -ENOTSUP. This is not desirable, so for the 'all' commit
case, only attempt the commit if both bs->drv and bs->backing_hd are
present.
More succinctly: 'commit all' now means a commit on all COW drives.
This means an individual commit to a specific non-COW drive will still
return the appropriate error (-ENOMEDIUM if eject / not present, -ENOTSUP
if no backing file).
Reported-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
block.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/block.c b/block.c
index 4582961..124a9eb 100644
--- a/block.c
+++ b/block.c
@@ -1640,9 +1640,11 @@ int bdrv_commit_all(void)
BlockDriverState *bs;
QTAILQ_FOREACH(bs, &bdrv_states, list) {
- int ret = bdrv_commit(bs);
- if (ret < 0) {
- return ret;
+ if (bs->drv && bs->backing_hd) {
+ int ret = bdrv_commit(bs);
+ if (ret < 0) {
+ return ret;
+ }
}
}
return 0;
--
1.8.1.4
^ permalink raw reply related [flat|nested] 23+ messages in thread