V9fsState now only contains generic fields. Introduce V9fsVirtioState
for virtio transport.  Change virtio-pci and virtio-ccw to use
V9fsVirtioState.
Signed-off-by: NWei Liu <wei.liu2@citrix.com>
Signed-off-by: NAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>

There's no such thing as "PCI queues" in the virtio core.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

The 'virtqueue_state' and 'ringsize' can be saved using VMSTATE
macros rather than hand coded .get/.put
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NAmit Shah <amit.shah@redhat.com>

The deleted file only contained V9fsConf which wasn't virtio specific.
Merge that to the general header of 9pfs.

Fixed header inclusions as I went along.
Signed-off-by: NWei Liu <wei.liu2@citrix.com>
Signed-off-by: NAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>

In 1811e64c 'hw/virtio: Add PCIe capability to virtio devices', the
QEMU_PCI_CAP_EXPRESS capability was added to virtio's pci_dev, within
'virtio_pci_realize' - the pci device object realization method.

This occurs to late, as 'pci_qdev_realize' (DeviceClass.realize of
TYPE_PCI_DEVICE) has already been called, without knowing that the
device instance is indeed an "express" instance, thus allocating
insufficient pci config space.

As a result, device may crash upon attempt to write to the PCIE config
space.

Fix, by arming the QEMU_PCI_CAP_EXPRESS capability early in virtio-pci's
own DeviceClass realize method.

This also makes code cleaner, as 'virtio_pci_realize' may now access the
'pci_is_express' predicate when needed.
Signed-off-by: NShmulik Ladkani <shmulik.ladkani@ravellosystems.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NMarcel Apfelbaum <marcel@redhat.com>
Tested-by: NMarcel Apfelbaum <marcel@redhat.com>

If you run a qemu advertising VERSION_1 with an old kernel where
vhost did not yet support VERSION_1, you'll end up with a device
that is {modern pci|ccw revision 1} but does not advertise VERSION_1.
This is not a sensible configuration and is rejected by the Linux
guest drivers.

To fix this, add a ->post_plugged() callback invoked after features
have been queried that can handle the VERSION_1 bit being withdrawn
and change ccw to fall back to revision 0 if VERSION_1 is gone.

Note that pci is _not_ fixed; we'll need to rethink the approach
for the next release but at least for pci it's not a regression.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

This reverts commit 3a12f322.

In case of live migration several queues can be enabled and not only the
first one. So informing backend that only the first queue is enabled is
wrong.
Reported-by: NThibaut Collet <thibaut.collet@6wind.com>
Cc: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NYuanhan Liu <yuanhan.liu@linux.intel.com>

commit 2b8819c6
("vhost-user: modify SET_LOG_BASE to pass mmap size and offset")
passes log size in units of 4 byte chunks instead of the
expected size in bytes.

Fix this up.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We are currently only sending VRING_ENABLE message for the first ring,
that's wrong: we must start/stop them all.
Reported-by: NVictor Kaplansky <victork@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

When we get an unexpected response, print out
the original request.
Helps debug protocol errors tremendously.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

But not depend on PROTOCOL_F_MQ feature bit. So that we could use
SET_VRING_ENABLE to sign the backend on stop, even if MQ is disabled.

That's reasonable, since we will have one queue pair at least.
Signed-off-by: NYuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

The virtio devices are converted to PCI-Express
if they are plugged into a PCI-Express bus and
the 'modern' protocol is enabled.

Devices plugged directly into the Root Complex as
Integrated Endpoints remain PCI.
Signed-off-by: NMarcel Apfelbaum <marcel@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Send SET_VRING_ENABLE at start/stop, to give the backend
an explicit sign of our state.
Signed-off-by: NYuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

This patch basically reverts commit d1f8b30e.

It turned out that it breaks stuff, so revert it:
    http://lists.nongnu.org/archive/html/qemu-devel/2015-10/msg00949.html

CC: "Michael S. Tsirkin" <mst@redhat.com>
Reported-by: NPaolo Bonzini <pbonzini@redhat.com>
Signed-off-by: NYuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Unlike the kernel, vhost-user application accesses log table by
mmaping it to its user space. This change adds two new fields to
VhostUserMsg payload: mmap_size, and mmap_offset and make QEMU to
pass the to vhost-user application in VHOST_USER_SET_LOG_BASE
request.
Signed-off-by: NVictor Kaplansky <victork@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Guest always get zero when reading queue_enable. This violates
spec. Fixing this by setting the queue_enable to true during any guest
writing and setting it to zero during reset.

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We used to use mmio for notification. This could be slow on some arch
(e.g on x86 without EPT). So this patch introduces pio bar and a pio
notification cap for modern device. This ability is enabled through
property "modern-pio-notify" for virtio pci devices and was disabled
by default. Management can enable when it thinks it was needed.

Benchmarks shows almost no obvious difference compared to legacy
device on machines without ept. Thanks Wenli Quan <wquan@redhat.com>
for the benchmarking.

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We use data match eventfd for 1.0 notification currently. This could
be slow since software decoding is needed for mmio exit. To speed this
up, we can switch to use zero length mmio eventfd for 1.0 notification
since we can examine the queue index directly from the writing
address. KVM kernel module can utilize this by registering it to fast
mmio bus which could be as fast as pio on ept capable machine when
fast mmio is supported by host kernel.

Lots of improvements were seen on a ept capable machine:

Guest RX:(TCP)
size/session/+throughput%/+cpu%/-+per cpu%/
64/1/+1.6807%/[-16.2421%]/[+21.3984%]/
64/2/+0.6091%/[-11.0187%]/[+13.0678%]/
64/4/+0.0553%/[-5.9768%]/[+6.4155%]/
64/8/+0.1206%/[-4.0057%]/[+4.2984%]/
256/1/-0.0031%/[-10.1166%]/[+11.2517%]/
256/2/-0.5058%/[-6.1656%]/+6.0317%]/
...

Guest TX:(TCP)
size/session/+throughput%/+cpu%/-+per cpu%/
64/1/[+18.9183%]/-0.2823%/[+19.2550%]/
64/2/[+13.5714%]/[+2.2675%]/[+11.0533%]/
64/4/[+13.1070%]/[+2.1817%]/[+10.6920%]/
64/8/[+13.0426%]/[+2.0887%]/[+10.7299%]/
256/1/[+36.2761%]/+6.3434%/[+28.1471%]/
...
1024/1/[+44.8873%]/+2.0811%/[+41.9335%]/
...
1024/4/+0.0228%/[-2.2044%]/[+2.2774%]/
...
16384/2/+0.0127%/[-5.0346%]/[+5.3148%]/
...
65535/1/[+0.0062%]/[-4.1183%]/[+4.3017%]/
65535/2/+0.0004%/[-4.2311%]/[+4.4185%]/
65535/4/+0.0107%/[-4.6106%]/[+4.8446%]/
65535/8/-0.0090%/[-5.5178%]/[+5.8306%]/

Latency:(TCP_RR)
size/session/+transaction rate%/+cpu%/-+per cpu%/
64/1/[+6.5248%]/[-9.2882%]/[+17.4322%]/
64/25/[+11.0854%]/[+0.8000%]/[+10.2038%]/
64/50/[+12.1076%]/[+2.4627%]/[+9.4131%]/
256/1/[+5.3677%]/[+10.5669%]/-4.7024%/
256/25/[+5.6402%]/-0.8962%/[+6.5955%]/
256/50/[+5.9685%]/[+1.7766%]/[+4.1188%]/
4096/1/+0.2508%/[-10.4941%]/[+12.0047%]/
4096/25/[+1.8533%]/-0.0273%/+1.8812%/
4096/50/[+1.2156%]/-1.4134%/+2.6667%/

Notes: data with '[]' is the one whose significance is greater than 95%.

Thanks Wenli Quan <wquan@redhat.com> for the benchmarking.

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We don't migrate the followings fields for virtio-pci:

uint32_t dfselect;
uint32_t gfselect;
uint32_t guest_features[2];
struct {
    uint16_t num;
    bool enabled;
    uint32_t desc[2];
    uint32_t avail[2];
    uint32_t used[2];
} vqs[VIRTIO_QUEUE_MAX];

This will confuse driver if migrating during initialization. Solves
this issue by:

- introduce transport specific callbacks to load and store extra
  virtqueue states.
- add a new subsection for virtio to migrate transport specific modern
  device state.
- implement pci specific callbacks.
- add a new property for virtio-pci for whether or not to migrate
  extra state.
- compat the migration for 2.4 and elder machine types

Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NCornelia Huck <cornelia.huck@de.ibm.com>

Postcopy detects accesses to pages that haven't been transferred yet
using userfaultfd, and it causes exceptions on pages that are 'not
present'.
Ballooning also causes pages to be marked as 'not present' when the
guest inflates the balloon.
Potentially a balloon could be inflated to discard pages that are
currently inflight during postcopy and that may be arriving at about
the same time.

To avoid this confusion, disable ballooning during postcopy.

When disabled we drop balloon requests from the guest.  Since ballooning
is generally initiated by the host, the management system should avoid
initiating any balloon instructions to the guest during migration,
although it's not possible to know how long it would take a guest to
process a request made prior to the start of migration.
Guest initiated ballooning will not know if it's really freed a page
of host memory or not.

Queueing the requests until after migration would be nice, but is
non-trivial, since the set of inflate/deflate requests have to
be compared with the state of the page to know what the final
outcome is allowed to be.
Signed-off-by: NDr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: NJuan Quintela <quintela@redhat.com>
Reviewed-by: NAmit Shah <amit.shah@redhat.com>
Signed-off-by: NJuan Quintela <quintela@redhat.com>

bring_map currently fails if one of the entries it's mapping is
contigious in GPA but not HVA address space.  Introduce a mapped_len
parameter so it can handle this, returning the actual mapped length.

This will still fail if there's no space left in the sg, but luckily max
queue size in use is currently 256, while max sg size is 1024, so we
should be OK even is all entries happen to cross a single DIMM boundary.

Won't work well with very small DIMM sizes, unfortunately:
e.g. this will fail with 4K DIMMs where a single
request might span a large number of DIMMs.

Let's hope these are uncommon - at least we are not breaking things.
Reported-by: NStefan Hajnoczi <stefanha@redhat.com>
Reported-by: NIgor Mammedov <imammedo@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NIgor Mammedov <imammedo@redhat.com>
Message-id: 1446047243-3221-2-git-send-email-mst@redhat.com
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>

Use address_space_read to make sure we handle the case of an indirect
descriptor crossing DIMM boundary correctly.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NStefan Hajnoczi <stefanha@redhat.com>
Tested-by: NIgor Mammedov <imammedo@redhat.com>
Message-id: 1446047243-3221-1-git-send-email-mst@redhat.com
Signed-off-by: NStefan Hajnoczi <stefanha@redhat.com>

Deprecated in favor of virtqueue_map.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NIgor Mammedov <imammedo@redhat.com>

Drop use of the deprecated virtio_map_sg in virtio core.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NIgor Mammedov <imammedo@redhat.com>

virtio_map_sg currently fails if one of the entries it's mapping is
contigious in GPA but not HVA address space.  Introduce virtio_map which
handles this by splitting sg entries.

This new API generally turns out to be a good idea since it's harder to
misuse: at least in one case the existing one was used incorrectly.

This will still fail if there's no space left in the sg, but luckily max
queue size in use is currently 256, while max sg size is 1024, so we
should be OK even is all entries happen to cross a single DIMM boundary.

Won't work well with very small DIMM sizes, unfortunately:
e.g. this will fail with 4K DIMMs where a single
request might span a large number of DIMMs.

Let's hope these are uncommon - at least we are not breaking things.

Note: virtio-scsi calls virtio_map_sg on data loaded from network, and
validates input, asserting on failure.  Copy the validating code here -
it will be dropped from virtio-scsi in a follow-up patch.
Reported-by: NIgor Mammedov <imammedo@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NStefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: NIgor Mammedov <imammedo@redhat.com>

We are sending msg fields, use sizeof on these
and not on local variables which happen to
have a matching type.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

We are using local msg structures everywhere, use them
for sizeof as well.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Build on RHEL6 fails:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=42875

Apparently unnamed unions couldn't use C99  named field initializers.
Let's just name the payload union field.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

When a live migration is started the log address to mark dirty pages is provided
to the vhost backend through the vhost_dev_set_log function.
This function is called for each queue pairs but the queue index is wrongly set:
always set to the first queue pair. Then vhost backend lost descriptor addresses
of the queue pairs greater than 1 and behaviour of the vhost backend is
unpredictable.

The queue index is computed by taking account of the vq_index (to retrieve the
queue pair index) and calling the vhost_get_vq_index method of the backend.
Signed-off-by: NThibaut Collet <thibaut.collet@6wind.com>
Cc: qemu-stable@nongnu.org
Acked-by: NMichael S. Tsirkin <mst@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>

Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

The VHOST_USER_PROTOCOL_FEATURE_MASK will be automatically updated when
adding new features to the enum.
Signed-off-by: NThibaut Collet <thibaut.collet@6wind.com>
[Adapted from mailing list discussion - Marc-André]
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

A new vhost user message is added to allow QEMU to ask to vhost user backend to
broadcast a fake RARP after live migration for guest without GUEST_ANNOUNCE
capability.

This new message is sent only if the backend supports the new
VHOST_USER_PROTOCOL_F_RARP protocol feature.
The payload of this new message is the MAC address of the guest (not known by
the backend). The MAC address is copied in the first 6 bytes of a u64 to avoid
to create a new payload message type.

This new message has no equivalent ioctl so a new callback is added in the
userOps structure to send the request.

Upon reception of this new message the vhost user backend must generate and
broadcast a fake RARP request to notify the migration is terminated.
Signed-off-by: NThibaut Collet <thibaut.collet@6wind.com>
[Rebased and fixed checkpatch errors - Marc-André]
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

Replace the generic vhost_call() by specific functions for each
function call to help with type safety and changing arguments.

While doing this, I found that "unsigned long long" and "uint64_t" were
used interchangeably and causing compilation warnings, using uint64_t
instead, as the vhost & protocol specifies.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
[Fix enum usage and MQ - Thibaut Collet]
Signed-off-by: NThibaut Collet <thibaut.collet@6wind.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

If VHOST_USER_PROTOCOL_F_LOG_SHMFD is not announced, block vhost-user
migration. The blocker is removed in vhost_dev_cleanup().
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

Send the shm for the dirty pages logging if the backend supports
VHOST_USER_PROTOCOL_F_LOG_SHMFD. Wait for a reply to make sure
the old log is no longer used.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

If the backend is requires it, allocate shareable memory.

vhost_log_get() now uses 2 globals "vhost_log" and "vhost_log_shm", that
way there is a common non-shareable log and a common shareable one.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

Check if the backend has VHOST_USER_PROTOCOL_F_LOG_SHMFD feature and
require a shared log.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

Split VHOST_SET_LOG_BASE call in a seperate function callback, so that
type safety works and more arguments can be added in the next patches.
Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

Signed-off-by: NMarc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Tested-by: NThibaut Collet <thibaut.collet@6wind.com>

Don't initialize vhost backend if memslots number exceeds the supported
limit. This prevents failures down the road when backend
is actually started.

[MST: rewrite commit log]
Signed-off-by: NIgor Mammedov <imammedo@redhat.com>
Reviewed-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>