Add some basic test for rocker to test L2/L3/L4 functionality. Requires an
external test environment, simp, located here:
https://github.com/scottfeldman/simp
To run tests, simp environment must be installed and a suitable VM image built
and installed with a Linux 3.18 (or greater) kernel with rocker driver support
enabled.
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Message-id: 1426306173-24884-8-git-send-email-sfeldma@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Rocker is a simulated ethernet switch device. The device supports up to 62
front-panel ports and supports L2 switching and L3 routing functions, as well
as L2/L3/L4 ACLs. The device presents a single PCI device for each switch,
with a memory-mapped register space for device driver access.
Rocker device is invoked with -device, for example a 4-port switch:
-device rocker,name=sw1,len-ports=4,ports[0]=dev0,ports[1]=dev1, \
ports[2]=dev2,ports[3]=dev3
Each port is a netdev and can be paired with using -netdev id=<port name>.
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Scott Feldman <sfeldma@gmail.com>
Acked-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David Ahern <dsahern@gmail.com>
Message-id: 1426306173-24884-7-git-send-email-sfeldma@gmail.com
rocker: fix clang compiler errors
Consolidate all forward typedef declarations to rocker.h.
Signed-off-by: David Ahern <dsahern@gmail.com>
Acked-by: Scott Feldman <sfeldma@gmail.com>
Acked-by: Jiri Pirko <jiri@resnulli.us>
rocker: add support for flow modification
We had support for flow add/del. This adds support for flow mod. I needed
this for L3 support where an existing route is modified using NLM_F_REPLACE.
For example:
ip route add 12.0.0.0/30 nexthop via 11.0.0.1 dev swp1
ip route change 12.0.0.0/30 nexthop via 11.0.0.9 dev swp2
The first cmd adds the route. The second cmd changes the existing route by
changing its nexthop info.
In the device, a mod operation results in the matching flow enty being modified
with the new settings. This is atomic to the device.
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Rocker is an ethernet switch device, so add 'other' network device class as
defined by PCI to cover these types of devices.
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Message-id: 1426306173-24884-6-git-send-email-sfeldma@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This is the register programming guide for the Rocker device. It's intended
for driver writers and device writers. It covers the device's PCI space,
the register set, DMA interface, and interrupts.
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Message-id: 1426306173-24884-4-git-send-email-sfeldma@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1426306173-24884-3-git-send-email-sfeldma@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
We can use this in virtio-net code as well as new Rocker driver code, so
up-level this.
Signed-off-by: Scott Feldman <sfeldma@gmail.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1426306173-24884-2-git-send-email-sfeldma@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
- two improvements to "info mtere" from Gerd
- KVM support for memory transaction attributes
- one more small step towards unlocked MMIO dispatch
- one piece of the qemu-nbd errno fixes
- trivial-ish patches from Denis and Thomas
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
iQEcBAABCAAGBQJVTLBhAAoJEL/70l94x66DkGIH/jlNJBMBGhlH/lwb1LzxtAMX
OxyDxsiwJpSxsOiZiY3oRz7d6VV6TCrmx5L+1HgG5IzU3WC61Tq6/FK4EXLepZIH
GSYlLtAALWny+2Uwsyh1Z7MMr5yxyhgeORk/l7O8pncmMeysludbEaJqQg8Aa9A2
j0B2pv4tvcl/qhuIFXL1YlbYMVXMsZy5W65D8jq+B6qf3q8kUcdKvgvbUMrxAiSH
JMISo4Z32t9w8SGnhlBa9s4HfN2yOvULRAozzkDBAu4c41cZrw16lvTV8XotamnU
LrG6eQ+2PFeIrcGhuIu7z5Bi4yiRRiThfRLCAvVApVTQYUf7IwvPNa5K1FrP9YU=
=Z/UD
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
- build bugfix from Fam and new configure check from Emilio
- two improvements to "info mtere" from Gerd
- KVM support for memory transaction attributes
- one more small step towards unlocked MMIO dispatch
- one piece of the qemu-nbd errno fixes
- trivial-ish patches from Denis and Thomas
# gpg: Signature made Fri May 8 13:47:29 2015 BST using RSA key ID 78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg: aka "Paolo Bonzini <pbonzini@redhat.com>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4 E2F7 7E15 100C CD36 69B1
# Subkey fingerprint: F133 3857 4B66 2389 866C 7682 BFFB D25F 78C7 AE83
* remotes/bonzini/tags/for-upstream:
qemu-nbd: only send a limited number of errno codes on the wire
rules.mak: Force CFLAGS for all objects in DSO
configure: require __thread support
exec: move rcu_read_lock/unlock to address_space_translate callers
kvm: add support for memory transaction attributes
mtree: also print disabled regions
mtree: tag & indent a bit better
apic_common: improve readability of apic_reset_common
kvm: Silence warning from valgrind
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
- updates for virtio-ccw and s390-virtio, making them more similar
to virtio-pci
- improvements regarding per-vcpu interrupts and migration
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)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=qrI/
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20150508' into staging
Assorted s390x patches:
- updates for virtio-ccw and s390-virtio, making them more similar
to virtio-pci
- improvements regarding per-vcpu interrupts and migration
# gpg: Signature made Fri May 8 09:45:09 2015 BST using RSA key ID C6F02FAF
# gpg: Good signature from "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg: aka "Cornelia Huck <cornelia.huck@de.ibm.com>"
* remotes/cohuck/tags/s390x-20150508:
s390x/kvm: migrate vcpu interrupt state
s390x: move fpu regs into a subsection of the vmstate
s390x/kvm: use ioctl KVM_S390_IRQ for vcpu interrupts
virtio-ccw: implement ->device_plugged
virtio-ccw: change realization sequence
s390-virtio: clear {used,avail}_event_idx on reset as well
s390-virtio: use common features
s390-virtio: Accommodate guests using virtqueues too early
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Right now, NBD includes potentially platform-specific error values in
the wire protocol.
Luckily, most common error values are more or less universal: in
particular, of all errno values <= 34 (up to ERANGE), they are all the
same on supported platforms except for 11 (which is EAGAIN on Windows and
Linux, but EDEADLK on Darwin and the *BSDs). So, in order to guarantee
some portability, only keep a handful of possible error codes and squash
everything else to EINVAL.
This patch defines a limited set of errno values that are valid for the
NBD protocol, and specifies recommendations for what error to return
in specific corner cases. The set of errno values is roughly based on
the errors listed in the read(2) and write(2) man pages, with some
exceptions:
- ENOMEM is added for servers that implement copy-on-write or other
formats that require dynamic allocation.
- EDQUOT is not part of the universal set of errors; it can be changed
to ENOSPC on the wire format.
- EFBIG is part of the universal set of errors, but it is also changed
to ENOSPC because it is pretty similar to ENOSPC or EDQUOT.
Incoming values will in general match system errno values, but not
on the Hurd which has different errno values (they have a "subsystem
code" equal to 0x10 in bits 24-31). The Hurd is probably not something
to which QEMU has been ported, but still do the right thing and
reverse-map the NBD errno values to the system errno values.
The corresponding patch to the NBD protocol description can be found at
http://article.gmane.org/gmane.linux.drivers.nbd.general/3154.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Because of the trick of process-archive-undefs, all .mo objects, even
with --enable-modules, are dependencies of executables.
This breaks CFLAGS propogation because the compiling of module object
will happen too early before building for DSO.
With GCC 5, the linking would fail because .o doesn't have -fPIC. Also,
BUILD_DSO will be missed. (module-common.o will have it, so the stamp
symbol was still liked in .so).
Fix the problem by forcing the CFLAGS on individual .o-cflags during
unnest-vars.
Reported-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Fam Zheng <famz@redhat.com>
Cc: qemu-stable@nongnu.org # 2.3
Message-Id: <1430981715-31465-1-git-send-email-famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
... and the status register should say so.
Fixes "usbus0: controller did not stop" error printed by freebsd.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
When x-root property not be configured, will cause segfault
because of null pointer accessing. Add a check for s->root
property avoid segfault.
Signed-off-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
When we find a IOC bit set on a setup trb and therefore queue an event,
that should not stop events being generated for following data trbs.
So clear the 'reported' flag.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This makes xhci generate multiple short packet events in case of
multi-trb transfers. Which is wrong. We need to fix this in a
different way.
This reverts commit aa6857891d.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The dev->config pointer isn't set until guest
system initializes usb devices (via
usb_desc_set_config). However qemu networking can
go through some motions prior to that, e.g.:
#0 is_rndis (s=0x555557261970) at hw/usb/dev-network.c:653
#1 0x000055555585f723 in usbnet_can_receive (nc=0x55555641e820) at hw/usb/dev-network.c:1315
#2 0x000055555587635e in qemu_can_send_packet (sender=0x5555572660a0) at net/net.c:470
#3 0x0000555555878e34 in net_hub_port_can_receive (nc=0x5555562d7800) at net/hub.c:101
#4 0x000055555587635e in qemu_can_send_packet (sender=0x5555562d7980) at net/net.c:470
#5 0x000055555587dbca in tap_can_send (opaque=0x5555562d7980) at net/tap.c:172
The command to reproduce most reliably was:
qemu-system-i386 -usb -device usb-net,vlan=0 -net tap,vlan=0
This wasn't strictly a problem with tap. Other
networking endpoints (vde, user) could trigger
this problem as well.
Fixes: https://bugs.launchpad.net/qemu/+bug/1050823
Cc: qemu-stable@nongnu.org
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The codebase doesn't build without __thread support.
Formalise this requirement by adding a check for it in the
configure script.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This patch adds support to migrate vcpu interrupts.
We use ioctl KVM_S390_GET_IRQ_STATE and _SET_IRQ_STATE
to get/set the complete interrupt state for a vcpu.
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Let's move the floating point registers into a seperate subsection and
bump up the version id. This cleans up the current vmstate and will
allow for a future extension with vector registers in a compatible way.
This patch is based on a patch from Eric Farman.
Reviewed-by: Eric Farman <farman@linux.vnet.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
KVM_S390_INT uses only two parameter fields. This is not
enough to pass all required information for certain interrupts.
A new ioctl KVM_S390_IRQ is available which allows us to
inject all local interrupts as defined in the Principles of
Operation. It takes a struct kvm_s390_irq as a parameter
which can store interrupt payload data for all interrupts.
Let's use the new ioctl for injecting vcpu interrupts.
Tested-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Let's move operations that are only valid after the backend has been
realized to a ->device_plugged callback, just as virtio-pci does.
Also reorder setting up the host feature bits to the sequence used
by virtio-pci.
While we're at it, also add a ->device_unplugged callback to stop
ioeventfd, just to be on the safe side.
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Message-Id: <1429627016-30656-3-git-send-email-cornelia.huck@de.ibm.com>
virtio-ccw has an odd sequence of realizing devices: first the
device-specific relization (net, block, ...), then the generic
realization. It feels less odd to have the generic realization
callback trigger the device-specific realization instead (and this
also matches what virtio-pci does).
One thing to note: We need to defer initializing the cu model in the
sense id data until after the device-specific realization has been
performed, as we need to refer to the virtio device's device_id.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Message-Id: <1429627016-30656-2-git-send-email-cornelia.huck@de.ibm.com>
The old s390-virtio transport clears the vring used/avail indices in
the shared area on reset. When we enabled event_idx for virtio-blk, we
noticed that this is not enough: We also need to clear the published
used/avail event indices, or reboot will fail.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
We used to avoid enabling event_idx for virtio-blk devices via
s390-virtio, but we now have a workaround in place for guests trying
to use the device before setting DRIVER_OK. Therefore, let's add
DEFINE_VIRTIO_COMMON_FEATURES to the base device so all devices get
those common features - and make s390-virtio use the same mechanism
as the other transports do.
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Shannon Zhao <shannon.zhao@linaro.org>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Feature updates are not a synchronuous operation for the legacy
s390-virtio transport. This transport syncs the guest feature bits
(those from finalize) on the set_status hypercall. Before that qemu
thinks that features are zero, which means QEMU will misbehave, e.g.
it will not write the event index, even if the guest asks for it.
Let's detect the case where a kick happens before the driver is ready
and force sync the features.
With this workaround, it is now safe to switch to the common feature
bit handling code as used by all other transports.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=uSLC
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/juanquintela/tags/migration/20150507-1' into staging
migration/next for 20150507
# gpg: Signature made Thu May 7 17:42:19 2015 BST using RSA key ID 5872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>"
# gpg: aka "Juan Quintela <quintela@trasno.org>"
* remotes/juanquintela/tags/migration/20150507-1:
migration: Fix migration state update issue
migration: avoid divide by zero in xbzrle cache miss rate
migration: Add hmp interface to set and query parameters
migration: Add qmp commands to set and query parameters
migration: Use an array instead of 3 parameters
migration: Add interface to control compression
migration: Add the core code for decompression
migration: Make compression co-work with xbzrle
migration: Add the core code of multi-thread compression
migration: Split save_zero_page from ram_save_page
arch_init: Add and free data struct for decompression
arch_init: Alloc and free data struct for compression
qemu-file: Add compression functions to QEMUFile
migration: Add the framework of multi-thread decompression
migration: Add the framework of multi-thread compression
docs: Add a doc about multiple thread compression
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
If live migration is very fast and can be completed in 1 second,
the dirty_sync_count of MigrationState will not be updated.
Then you will see "dirty sync count: 0" in qemu monitor even if
the actual dirty sync count is not 0.
Signed-off-by: Liang Li <liang.z.li@intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr.David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This bug manifested itself as a VM that could not be resumed by libvirt
following a migration:
# virsh resume example
error: Failed to resume domain example
error: internal error: cannot parse json {"return":
{"xbzrle-cache":
{..., "cache-miss-rate": -nan, ...},
...
}
}: lexical error: malformed number, a digit is required after the minus sign.
This patch also ensures xbzrle_cache_miss_prev and iterations_prev are
cleared at the start of the migration.
Signed-off-by: Michael Chapman <mike@very.puzzling.org>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Add the hmp interface to tune and query the parameters used in
live migration.
Signed-off-by: Liang Li <liang.z.li@intel.com>
Signed-off-by: Yang Zhang <yang.z.zhang@intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Add the qmp commands to tune and query the parameters used in live
migration.
Signed-off-by: Liang Li <liang.z.li@intel.com>
Signed-off-by: Yang Zhang <yang.z.zhang@intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Put the three parameters related to multiple thread (de)compression
into an int array, and use an enum type to index the parameter.
Signed-off-by: Liang Li <liang.z.li@intel.com>
Signed-off-by: Yang Zhang <yang.z.zhang@intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
