migration/tls: save hostname into MigrationState
migration/tls: extract migration_tls_client_create for common-use
migration/tls: add tls_hostname into MultiFDSendParams
migration/tls: extract cleanup function for common-use
migration/tls: add support for multifd tls-handshake
migration/tls: add trace points for multifd-tls
qemu-file: Don't do IO after shutdown
multifd: Make sure that we don't do any IO after an error
migration: Don't send data if we have stopped
migration: Create migration_is_running()
migration: fix COLO broken caused by a previous commit
migration/multifd: fix hangup with TLS-Multifd due to blocking handshake
multifd/tls: fix memoryleak of the QIOChannelSocket object when cancelling migration
Signed-off-by: Alex Chen <alex.chen@huawei.com>
migration/dirtyrate: setup up query-dirtyrate framwork
migration/dirtyrate: add DirtyRateStatus to denote calculation status
migration/dirtyrate: Add RamblockDirtyInfo to store sampled page info
migration/dirtyrate: Add dirtyrate statistics series functions
migration/dirtyrate: move RAMBLOCK_FOREACH_MIGRATABLE into ram.h
migration/dirtyrate: Record hash results for each sampled page
migration/dirtyrate: Compare page hash results for recorded sampled page
migration/dirtyrate: skip sampling ramblock with size below MIN_RAMBLOCK_SIZE
migration/dirtyrate: Implement set_sample_page_period() and is_sample_period_valid()
migration/dirtyrate: Implement calculate_dirtyrate() function
migration/dirtyrate: Implement qmp_cal_dirty_rate()/qmp_get_dirty_rate() function
migration/dirtyrate: Add trace_calls to make it easier to debug
migration/dirtyrate: record start_time and calc_time while at the measuring state
migration/dirtyrate: present dirty rate only when querying the rate has completed
migration/dirtyrate: simplify includes in dirtyrate.c
Signed-off-by: Alex Chen <alex.chen@huawei.com>
elf2dmp: Fix memory leak on main() error paths
io: Don't use '#' flag of printf format
hw/display/omap_lcdc: Fix potential NULL pointer dereference
hw/display/exynos4210_fimd: Fix potential NULL pointer dereference
block/vvfat: Fix bad printf format specifiers
block: Remove unused include
ssi: Fix bad printf format specifiers
net/l2tpv3: Remove redundant check in net_init_l2tpv3()
Signed-off-by: Alex Chen <alex.chen@huawei.com>
Drop IPv6 message shorter than what's mentioned in the payload
length header (+ the size of the IPv6 header). They're invalid and could
lead to data leakage in icmp6_send_echoreply().
evtstrm and cpuid cann't be controlled by VMM:
1. evtstrm: The generic timer is configured to generate events at a
frequency of approximately 100KHz. It's controlled by the linux
kernel config CONFIG_ARM_ARCH_TIMER_EVTSTREAM.
2. cpuid: EL0 access to certain ID registers is available. It's always
set by linux kernel after 77c97b4ee2129 ("arm64: cpufeature: Expose
CPUID registers by emulation").
However, they are exposed by getauxval() and /proc/cpuinfo. Hence,
let's report and ignore the CPU features if someone set them.
Signed-off-by: Peng Liang <liangpeng10@huawei.com>
QEMU does not support disable/enable CPU features in AArch64 for now.
This patch series add support for CPU features in AArch64.
Firstly, we change the isar struct in ARMCPU to an array for
convenience. Secondly, we add support to configure CPU feautres in
AArch64 and make sure that the ID registers can be synchronized to KVM
so that guest can read the value we configure. Thirdly, we add a
mechanism to solve the dependency relationship of some CPU features.
Last, we add a KVM_CAP_ARM_CPU_FEATURE to check whether KVM supports to
set CPU features in AArch64.
Also export CPU features to the result of qmp query-cpu-model-expansion
so that libvirt can get the supported CPU features.
Update the ID fields to ARMv8.6 and add some CPU features according to
the new ID fields.
With related KVM patch set[1], we can disable/enable CPU features in
AArch64.
[1] https://patchwork.kernel.org/cover/11711693/
Signed-off-by: Peng Liang <liangpeng10@huawei.com>
To support CPU feature in AArch64, we need to move some field from
ARMCPU to ARMISARegisters, add more definitions of ID fields, and add
suport query-cpu-model-expansion qmp command. Let's backport upstream
patches to do these.
Signed-off-by: Peng Liang <liangpeng10@huawei.com>
if vcpu support smt, create new smt hierarchy according to
Processor Properties Topology Table(PPTT) in acpi spec 6.3.
Threads sharing a core must be grouped under a unique Processor
hierarchy node structure for each group of threads
Signed-off-by: Henglong Fan <fanhenglong@huawei.com>
Since filemonitor testcase requires that host kernel being a LTS version,
we cannot guarantee that on OBS system. Let's disable it by default.
Signed-off-by: Ying Fang <fangying1@huawei.com>
Add the kvm_adjvtime vcpu property for ARM Cortex-A72 cpu model,
so that virtual time adjust will be enabled for it.
Signed-off-by: Ying Fang <fangying1@huawei.com>
This patch drops the vtimer virtual timer adjust, cross version migration
from openEuler qemu-4.0.1 to qemu-4.1.0 is not supported as a consequence.
By default openEuler qemu-4.1.0 use kvm_adjvtime as the virtual timer.
Signed-off-by: Ying Fang <fangying1@huawei.com>
Vtimer adjust is used in openEuler qemu-4.0.1, however kvm_adjvtime
is introduced in openEuler qemu-4.1.0. To maintain the compatibility
and enable cross version migration, let's enable vtimer adjust only
if kvm_adjvtime is not enabled, otherwise there may be conflicts
between vtimer adjust and kvm_adjvtime.
After this modification:
1: openEuler qemu-4.0.1 use vtimer as the default virtual timer
2: openEuler qemu-4.1.0 use kvm_adjvtime as the defaut virtual timer
Migration from openEuler qemu-4.0.1 to openEuler qemu-4.1.0 will
be ok, but migration path from upstream qemu-4.0.1 to openEuler
qemu-4..0.1 will be broken.
Since openEuler qemu-4.1.0, kvm_adjvtime is used as the default
virtual timer. So please upgrade to openEuler qemu-4.1.0 and
use the virt-4.1 machine.
Signed-off-by: Ying Fang <fangying1@huawei.com>
Machine compatibility for kvm-no-adjvtime is missed,
let's add it for virt machine 4.0, thus kvm-no-adjvtime
is supported in v4.1.0.
Signed-off-by: Ying Fang <fangying1@huawei.com>
To support cross version migration, we had to add the vtimer back
which was introduced in openEuler qemu-4.0.1.
Signed-off-by: Ying Fang <fangying1@huawei.com>
This adds CPU cold plug support to arm virt machine board.
CPU cold plug means adding CPU by using "-device xx-arm-cpu"
when we bring up Qemu.
Signed-off-by: Keqian Zhu <zhukeqian1@huawei.com>
