We declare incomplete struct VMStateDescription in a couple of places
so we don't have to include migration/vmstate.h for the typedef.
That's fine with me. However, the next commit will drop
migration/vmstate.h from a massive number of compiles. Move the
typedef to qemu/typedefs.h now, so I don't have to insert struct in
front of VMStateDescription all over the place then.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190812052359.30071-15-armbru@redhat.com>
Previous to this change, a vCPU exposed with VMX running on a kernel
without KVM_CAP_NESTED_STATE or KVM_CAP_EXCEPTION_PAYLOAD resulted in
adding a migration blocker. This was because when the code was written
it was thought there is no way to reliably know if a vCPU is utilising
VMX or not at runtime. However, it turns out that this can be known to
some extent:
In order for a vCPU to enter VMX operation it must have CR4.VMXE set.
Since it was set, CR4.VMXE must remain set as long as the vCPU is in
VMX operation. This is because CR4.VMXE is one of the bits set
in MSR_IA32_VMX_CR4_FIXED1.
There is one exception to the above statement when vCPU enters SMM mode.
When a vCPU enters SMM mode, it temporarily exits VMX operation and
may also reset CR4.VMXE during execution in SMM mode.
When the vCPU exits SMM mode, vCPU state is restored to be in VMX operation
and CR4.VMXE is restored to its original state of being set.
Therefore, when the vCPU is not in SMM mode, we can infer whether
VMX is being used by examining CR4.VMXE. Otherwise, we cannot
know for certain but assume the worse that vCPU may utilise VMX.
Summaring all the above, a vCPU may have enabled VMX in case
CR4.VMXE is set or vCPU is in SMM mode.
Therefore, remove migration blocker and check before migration
(cpu_pre_save()) if the vCPU may have enabled VMX. If true, only then
require relevant kernel capabilities.
While at it, demand KVM_CAP_EXCEPTION_PAYLOAD only when the vCPU is in
guest-mode and there is a pending/injected exception. Otherwise, this
kernel capability is not required for proper migration.
Reviewed-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Maran Wilson <maran.wilson@oracle.com>
Tested-by: Maran Wilson <maran.wilson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This will make unversioned CPU models behavior depend on the
machine type:
* "pc-*-4.0" and older will not report them as aliases.
This is done to keep compatibility with older QEMU versions
after management software starts translating aliases.
* "pc-*-4.1" will translate unversioned CPU models to -v1.
This is done to keep compatibility with existing management
software, that still relies on CPU model runnability promises.
* "none" will translate unversioned CPU models to their latest
version. This is planned become the default in future machine
types (probably in pc-*-4.3).
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20190628002844.24894-8-ehabkost@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Add support for registration of multiple versions of CPU models.
The existing CPU models will be registered with a "-v1" suffix.
The -noTSX, -IBRS, and -IBPB CPU model variants will become
versions of the original models in a separate patch, so
make sure we register no versions for them.
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20190628002844.24894-5-ehabkost@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
The CPUID.1F as Intel V2 Extended Topology Enumeration Leaf would be
exposed if guests want to emulate multiple software-visible die within
each package. Per Intel's SDM, the 0x1f is a superset of 0xb, thus they
can be generated by almost same code as 0xb except die_offset setting.
If the number of dies per package is greater than 1, the cpuid_min_level
would be adjusted to 0x1f regardless of whether the host supports CPUID.1F.
Likewise, the CPUID.1F wouldn't be exposed if env->nr_dies < 2.
Suggested-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Like Xu <like.xu@linux.intel.com>
Message-Id: <20190620054525.37188-2-like.xu@linux.intel.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
The current default value for hv-spinlocks is 0xFFFFFFFF (meaning
"never retry"). However, the value is stored as a signed
integer, making the getter of the hv-spinlocks QOM property
return -1 instead of 0xFFFFFFFF.
Fix this by changing the type of X86CPU::hyperv_spinlock_attempts
to uint32_t. This has no visible effect to guest operating
systems, affecting just the behavior of the QOM getter.
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20190615200505.31348-1-ehabkost@redhat.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
The field die_id (default as 0) and has_die_id are introduced to X86CPU.
Following the legacy smp check rules, the die_id validity is added to
the same contexts as leagcy smp variables such as hmp_hotpluggable_cpus(),
machine_set_cpu_numa_node(), cpu_slot_to_string() and pc_cpu_pre_plug().
Acked-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Like Xu <like.xu@linux.intel.com>
Message-Id: <20190612084104.34984-4-like.xu@linux.intel.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
The die-level as the first PC-specific cpu topology is added to the leagcy
cpu topology model, which has one die per package implicitly and only the
numbers of sockets/cores/threads are configurable.
In the new model with die-level support, the total number of logical
processors (including offline) on board will be calculated as:
#cpus = #sockets * #dies * #cores * #threads
and considering compatibility, the default value for #dies would be
initialized to one in x86_cpu_initfn() and pc_machine_initfn().
Signed-off-by: Like Xu <like.xu@linux.intel.com>
Message-Id: <20190612084104.34984-2-like.xu@linux.intel.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Kernel commit c4f55198c7c2 ("kvm: x86: Introduce KVM_CAP_EXCEPTION_PAYLOAD")
introduced a new KVM capability which allows userspace to correctly
distinguish between pending and injected exceptions.
This distinguish is important in case of nested virtualization scenarios
because a L2 pending exception can still be intercepted by the L1 hypervisor
while a L2 injected exception cannot.
Furthermore, when an exception is attempted to be injected by QEMU,
QEMU should specify the exception payload (CR2 in case of #PF or
DR6 in case of #DB) instead of having the payload already delivered in
the respective vCPU register. Because in case exception is injected to
L2 guest and is intercepted by L1 hypervisor, then payload needs to be
reported to L1 intercept (VMExit handler) while still preserving
respective vCPU register unchanged.
This commit adds support for QEMU to properly utilise this new KVM
capability (KVM_CAP_EXCEPTION_PAYLOAD).
Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20190619162140.133674-10-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Commit d98f26073b ("target/i386: kvm: add VMX migration blocker")
added a migration blocker for vCPU exposed with Intel VMX.
However, migration should also be blocked for vCPU exposed with
AMD SVM.
Both cases should be blocked because QEMU should extract additional
vCPU state from KVM that should be migrated as part of vCPU VMState.
E.g. Whether vCPU is running in guest-mode or host-mode.
Fixes: d98f26073b ("target/i386: kvm: add VMX migration blocker")
Reviewed-by: Maran Wilson <maran.wilson@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20190619162140.133674-6-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hyper-V on KVM can only use Synthetic timers with Direct Mode (opting for
an interrupt instead of VMBus message). This new capability is only
announced in KVM_GET_SUPPORTED_HV_CPUID.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20190517141924.19024-10-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In many case we just want to give Windows guests all currently supported
Hyper-V enlightenments and that's where this new mode may come handy. We
pass through what was returned by KVM_GET_SUPPORTED_HV_CPUID.
hv_cpuid_check_and_set() is modified to also set cpu->hyperv_* flags as
we may want to check them later (and we actually do for hv_runtime,
hv_synic,...).
'hv-passthrough' is a development only feature, a migration blocker is
added to prevent issues while migrating between hosts with different
feature sets.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20190517141924.19024-6-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Other accelerators have their own headers: sysemu/hax.h, sysemu/hvf.h,
sysemu/kvm.h, sysemu/whpx.h. Only tcg_enabled() & friends sit in
qemu-common.h. This necessitates inclusion of qemu-common.h into
headers, which is against the rules spelled out in qemu-common.h's
file comment.
Move tcg_enabled() & friends into their own header sysemu/tcg.h, and
adjust #include directives.
Cc: Richard Henderson <rth@twiddle.net>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190523143508.25387-2-armbru@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
[Rebased with conflicts resolved automatically, except for
accel/tcg/tcg-all.c]
This macro is now always empty, so remove it. This leaves the
entire contents of CPUArchState under the control of the guest
architecture.
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Cleanup in the boilerplate that each target must define.
Replace x86_env_get_cpu with env_archcpu. The combination
CPU(x86_env_get_cpu) should have used ENV_GET_CPU to begin;
use env_cpu now.
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
For all targets, into this new file move TARGET_LONG_BITS,
TARGET_PAGE_BITS, TARGET_PHYS_ADDR_SPACE_BITS,
TARGET_VIRT_ADDR_SPACE_BITS, and NB_MMU_MODES.
Include this new file from exec/cpu-defs.h.
This now removes the somewhat odd requirement that target/arch/cpu.h
defines TARGET_LONG_BITS before including exec/cpu-defs.h, so push the
bulk of the includes within target/arch/cpu.h to the top.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Add a new base CPU model called 'Dhyana' to model processors from Hygon
Dhyana(family 18h), which derived from AMD EPYC(family 17h).
The following features bits have been removed compare to AMD EPYC:
aes, pclmulqdq, sha_ni
The Hygon Dhyana support to KVM in Linux is already accepted upstream[1].
So add Hygon Dhyana support to Qemu is necessary to create Hygon's own
CPU model.
Reference:
[1] https://git.kernel.org/tip/fec98069fb72fb656304a3e52265e0c2fc9adf87
Signed-off-by: Pu Wen <puwen@hygon.cn>
Message-Id: <1555416373-28690-1-git-send-email-puwen@hygon.cn>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
CPUClass method dump_statistics() takes an fprintf()-like callback and
a FILE * to pass to it. Most callers pass fprintf() and stderr.
log_cpu_state() passes fprintf() and qemu_log_file.
hmp_info_registers() passes monitor_fprintf() and the current monitor
cast to FILE *. monitor_fprintf() casts it right back, and is
otherwise identical to monitor_printf().
The callback gets passed around a lot, which is tiresome. The
type-punning around monitor_fprintf() is ugly.
Drop the callback, and call qemu_fprintf() instead. Also gets rid of
the type-punning, since qemu_fprintf() takes NULL instead of the
current monitor cast to FILE *.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20190417191805.28198-15-armbru@redhat.com>
x86_cpu_dump_local_apic_state() takes an fprintf()-like callback and a
FILE * to pass to it, and so do its helper functions.
Its only caller hmp_info_local_apic() passes monitor_fprintf() and the
current monitor cast to FILE *. monitor_fprintf() casts it right
back, and is otherwise identical to monitor_printf(). The
type-punning is ugly.
Drop the callback, and call qemu_printf() instead.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20190417191805.28198-12-armbru@redhat.com>
The various TARGET_cpu_list() take an fprintf()-like callback and a
FILE * to pass to it. Their callers (vl.c's main() via list_cpus(),
bsd-user/main.c's main(), linux-user/main.c's main()) all pass
fprintf() and stdout. Thus, the flexibility provided by the (rather
tiresome) indirection isn't actually used.
Drop the callback, and call qemu_printf() instead.
Calling printf() would also work, but would make the code unsuitable
for monitor context without making it simpler.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20190417191805.28198-10-armbru@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Some downstream distributions of QEMU set host-phys-bits=on by
default. This worked very well for most use cases, because
phys-bits really didn't have huge consequences. The only
difference was on the CPUID data seen by guests, and on the
handling of reserved bits.
This changed in KVM commit 855feb673640 ("KVM: MMU: Add 5 level
EPT & Shadow page table support"). Now choosing a large
phys-bits value for a VM has bigger impact: it will make KVM use
5-level EPT even when it's not really necessary. This means
using the host phys-bits value may not be the best choice.
Management software could address this problem by manually
configuring phys-bits depending on the size of the VM and the
amount of MMIO address space required for hotplug. But this is
not trivial to implement.
However, there's another workaround that would work for most
cases: keep using the host phys-bits value, but only if it's
smaller than 48. This patch makes this possible by introducing a
new "-cpu" option: "host-phys-bits-limit". Management software
or users can make sure they will always use 4-level EPT using:
"host-phys-bits=on,host-phys-bits-limit=48".
This behavior is still not enabled by default because QEMU
doesn't enable host-phys-bits=on by default. But users,
management software, or downstream distributions may choose to
change their defaults using the new option.
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20181211192527.13254-1-ehabkost@redhat.com>
[ehabkost: removed test code while some issues are addressed]
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
It was found that QMP users of QEMU (e.g. libvirt) may need
HV_CPUID_ENLIGHTMENT_INFO.EAX/HV_CPUID_NESTED_FEATURES.EAX information. In
particular, 'hv_tlbflush' and 'hv_evmcs' enlightenments are only exposed in
HV_CPUID_ENLIGHTMENT_INFO.EAX.
HV_CPUID_NESTED_FEATURES.EAX is exposed for two reasons: convenience
(we don't need to export it from hyperv_handle_properties() and as
future-proof for Enlightened MSR-Bitmap, PV EPT invalidation and
direct virtual flush features.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20181126135958.20956-1-vkuznets@redhat.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Certain configurations do not allow SynIC to be used in QEMU. In
particular,
- when hyperv_vpindex is off, SINT routes can't be used as they refer to
the destination vCPU by vp_index
- older KVM (which doesn't expose KVM_CAP_HYPERV_SYNIC2) zeroes out
SynIC message and event pages on every msr load, breaking migration
OTOH in-KVM users of SynIC -- SynIC timers -- do work in those
configurations, and we shouldn't stop the guest from using them.
To cover both scenarios, introduce an X86CPU property that makes CPU
init code to skip creation of the SynIC object (and thus disables any
SynIC use in QEMU) but keeps the KVM part of the SynIC working.
The property is clear by default but is set via compat logic for older
machine types.
As a result, when hv_synic and a modern machine type are specified, QEMU
will refuse to run unless vp_index is on and the kernel is recent
enough. OTOH with an older machine type QEMU will run fine with
hv_synic=on against an older kernel and/or without vp_index enabled but
will disallow the in-QEMU uses of SynIC (in e.g. VMBus).
Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Message-Id: <20180921082217.29481-4-rkagan@virtuozzo.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hyper-V PV IPI support is merged to KVM, enable the feature in Qemu. When
enabled, this allows Windows guests to send IPIs to other vCPUs with a
single hypercall even when there are >64 vCPUs in the request.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Roman Kagan <rkagan@virtuozzo.com>
Message-Id: <20181009130853.6412-3-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This flag will be used for KVM's nested VMX migration; the HF_GUEST_MASK name
is already used in KVM, adopt it in QEMU as well.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Interrupt handling depends on various flags in env->hflags or env->hflags2,
and the exact detail were not exactly replicated between x86_cpu_has_work
and x86_cpu_exec_interrupt. Create a new function that extracts the
highest-priority non-masked interrupt, and use it in both functions.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
