Both lock holder vCPU and IPI receiver that has halted are condidate for
boost. However, the PLE handler was originally designed to deal with the
lock holder preemption problem. The Intel PLE occurs when the spinlock
waiter is in kernel mode. This assumption doesn't hold for IPI receiver,
they can be in either kernel or user mode. the vCPU candidate in user mode
will not be boosted even if they should respond to IPIs. Some benchmarks
like pbzip2, swaptions etc do the TLB shootdown in kernel mode and most
of the time they are running in user mode. It can lead to a large number
of continuous PLE events because the IPI sender causes PLE events
repeatedly until the receiver is scheduled while the receiver is not
candidate for a boost.
This patch boosts the vCPU candidiate in user mode which is delivery
interrupt. We can observe the speed of pbzip2 improves 10% in 96 vCPUs
VM in over-subscribe scenario (The host machine is 2 socket, 48 cores,
96 HTs Intel CLX box). There is no performance regression for other
benchmarks like Unixbench spawn (most of the time contend read/write
lock in kernel mode), ebizzy (most of the time contend read/write sem
and TLB shoodtdown in kernel mode).
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Message-Id: <1618542490-14756-1-git-send-email-wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add a capability for userspace to mirror SEV encryption context from
one vm to another. On our side, this is intended to support a
Migration Helper vCPU, but it can also be used generically to support
other in-guest workloads scheduled by the host. The intention is for
the primary guest and the mirror to have nearly identical memslots.
The primary benefits of this are that:
1) The VMs do not share KVM contexts (think APIC/MSRs/etc), so they
can't accidentally clobber each other.
2) The VMs can have different memory-views, which is necessary for post-copy
migration (the migration vCPUs on the target need to read and write to
pages, when the primary guest would VMEXIT).
This does not change the threat model for AMD SEV. Any memory involved
is still owned by the primary guest and its initial state is still
attested to through the normal SEV_LAUNCH_* flows. If userspace wanted
to circumvent SEV, they could achieve the same effect by simply attaching
a vCPU to the primary VM.
This patch deliberately leaves userspace in charge of the memslots for the
mirror, as it already has the power to mess with them in the primary guest.
This patch does not support SEV-ES (much less SNP), as it does not
handle handing off attested VMSAs to the mirror.
For additional context, we need a Migration Helper because SEV PSP
migration is far too slow for our live migration on its own. Using
an in-guest migrator lets us speed this up significantly.
Signed-off-by: Nathan Tempelman <natet@google.com>
Message-Id: <20210408223214.2582277-1-natet@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
According to section "Canonicalization and Consistency Checks" in APM vol 2,
the following guest state is illegal:
"The MSR or IOIO intercept tables extend to a physical address that
is greater than or equal to the maximum supported physical address."
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Message-Id: <20210412215611.110095-5-krish.sadhukhan@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Define the actual size of the IOPM and MSRPM tables so that the actual size
can be used when initializing them and when checking the consistency of their
physical address.
These #defines are placed in svm.h so that they can be shared.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Message-Id: <20210412215611.110095-2-krish.sadhukhan@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add a capability, KVM_CAP_SGX_ATTRIBUTE, that can be used by userspace
to grant a VM access to a priveleged attribute, with args[0] holding a
file handle to a valid SGX attribute file.
The SGX subsystem restricts access to a subset of enclave attributes to
provide additional security for an uncompromised kernel, e.g. to prevent
malware from using the PROVISIONKEY to ensure its nodes are running
inside a geniune SGX enclave and/or to obtain a stable fingerprint.
To prevent userspace from circumventing such restrictions by running an
enclave in a VM, KVM restricts guest access to privileged attributes by
default.
Cc: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <0b099d65e933e068e3ea934b0523bab070cb8cea.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Enable SGX virtualization now that KVM has the VM-Exit handlers needed
to trap-and-execute ENCLS to ensure correctness and/or enforce the CPU
model exposed to the guest. Add a KVM module param, "sgx", to allow an
admin to disable SGX virtualization independent of the kernel.
When supported in hardware and the kernel, advertise SGX1, SGX2 and SGX
LC to userspace via CPUID and wire up the ENCLS_EXITING bitmap based on
the guest's SGX capabilities, i.e. to allow ENCLS to be executed in an
SGX-enabled guest. With the exception of the provision key, all SGX
attribute bits may be exposed to the guest. Guest access to the
provision key, which is controlled via securityfs, will be added in a
future patch.
Note, KVM does not yet support exposing ENCLS_C leafs or ENCLV leafs.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <a99e9c23310c79f2f4175c1af4c4cbcef913c3e5.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add a VM-Exit handler to trap-and-execute EINIT when SGX LC is enabled
in the host. When SGX LC is enabled, the host kernel may rewrite the
hardware values at will, e.g. to launch enclaves with different signers,
thus KVM needs to intercept EINIT to ensure it is executed with the
correct LE hash (even if the guest sees a hardwired hash).
Switching the LE hash MSRs on VM-Enter/VM-Exit is not a viable option as
writing the MSRs is prohibitively expensive, e.g. on SKL hardware each
WRMSR is ~400 cycles. And because EINIT takes tens of thousands of
cycles to execute, the ~1500 cycle overhead to trap-and-execute EINIT is
unlikely to be noticed by the guest, let alone impact its overall SGX
performance.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <57c92fa4d2083eb3be9e6355e3882fc90cffea87.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Emulate the four Launch Enclave public key hash MSRs (LE hash MSRs) that
exist on CPUs that support SGX Launch Control (LC). SGX LC modifies the
behavior of ENCLS[EINIT] to use the LE hash MSRs when verifying the key
used to sign an enclave. On CPUs without LC support, the LE hash is
hardwired into the CPU to an Intel controlled key (the Intel key is also
the reset value of the LE hash MSRs). Track the guest's desired hash so
that a future patch can stuff the hash into the hardware MSRs when
executing EINIT on behalf of the guest, when those MSRs are writable in
host.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Co-developed-by: Kai Huang <kai.huang@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <c58ef601ddf88f3a113add837969533099b1364a.1618196135.git.kai.huang@intel.com>
[Add a comment regarding the MSRs being available until SGX is locked.
- Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add an ECREATE handler that will be used to intercept ECREATE for the
purpose of enforcing and enclave's MISCSELECT, ATTRIBUTES and XFRM, i.e.
to allow userspace to restrict SGX features via CPUID. ECREATE will be
intercepted when any of the aforementioned masks diverges from hardware
in order to enforce the desired CPUID model, i.e. inject #GP if the
guest attempts to set a bit that hasn't been enumerated as allowed-1 in
CPUID.
Note, access to the PROVISIONKEY is not yet supported.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Co-developed-by: Kai Huang <kai.huang@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <c3a97684f1b71b4f4626a1fc3879472a95651725.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Introduce sgx.c and sgx.h, along with the framework for handling ENCLS
VM-Exits. Add a bool, enable_sgx, that will eventually be wired up to a
module param to control whether or not SGX virtualization is enabled at
runtime.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <1c782269608b2f5e1034be450f375a8432fb705d.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add support for handling VM-Exits that originate from a guest SGX
enclave. In SGX, an "enclave" is a new CPL3-only execution environment,
wherein the CPU and memory state is protected by hardware to make the
state inaccesible to code running outside of the enclave. When exiting
an enclave due to an asynchronous event (from the perspective of the
enclave), e.g. exceptions, interrupts, and VM-Exits, the enclave's state
is automatically saved and scrubbed (the CPU loads synthetic state), and
then reloaded when re-entering the enclave. E.g. after an instruction
based VM-Exit from an enclave, vmcs.GUEST_RIP will not contain the RIP
of the enclave instruction that trigered VM-Exit, but will instead point
to a RIP in the enclave's untrusted runtime (the guest userspace code
that coordinates entry/exit to/from the enclave).
To help a VMM recognize and handle exits from enclaves, SGX adds bits to
existing VMCS fields, VM_EXIT_REASON.VMX_EXIT_REASON_FROM_ENCLAVE and
GUEST_INTERRUPTIBILITY_INFO.GUEST_INTR_STATE_ENCLAVE_INTR. Define the
new architectural bits, and add a boolean to struct vcpu_vmx to cache
VMX_EXIT_REASON_FROM_ENCLAVE. Clear the bit in exit_reason so that
checks against exit_reason do not need to account for SGX, e.g.
"if (exit_reason == EXIT_REASON_EXCEPTION_NMI)" continues to work.
KVM is a largely a passive observer of the new bits, e.g. KVM needs to
account for the bits when propagating information to a nested VMM, but
otherwise doesn't need to act differently for the majority of VM-Exits
from enclaves.
The one scenario that is directly impacted is emulation, which is for
all intents and purposes impossible[1] since KVM does not have access to
the RIP or instruction stream that triggered the VM-Exit. The inability
to emulate is a non-issue for KVM, as most instructions that might
trigger VM-Exit unconditionally #UD in an enclave (before the VM-Exit
check. For the few instruction that conditionally #UD, KVM either never
sets the exiting control, e.g. PAUSE_EXITING[2], or sets it if and only
if the feature is not exposed to the guest in order to inject a #UD,
e.g. RDRAND_EXITING.
But, because it is still possible for a guest to trigger emulation,
e.g. MMIO, inject a #UD if KVM ever attempts emulation after a VM-Exit
from an enclave. This is architecturally accurate for instruction
VM-Exits, and for MMIO it's the least bad choice, e.g. it's preferable
to killing the VM. In practice, only broken or particularly stupid
guests should ever encounter this behavior.
Add a WARN in skip_emulated_instruction to detect any attempt to
modify the guest's RIP during an SGX enclave VM-Exit as all such flows
should either be unreachable or must handle exits from enclaves before
getting to skip_emulated_instruction.
[1] Impossible for all practical purposes. Not truly impossible
since KVM could implement some form of para-virtualization scheme.
[2] PAUSE_LOOP_EXITING only affects CPL0 and enclaves exist only at
CPL3, so we also don't need to worry about that interaction.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <315f54a8507d09c292463ef29104e1d4c62e9090.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Define a new KVM-only feature word for advertising and querying SGX
sub-features in CPUID.0x12.0x0.EAX. Because SGX1 and SGX2 are scattered
in the kernel's feature word, they need to be translated so that the
bit numbers match those of hardware.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <e797c533f4c71ae89265bbb15a02aef86b67cbec.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Introduce a scheme that allows KVM's CPUID magic to support features
that are scattered in the kernel's feature words. To advertise and/or
query guest support for CPUID-based features, KVM requires the bit
number of an X86_FEATURE_* to match the bit number in its associated
CPUID entry. For scattered features, this does not hold true.
Add a framework to allow defining KVM-only words, stored in
kvm_cpu_caps after the shared kernel caps, that can be used to gather
the scattered feature bits by translating X86_FEATURE_* flags into their
KVM-defined feature.
Note, because reverse_cpuid_check() effectively forces kvm_cpu_caps
lookups to be resolved at compile time, there is no runtime cost for
translating from kernel-defined to kvm-defined features.
More details here: https://lkml.kernel.org/r/X/jxCOLG+HUO4QlZ@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <16cad8d00475f67867fb36701fc7fb7c1ec86ce1.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Export the gva_to_gpa() helpers for use by SGX virtualization when
executing ENCLS[ECREATE] and ENCLS[EINIT] on behalf of the guest.
To execute ECREATE and EINIT, KVM must obtain the GPA of the target
Secure Enclave Control Structure (SECS) in order to get its
corresponding HVA.
Because the SECS must reside in the Enclave Page Cache (EPC), copying
the SECS's data to a host-controlled buffer via existing exported
helpers is not a viable option as the EPC is not readable or writable
by the kernel.
SGX virtualization will also use gva_to_gpa() to obtain HVAs for
non-EPC pages in order to pass user pointers directly to ECREATE and
EINIT, which avoids having to copy pages worth of data into the kernel.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Acked-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <02f37708321bcdfaa2f9d41c8478affa6e84b04d.1618196135.git.kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add compile-time assertions in vmcs_check32() to disallow accesses to
64-bit and 64-bit high fields via vmcs_{read,write}32(). Upper level KVM
code should never do partial accesses to VMCS fields. KVM handles the
split accesses automatically in vmcs_{read,write}64() when running as a
32-bit kernel.
Reviewed-and-tested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Haiwei Li <lihaiwei@tencent.com>
Message-Id: <20210409022456.23528-1-lihaiwei.kernel@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Explicitly document why a vmcb must be marked dirty and assigned a new
asid when it will be run on a different cpu. The "what" is relatively
obvious, whereas the "why" requires reading the APM and/or KVM code.
Opportunistically remove a spurious period and several unnecessary
newlines in the comment.
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210406171811.4043363-5-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add a comment above the declaration of vcpu_svm.vmcb to call out that it
is simply a shorthand for current_vmcb->ptr. The myriad accesses to
svm->vmcb are quite confusing without this crucial detail.
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210406171811.4043363-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Remove vmcb_pa from vcpu_svm and simply read current_vmcb->pa directly in
the one path where it is consumed. Unlike svm->vmcb, use of the current
vmcb's address is very limited, as evidenced by the fact that its use
can be trimmed to a single dereference.
Opportunistically add a comment about using vmcb01 for VMLOAD/VMSAVE, at
first glance using vmcb01 instead of vmcb_pa looks wrong.
No functional change intended.
Cc: Maxim Levitsky <mlevitsk@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210406171811.4043363-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Do not update the new vmcb's last-run cpu when switching to a different
vmcb. If the vCPU is migrated between its last run and a vmcb switch,
e.g. for nested VM-Exit, then setting the cpu without marking the vmcb
dirty will lead to KVM running the vCPU on a different physical cpu with
stale clean bit settings.
vcpu->cpu current_vmcb->cpu hardware
pre_svm_run() cpu0 cpu0 cpu0,clean
kvm_arch_vcpu_load() cpu1 cpu0 cpu0,clean
svm_switch_vmcb() cpu1 cpu1 cpu0,clean
pre_svm_run() cpu1 cpu1 kaboom
Simply delete the offending code; unlike VMX, which needs to update the
cpu at switch time due to the need to do VMPTRLD, SVM only cares about
which cpu last ran the vCPU.
Fixes: af18fa775d ("KVM: nSVM: Track the physical cpu of the vmcb vmrun through the vmcb")
Cc: Cathy Avery <cavery@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210406171811.4043363-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Access to the GHCB is mainly in the VMGEXIT path and it is known that the
GHCB will be mapped. But there are two paths where it is possible the GHCB
might not be mapped.
The sev_vcpu_deliver_sipi_vector() routine will update the GHCB to inform
the caller of the AP Reset Hold NAE event that a SIPI has been delivered.
However, if a SIPI is performed without a corresponding AP Reset Hold,
then the GHCB might not be mapped (depending on the previous VMEXIT),
which will result in a NULL pointer dereference.
The svm_complete_emulated_msr() routine will update the GHCB to inform
the caller of a RDMSR/WRMSR operation about any errors. While it is likely
that the GHCB will be mapped in this situation, add a safe guard
in this path to be certain a NULL pointer dereference is not encountered.
Fixes: f1c6366e30 ("KVM: SVM: Add required changes to support intercepts under SEV-ES")
Fixes: 647daca25d ("KVM: SVM: Add support for booting APs in an SEV-ES guest")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Cc: stable@vger.kernel.org
Message-Id: <a5d3ebb600a91170fc88599d5a575452b3e31036.1617979121.git.thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If the target is self we do not need to yield, we can avoid malicious
guest to play this.
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Message-Id: <1617941911-5338-3-git-send-email-wanpengli@tencent.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
To analyze some performance issues with lock contention and scheduling,
it is nice to know when directed yield are successful or failing.
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Message-Id: <1617941911-5338-2-git-send-email-wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
To avoid saddling a vCPU thread with the work of tearing down an entire
paging structure, take a reference on each root before they become
obsolete, so that the thread initiating the fast invalidation can tear
down the paging structure and (most likely) release the last reference.
As a bonus, this teardown can happen under the MMU lock in read mode so
as not to block the progress of vCPU threads.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-14-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Provide a real mechanism for fast invalidation by marking roots as
invalid so that their reference count will quickly fall to zero
and they will be torn down.
One negative side affect of this approach is that a vCPU thread will
likely drop the last reference to a root and be saddled with the work of
tearing down an entire paging structure. This issue will be resolved in
a later commit.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-13-bgardon@google.com>
[Move the loop to tdp_mmu.c, otherwise compilation fails on 32-bit. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
To reduce lock contention and interference with page fault handlers,
allow the TDP MMU functions which enable and disable dirty logging
to operate under the MMU read lock.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-12-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
To reduce the impact of disabling dirty logging, change the TDP MMU
function which zaps collapsible SPTEs to run under the MMU read lock.
This way, page faults on zapped SPTEs can proceed in parallel with
kvm_mmu_zap_collapsible_sptes.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-11-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
To reduce lock contention and interference with page fault handlers,
allow the TDP MMU function to zap a GFN range to operate under the MMU
read lock.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-10-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Protect the contents of the TDP MMU roots list with RCU in preparation
for a future patch which will allow the iterator macro to be used under
the MMU lock in read mode.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-9-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
To reduce dependence on the MMU write lock, don't rely on the assumption
that the atomic operation in kvm_tdp_mmu_get_root will always succeed.
By not relying on that assumption, threads do not need to hold the MMU
lock in write mode in order to take a reference on a TDP MMU root.
In the root iterator, this change means that some roots might have to be
skipped if they are found to have a zero refcount. This will still never
happen as of this patch, but a future patch will need that flexibility to
make the root iterator safe under the MMU read lock.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-8-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In order to parallelize more operations for the TDP MMU, make the
refcount on TDP MMU roots atomic, so that a future patch can allow
multiple threads to take a reference on the root concurrently, while
holding the MMU lock in read mode.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-7-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Refactor the yield safe TDP MMU root iterator to be more amenable to
changes in future commits which will allow it to be used under the MMU
lock in read mode. Currently the iterator requires a complicated dance
between the helper functions and different parts of the for loop which
makes it hard to reason about. Moving all the logic into a single function
simplifies the iterator substantially.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-6-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
kvm_tdp_mmu_put_root and kvm_tdp_mmu_free_root are always called
together, so merge the functions to simplify TDP MMU root refcounting /
freeing.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-5-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Minor cleanup to deduplicate the code used to free a struct kvm_mmu_page
in the TDP MMU.
No functional change intended.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-4-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The TDP MMU is almost the only user of kvm_mmu_get_root and
kvm_mmu_put_root. There is only one use of put_root in mmu.c for the
legacy / shadow MMU. Open code that one use and move the get / put
functions to the TDP MMU so they can be extended in future commits.
No functional change intended.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-3-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
kvm_tdp_mmu_zap_collapsible_sptes unnecessarily removes the const
qualifier from its memlsot argument, leading to a compiler warning. Add
the const annotation and pass it to subsequent functions.
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210401233736.638171-2-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Let the TDP MMU yield when unmapping a range in response to a MMU
notification, if yielding is allowed by said notification. There is no
reason to disallow yielding in this case, and in theory the range being
invalidated could be quite large.
Cc: Ben Gardon <bgardon@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210402005658.3024832-11-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Move the hva->gfn lookup for MMU notifiers into common code. Every arch
does a similar lookup, and some arch code is all but identical across
multiple architectures.
In addition to consolidating code, this will allow introducing
optimizations that will benefit all architectures without incurring
multiple walks of the memslots, e.g. by taking mmu_lock if and only if a
relevant range exists in the memslots.
The use of __always_inline to avoid indirect call retpolines, as done by
x86, may also benefit other architectures.
Consolidating the lookups also fixes a wart in x86, where the legacy MMU
and TDP MMU each do their own memslot walks.
Lastly, future enhancements to the memslot implementation, e.g. to add an
interval tree to track host address, will need to touch far less arch
specific code.
MIPS, PPC, and arm64 will be converted one at a time in future patches.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210402005658.3024832-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When using manual protection of dirty pages, it is not necessary
to protect nested page tables down to the 4K level; instead KVM
can protect only hugepages in order to split them lazily, and
delay write protection at 4K-granularity until KVM_CLEAR_DIRTY_LOG.
This was overlooked in the TDP MMU, so do it there as well.
Fixes: a6a0b05da9 ("kvm: x86/mmu: Support dirty logging for the TDP MMU")
Cc: Ben Gardon <bgardon@google.com>
Reviewed-by: Keqian Zhu <zhukeqian1@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Store the supported bits into KVM_GUESTDBG_VALID_MASK
macro, similar to how arm does this.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20210401135451.1004564-4-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Injected interrupts/nmi should not block a pending exception,
but rather be either lost if nested hypervisor doesn't
intercept the pending exception (as in stock x86), or be delivered
in exitintinfo/IDT_VECTORING_INFO field, as a part of a VMexit
that corresponds to the pending exception.
The only reason for an exception to be blocked is when nested run
is pending (and that can't really happen currently
but still worth checking for).
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20210401143817.1030695-2-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
While KVM's MMU should be fully reset by loading of nested CR0/CR3/CR4
by KVM_SET_SREGS, we are not in nested mode yet when we do it and therefore
only root_mmu is reset.
On regular nested entries we call nested_svm_load_cr3 which both updates
the guest's CR3 in the MMU when it is needed, and it also initializes
the mmu again which makes it initialize the walk_mmu as well when nested
paging is enabled in both host and guest.
Since we don't call nested_svm_load_cr3 on nested state load,
the walk_mmu can be left uninitialized, which can lead to a NULL pointer
dereference while accessing it if we happen to get a nested page fault
right after entering the nested guest first time after the migration and
we decide to emulate it, which leads to the emulator trying to access
walk_mmu->gva_to_gpa which is NULL.
Therefore we should call this function on nested state load as well.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20210401141814.1029036-3-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When dumping the current VMCS state, include the MSRs that are being
automatically loaded/stored during VM entry/exit.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210318120841.133123-6-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If EFER is not being loaded from the VMCS, show the effective value by
reference to the MSR autoload list or calculation.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210318120841.133123-5-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When deciding whether to dump the GUEST_IA32_EFER and GUEST_IA32_PAT
fields of the VMCS, examine only the VM entry load controls, as saving
on VM exit has no effect on whether VM entry succeeds or fails.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210318120841.133123-4-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Show EFER and PAT based on their individual entry/exit controls.
Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210318120841.133123-3-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If the VM entry/exit controls for loading/saving MSR_EFER are either
not available (an older processor or explicitly disabled) or not
used (host and guest values are the same), reading GUEST_IA32_EFER
from the VMCS returns an inaccurate value.
Because of this, in dump_vmcs() don't use GUEST_IA32_EFER to decide
whether to print the PDPTRs - always do so if the fields exist.
Fixes: 4eb64dce8d ("KVM: x86: dump VMCS on invalid entry")
Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Message-Id: <20210318120841.133123-2-david.edmondson@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Currently to support Intel->AMD migration, if CPU vendor is GenuineIntel,
we emulate the full 64 value for MSR_IA32_SYSENTER_{EIP|ESP}
msrs, and we also emulate the sysenter/sysexit instruction in long mode.
(Emulator does still refuse to emulate sysenter in 64 bit mode, on the
ground that the code for that wasn't tested and likely has no users)
However when virtual vmload/vmsave is enabled, the vmload instruction will
update these 32 bit msrs without triggering their msr intercept,
which will lead to having stale values in kvm's shadow copy of these msrs,
which relies on the intercept to be up to date.
Fix/optimize this by doing the following:
1. Enable the MSR intercepts for SYSENTER MSRs iff vendor=GenuineIntel
(This is both a tiny optimization and also ensures that in case
the guest cpu vendor is AMD, the msrs will be 32 bit wide as
AMD defined).
2. Store only high 32 bit part of these msrs on interception and combine
it with hardware msr value on intercepted read/writes
iff vendor=GenuineIntel.
3. Disable vmload/vmsave virtualization if vendor=GenuineIntel.
(It is somewhat insane to set vendor=GenuineIntel and still enable
SVM for the guest but well whatever).
Then zero the high 32 bit parts when kvm intercepts and emulates vmload.
Thanks a lot to Paulo Bonzini for helping me with fixing this in the most
correct way.
This patch fixes nested migration of 32 bit nested guests, that was
broken because incorrect cached values of SYSENTER msrs were stored in
the migration stream if L1 changed these msrs with
vmload prior to L2 entry.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20210401111928.996871-3-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This is similar to existing 'guest_cpuid_is_amd_or_hygon'
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20210401111928.996871-2-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Switch to GFP_KERNEL_ACCOUNT for a handful of allocations that are
clearly associated with a single task/VM.
Note, there are a several SEV allocations that aren't accounted, but
those can (hopefully) be fixed by using the local stack for memory.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210331023025.2485960-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reject KVM_SEV_INIT and KVM_SEV_ES_INIT if they are attempted after one
or more vCPUs have been created. KVM assumes a VM is tagged SEV/SEV-ES
prior to vCPU creation, e.g. init_vmcb() needs to mark the VMCB as SEV
enabled, and svm_create_vcpu() needs to allocate the VMSA. At best,
creating vCPUs before SEV/SEV-ES init will lead to unexpected errors
and/or behavior, and at worst it will crash the host, e.g.
sev_launch_update_vmsa() will dereference a null svm->vmsa pointer.
Fixes: 1654efcbc4 ("KVM: SVM: Add KVM_SEV_INIT command")
Fixes: ad73109ae7 ("KVM: SVM: Provide support to launch and run an SEV-ES guest")
Cc: stable@vger.kernel.org
Cc: Brijesh Singh <brijesh.singh@amd.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210331031936.2495277-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Set sev->es_active only after the guts of KVM_SEV_ES_INIT succeeds. If
the command fails, e.g. because SEV is already active or there are no
available ASIDs, then es_active will be left set even though the VM is
not fully SEV-ES capable.
Refactor the code so that "es_active" is passed on the stack instead of
being prematurely shoved into sev_info, both to avoid having to unwind
sev_info and so that it's more obvious what actually consumes es_active
in sev_guest_init() and its helpers.
Fixes: ad73109ae7 ("KVM: SVM: Provide support to launch and run an SEV-ES guest")
Cc: stable@vger.kernel.org
Cc: Brijesh Singh <brijesh.singh@amd.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210331031936.2495277-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Use the kvm_for_each_vcpu() helper to iterate over vCPUs when encrypting
VMSAs for SEV, which effectively switches to use online_vcpus instead of
created_vcpus. This fixes a possible null-pointer dereference as
created_vcpus does not guarantee a vCPU exists, since it is updated at
the very beginning of KVM_CREATE_VCPU. created_vcpus exists to allow the
bulk of vCPU creation to run in parallel, while still correctly
restricting the max number of max vCPUs.
Fixes: ad73109ae7 ("KVM: SVM: Provide support to launch and run an SEV-ES guest")
Cc: stable@vger.kernel.org
Cc: Brijesh Singh <brijesh.singh@amd.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210331031936.2495277-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Use a basic NOT+AND sequence to clear the Accessed bit in TDP MMU SPTEs,
as opposed to the fancy ffs()+clear_bit() logic that was copied from the
legacy MMU. The legacy MMU uses clear_bit() because it is operating on
the SPTE itself, i.e. clearing needs to be atomic. The TDP MMU operates
on a local variable that it later writes to the SPTE, and so doesn't need
to be atomic or even resident in memory.
Opportunistically drop unnecessary initialization of new_spte, it's
guaranteed to be written before being accessed.
Using NOT+AND instead of ffs()+clear_bit() reduces the sequence from:
0x0000000000058be6 <+134>: test %rax,%rax
0x0000000000058be9 <+137>: je 0x58bf4 <age_gfn_range+148>
0x0000000000058beb <+139>: test %rax,%rdi
0x0000000000058bee <+142>: je 0x58cdc <age_gfn_range+380>
0x0000000000058bf4 <+148>: mov %rdi,0x8(%rsp)
0x0000000000058bf9 <+153>: mov $0xffffffff,%edx
0x0000000000058bfe <+158>: bsf %eax,%edx
0x0000000000058c01 <+161>: movslq %edx,%rdx
0x0000000000058c04 <+164>: lock btr %rdx,0x8(%rsp)
0x0000000000058c0b <+171>: mov 0x8(%rsp),%r15
to:
0x0000000000058bdd <+125>: test %rax,%rax
0x0000000000058be0 <+128>: je 0x58beb <age_gfn_range+139>
0x0000000000058be2 <+130>: test %rax,%r8
0x0000000000058be5 <+133>: je 0x58cc0 <age_gfn_range+352>
0x0000000000058beb <+139>: not %rax
0x0000000000058bee <+142>: and %r8,%rax
0x0000000000058bf1 <+145>: mov %rax,%r15
thus eliminating several memory accesses, including a locked access.
Cc: Ben Gardon <bgardon@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210331004942.2444916-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Don't clear the dirty bit when aging a TDP MMU SPTE (in response to a MMU
notifier event). Prematurely clearing the dirty bit could cause spurious
PML updates if aging a page happened to coincide with dirty logging.
Note, tdp_mmu_set_spte_no_acc_track() flows into __handle_changed_spte(),
so the host PFN will be marked dirty, i.e. there is no potential for data
corruption.
Fixes: a6a0b05da9 ("kvm: x86/mmu: Support dirty logging for the TDP MMU")
Cc: Ben Gardon <bgardon@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210331004942.2444916-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Remove x86's trace_kvm_age_page() tracepoint. It's mostly redundant with
the common trace_kvm_age_hva() tracepoint, and if there is a need for the
extra details, e.g. gfn, referenced, etc... those details should be added
to the common tracepoint so that all architectures and MMUs benefit from
the info.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-19-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Use the leaf-only TDP iterator when changing the SPTE in reaction to a
MMU notifier. Practically speaking, this is a nop since the guts of the
loop explicitly looks for 4k SPTEs, which are always leaf SPTEs. Switch
the iterator to match age_gfn_range() and test_age_gfn() so that a future
patch can consolidate the core iterating logic.
No real functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-8-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Move the address space ID check that is performed when iterating over
roots into the macro helpers to consolidate code.
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-7-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Pass the address space ID to TDP MMU's primary "zap gfn range" helper to
allow the MMU notifier paths to iterate over memslots exactly once.
Currently, both the legacy MMU and TDP MMU iterate over memslots when
looking for an overlapping hva range, which can be quite costly if there
are a large number of memslots.
Add a "flush" parameter so that iterating over multiple address spaces
in the caller will continue to do the right thing when yielding while a
flush is pending from a previous address space.
Note, this also has a functional change in the form of coalescing TLB
flushes across multiple address spaces in kvm_zap_gfn_range(), and also
optimizes the TDP MMU to utilize range-based flushing when running as L1
with Hyper-V enlightenments.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-6-seanjc@google.com>
[Keep separate for loops to prepare for other incoming patches. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Gather pending TLB flushes across both address spaces when zapping a
given gfn range. This requires feeding "flush" back into subsequent
calls, but on the plus side sets the stage for further batching
between the legacy MMU and TDP MMU. It also allows refactoring the
address space iteration to cover the legacy and TDP MMUs without
introducing truly ugly code.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-5-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Gather pending TLB flushes across both the legacy and TDP MMUs when
zapping collapsible SPTEs to avoid multiple flushes if both the legacy
MMU (for nested guests) and TDP MMU have mappings for the memslot.
Note, this also optimizes the TDP MMU to flush only the relevant range
when running as L1 with Hyper-V enlightenments.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Place the onus on the caller of slot_handle_*() to flush the TLB, rather
than handling the flush in the helper, and rename parameters accordingly.
This will allow future patches to coalesce flushes between address spaces
and between the legacy and TDP MMUs.
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When zapping collapsible SPTEs across multiple roots, gather pending
flushes and perform a single remote TLB flush at the end, as opposed to
flushing after processing every root.
Note, flush may be cleared by the result of zap_collapsible_spte_range().
This is intended and correct, e.g. yielding may have serviced a prior
pending flush.
Cc: Ben Gardon <bgardon@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210326021957.1424875-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
MSR_F15H_PERF_CTL0-5, MSR_F15H_PERF_CTR0-5 MSRs have a CPUID bit assigned
to them (X86_FEATURE_PERFCTR_CORE) and when it wasn't exposed to the guest
the correct behavior is to inject #GP an not just return zero.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210329124804.170173-1-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
According to APM, the #DB intercept for a single-stepped VMRUN must happen
after the completion of that instruction, when the guest does #VMEXIT to
the host. However, in the current implementation of KVM, the #DB intercept
for a single-stepped VMRUN happens after the completion of the instruction
that follows the VMRUN instruction. When the #DB intercept handler is
invoked, it shows the RIP of the instruction that follows VMRUN, instead of
of VMRUN itself. This is an incorrect RIP as far as single-stepping VMRUN
is concerned.
This patch fixes the problem by checking, in nested_svm_vmexit(), for the
condition that the VMRUN instruction is being single-stepped and if so,
queues the pending #DB intercept so that the #DB is accounted for before
we execute L1's next instruction.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Krish Sadhukhan <krish.sadhukhan@oraacle.com>
Message-Id: <20210323175006.73249-2-krish.sadhukhan@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
On SVM, reading PDPTRs might access guest memory, which might fault
and thus might sleep. On the other hand, it is not possible to
release the lock after make_mmu_pages_available has been called.
Therefore, push the call to make_mmu_pages_available and the
mmu_lock critical section within mmu_alloc_direct_roots and
mmu_alloc_shadow_roots.
Reported-by: Wanpeng Li <wanpengli@tencent.com>
Co-developed-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
__vmx_handle_exit() uses vcpu->run->internal.ndata as an index for
an array access. Since vcpu->run is (can be) mapped to a user address
space with a writer permission, the 'ndata' could be updated by the
user process at anytime (the user process can set it to outside the
bounds of the array).
So, it is not safe that __vmx_handle_exit() uses the 'ndata' that way.
Fixes: 1aa561b1a4 ("kvm: x86: Add "last CPU" to some KVM_EXIT information")
Signed-off-by: Reiji Watanabe <reijiw@google.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
Message-Id: <20210413154739.490299-1-reijiw@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Right now, if a call to kvm_tdp_mmu_zap_sp returns false, the caller
will skip the TLB flush, which is wrong. There are two ways to fix
it:
- since kvm_tdp_mmu_zap_sp will not yield and therefore will not flush
the TLB itself, we could change the call to kvm_tdp_mmu_zap_sp to
use "flush |= ..."
- or we can chain the flush argument through kvm_tdp_mmu_zap_sp down
to __kvm_tdp_mmu_zap_gfn_range. Note that kvm_tdp_mmu_zap_sp will
neither yield nor flush, so flush would never go from true to
false.
This patch does the former to simplify application to stable kernels,
and to make it further clearer that kvm_tdp_mmu_zap_sp will not flush.
Cc: seanjc@google.com
Fixes: 048f49809c ("KVM: x86/mmu: Ensure TLBs are flushed for TDP MMU during NX zapping")
Cc: <stable@vger.kernel.org> # 5.10.x: 048f49809c: KVM: x86/mmu: Ensure TLBs are flushed for TDP MMU during NX zapping
Cc: <stable@vger.kernel.org> # 5.10.x: 33a3164161: KVM: x86/mmu: Don't allow TDP MMU to yield when recovering NX pages
Cc: <stable@vger.kernel.org>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add a misc device /dev/sgx_vepc to allow userspace to allocate "raw"
Enclave Page Cache (EPC) without an associated enclave. The intended
and only known use case for raw EPC allocation is to expose EPC to a
KVM guest, hence the 'vepc' moniker, virt.{c,h} files and X86_SGX_KVM
Kconfig.
The SGX driver uses the misc device /dev/sgx_enclave to support
userspace in creating an enclave. Each file descriptor returned from
opening /dev/sgx_enclave represents an enclave. Unlike the SGX driver,
KVM doesn't control how the guest uses the EPC, therefore EPC allocated
to a KVM guest is not associated with an enclave, and /dev/sgx_enclave
is not suitable for allocating EPC for a KVM guest.
Having separate device nodes for the SGX driver and KVM virtual EPC also
allows separate permission control for running host SGX enclaves and KVM
SGX guests.
To use /dev/sgx_vepc to allocate a virtual EPC instance with particular
size, the hypervisor opens /dev/sgx_vepc, and uses mmap() with the
intended size to get an address range of virtual EPC. Then it may use
the address range to create one KVM memory slot as virtual EPC for
a guest.
Implement the "raw" EPC allocation in the x86 core-SGX subsystem via
/dev/sgx_vepc rather than in KVM. Doing so has two major advantages:
- Does not require changes to KVM's uAPI, e.g. EPC gets handled as
just another memory backend for guests.
- EPC management is wholly contained in the SGX subsystem, e.g. SGX
does not have to export any symbols, changes to reclaim flows don't
need to be routed through KVM, SGX's dirty laundry doesn't have to
get aired out for the world to see, and so on and so forth.
The virtual EPC pages allocated to guests are currently not reclaimable.
Reclaiming an EPC page used by enclave requires a special reclaim
mechanism separate from normal page reclaim, and that mechanism is not
supported for virutal EPC pages. Due to the complications of handling
reclaim conflicts between guest and host, reclaiming virtual EPC pages
is significantly more complex than basic support for SGX virtualization.
[ bp:
- Massage commit message and comments
- use cpu_feature_enabled()
- vertically align struct members init
- massage Virtual EPC clarification text
- move Kconfig prompt to Virtualization ]
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Co-developed-by: Kai Huang <kai.huang@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Dave Hansen <dave.hansen@intel.com>
Acked-by: Jarkko Sakkinen <jarkko@kernel.org>
Link: https://lkml.kernel.org/r/0c38ced8c8e5a69872db4d6a1c0dabd01e07cad7.1616136308.git.kai.huang@intel.com
Secure Encrypted Virtualization (SEV) and Secure Encrypted
Virtualization - Encrypted State (SEV-ES) ASIDs are used to encrypt KVMs
on AMD platform. These ASIDs are available in the limited quantities on
a host.
Register their capacity and usage to the misc controller for tracking
via cgroups.
Signed-off-by: Vipin Sharma <vipinsh@google.com>
Reviewed-by: David Rientjes <rientjes@google.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
When guest time is reset with KVM_SET_CLOCK(0), it is possible for
'hv_clock->system_time' to become a small negative number. This happens
because in KVM_SET_CLOCK handling we set 'kvm->arch.kvmclock_offset' based
on get_kvmclock_ns(kvm) but when KVM_REQ_CLOCK_UPDATE is handled,
kvm_guest_time_update() does (masterclock in use case):
hv_clock.system_time = ka->master_kernel_ns + v->kvm->arch.kvmclock_offset;
And 'master_kernel_ns' represents the last time when masterclock
got updated, it can precede KVM_SET_CLOCK() call. Normally, this is not a
problem, the difference is very small, e.g. I'm observing
hv_clock.system_time = -70 ns. The issue comes from the fact that
'hv_clock.system_time' is stored as unsigned and 'system_time / 100' in
compute_tsc_page_parameters() becomes a very big number.
Use 'master_kernel_ns' instead of get_kvmclock_ns() when masterclock is in
use and get_kvmclock_base_ns() when it's not to prevent 'system_time' from
going negative.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210331124130.337992-2-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
pvclock_gtod_sync_lock can be taken with interrupts disabled if the
preempt notifier calls get_kvmclock_ns to update the Xen
runstate information:
spin_lock include/linux/spinlock.h:354 [inline]
get_kvmclock_ns+0x25/0x390 arch/x86/kvm/x86.c:2587
kvm_xen_update_runstate+0x3d/0x2c0 arch/x86/kvm/xen.c:69
kvm_xen_update_runstate_guest+0x74/0x320 arch/x86/kvm/xen.c:100
kvm_xen_runstate_set_preempted arch/x86/kvm/xen.h:96 [inline]
kvm_arch_vcpu_put+0x2d8/0x5a0 arch/x86/kvm/x86.c:4062
So change the users of the spinlock to spin_lock_irqsave and
spin_unlock_irqrestore.
Reported-by: syzbot+b282b65c2c68492df769@syzkaller.appspotmail.com
Fixes: 30b5c851af ("KVM: x86/xen: Add support for vCPU runstate information")
Cc: David Woodhouse <dwmw@amazon.co.uk>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
There is no need to include changes to vcpu->requests into
the pvclock_gtod_sync_lock critical section. The changes to
the shared data structures (in pvclock_update_vm_gtod_copy)
already occur under the lock.
Cc: David Woodhouse <dwmw@amazon.co.uk>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fixing nested_vmcb_check_save to avoid all TOC/TOU races
is a bit harder in released kernels, so do the bare minimum
by avoiding that EFER.SVME is cleared. This is problematic
because svm_set_efer frees the data structures for nested
virtualization if EFER.SVME is cleared.
Also check that EFER.SVME remains set after a nested vmexit;
clearing it could happen if the bit is zero in the save area
that is passed to KVM_SET_NESTED_STATE (the save area of the
nested state corresponds to the nested hypervisor's state
and is restored on the next nested vmexit).
Cc: stable@vger.kernel.org
Fixes: 2fcf4876ad ("KVM: nSVM: implement on demand allocation of the nested state")
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Avoid races between check and use of the nested VMCB controls. This
for example ensures that the VMRUN intercept is always reflected to the
nested hypervisor, instead of being processed by the host. Without this
patch, it is possible to end up with svm->nested.hsave pointing to
the MSR permission bitmap for nested guests.
This bug is CVE-2021-29657.
Reported-by: Felix Wilhelm <fwilhelm@google.com>
Cc: stable@vger.kernel.org
Fixes: 2fcf4876ad ("KVM: nSVM: implement on demand allocation of the nested state")
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Prevent the TDP MMU from yielding when zapping a gfn range during NX
page recovery. If a flush is pending from a previous invocation of the
zapping helper, either in the TDP MMU or the legacy MMU, but the TDP MMU
has not accumulated a flush for the current invocation, then yielding
will release mmu_lock with stale TLB entries.
That being said, this isn't technically a bug fix in the current code, as
the TDP MMU will never yield in this case. tdp_mmu_iter_cond_resched()
will yield if and only if it has made forward progress, as defined by the
current gfn vs. the last yielded (or starting) gfn. Because zapping a
single shadow page is guaranteed to (a) find that page and (b) step
sideways at the level of the shadow page, the TDP iter will break its loop
before getting a chance to yield.
But that is all very, very subtle, and will break at the slightest sneeze,
e.g. zapping while holding mmu_lock for read would break as the TDP MMU
wouldn't be guaranteed to see the present shadow page, and thus could step
sideways at a lower level.
Cc: Ben Gardon <bgardon@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210325200119.1359384-4-seanjc@google.com>
[Add lockdep assertion. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Honor the "flush needed" return from kvm_tdp_mmu_zap_gfn_range(), which
does the flush itself if and only if it yields (which it will never do in
this particular scenario), and otherwise expects the caller to do the
flush. If pages are zapped from the TDP MMU but not the legacy MMU, then
no flush will occur.
Fixes: 29cf0f5007 ("kvm: x86/mmu: NX largepage recovery for TDP MMU")
Cc: stable@vger.kernel.org
Cc: Ben Gardon <bgardon@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210325200119.1359384-3-seanjc@google.com>
Reviewed-by: Ben Gardon <bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When flushing a range of GFNs across multiple roots, ensure any pending
flush from a previous root is honored before yielding while walking the
tables of the current root.
Note, kvm_tdp_mmu_zap_gfn_range() now intentionally overwrites its local
"flush" with the result to avoid redundant flushes. zap_gfn_range()
preserves and return the incoming "flush", unless of course the flush was
performed prior to yielding and no new flush was triggered.
Fixes: 1af4a96025 ("KVM: x86/mmu: Yield in TDU MMU iter even if no SPTES changed")
Cc: stable@vger.kernel.org
Reviewed-by: Ben Gardon <bgardon@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210325200119.1359384-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Building kvm module out-of-source with,
make -C $SRC O=$BIN M=arch/x86/kvm
fails to find "irq.h" as the include dir passed to cflags-y does not
prefix the source dir. Fix this by prefixing $(srctree) to the include
dir path.
Signed-off-by: Siddharth Chandrasekaran <sidcha@amazon.de>
Message-Id: <20210324124347.18336-1-sidcha@amazon.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
MSR_F15H_PERF_CTL0-5, MSR_F15H_PERF_CTR0-5 MSRs are only available when
X86_FEATURE_PERFCTR_CORE CPUID bit was exposed to the guest. KVM, however,
allows these MSRs unconditionally because kvm_pmu_is_valid_msr() ->
amd_msr_idx_to_pmc() check always passes and because kvm_pmu_set_msr() ->
amd_pmu_set_msr() doesn't fail.
In case of a counter (CTRn), no big harm is done as we only increase
internal PMC's value but in case of an eventsel (CTLn), we go deep into
perf internals with a non-existing counter.
Note, kvm_get_msr_common() just returns '0' when these MSRs don't exist
and this also seems to contradict architectural behavior which is #GP
(I did check one old Opteron host) but changing this status quo is a bit
scarier.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210323084515.1346540-1-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
kvm_write_tsc() was renamed and made static since commit 0c899c25d7
("KVM: x86: do not attempt TSC synchronization on guest writes"). Remove
its unused declaration.
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Message-Id: <20210326070334.12310-1-dongli.zhang@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
kvm_msr_ignored_check function never uses vcpu argument. Clean up the
function and invokers.
Signed-off-by: Haiwei Li <lihaiwei@tencent.com>
Message-Id: <20210313051032.4171-1-lihaiwei.kernel@gmail.com>
Reviewed-by: Keqian Zhu <zhukeqian1@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fix another ~42 single-word typos in arch/x86/ code comments,
missed a few in the first pass, in particular in .S files.
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: linux-kernel@vger.kernel.org
In order to deal with noncoherent DMA, we should execute wbinvd on
all dirty pCPUs when guest wbinvd exits to maintain data consistency.
smp_call_function_many() does not execute the provided function on the
local core, therefore replace it by on_each_cpu_mask().
Reported-by: Nadav Amit <namit@vmware.com>
Cc: Nadav Amit <namit@vmware.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Message-Id: <1615517151-7465-1-git-send-email-wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fix a plethora of issues with MSR filtering by installing the resulting
filter as an atomic bundle instead of updating the live filter one range
at a time. The KVM_X86_SET_MSR_FILTER ioctl() isn't truly atomic, as
the hardware MSR bitmaps won't be updated until the next VM-Enter, but
the relevant software struct is atomically updated, which is what KVM
really needs.
Similar to the approach used for modifying memslots, make arch.msr_filter
a SRCU-protected pointer, do all the work configuring the new filter
outside of kvm->lock, and then acquire kvm->lock only when the new filter
has been vetted and created. That way vCPU readers either see the old
filter or the new filter in their entirety, not some half-baked state.
Yuan Yao pointed out a use-after-free in ksm_msr_allowed() due to a
TOCTOU bug, but that's just the tip of the iceberg...
- Nothing is __rcu annotated, making it nigh impossible to audit the
code for correctness.
- kvm_add_msr_filter() has an unpaired smp_wmb(). Violation of kernel
coding style aside, the lack of a smb_rmb() anywhere casts all code
into doubt.
- kvm_clear_msr_filter() has a double free TOCTOU bug, as it grabs
count before taking the lock.
- kvm_clear_msr_filter() also has memory leak due to the same TOCTOU bug.
The entire approach of updating the live filter is also flawed. While
installing a new filter is inherently racy if vCPUs are running, fixing
the above issues also makes it trivial to ensure certain behavior is
deterministic, e.g. KVM can provide deterministic behavior for MSRs with
identical settings in the old and new filters. An atomic update of the
filter also prevents KVM from getting into a half-baked state, e.g. if
installing a filter fails, the existing approach would leave the filter
in a half-baked state, having already committed whatever bits of the
filter were already processed.
[*] https://lkml.kernel.org/r/20210312083157.25403-1-yaoyuan0329os@gmail.com
Fixes: 1a155254ff ("KVM: x86: Introduce MSR filtering")
Cc: stable@vger.kernel.org
Cc: Alexander Graf <graf@amazon.com>
Reported-by: Yuan Yao <yaoyuan0329os@gmail.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210316184436.2544875-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fix ~144 single-word typos in arch/x86/ code comments.
Doing this in a single commit should reduce the churn.
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: linux-kernel@vger.kernel.org
When guest opts for re-enlightenment notifications upon migration, it is
in its right to assume that TSC page values never change (as they're only
supposed to change upon migration and the host has to keep things as they
are before it receives confirmation from the guest). This is mostly true
until the guest is migrated somewhere. KVM userspace (e.g. QEMU) will
trigger masterclock update by writing to HV_X64_MSR_REFERENCE_TSC, by
calling KVM_SET_CLOCK,... and as TSC value and kvmclock reading drift
apart (even slightly), the update causes TSC page values to change.
The issue at hand is that when Hyper-V is migrated, it uses stale (cached)
TSC page values to compute the difference between its own clocksource
(provided by KVM) and its guests' TSC pages to program synthetic timers
and in some cases, when TSC page is updated, this puts all stimer
expirations in the past. This, in its turn, causes an interrupt storm
and L2 guests not making much forward progress.
Note, KVM doesn't fully implement re-enlightenment notification. Basically,
the support for reenlightenment MSRs is just a stub and userspace is only
expected to expose the feature when TSC scaling on the expected destination
hosts is available. With TSC scaling, no real re-enlightenment is needed
as TSC frequency doesn't change. With TSC scaling becoming ubiquitous, it
likely makes little sense to fully implement re-enlightenment in KVM.
Prevent TSC page from being updated after migration. In case it's not the
guest who's initiating the change and when TSC page is already enabled,
just keep it as it is: TSC value is supposed to be preserved across
migration and TSC frequency can't change with re-enlightenment enabled.
The guest is doomed anyway if any of this is not true.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210316143736.964151-5-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Create an infrastructure for tracking Hyper-V TSC page status, i.e. if it
was updated from guest/host side or if we've failed to set it up (because
e.g. guest wrote some garbage to HV_X64_MSR_REFERENCE_TSC) and there's no
need to retry.
Also, in a hypothetical situation when we are in 'always catchup' mode for
TSC we can now avoid contending 'hv->hv_lock' on every guest enter by
setting the state to HV_TSC_PAGE_BROKEN after compute_tsc_page_parameters()
returns false.
Check for HV_TSC_PAGE_SET state instead of '!hv->tsc_ref.tsc_sequence' in
get_time_ref_counter() to properly handle the situation when we failed to
write the updated TSC page values to the guest.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210316143736.964151-4-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When KVM_REQ_MASTERCLOCK_UPDATE request is issued (e.g. after migration)
we need to make sure no vCPU sees stale values in PV clock structures and
thus all vCPUs are kicked with KVM_REQ_CLOCK_UPDATE. Hyper-V TSC page
clocksource is global and kvm_guest_time_update() only updates in on vCPU0
but this is not entirely correct: nothing blocks some other vCPU from
entering the guest before we finish the update on CPU0 and it can read
stale values from the page.
Invalidate TSC page in kvm_gen_update_masterclock() to switch all vCPUs
to using MSR based clocksource (HV_X64_MSR_TIME_REF_COUNT).
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210316143736.964151-3-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
HV_X64_MSR_TSC_EMULATION_STATUS indicates whether TSC accesses are emulated
after migration (to accommodate for a different host TSC frequency when TSC
scaling is not supported; we don't implement this in KVM). Guest can use
the same MSR to stop TSC access emulation by writing zero. Writing anything
else is forbidden.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20210316143736.964151-2-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Store the address space ID in the TDP iterator so that it can be
retrieved without having to bounce through the root shadow page. This
streamlines the code and fixes a Sparse warning about not properly using
rcu_dereference() when grabbing the ID from the root on the fly.
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Ben Gardon <bgardon@google.com>
Message-Id: <20210315233803.2706477-5-bgardon@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
