Pull Xen-for-KVM changes from David Woodhouse:
* add support for 32-bit guests in SCHEDOP_poll
* the rest of the gfn-to-pfn cache API cleanup
"I still haven't reinstated the last of those patches to make gpc->len
immutable."
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
CPUID.80000021H:EAX[bit 9] indicates that the SMM_CTL MSR (0xc0010116) is
not supported. This defeature can be advertised by KVM_GET_SUPPORTED_CPUID
regardless of whether or not the host enumerates it; currently it will be
included only if the host enumerates at least leaf 8000001DH, due to a
preexisting bug in QEMU that KVM has to work around (commit f751d8eac1,
"KVM: x86: work around QEMU issue with synthetic CPUID leaves", 2022-04-29).
Signed-off-by: Jim Mattson <jmattson@google.com>
Message-Id: <20221007221644.138355-1-jmattson@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Several symbols are not used by vendor modules but still exported.
Removing them ensures that new coupling between kvm.ko and kvm-*.ko
is noticed and reviewed.
Co-developed-by: Sean Christopherson <seanjc@google.com>
Co-developed-by: Like Xu <like.xu.linux@gmail.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Like Xu <like.xu.linux@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Convert {clear,set}_bit() to atomics as KVM's ucall implementation relies
on clear_bit() being atomic, they are defined in atomic.h, and the same
helpers in the kernel proper are atomic.
KVM's ucall infrastructure is the only user of clear_bit() in tools/, and
there are no true set_bit() users. tools/testing/nvdimm/ does make heavy
use of set_bit(), but that code builds into a kernel module of sorts, i.e.
pulls in all of the kernel's header and so is already getting the kernel's
atomic set_bit().
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221119013450.2643007-10-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Drop the "atomic_" prefix from tools' atomic_test_and_set_bit() to
match the kernel nomenclature where test_and_set_bit() is atomic,
and __test_and_set_bit() provides the non-atomic variant.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221119013450.2643007-9-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Drop tools' non-atomic test_and_set_bit() and test_and_clear_bit() helpers
now that all users are gone. The names will be claimed in the future for
atomic versions.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221119013450.2643007-8-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Use the dedicated non-atomic helpers for {clear,set}_bit() and their
test variants, i.e. the double-underscore versions. Depsite being
defined in atomic.h, and despite the kernel versions being atomic in the
kernel, tools' {clear,set}_bit() helpers aren't actually atomic. Move
to the double-underscore versions so that the versions that are expected
to be atomic (for kernel developers) can be made atomic without affecting
users that don't want atomic operations.
Leave the usage in ucall_free() as-is, it's the one place in tools/ that
actually wants/needs atomic behavior.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221119013450.2643007-7-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Use the dedicated non-atomic helpers for {clear,set}_bit() and their
test variants, i.e. the double-underscore versions. Depsite being
defined in atomic.h, and despite the kernel versions being atomic in the
kernel, tools' {clear,set}_bit() helpers aren't actually atomic. Move
to the double-underscore versions so that the versions that are expected
to be atomic (for kernel developers) can be made atomic without affecting
users that don't want atomic operations.
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Message-Id: <20221119013450.2643007-6-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Take @bit as an unsigned long instead of a signed int in clear_bit() and
set_bit() so that they match the double-underscore versions, __clear_bit()
and __set_bit(). This will allow converting users that really don't want
atomic operations to the double-underscores without introducing a
functional change, which will in turn allow making {clear,set}_bit()
atomic (as advertised).
Practically speaking, this _should_ have no functional impact. KVM's
selftests usage is either hardcoded (Hyper-V tests) or is artificially
limited (arch_timer test and dirty_log test). In KVM, dirty_log test is
the only mildly interesting case as it's use indirectly restricted to
unsigned 32-bit values, but in theory it could generate a negative value
when cast to a signed int. But in that case, taking an "unsigned long"
is actually a bug fix.
Perf's usage is more difficult to audit, but any code that is affected
by the switch is likely already broken. perf_header__{set,clear}_feat()
and perf_file_header__read() effectively use only hardcoded enums with
small, positive values, atom_new() passes an unsigned long, but its value
is capped at 128 via NR_ATOM_PER_PAGE, etc...
The only real potential for breakage is in the perf flows that take a
"cpu", but it's unlikely perf is subtly relying on a negative index into
bitmaps, e.g. "cpu" can be "-1", but only as "not valid" placeholder.
Note, tools/testing/nvdimm/ makes heavy use of set_bit(), but that code
builds into a kernel module of sorts, i.e. pulls in all of the kernel's
header and so is getting the kernel's atomic set_bit(). The NVDIMM test
usage of atomics is likely unnecessary, e.g. ndtest_dimm_register() sets
bits in a local variable, but that's neither here nor there as far as
this change is concerned.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221119013450.2643007-5-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add a new ucall hook, GUEST_UCALL_NONE(), to allow tests to make ucalls
without allocating a ucall struct, and use it to enable single-step
in ARM's debug-exceptions test. Like the disable single-step path, the
enabling path also needs to ensure that no exclusive access sequences are
attempted after enabling single-step, as the exclusive monitor is cleared
on ERET from the debug exception taken to EL2.
The test currently "works" because clear_bit() isn't actually an atomic
operation... yet.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221119013450.2643007-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When a VM reboots itself, the reset process will result in
an ioctl(KVM_SET_LAPIC, ...) to disable x2APIC mode and set
the xAPIC id of the vCPU to its default value, which is the
vCPU id.
That will be handled in KVM as follows:
kvm_vcpu_ioctl_set_lapic
kvm_apic_set_state
kvm_lapic_set_base => disable X2APIC mode
kvm_apic_state_fixup
kvm_lapic_xapic_id_updated
kvm_xapic_id(apic) != apic->vcpu->vcpu_id
kvm_set_apicv_inhibit(APICV_INHIBIT_REASON_APIC_ID_MODIFIED)
memcpy(vcpu->arch.apic->regs, s->regs, sizeof(*s)) => update APIC_ID
When kvm_apic_set_state invokes kvm_lapic_set_base to disable
x2APIC mode, the old 32-bit x2APIC id is still present rather
than the 8-bit xAPIC id. kvm_lapic_xapic_id_updated will set the
APICV_INHIBIT_REASON_APIC_ID_MODIFIED bit and disable APICv/x2AVIC.
Instead, kvm_lapic_xapic_id_updated must be called after APIC_ID is
changed.
In fact, this fixes another small issue in the code in that
potential changes to a vCPU's xAPIC ID need not be tracked for
KVM_GET_LAPIC.
Fixes: 3743c2f025 ("KVM: x86: inhibit APICv/AVIC on changes to APIC ID or APIC base")
Signed-off-by: Yuan ZhaoXiong <yuanzhaoxiong@baidu.com>
Message-Id: <1669984574-32692-1-git-send-email-yuanzhaoxiong@baidu.com>
Cc: stable@vger.kernel.org
Reported-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Misc KVM x86 fixes and cleanups for 6.2:
- One-off fixes for various emulation flows (SGX, VMXON, NRIPS=0).
- Reinstate IBPB on emulated VM-Exit that was incorrectly dropped a few
years back when eliminating unnecessary barriers when switching between
vmcs01 and vmcs02.
- Clean up the MSR filter docs.
- Clean up vmread_error_trampoline() to make it more obvious that params
must be passed on the stack, even for x86-64.
- Let userspace set all supported bits in MSR_IA32_FEAT_CTL irrespective
of the current guest CPUID.
- Fudge around a race with TSC refinement that results in KVM incorrectly
thinking a guest needs TSC scaling when running on a CPU with a
constant TSC, but no hardware-enumerated TSC frequency.
KVM selftests fixes for 6.2
- Fix an inverted check in the access tracking perf test, and restore
support for asserting that there aren't too many idle pages when
running on bare metal.
- Fix an ordering issue in the AMX test introduced by recent conversions
to use kvm_cpu_has(), and harden the code to guard against similar bugs
in the future. Anything that tiggers caching of KVM's supported CPUID,
kvm_cpu_has() in this case, effectively hides opt-in XSAVE features if
the caching occurs before the test opts in via prctl().
- Fix build errors that occur in certain setups (unsure exactly what is
unique about the problematic setup) due to glibc overriding
static_assert() to a variant that requires a custom message.
There are still references to the removed kvm_memory_region data structure
but the doc and comments should mention struct kvm_userspace_memory_region
instead, since that is what's used by the ioctl that replaced the old one
and this data structure support the same set of flags.
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Message-Id: <20221202105011.185147-4-javierm@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Define and use kvm_static_assert() in the common KVM selftests headers to
provide deterministic behavior, and to allow creating static asserts
without dummy messages.
The kernel's static_assert() makes the message param optional, and on the
surface, tools/include/linux/build_bug.h appears to follow suit. However,
glibc may override static_assert() and redefine it as a direct alias of
_Static_assert(), which makes the message parameter mandatory. This leads
to non-deterministic behavior as KVM selftests code that utilizes
static_assert() without a custom message may or not compile depending on
the order of includes. E.g. recently added asserts in
x86_64/processor.h fail on some systems with errors like
In file included from lib/memstress.c:11:0:
include/x86_64/processor.h: In function ‘this_cpu_has_p’:
include/x86_64/processor.h:193:34: error: expected ‘,’ before ‘)’ token
static_assert(low_bit < high_bit); \
^
due to _Static_assert() expecting a comma before a message. The "message
optional" version of static_assert() uses macro magic to strip away the
comma when presented with empty an __VA_ARGS__
#ifndef static_assert
#define static_assert(expr, ...) __static_assert(expr, ##__VA_ARGS__, #expr)
#define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
#endif // static_assert
and effectively generates "_Static_assert(expr, #expr)".
The incompatible version of static_assert() gets defined by this snippet
in /usr/include/assert.h:
#if defined __USE_ISOC11 && !defined __cplusplus
# undef static_assert
# define static_assert _Static_assert
#endif
which yields "_Static_assert(expr)" and thus fails as above.
KVM selftests don't actually care about using C11, but __USE_ISOC11 gets
defined because of _GNU_SOURCE, which many tests do #define. _GNU_SOURCE
triggers a massive pile of defines in /usr/include/features.h, including
_ISOC11_SOURCE:
/* If _GNU_SOURCE was defined by the user, turn on all the other features. */
#ifdef _GNU_SOURCE
# undef _ISOC95_SOURCE
# define _ISOC95_SOURCE 1
# undef _ISOC99_SOURCE
# define _ISOC99_SOURCE 1
# undef _ISOC11_SOURCE
# define _ISOC11_SOURCE 1
# undef _POSIX_SOURCE
# define _POSIX_SOURCE 1
# undef _POSIX_C_SOURCE
# define _POSIX_C_SOURCE 200809L
# undef _XOPEN_SOURCE
# define _XOPEN_SOURCE 700
# undef _XOPEN_SOURCE_EXTENDED
# define _XOPEN_SOURCE_EXTENDED 1
# undef _LARGEFILE64_SOURCE
# define _LARGEFILE64_SOURCE 1
# undef _DEFAULT_SOURCE
# define _DEFAULT_SOURCE 1
# undef _ATFILE_SOURCE
# define _ATFILE_SOURCE 1
#endif
which further down in /usr/include/features.h leads to:
/* This is to enable the ISO C11 extension. */
#if (defined _ISOC11_SOURCE \
|| (defined __STDC_VERSION__ && __STDC_VERSION__ >= 201112L))
# define __USE_ISOC11 1
#endif
To make matters worse, /usr/include/assert.h doesn't guard against
multiple inclusion by turning itself into a nop, but instead #undefs a
few macros and continues on. As a result, it's all but impossible to
ensure the "message optional" version of static_assert() will actually be
used, e.g. explicitly including assert.h and #undef'ing static_assert()
doesn't work as a later inclusion of assert.h will again redefine its
version.
#ifdef _ASSERT_H
# undef _ASSERT_H
# undef assert
# undef __ASSERT_VOID_CAST
# ifdef __USE_GNU
# undef assert_perror
# endif
#endif /* assert.h */
#define _ASSERT_H 1
#include <features.h>
Fixes: fcba483e82 ("KVM: selftests: Sanity check input to ioctls() at build time")
Fixes: ee37955366 ("KVM: selftests: Refactor X86_FEATURE_* framework to prep for X86_PROPERTY_*")
Fixes: 53a7dc0f21 ("KVM: selftests: Add X86_PROPERTY_* framework to retrieve CPUID values")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221122013309.1872347-1-seanjc@google.com
Move the AMX test's kvm_cpu_has() checks before creating the VM+vCPU,
there are no dependencies between the two operations. Opportunistically
add a comment to call out that enabling off-by-default XSAVE-managed
features must be done before KVM_GET_SUPPORTED_CPUID is cached.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221128225735.3291648-5-seanjc@google.com
Disallow using kvm_get_supported_cpuid() and thus caching KVM's supported
CPUID info before enabling XSAVE-managed features that are off-by-default
and must be enabled by ARCH_REQ_XCOMP_GUEST_PERM. Caching the supported
CPUID before all XSAVE features are enabled can result in false negatives
due to testing features that were cached before they were enabled.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221128225735.3291648-4-seanjc@google.com
Move __vm_xsave_require_permission() below the CPUID helpers so that a
future change can reference the cached result of KVM_GET_SUPPORTED_CPUID
while keeping the definition of the variable close to its intended user,
kvm_get_supported_cpuid().
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221128225735.3291648-3-seanjc@google.com
Move the kvm_cpu_has() check on X86_FEATURE_XFD out of the helper to
enable off-by-default XSAVE-managed features and into the one test that
currenty requires XFD (XFeature Disable) support. kvm_cpu_has() uses
kvm_get_supported_cpuid() and thus caches KVM_GET_SUPPORTED_CPUID, and so
using kvm_cpu_has() before ARCH_REQ_XCOMP_GUEST_PERM effectively results
in the test caching stale values, e.g. subsequent checks on AMX_TILE will
get false negatives.
Although off-by-default features are nonsensical without XFD, checking
for XFD virtualization prior to enabling such features isn't strictly
required.
Signed-off-by: Lei Wang <lei4.wang@intel.com>
Fixes: 7fbb653e01 ("KVM: selftests: Check KVM's supported CPUID, not host CPUID, for XFD")
Link: https://lore.kernel.org/r/20221125023839.315207-1-lei4.wang@intel.com
[sean: add Fixes, reword changelog]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221128225735.3291648-2-seanjc@google.com
Restore the assert (on x86-64) that <10% of pages are still idle when NOT
running as a nested VM in the access tracking test. The original assert
was converted to a "warning" to avoid false failures when running the
test in a VM, but the non-nested case does not suffer from the same
"infinite TLB size" issue.
Using the HYPERVISOR flag isn't infallible as VMMs aren't strictly
required to enumerate the "feature" in CPUID, but practically speaking
anyone that is running KVM selftests in VMs is going to be using a VMM
and hypervisor that sets the HYPERVISOR flag.
Cc: David Matlack <dmatlack@google.com>
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221129175300.4052283-3-seanjc@google.com
Warn if the number of idle pages is greater than or equal to 10% of the
total number of pages, not if the percentage of idle pages is less than
10%. The original code asserted that less than 10% of pages were still
idle, but the check got inverted when the assert was converted to a
warning.
Opportunistically clean up the warning; selftests are 64-bit only, there
is no need to use "%PRIu64" instead of "%lu".
Fixes: 6336a810db ("KVM: selftests: replace assertion with warning in access_tracking_perf_test")
Reviewed-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221129175300.4052283-2-seanjc@google.com
Don't snapshot tsc_khz into per-cpu cpu_tsc_khz if the host TSC is
constant, in which case the actual TSC frequency will never change and thus
capturing TSC during initialization is unnecessary, KVM can simply use
tsc_khz. This value is snapshotted from
kvm_timer_init->kvmclock_cpu_online->tsc_khz_changed(NULL)
On CPUs with constant TSC, but not a hardware-specified TSC frequency,
snapshotting cpu_tsc_khz and using that to set a VM's target TSC frequency
can lead to VM to think its TSC frequency is not what it actually is if
refining the TSC completes after KVM snapshots tsc_khz. The actual
frequency never changes, only the kernel's calculation of what that
frequency is changes.
Ideally, KVM would not be able to race with TSC refinement, or would have
a hook into tsc_refine_calibration_work() to get an alert when refinement
is complete. Avoiding the race altogether isn't practical as refinement
takes a relative eternity; it's deliberately put on a work queue outside of
the normal boot sequence to avoid unnecessarily delaying boot.
Adding a hook is doable, but somewhat gross due to KVM's ability to be
built as a module. And if the TSC is constant, which is likely the case
for every VMX/SVM-capable CPU produced in the last decade, the race can be
hit if and only if userspace is able to create a VM before TSC refinement
completes; refinement is slow, but not that slow.
For now, punt on a proper fix, as not taking a snapshot can help some uses
cases and not taking a snapshot is arguably correct irrespective of the
race with refinement.
Signed-off-by: Anton Romanov <romanton@google.com>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220608183525.1143682-1-romanton@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Verify the KVM allows userspace to set all supported bits in the
IA32_FEATURE_CONTROL MSR irrespective of the current guest CPUID, and
that all unsupported bits are rejected.
Throw the testcase into vmx_msrs_test even though it's not technically a
VMX MSR; it's close enough, and the most frequently feature controlled by
the MSR is VMX.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220607232353.3375324-4-seanjc@google.com
Move the check on IA32_FEATURE_CONTROL being locked, i.e. read-only from
the guest, into the helper to check the overall validity of the incoming
value. Opportunistically rename the helper to make it clear that it
returns a bool.
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220607232353.3375324-3-seanjc@google.com
Allow userspace to set all supported bits in MSR IA32_FEATURE_CONTROL
irrespective of the guest CPUID model, e.g. via KVM_SET_MSRS. KVM's ABI
is that userspace is allowed to set MSRs before CPUID, i.e. can set MSRs
to values that would fault according to the guest CPUID model.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220607232353.3375324-2-seanjc@google.com
Declare vmread_error_trampoline() as an opaque symbol so that it cannot
be called from C code, at least not without some serious fudging. The
trampoline always passes parameters on the stack so that the inline
VMREAD sequence doesn't need to clobber registers. regparm(0) was
originally added to document the stack behavior, but it ended up being
confusing because regparm(0) is a nop for 64-bit targets.
Opportunustically wrap the trampoline and its declaration in #ifdeffery
to make it even harder to invoke incorrectly, to document why it exists,
and so that it's not left behind if/when CONFIG_CC_HAS_ASM_GOTO_OUTPUT
is true for all supported toolchains.
No functional change intended.
Cc: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220928232015.745948-1-seanjc@google.com
Reword the comments that (attempt to) document nVMX's overrides of the
CR0/4 read shadows for L2 after calling vmx_set_cr0/4(). The important
behavior that needs to be documented is that KVM needs to override the
shadows to account for L1's masks even though the shadows are set by the
common helpers (and that setting the shadows first would result in the
correct shadows being clobbered).
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
Link: https://lore.kernel.org/r/20220831000721.4066617-1-seanjc@google.com
Clean up the KVM_CAP_X86_USER_SPACE_MSR documentation to eliminate
misleading and/or inconsistent verbiage, and to actually document what
accesses are intercepted by which flags.
- s/will/may since not all #GPs are guaranteed to be intercepted
- s/deflect/intercept to align with common KVM terminology
- s/user space/userspace to align with the majority of KVM docs
- Avoid using "trap" terminology, as KVM exits to userspace _before_
stepping, i.e. doesn't exhibit trap-like behavior
- Actually document the flags
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220831001706.4075399-4-seanjc@google.com
Reword the MSR filtering documentatiion to more precisely define the
behavior of filtering using common virtualization terminology.
- Explicitly document KVM's behavior when an MSR is denied
- s/handled/allowed as there is no guarantee KVM will "handle" the
MSR access
- Drop the "fall back" terminology, which incorrectly suggests that
there is existing KVM behavior to fall back to
- Fix an off-by-one error in the range (the end is exclusive)
- Call out the interaction between MSR filtering and
KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER
- Delete the redundant paragraph on what '0' and '1' in the bitmap
means, it's covered by the sections on KVM_MSR_FILTER_{READ,WRITE}
- Delete the clause on x2APIC MSR behavior depending on APIC base, this
is covered by stating that KVM follows architectural behavior when
emulating/virtualizing MSR accesses
Reported-by: Aaron Lewis <aaronlewis@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220831001706.4075399-3-seanjc@google.com
Delete the paragraph that describes the behavior when both
KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE are set for a range. There is
nothing special about KVM's handling of this combination, whereas
explicitly documenting the combination suggests that there is some magic
behavior the user needs to be aware of.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220831001706.4075399-2-seanjc@google.com
According to Intel's document on Indirect Branch Restricted
Speculation, "Enabling IBRS does not prevent software from controlling
the predicted targets of indirect branches of unrelated software
executed later at the same predictor mode (for example, between two
different user applications, or two different virtual machines). Such
isolation can be ensured through use of the Indirect Branch Predictor
Barrier (IBPB) command." This applies to both basic and enhanced IBRS.
Since L1 and L2 VMs share hardware predictor modes (guest-user and
guest-kernel), hardware IBRS is not sufficient to virtualize
IBRS. (The way that basic IBRS is implemented on pre-eIBRS parts,
hardware IBRS is actually sufficient in practice, even though it isn't
sufficient architecturally.)
For virtual CPUs that support IBRS, add an indirect branch prediction
barrier on emulated VM-exit, to ensure that the predicted targets of
indirect branches executed in L1 cannot be controlled by software that
was executed in L2.
Since we typically don't intercept guest writes to IA32_SPEC_CTRL,
perform the IBPB at emulated VM-exit regardless of the current
IA32_SPEC_CTRL.IBRS value, even though the IBPB could technically be
deferred until L1 sets IA32_SPEC_CTRL.IBRS, if IA32_SPEC_CTRL.IBRS is
clear at emulated VM-exit.
This is CVE-2022-2196.
Fixes: 5c911beff2 ("KVM: nVMX: Skip IBPB when switching between vmcs01 and vmcs02")
Cc: Sean Christopherson <seanjc@google.com>
Signed-off-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221019213620.1953281-3-jmattson@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Inject #GP for if VMXON is attempting with a CR0/CR4 that fails the
generic "is CRx valid" check, but passes the CR4.VMXE check, and do the
generic checks _after_ handling the post-VMXON VM-Fail.
The CR4.VMXE check, and all other #UD cases, are special pre-conditions
that are enforced prior to pivoting on the current VMX mode, i.e. occur
before interception if VMXON is attempted in VMX non-root mode.
All other CR0/CR4 checks generate #GP and effectively have lower priority
than the post-VMXON check.
Per the SDM:
IF (register operand) or (CR0.PE = 0) or (CR4.VMXE = 0) or ...
THEN #UD;
ELSIF not in VMX operation
THEN
IF (CPL > 0) or (in A20M mode) or
(the values of CR0 and CR4 are not supported in VMX operation)
THEN #GP(0);
ELSIF in VMX non-root operation
THEN VMexit;
ELSIF CPL > 0
THEN #GP(0);
ELSE VMfail("VMXON executed in VMX root operation");
FI;
which, if re-written without ELSIF, yields:
IF (register operand) or (CR0.PE = 0) or (CR4.VMXE = 0) or ...
THEN #UD
IF in VMX non-root operation
THEN VMexit;
IF CPL > 0
THEN #GP(0)
IF in VMX operation
THEN VMfail("VMXON executed in VMX root operation");
IF (in A20M mode) or
(the values of CR0 and CR4 are not supported in VMX operation)
THEN #GP(0);
Note, KVM unconditionally forwards VMXON VM-Exits that occur in L2 to L1,
i.e. there is no need to check the vCPU is not in VMX non-root mode. Add
a comment to explain why unconditionally forwarding such exits is
functionally correct.
Reported-by: Eric Li <ercli@ucdavis.edu>
Fixes: c7d855c2af ("KVM: nVMX: Inject #UD if VMXON is attempted with incompatible CR0/CR4")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221006001956.329314-1-seanjc@google.com
The use of kmap_atomic() is being deprecated in favor of
kmap_local_page()[1].
The main difference between atomic and local mappings is that local
mappings don't disable page faults or preemption.
There're 2 reasons we can use kmap_local_page() here:
1. SEV is 64-bit only and kmap_local_page() doesn't disable migration in
this case, but here the function clflush_cache_range() uses CLFLUSHOPT
instruction to flush, and on x86 CLFLUSHOPT is not CPU-local and flushes
the page out of the entire cache hierarchy on all CPUs (APM volume 3,
chapter 3, CLFLUSHOPT). So there's no need to disable preemption to ensure
CPU-local.
2. clflush_cache_range() doesn't need to disable pagefault and the mapping
is still valid even if sleeps. This is also true for sched out/in when
preempted.
In addition, though kmap_local_page() is a thin wrapper around
page_address() on 64-bit, kmap_local_page() should still be used here in
preference to page_address() since page_address() isn't suitable to be used
in a generic function (like sev_clflush_pages()) where the page passed in
is not easy to determine the source of allocation. Keeping the kmap* API in
place means it can be used for things other than highmem mappings[2].
Therefore, sev_clflush_pages() is a function that should use
kmap_local_page() in place of kmap_atomic().
Convert the calls of kmap_atomic() / kunmap_atomic() to kmap_local_page() /
kunmap_local().
[1]: https://lore.kernel.org/all/20220813220034.806698-1-ira.weiny@intel.com
[2]: https://lore.kernel.org/lkml/5d667258-b58b-3d28-3609-e7914c99b31b@intel.com/
Suggested-by: Dave Hansen <dave.hansen@intel.com>
Suggested-by: Ira Weiny <ira.weiny@intel.com>
Suggested-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220928092748.463631-1-zhao1.liu@linux.intel.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Skip the WRMSR fastpath in SVM's VM-Exit handler if the next RIP isn't
valid, e.g. because KVM is running with nrips=false. SVM must decode and
emulate to skip the WRMSR if the CPU doesn't provide the next RIP.
Getting the instruction bytes to decode the WRMSR requires reading guest
memory, which in turn means dereferencing memslots, and that isn't safe
because KVM doesn't hold SRCU when the fastpath runs.
Don't bother trying to enable the fastpath for this case, e.g. by doing
only the WRMSR and leaving the "skip" until later. NRIPS is supported on
all modern CPUs (KVM has considered making it mandatory), and the next
RIP will be valid the vast, vast majority of the time.
=============================
WARNING: suspicious RCU usage
6.0.0-smp--4e557fcd3d80-skip #13 Tainted: G O
-----------------------------
include/linux/kvm_host.h:954 suspicious rcu_dereference_check() usage!
other info that might help us debug this:
rcu_scheduler_active = 2, debug_locks = 1
1 lock held by stable/206475:
#0: ffff9d9dfebcc0f0 (&vcpu->mutex){+.+.}-{3:3}, at: kvm_vcpu_ioctl+0x8b/0x620 [kvm]
stack backtrace:
CPU: 152 PID: 206475 Comm: stable Tainted: G O 6.0.0-smp--4e557fcd3d80-skip #13
Hardware name: Google, Inc. Arcadia_IT_80/Arcadia_IT_80, BIOS 10.48.0 01/27/2022
Call Trace:
<TASK>
dump_stack_lvl+0x69/0xaa
dump_stack+0x10/0x12
lockdep_rcu_suspicious+0x11e/0x130
kvm_vcpu_gfn_to_memslot+0x155/0x190 [kvm]
kvm_vcpu_gfn_to_hva_prot+0x18/0x80 [kvm]
paging64_walk_addr_generic+0x183/0x450 [kvm]
paging64_gva_to_gpa+0x63/0xd0 [kvm]
kvm_fetch_guest_virt+0x53/0xc0 [kvm]
__do_insn_fetch_bytes+0x18b/0x1c0 [kvm]
x86_decode_insn+0xf0/0xef0 [kvm]
x86_emulate_instruction+0xba/0x790 [kvm]
kvm_emulate_instruction+0x17/0x20 [kvm]
__svm_skip_emulated_instruction+0x85/0x100 [kvm_amd]
svm_skip_emulated_instruction+0x13/0x20 [kvm_amd]
handle_fastpath_set_msr_irqoff+0xae/0x180 [kvm]
svm_vcpu_run+0x4b8/0x5a0 [kvm_amd]
vcpu_enter_guest+0x16ca/0x22f0 [kvm]
kvm_arch_vcpu_ioctl_run+0x39d/0x900 [kvm]
kvm_vcpu_ioctl+0x538/0x620 [kvm]
__se_sys_ioctl+0x77/0xc0
__x64_sys_ioctl+0x1d/0x20
do_syscall_64+0x3d/0x80
entry_SYSCALL_64_after_hwframe+0x63/0xcd
Fixes: 404d5d7bff ("KVM: X86: Introduce more exit_fastpath_completion enum values")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220930234031.1732249-1-seanjc@google.com
Treat any exception during instruction decode for EMULTYPE_SKIP as a
"full" emulation failure, i.e. signal failure instead of queuing the
exception. When decoding purely to skip an instruction, KVM and/or the
CPU has already done some amount of emulation that cannot be unwound,
e.g. on an EPT misconfig VM-Exit KVM has already processeed the emulated
MMIO. KVM already does this if a #UD is encountered, but not for other
exceptions, e.g. if a #PF is encountered during fetch.
In SVM's soft-injection use case, queueing the exception is particularly
problematic as queueing exceptions while injecting events can put KVM
into an infinite loop due to bailing from VM-Enter to service the newly
pending exception. E.g. multiple warnings to detect such behavior fire:
------------[ cut here ]------------
WARNING: CPU: 3 PID: 1017 at arch/x86/kvm/x86.c:9873 kvm_arch_vcpu_ioctl_run+0x1de5/0x20a0 [kvm]
Modules linked in: kvm_amd ccp kvm irqbypass
CPU: 3 PID: 1017 Comm: svm_nested_soft Not tainted 6.0.0-rc1+ #220
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
RIP: 0010:kvm_arch_vcpu_ioctl_run+0x1de5/0x20a0 [kvm]
Call Trace:
kvm_vcpu_ioctl+0x223/0x6d0 [kvm]
__x64_sys_ioctl+0x85/0xc0
do_syscall_64+0x2b/0x50
entry_SYSCALL_64_after_hwframe+0x46/0xb0
---[ end trace 0000000000000000 ]---
------------[ cut here ]------------
WARNING: CPU: 3 PID: 1017 at arch/x86/kvm/x86.c:9987 kvm_arch_vcpu_ioctl_run+0x12a3/0x20a0 [kvm]
Modules linked in: kvm_amd ccp kvm irqbypass
CPU: 3 PID: 1017 Comm: svm_nested_soft Tainted: G W 6.0.0-rc1+ #220
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
RIP: 0010:kvm_arch_vcpu_ioctl_run+0x12a3/0x20a0 [kvm]
Call Trace:
kvm_vcpu_ioctl+0x223/0x6d0 [kvm]
__x64_sys_ioctl+0x85/0xc0
do_syscall_64+0x2b/0x50
entry_SYSCALL_64_after_hwframe+0x46/0xb0
---[ end trace 0000000000000000 ]---
Fixes: 6ea6e84309 ("KVM: x86: inject exceptions produced by x86_decode_insn")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220930233632.1725475-1-seanjc@google.com
Acquire SRCU before taking the gpc spinlock in wait_pending_event() so as
to be consistent with all other functions that acquire both locks. It's
not illegal to acquire SRCU inside a spinlock, nor is there deadlock
potential, but in general it's preferable to order locks from least
restrictive to most restrictive, e.g. if wait_pending_event() needed to
sleep for whatever reason, it could do so while holding SRCU, but would
need to drop the spinlock.
Signed-off-by: Peng Hao <flyingpeng@tencent.com>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/CAPm50a++Cb=QfnjMZ2EnCj-Sb9Y4UM-=uOEtHAcjnNLCAAf-dQ@mail.gmail.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
When refreshing a gfn=>pfn cache, skip straight to unlocking if the cache
already valid instead of stuffing the "old" variables to turn the
unmapping outro into a nop.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Drop the @gpa param from the exported check()+refresh() helpers and limit
changing the cache's GPA to the activate path. All external users just
feed in gpc->gpa, i.e. this is a fancy nop.
Allowing users to change the GPA at check()+refresh() is dangerous as
those helpers explicitly allow concurrent calls, e.g. KVM could get into
a livelock scenario. It's also unclear as to what the expected behavior
should be if multiple tasks attempt to refresh with different GPAs.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Don't partially reinitialize a gfn=>pfn cache when activating the cache,
and instead assert that the cache is not valid during activation. Bug
the VM if the assertion fails, as use-after-free and/or data corruption
is all but guaranteed if KVM ends up with a valid-but-inactive cache.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Drop kvm_gpc_unmap() as it has no users and unclear requirements. The
API was added as part of the original gfn_to_pfn_cache support, but its
sole usage[*] was never merged. Fold the guts of kvm_gpc_unmap() into
the deactivate path and drop the API. Omit acquiring refresh_lock as
as concurrent calls to kvm_gpc_deactivate() are not allowed (this is
not enforced, e.g. via lockdep. due to it being called during vCPU
destruction).
If/when temporary unmapping makes a comeback, the desirable behavior is
likely to restrict temporary unmapping to vCPU-exclusive mappings and
require the vcpu->mutex be held to serialize unmap. Use of the
refresh_lock to protect unmapping was somewhat specuatively added by
commit 93984f19e7 ("KVM: Fully serialize gfn=>pfn cache refresh via
mutex") to guard against concurrent unmaps, but the primary use case of
the temporary unmap, nested virtualization[*], doesn't actually need or
want concurrent unmaps.
[*] https://lore.kernel.org/all/20211210163625.2886-7-dwmw2@infradead.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Make kvm_gpc_refresh() use kvm instance cached in gfn_to_pfn_cache.
No functional change intended.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Michal Luczaj <mhal@rbox.co>
[sean: leave kvm_gpc_unmap() as-is]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Make hva_to_pfn_retry() use kvm instance cached in gfn_to_pfn_cache.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Michal Luczaj <mhal@rbox.co>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Make kvm_gpc_check() use kvm instance cached in gfn_to_pfn_cache.
Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Michal Luczaj <mhal@rbox.co>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
