When translating virtual to physical address with a guest CPU that
supports nested paging (NPT), we need to perform every page table walk
access indirectly through the NPT, which we correctly do.
However, we treat real mode (no page table walk) special: In that case,
we currently just skip any walks and translate VA -> PA. With NPT
enabled, we also need to then perform NPT walk to do GVA -> GPA -> HPA
which we fail to do so far.
The net result of that is that TCG VMs with NPT enabled that execute
real mode code (like SeaBIOS) end up with GPA==HPA mappings which means
the guest accesses host code and data. This typically shows as failure
to boot guests.
This patch changes the page walk logic for NPT enabled guests so that we
always perform a GVA -> GPA translation and then skip any logic that
requires an actual PTE.
That way, all remaining logic to walk the NPT stays and we successfully
walk the NPT in real mode.
Cc: qemu-stable@nongnu.org
Fixes: fe441054bb ("target-i386: Add NPT support")
Signed-off-by: Alexander Graf <graf@amazon.com>
Reported-by: Eduard Vlad <evlad@amazon.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20240921085712.28902-1-graf@amazon.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The error message for a "stepping" value that is out of bounds is a
bit odd:
$ qemu-system-x86_64 -cpu qemu64,stepping=16
qemu-system-x86_64: can't apply global qemu64-x86_64-cpu.stepping=16: Property .stepping doesn't take value 16 (minimum: 0, maximum: 15)
The "can't apply global" part is an unfortunate artifact of -cpu's
implementation. Left for another day.
The remainder feels overly verbose. Change it to
qemu64-x86_64-cpu: can't apply global qemu64-x86_64-cpu.stepping=16: parameter 'stepping' can be at most 15
Likewise for "family", "model", and "tsc-frequency".
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20241010150144.986655-6-armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Properties "family", "model", and "stepping" are visited as signed
integers. They are backed by bits in CPUX86State member
@cpuid_version. The code to extract and insert these bits mixes
signed and unsigned. Not actually wrong, but avoiding such mixing is
good practice.
Visit them as unsigned integers instead.
This adds a few mildly ugly cast in arguments of error_setg(). The
next commit will get rid of them again.
Property "tsc-frequency" is also visited as signed integer. The value
ultimately flows into the kernel, where it is 31 bits unsigned. The
QEMU code freely mixes int, uint32_t, int64_t. I elect not to attempt
draining this swamp today.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20241010150144.986655-5-armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Stack accesses should be explicit and use the privilege level of the
target stack. This ensures that SMAP is not applied when the target
stack is in ring 3.
This fixes a bug wherein i386/tcg assumed that an interrupt return, or a
far call using the CALL or JMP instruction, was always going from kernel
or user mode to kernel mode when using a call gate. This assumption is
violated if the call gate has a DPL that is greater than 0.
Analyzed-by: Robert R. Henry <rrh.henry@gmail.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/249
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Now all decoding has been done before any code generation.
There is no need anymore to save and restore cc_op* and
pc_save but, for the time being, assert that this is indeed
the case.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Group them so that it is easier to figure out which two-byte opcodes to
tackle together.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The gen_cmpxchg8b and gen_cmpxchg16b functions even have the correct
prototype already; the only thing that needs to be done is removing the
gen_lea_modrm() call.
This moves the last LOCK-enabled instructions to the new decoder. It is
now possible to assume that gen_multi0F is called only after checking
that PREFIX_LOCK was not specified.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
There are now relatively few unconverted opcodes in translate.c (there
are 13 of them including 8 for x87), and all of them have the same
format with a mod/rm byte and no immediate. A good next step is
to remove the early bail out to disas_insn_x87/disas_insn_old,
instead giving these legacy translator functions the same prototype
as the other gen_* functions.
To do this, the X86DecodeInsn can be passed down to the places that
used to fetch address bytes from the instruction stream. To make
sure that everything is done cleanly, the CPUX86State* argument is
removed.
As part of the unification, the gen_lea_modrm() name is now free,
so rename gen_load_ea() to gen_lea_modrm(). This is as good a name
and it makes the changes to translate.c easier to review.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Code generation was rewritten; it reuses the same trick to use the
CC_OP_SAR values for cc_op, but it tries to use CC_OP_ADCX or CC_OP_ADCOX
instead of CC_OP_EFLAGS. This is a tiny bit more efficient in the
common case where only CF is checked in the resulting flags.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
'hyperv_synic' test from KVM unittests was observed to be flaky on certain
hardware (hangs sometimes). Debugging shows that the problem happens in
hyperv_sint_route_new() when the test tries to set up a new SynIC
route. The function bails out on:
if (!synic->sctl_enabled) {
goto cleanup;
}
but the test writes to HV_X64_MSR_SCONTROL just before it starts
establishing SINT routes. Further investigation shows that
synic_update() (called from async_synic_update()) happens after the SINT
setup attempt and not before. Apparently, the comment before
async_safe_run_on_cpu() in kvm_hv_handle_exit() does not correctly describe
the guarantees async_safe_run_on_cpu() gives. In particular, async worked
added to a CPU is actually processed from qemu_wait_io_event() which is not
always called before KVM_RUN, i.e. kvm_cpu_exec() checks whether an exit
request is pending for a CPU and if not, keeps running the vCPU until it
meets an exit it can't handle internally. Hyper-V specific MSR writes are
not automatically trigger an exit.
Fix the issue by simply raising an exit request for the vCPU where SynIC
update was queued. This is not a performance critical path as SynIC state
does not get updated so often (and async_safe_run_on_cpu() is a big hammer
anyways).
Reported-by: Jan Richter <jarichte@redhat.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Link: https://lore.kernel.org/r/20240917160051.2637594-4-vkuznets@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Windows with Hyper-V role enabled doesn't boot with 'hv-passthrough' when
no debugger is configured, this significantly limits the usefulness of the
feature as there's no support for subtracting Hyper-V features from CPU
flags at this moment (e.g. "-cpu host,hv-passthrough,-hv-syndbg" does not
work). While this is also theoretically fixable, 'hv-syndbg' is likely
very special and unneeded in the default set. Genuine Hyper-V doesn't seem
to enable it either.
Introduce 'skip_passthrough' flag to 'kvm_hyperv_properties' and use it as
one-off to skip 'hv-syndbg' when enabling features in 'hv-passthrough'
mode. Note, "-cpu host,hv-passthrough,hv-syndbg" can still be used if
needed.
As both 'hv-passthrough' and 'hv-syndbg' are debug features, the change
should not have any effect on production environments.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Link: https://lore.kernel.org/r/20240917160051.2637594-3-vkuznets@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Putting HYPERV_FEAT_SYNDBG entry under "#ifdef CONFIG_SYNDBG" in
'kvm_hyperv_properties' array is wrong: as HYPERV_FEAT_SYNDBG is not
the highest feature number, the result is an empty (zeroed) entry in
the array (and not a skipped entry!). hyperv_feature_supported() is
designed to check that all CPUID bits are set but for a zeroed
feature in 'kvm_hyperv_properties' it returns 'true' so QEMU considers
HYPERV_FEAT_SYNDBG as always supported, regardless of whether KVM host
actually supports it.
To fix the issue, leave HYPERV_FEAT_SYNDBG's definition in
'kvm_hyperv_properties' array, there's nothing wrong in having it defined
even when 'CONFIG_SYNDBG' is not set. Instead, put "hv-syndbg" CPU property
under '#ifdef CONFIG_SYNDBG' to alter the existing behavior when the flag
is silently skipped in !CONFIG_SYNDBG builds.
Leave an 'assert' sentinel in hyperv_feature_supported() making sure there
are no 'holes' or improperly defined features in 'kvm_hyperv_properties'.
Fixes: d8701185f4 ("hw: hyperv: Initial commit for Synthetic Debugging device")
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Link: https://lore.kernel.org/r/20240917160051.2637594-2-vkuznets@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Following 5 bits in CPUID.7.2.EDX are supported by KVM. Add their
supports in QEMU. Each of them indicates certain bits of IA32_SPEC_CTRL
are supported. Those bits can control CPU speculation behavior which can
be used to defend against side-channel attacks.
bit0: intel-psfd
if 1, indicates bit 7 of the IA32_SPEC_CTRL MSR is supported. Bit 7 of
this MSR disables Fast Store Forwarding Predictor without disabling
Speculative Store Bypass
bit1: ipred-ctrl
If 1, indicates bits 3 and 4 of the IA32_SPEC_CTRL MSR are supported.
Bit 3 of this MSR enables IPRED_DIS control for CPL3. Bit 4 of this
MSR enables IPRED_DIS control for CPL0/1/2
bit2: rrsba-ctrl
If 1, indicates bits 5 and 6 of the IA32_SPEC_CTRL MSR are supported.
Bit 5 of this MSR disables RRSBA behavior for CPL3. Bit 6 of this MSR
disables RRSBA behavior for CPL0/1/2
bit3: ddpd-u
If 1, indicates bit 8 of the IA32_SPEC_CTRL MSR is supported. Bit 8 of
this MSR disables Data Dependent Prefetcher.
bit4: bhi-ctrl
if 1, indicates bit 10 of the IA32_SPEC_CTRL MSR is supported. Bit 10
of this MSR enables BHI_DIS_S behavior.
Signed-off-by: Chao Gao <chao.gao@intel.com>
Link: https://lore.kernel.org/r/20240919051011.118309-1-chao.gao@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
- CPUID.(EAX=07H,ECX=0H):EBX[bit 6]: x87 FPU Data Pointer updated only
on x87 exceptions if 1.
- CPUID.(EAX=07H,ECX=0H):EBX[bit 13]: Deprecates FPU CS and FPU DS
values if 1. i.e., X87 FCS and FDS are always zero.
Define names for them so that they can be exposed to guest with -cpu host.
Also define the bit field MACROs so that named cpu models can add it as
well in the future.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Link: https://lore.kernel.org/r/20240814075431.339209-3-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
According to AMD's Speculative Return Stack Overflow whitepaper (link
below), the hypervisor should synthesize the value of IBPB_BRTYPE and
SBPB CPUID bits to the guest.
Support for this is already present in the kernel with commit
e47d86083c66 ("KVM: x86: Add SBPB support") and commit 6f0f23ef76be
("KVM: x86: Add IBPB_BRTYPE support").
Add support in QEMU to expose the bits to the guest OS.
host:
# cat /sys/devices/system/cpu/vulnerabilities/spec_rstack_overflow
Mitigation: Safe RET
before (guest):
$ cpuid -l 0x80000021 -1 -r
0x80000021 0x00: eax=0x00000045 ebx=0x00000000 ecx=0x00000000 edx=0x00000000
^
$ cat /sys/devices/system/cpu/vulnerabilities/spec_rstack_overflow
Vulnerable: Safe RET, no microcode
after (guest):
$ cpuid -l 0x80000021 -1 -r
0x80000021 0x00: eax=0x18000045 ebx=0x00000000 ecx=0x00000000 edx=0x00000000
^
$ cat /sys/devices/system/cpu/vulnerabilities/spec_rstack_overflow
Mitigation: Safe RET
Reported-by: Fabian Vogt <fvogt@suse.de>
Link: https://www.amd.com/content/dam/amd/en/documents/corporate/cr/speculative-return-stack-overflow-whitepaper.pdf
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240805202041.5936-1-farosas@suse.de
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
identity_base variable is first initialzied to address 0xfffbc000 and then
kvm_vm_set_identity_map_addr() overrides this value to address 0xfeffc000.
The initial address to which the variable was initialized was never used. Clean
everything up, placing 0xfeffc000 in a preprocessor constant.
Reported-by: Ani Sinha <anisinha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
is_host_cpu_intel() should return TRUE if the host cpu in Intel based, otherwise
it should return FALSE. Currently, it returns zero (FALSE) when the host CPU
is INTEL and non-zero otherwise. Fix the function so that it agrees more with
the semantics. Adjust the calling logic accordingly. RAPL needs Intel host cpus.
If the host CPU is not Intel baseed, we should report error.
Signed-off-by: Ani Sinha <anisinha@redhat.com>
Link: https://lore.kernel.org/r/20240903080004.33746-1-anisinha@redhat.com
[While touching the code remove too many spaces from the second part of the
error. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Because the index value of the VMCS field encoding of FRED injected-event
data (one of the newly added VMCS fields for FRED transitions), 0x52, is
larger than any existing index value, raise the highest index value used
for any VMCS encoding to 0x52.
Because the index value of the VMCS field encoding of Secondary VM-exit
controls, 0x44, is larger than any existing index value, raise the highest
index value used for any VMCS encoding to 0x44.
Co-developed-by: Xin Li <xin3.li@intel.com>
Signed-off-by: Xin Li <xin3.li@intel.com>
Signed-off-by: Lei Wang <lei4.wang@intel.com>
Signed-off-by: Xin Li (Intel) <xin@zytor.com>
Link: https://lore.kernel.org/r/20240807081813.735158-4-xin@zytor.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This is preliminary work to split up hv_vm_create
logic per platform so we can support creating VMs
with > 64GB of RAM on Apple Silicon machines. This
is done via ARM HVF's hv_vm_config_create() (and
other APIs that modify this config that will be
coming in future patches). This should have no
behavioral difference at all as hv_vm_config_create()
just assigns the same default values as if you just
passed NULL to the function.
Signed-off-by: Danny Canter <danny_canter@apple.com>
Message-id: 20240828111552.93482-3-danny_canter@apple.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Change the data type of the ioctl _request_ argument from 'int' to
'unsigned long' for the various accel/kvm functions which are
essentially wrappers around the ioctl() syscall.
The correct type for ioctl()'s 'request' argument is confused:
* POSIX defines the request argument as 'int'
* glibc uses 'unsigned long' in the prototype in sys/ioctl.h
* the glibc info documentation uses 'int'
* the Linux manpage uses 'unsigned long'
* the Linux implementation of the syscall uses 'unsigned int'
If we wrap ioctl() with another function which uses 'int' as the
type for the request argument, then requests with the 0x8000_0000
bit set will be sign-extended when the 'int' is cast to
'unsigned long' for the call to ioctl().
On x86_64 one such example is the KVM_IRQ_LINE_STATUS request.
Bit requests with the _IOC_READ direction bit set, will have the high
bit set.
Fortunately the Linux Kernel truncates the upper 32bit of the request
on 64bit machines (because it uses 'unsigned int', and see also Linus
Torvalds' comments in
https://sourceware.org/bugzilla/show_bug.cgi?id=14362 )
so this doesn't cause active problems for us. However it is more
consistent to follow the glibc ioctl() prototype when we define
functions that are essentially wrappers around ioctl().
This resolves a Coverity issue where it points out that in
kvm_get_xsave() we assign a value (KVM_GET_XSAVE or KVM_GET_XSAVE2)
to an 'int' variable which can't hold it without overflow.
Resolves: Coverity CID 1547759
Signed-off-by: Johannes Stoelp <johannes.stoelp@gmail.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20240815122747.3053871-1-peter.maydell@linaro.org
[PMM: Rebased patch, adjusted commit message, included note about
Coverity fix, updated the type of the local var in kvm_get_xsave,
updated the comment in the KVMState struct definition]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious. It's best used with
restraint.
QCryptoHashAlgorithm has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTO_HASH_ALG.
We could simply drop 'prefix', but then the prefix becomes
QCRYPTO_HASH_ALGORITHM, which is rather long.
We could additionally rename the type to QCryptoHashAlg, but I think
the abbreviation "alg" is less than clear.
Rename the type to QCryptoHashAlgo instead. The prefix becomes to
QCRYPTO_HASH_ALGO.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-12-armbru@redhat.com>
[Conflicts with merge commit 7bbadc60b5 resolved]
When we are using TCG plugin memory callbacks probe_access_internal
will return TLB_MMIO to force the slow path for memory access. This
results in probe_access returning NULL but the x86 access_ptr function
happily accepts an empty haddr resulting in segfault hilarity.
Check for an empty haddr to prevent the segfault and enable plugins to
track all the memory operations for the x86 save/restore helpers. As
we also want to run the slow path when instrumenting *-user we should
also not have the short cutting test_ptr macro.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2489
Fixes: 6d03226b42 (plugins: force slow path when plugins instrument memory ops)
Reviewed-by: Alexandre Iooss <erdnaxe@crans.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240813202329.1237572-8-alex.bennee@linaro.org>
