The address translation logic in get_physical_address() will currently
truncate physical addresses to 32 bits unless long mode is enabled.
This is incorrect when using physical address extensions (PAE) outside
of long mode, with the result that a 32-bit operating system using PAE
to access memory above 4G will experience undefined behaviour.
The truncation code was originally introduced in commit 33dfdb5 ("x86:
only allow real mode to access 32bit without LMA"), where it applied
only to translations performed while paging is disabled (and so cannot
affect guests using PAE).
Commit 9828198 ("target/i386: Add MMU_PHYS_IDX and MMU_NESTED_IDX")
rearranged the code such that the truncation also applied to the use
of MMU_PHYS_IDX and MMU_NESTED_IDX. Commit 4a1e9d4 ("target/i386: Use
atomic operations for pte updates") brought this truncation into scope
for page table entry accesses, and is the first commit for which a
Windows 10 32-bit guest will reliably fail to boot if memory above 4G
is present.
The truncation code however is not completely redundant. Even though the
maximum address size for any executed instruction is 32 bits, helpers for
operations such as BOUND, FSAVE or XSAVE may ask get_physical_address()
to translate an address outside of the 32-bit range, if invoked with an
argument that is close to the 4G boundary. Likewise for processor
accesses, for example TSS or IDT accesses, when EFER.LMA==0.
So, move the address truncation in get_physical_address() so that it
applies to 32-bit MMU indexes, but not to MMU_PHYS_IDX and MMU_NESTED_IDX.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2040
Fixes: 4a1e9d4d11 ("target/i386: Use atomic operations for pte updates", 2022-10-18)
Cc: qemu-stable@nongnu.org
Co-developed-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Accesses from a 32-bit environment (32-bit code segment for instruction
accesses, EFER.LMA==0 for processor accesses) have to mask away the
upper 32 bits of the address. While a bit wasteful, the easiest way
to do so is to use separate MMU indexes. These days, QEMU anyway is
compiled with a fixed value for NB_MMU_MODES. Split MMU_USER_IDX,
MMU_KSMAP_IDX and MMU_KNOSMAP_IDX in two.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The processor tracing features in cpu_x86_cpuid() are hardcoded to a set
that should be safe on all processor that support PT virtualization.
But as an additional check, x86_cpu_filter_features() also checks
that the accelerator supports that safe subset, and if not it marks
CPUID_7_0_EBX_INTEL_PT as unavailable.
This check fails on accelerators other than KVM, but it is actually
unnecessary to do it because KVM is the only accelerator that uses the
safe subset. Everything else just provides nonzero values for CPUID
leaf 0x14 (TCG/HVF because processor tracing is not supported; qtest
because nothing is able to read CPUID anyway). Restricting the check
to KVM fixes a warning with the qtest accelerator:
$ qemu-system-x86_64 -display none -cpu max,mmx=off -accel qtest
qemu-system-x86_64: warning: TCG doesn't support requested feature: CPUID.07H:EBX.intel-pt [bit 25]
The warning also happens in the test-x86-cpuid-compat qtest.
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2096
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <20240221162910.101327-1-pbonzini@redhat.com>
Fixes: d047402436 ("target/i386: Call accel-agnostic x86_cpu_get_supported_cpuid()")
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
KVM_FEATURE_ASYNC_PF_VMEXIT has been introduced for years, however QEMU
doesn't support expose it to guest. Add support for it.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-ID: <20231024083354.1171308-1-xiaoyao.li@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The value of FEAT_XSAVE_XCR0_HI leaf and FEAT_XSAVE_XSS_HI leaf also
need to be masked by XCR0 and XSS mask respectively, to make it
logically correct.
Fixes: 301e90675c ("target/i386: Enable support for XSAVES based features")
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Reviewed-by: Yang Weijiang <weijiang.yang@intel.com>
Message-ID: <20240115091325.1904229-3-xiaoyao.li@intel.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Leaf FEAT_XSAVE_XSS_LO and FEAT_XSAVE_XSS_HI also need to be cleared
when CPUID_EXT_XSAVE is not set.
Fixes: 301e90675c ("target/i386: Enable support for XSAVES based features")
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Reviewed-by: Yang Weijiang <weijiang.yang@intel.com>
Message-ID: <20240115091325.1904229-2-xiaoyao.li@intel.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240106132546.21248-4-shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This commit adds support for x2APIC transitions when writing to
MSR_IA32_APICBASE register and finally adds CPUID_EXT_X2APIC to
TCG_EXT_FEATURES.
The set_base in APICCommonClass now returns an integer to indicate error in
execution. apic_set_base return -1 on invalid APIC state transition,
accelerator can use this to raise appropriate exception.
Signed-off-by: Bui Quang Minh <minhquangbui99@gmail.com>
Message-Id: <20240111154404.5333-4-minhquangbui99@gmail.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
For user-only mode, use MMU_USER_IDX.
For system mode, use CPUClass.mmu_index.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
'tcg_cflags' is specific to TCG.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231130075958.21285-1-philmd@linaro.org>
Use generic cpu_model_from_type() when the CPU model name needs to
be extracted from the CPU type name.
Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-23-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
The main difficulty here is that a page fault when writing to the destination
must not overwrite the flags. Therefore, the flags computation must be
inlined instead of using gen_jcc1*.
For simplicity, I am using an unconditional cmpxchg operation, that becomes
a NOP if the comparison fails.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
vCPU "reset" is only possible with system emulation.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20231212113640.30287-5-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Improve
$ qemu-system-x86_64 -device max-x86_64-cpu,vendor=me
qemu-system-x86_64: -device max-x86_64-cpu,vendor=me: Property '.vendor' doesn't take value 'me'
to
qemu-system-x86_64: -device max-x86_64-cpu,vendor=0123456789abc: value of property 'vendor' must consist of exactly 12 characters
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20231031111059.3407803-8-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
[Typo corrected]
From CPUState.nr_cores' comment, it represents "number of cores within
this CPU package".
After 003f230e37 ("machine: Tweak the order of topology members in
struct CpuTopology"), the meaning of smp.cores changed to "the number of
cores in one die", but this commit missed to change CPUState.nr_cores'
calculation, so that CPUState.nr_cores became wrong and now it
misses to consider numbers of clusters and dies.
At present, only i386 is using CPUState.nr_cores.
But as for i386, which supports die level, the uses of CPUState.nr_cores
are very confusing:
Early uses are based on the meaning of "cores per package" (before die
is introduced into i386), and later uses are based on "cores per die"
(after die's introduction).
This difference is due to that commit a94e142899 ("target/i386: Add
CPUID.1F generation support for multi-dies PCMachine") misunderstood
that CPUState.nr_cores means "cores per die" when calculated
CPUID.1FH.01H:EBX. After that, the changes in i386 all followed this
wrong understanding.
With the influence of 003f230e37 and a94e142899, for i386 currently
the result of CPUState.nr_cores is "cores per die", thus the original
uses of CPUState.cores based on the meaning of "cores per package" are
wrong when multiple dies exist:
1. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.01H:EBX[bits 23:16] is
incorrect because it expects "cpus per package" but now the
result is "cpus per die".
2. In cpu_x86_cpuid() of target/i386/cpu.c, for all leaves of CPUID.04H:
EAX[bits 31:26] is incorrect because they expect "cpus per package"
but now the result is "cpus per die". The error not only impacts the
EAX calculation in cache_info_passthrough case, but also impacts other
cases of setting cache topology for Intel CPU according to cpu
topology (specifically, the incoming parameter "num_cores" expects
"cores per package" in encode_cache_cpuid4()).
3. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.0BH.01H:EBX[bits
15:00] is incorrect because the EBX of 0BH.01H (core level) expects
"cpus per package", which may be different with 1FH.01H (The reason
is 1FH can support more levels. For QEMU, 1FH also supports die,
1FH.01H:EBX[bits 15:00] expects "cpus per die").
4. In cpu_x86_cpuid() of target/i386/cpu.c, when CPUID.80000001H is
calculated, here "cpus per package" is expected to be checked, but in
fact, now it checks "cpus per die". Though "cpus per die" also works
for this code logic, this isn't consistent with AMD's APM.
5. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.80000008H:ECX expects
"cpus per package" but it obtains "cpus per die".
6. In simulate_rdmsr() of target/i386/hvf/x86_emu.c, in
kvm_rdmsr_core_thread_count() of target/i386/kvm/kvm.c, and in
helper_rdmsr() of target/i386/tcg/sysemu/misc_helper.c,
MSR_CORE_THREAD_COUNT expects "cpus per package" and "cores per
package", but in these functions, it obtains "cpus per die" and
"cores per die".
On the other hand, these uses are correct now (they are added in/after
a94e142899):
1. In cpu_x86_cpuid() of target/i386/cpu.c, topo_info.cores_per_die
meets the actual meaning of CPUState.nr_cores ("cores per die").
2. In cpu_x86_cpuid() of target/i386/cpu.c, vcpus_per_socket (in CPUID.
04H's calculation) considers number of dies, so it's correct.
3. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.1FH.01H:EBX[bits
15:00] needs "cpus per die" and it gets the correct result, and
CPUID.1FH.02H:EBX[bits 15:00] gets correct "cpus per package".
When CPUState.nr_cores is correctly changed to "cores per package" again
, the above errors will be fixed without extra work, but the "currently"
correct cases will go wrong and need special handling to pass correct
"cpus/cores per die" they want.
Fix CPUState.nr_cores' calculation to fit the original meaning "cores
per package", as well as changing calculation of topo_info.cores_per_die,
vcpus_per_socket and CPUID.1FH.
Fixes: a94e142899 ("target/i386: Add CPUID.1F generation support for multi-dies PCMachine")
Fixes: 003f230e37 ("machine: Tweak the order of topology members in struct CpuTopology")
Signed-off-by: Zhuocheng Ding <zhuocheng.ding@intel.com>
Co-developed-by: Zhao Liu <zhao1.liu@intel.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Tested-by: Babu Moger <babu.moger@amd.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20231024090323.1859210-4-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
PAE mode in x86 supports 36 bit address space. Check the PAE CPUID on the
guest processor and set phys_bits to 36 if PAE feature is set. This is in
addition to checking the presence of PSE36 CPUID feature for setting 36 bit
phys_bits.
Signed-off-by: Ani Sinha <anisinha@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-ID: <20230912120650.371781-1-anisinha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The implementation was validated with OpenSSL and with the test vectors in
https://github.com/rust-lang/stdarch/blob/master/crates/core_arch/src/x86/sha.rs.
The instructions provide a ~25% improvement on hashing a 64 MiB file:
runtime goes down from 1.8 seconds to 1.4 seconds; instruction count on
the host goes down from 5.8 billion to 4.8 billion with slightly better
IPC too. Good job Intel. ;)
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When explicitly booting a multiple vcpus vm with "-cpu +ht", it gets
warning of
warning: host doesn't support requested feature: CPUID.01H:EDX.ht [bit 28]
Make CPUID_HT as supported unconditionally can resolve the warning.
However it introduces another issue that it also expose CPUID_HT to
guest when "-cpu host/max" with only 1 vcpu. To fix this, need mark
CPUID_HT as the no_autoenable_flags.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-ID: <20231010060539.210258-1-xiaoyao.li@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
All implementations of gdb_arch_name() returns dynamic duplicates of
static strings. It's also unlikely that there will be an implementation
of gdb_arch_name() that returns a truly dynamic value due to the nature
of the function returning a well-known identifiers. Qualify the value
gdb_arch_name() with const and make all of its implementations return
static strings.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230912224107.29669-8-akihiko.odaki@daynix.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20231009164104.369749-15-alex.bennee@linaro.org>
Since we *might* have user emulation with softmmu,
replace the system emulation check by !user emulation one.
(target/ was cleaned from invalid CONFIG_SOFTMMU uses at
commit cab35c73be, but these files were merged few days
after, thus missed the cleanup.)
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231004082239.27251-1-philmd@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
hvf_get_supported_cpuid() is only defined for x86 targets
(in target/i386/hvf/x86_cpuid.c).
Its declaration is pointless on all other targets.
All the calls to it in target/i386/cpu.c are guarded by
a call on hvf_enabled(), so are elided when HVF is not
built in. Therefore we can remove the unnecessary function
stub.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231004092510.39498-3-philmd@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This function is now empty, so remove it. In the case of
m68k and tricore, this empties the class instance initfn,
so remove those as well.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The omission of alignment has technically been wrong since
269bd5d8f6, where QEMU_ALIGNED was added to CPUTLBDescFast.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Gather Data Sampling (GDS) is a side-channel attack using Gather
instructions. Some Intel processors will set ARCH_CAP_GDS_NO bit in
MSR IA32_ARCH_CAPABILITIES to report that they are not vulnerable to
GDS.
Make this bit available to guests.
Closes: https://lore.kernel.org/qemu-devel/CAMGffEmG6TNq0n3+4OJAgXc8J0OevY60KHZekXCBs3LoK9vehA@mail.gmail.com/
Reported-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Tested-by: Jack Wang <jinpu.wang@ionos.com>
Tested-by: Daniel Sneddon <daniel.sneddon@linux.intel.com>
Message-ID: <fde42d81ce454477ca8e27d5429a190b7366fe86.1692074650.git.pawan.kumar.gupta@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
On parts that enumerate IA32_VMX_BASIC MSR bit as 1, any exception vector
can be delivered with or without an error code if the other consistency
checks are satisfied.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
x86_cpu_get_supported_cpuid() is generic and handles the different
accelerators. Use it instead of kvm_arch_get_supported_cpuid().
That fixes a link failure introduced by commit 3adce820cf
("target/i386: Remove unused KVM stubs") when QEMU is configured
as:
$ ./configure --cc=clang \
--target-list=x86_64-linux-user,x86_64-softmmu \
--enable-debug
We were getting:
[71/71] Linking target qemu-x86_64
FAILED: qemu-x86_64
/usr/bin/ld: libqemu-x86_64-linux-user.fa.p/target_i386_cpu.c.o: in function `cpu_x86_cpuid':
cpu.c:(.text+0x1374): undefined reference to `kvm_arch_get_supported_cpuid'
/usr/bin/ld: libqemu-x86_64-linux-user.fa.p/target_i386_cpu.c.o: in function `x86_cpu_filter_features':
cpu.c:(.text+0x81c2): undefined reference to `kvm_arch_get_supported_cpuid'
/usr/bin/ld: cpu.c:(.text+0x81da): undefined reference to `kvm_arch_get_supported_cpuid'
/usr/bin/ld: cpu.c:(.text+0x81f2): undefined reference to `kvm_arch_get_supported_cpuid'
/usr/bin/ld: cpu.c:(.text+0x820a): undefined reference to `kvm_arch_get_supported_cpuid'
/usr/bin/ld: libqemu-x86_64-linux-user.fa.p/target_i386_cpu.c.o:cpu.c:(.text+0x8225): more undefined references to `kvm_arch_get_supported_cpuid' follow
clang: error: linker command failed with exit code 1 (use -v to see invocation)
ninja: build stopped: subcommand failed.
For the record, this is because '--enable-debug' disables
optimizations (CFLAGS=-O0).
While at this (un)optimization level GCC eliminate the
following dead code (CPP output of mentioned build):
static void x86_cpu_get_supported_cpuid(uint32_t func, uint32_t index,
uint32_t *eax, uint32_t *ebx,
uint32_t *ecx, uint32_t *edx)
{
if ((0)) {
*eax = kvm_arch_get_supported_cpuid(kvm_state, func, index, R_EAX);
*ebx = kvm_arch_get_supported_cpuid(kvm_state, func, index, R_EBX);
*ecx = kvm_arch_get_supported_cpuid(kvm_state, func, index, R_ECX);
*edx = kvm_arch_get_supported_cpuid(kvm_state, func, index, R_EDX);
} else if (0) {
*eax = 0;
*ebx = 0;
*ecx = 0;
*edx = 0;
} else {
*eax = 0;
*ebx = 0;
*ecx = 0;
*edx = 0;
}
Clang does not (see commit 2140cfa51d "i386: Fix build by
providing stub kvm_arch_get_supported_cpuid()").
Cc: qemu-stable@nongnu.org
Fixes: 3adce820cf ("target/i386: Remove unused KVM stubs")
Reported-by: Kevin Wolf <kwolf@redhat.com>
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230913093009.83520-4-philmd@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
x86_cpu_get_supported_cpuid() already checks for KVM/HVF
accelerators, so it is not needed to manually check it via
a call to accel_uses_host_cpuid() before calling it.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230913093009.83520-3-philmd@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In case more code is added after the kvm_hyperv_expand_features()
call, check its return value (since it can fail).
Fixes: 071ce4b03b ("i386: expand Hyper-V features during CPU feature expansion time")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230913093009.83520-2-philmd@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Latest Intel platform GraniteRapids-D introduces AMX-COMPLEX, which adds
two instructions to perform matrix multiplication of two tiles containing
complex elements and accumulate the results into a packed single precision
tile.
AMX-COMPLEX is enumerated via CPUID.(EAX=7,ECX=1):EDX[bit 8]. Add the CPUID
definition for AMX-COMPLEX, AMX-COMPLEX will be enabled automatically when
using '-cpu host' and KVM advertises AMX-COMPLEX to userspace.
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-ID: <20230830074324.84059-1-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Current QEMU can expose waitpkg to guests when it is available. However,
VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE is still not recognized and
masked by QEMU. This can lead to an unexpected situation when a L1
hypervisor wants to expose waitpkg to a L2 guest. The L1 hypervisor can
assume that VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE exists as waitpkg is
available. The L1 hypervisor then can accidentally expose waitpkg to the
L2 guest. This will cause invalid opcode exception in the L2 guest when
it executes waitpkg related instructions.
This patch adds VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE support, and
sets up dependency between the bit and CPUID_7_0_ECX_WAITPKG. QEMU should
not expose waitpkg feature if VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE is
not available to avoid unexpected invalid opcode exception in L2 guests.
Signed-off-by: Ake Koomsin <ake@igel.co.jp>
Message-ID: <20230807093339.32091-2-ake@igel.co.jp>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The GraniteRapids CPU model mainly adds the following new features
based on SapphireRapids:
- PREFETCHITI CPUID.(EAX=7,ECX=1):EDX[bit 14]
- AMX-FP16 CPUID.(EAX=7,ECX=1):EAX[bit 21]
And adds the following security fix for corresponding vulnerabilities:
- MCDT_NO CPUID.(EAX=7,ECX=2):EDX[bit 5]
- SBDR_SSDP_NO MSR_IA32_ARCH_CAPABILITIES[bit 13]
- FBSDP_NO MSR_IA32_ARCH_CAPABILITIES[bit 14]
- PSDP_NO MSR_IA32_ARCH_CAPABILITIES[bit 15]
- PBRSB_NO MSR_IA32_ARCH_CAPABILITIES[bit 24]
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Tested-by: Xuelian Guo <xuelian.guo@intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-ID: <20230706054949.66556-7-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
SapphireRapids has bit 13, 14 and 15 of MSR_IA32_ARCH_CAPABILITIES
enabled, which are related to some security fixes.
Add version 2 of SapphireRapids CPU model with those bits enabled also.
Signed-off-by: Lei Wang <lei4.wang@intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Message-ID: <20230706054949.66556-6-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
CPUID.(EAX=7,ECX=2):EDX[bit 5] enumerates MCDT_NO. Processors enumerate
this bit as 1 do not exhibit MXCSR Configuration Dependent Timing (MCDT)
behavior and do not need to be mitigated to avoid data-dependent behavior
for certain instructions.
Since MCDT_NO is in a new sub-leaf, add a new CPUID feature word
FEAT_7_2_EDX. Also update cpuid_level_func7 by FEAT_7_2_EDX.
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-ID: <20230706054949.66556-3-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If FEAT_7_1_EAX is 0 and FEAT_7_1_EDX is non-zero, as is the case
with a Granite Rapids host and
'-cpu host,-avx-vnni,-avx512-bf16,-fzrm,-fsrs,-fsrc,-amx-fp16', we can't
get CPUID_7_1 leaf even though CPUID_7_1_EDX has non-zero value.
Update cpuid_level_func7 according to CPUID_7_1_EDX, otherwise
guest may report wrong maximum number sub-leaves in leaf 07H.
Fixes: eaaa197d5b ("target/i386: Add support for AVX-VNNI-INT8 in CPUID enumeration")
Cc: qemu-stable@nongnu.org
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-ID: <20230706054949.66556-2-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
32-bit binaries can run on a long mode processor even if the kernel
is 64-bit, of course, and this can have slightly different behavior;
for example, SYSCALL is allowed on Intel processors.
Allow reporting LM to programs running under user mode emulation,
so that "-cpu" can be used with named CPU models even for qemu-i386
and even without disabling LM by hand.
Fortunately, most of the runtime code in QEMU has to depend on HF_LMA_MASK
or on HF_CS64_MASK (which is anyway false for qemu-i386's 32-bit code
segment) rather than TARGET_X86_64, therefore all that is needed is an
update of linux-user's ring 0 setup.
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/1534
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Features such as PCID are only accessible through privileged operations,
and therefore have no impact on any user-mode operation. Allow reporting
them to programs running under user mode emulation, so that "-cpu" can be
used with more named CPU models.
XSAVES would be similar, but it doesn't make sense to provide it until
XSAVEC is implemented.
With this change, all CPUs up to Broadwell-v4 can be emulate. Skylake-Client
requires XSAVEC, while EPYC also requires SHA-NI, MISALIGNSSE and TOPOEXT.
MISALIGNSSE is not hard to implement, but I am not sure it is worth using
a precious hflags bit for it.
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/1534
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
ARCH_CAPABILITIES is only accessible through a read-only MSR, so it has
no impact on any user-mode operation (user-mode cannot read the MSR).
So do not bother printing warnings about it in user mode emulation.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
On Intel CPUs there are certain bits in MSR_ARCH_CAPABILITIES that
indicates if the CPU is not affected by a vulnerability. Without these
bits guests may try to deploy the mitigation even if the CPU is not
affected.
Export the bits to guests that indicate immunity to hardware
vulnerabilities.
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Message-ID: <63d85cc76d4cdc51e6c732478b81d8f13be11e5a.1687551881.git.pawan.kumar.gupta@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
AMD supports both 32-bit and 64-bit SYSCALL/SYSRET, but the TCG only
exposes it for 64-bit targets. For system emulation just reuse the
helper; for user-mode emulation the ABI is the same as "int $80".
The BSDs does not support any fast system call mechanism in 32-bit
mode so add to bsd-user the same stub that FreeBSD has for 64-bit
compatibility mode.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
RDPID corresponds to a RDMSR(TSC_AUX); however, it is unprivileged
so for user-mode emulation we must provide the value that the kernel
places in the MSR. For Linux, it is a combination of the current CPU
and the current NUMA node, both of which can be retrieved with getcpu(2).
Also try sched_getcpu(), which might be there on the BSDs. If there is
no portable way to retrieve the current CPU id from userspace, return 0.
RDTSCP is reimplemented as RDTSC + RDPID ECX; the differences in terms
of serializability are not relevant to QEMU.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
WBNOINVD is the same as INVD or WBINVD as far as TCG is concerned,
since there is no cache in TCG and therefore no invalidation side effect
in WBNOINVD.
With respect to SVM emulation, processors that do not support WBNOINVD
will ignore the prefix and treat it as WBINVD, while those that support
it will generate exactly the same vmexit.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
XSAVEERPTR is actually a fix for an errata; TCG does not have the issue.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
TCG implements RDSEED, and in fact uses qcrypto_random_bytes which is
secure enough to match hardware behavior. Expose it to guests.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The AMD prefetch(w) instructions have not been deprecated together with the rest
of 3DNow!, and in fact are even supported by newer Intel processor. Mark them
as supported by TCG, as it supports all of 3DNow!.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Based on the kernel commit "b0563468ee x86/CPU/AMD: Disable XSAVES on
AMD family 0x17", host system with EPYC-Rome can clear XSAVES capability
bit. In another words, EPYC-Rome host without XSAVES can occur. Thus, we
need an EPYC-Rome cpu model (without this feature) that matches the
solution of fixing this erratum
Signed-off-by: Maksim Davydov <davydov-max@yandex-team.ru>
Message-Id: <20230524213748.8918-1-davydov-max@yandex-team.ru>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
As reported by the Intel's doc:
"FB_CLEAR: The processor will overwrite fill buffer values as part of
MD_CLEAR operations with the VERW instruction.
On these processors, L1D_FLUSH does not overwrite fill buffer values."
If this cpu feature is present in host, allow QEMU to choose whether to
show it to the guest too.
One disadvantage of not exposing it is that the guest will report
a non existing vulnerability in
/sys/devices/system/cpu/vulnerabilities/mmio_stale_data
because the mitigation is present only when the cpu has
(FLUSH_L1D and MD_CLEAR) or FB_CLEAR
features enabled.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20230201135759.555607-3-eesposit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
As reported by Intel's doc:
"L1D_FLUSH: Writeback and invalidate the L1 data cache"
If this cpu feature is present in host, allow QEMU to choose whether to
show it to the guest too.
One disadvantage of not exposing it is that the guest will report
a non existing vulnerability in
/sys/devices/system/cpu/vulnerabilities/mmio_stale_data
because the mitigation is present only when the cpu has
(FLUSH_L1D and MD_CLEAR) or FB_CLEAR
features enabled.
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20230201135759.555607-2-eesposit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Adds the support for AMD EPYC Genoa generation processors. The model
display for the new processor will be EPYC-Genoa.
Adds the following new feature bits on top of the feature bits from
the previous generation EPYC models.
avx512f : AVX-512 Foundation instruction
avx512dq : AVX-512 Doubleword & Quadword Instruction
avx512ifma : AVX-512 Integer Fused Multiply Add instruction
avx512cd : AVX-512 Conflict Detection instruction
avx512bw : AVX-512 Byte and Word Instructions
avx512vl : AVX-512 Vector Length Extension Instructions
avx512vbmi : AVX-512 Vector Byte Manipulation Instruction
avx512_vbmi2 : AVX-512 Additional Vector Byte Manipulation Instruction
gfni : AVX-512 Galois Field New Instructions
avx512_vnni : AVX-512 Vector Neural Network Instructions
avx512_bitalg : AVX-512 Bit Algorithms, add bit algorithms Instructions
avx512_vpopcntdq: AVX-512 AVX-512 Vector Population Count Doubleword and
Quadword Instructions
avx512_bf16 : AVX-512 BFLOAT16 instructions
la57 : 57-bit virtual address support (5-level Page Tables)
vnmi : Virtual NMI (VNMI) allows the hypervisor to inject the NMI
into the guest without using Event Injection mechanism
meaning not required to track the guest NMI and intercepting
the IRET.
auto-ibrs : The AMD Zen4 core supports a new feature called Automatic IBRS.
It is a "set-and-forget" feature that means that, unlike e.g.,
s/w-toggled SPEC_CTRL.IBRS, h/w manages its IBRS mitigation
resources automatically across CPL transitions.
Signed-off-by: Babu Moger <babu.moger@amd.com>
Message-Id: <20230504205313.225073-8-babu.moger@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add the following featute bits.
vnmi: Virtual NMI (VNMI) allows the hypervisor to inject the NMI into the
guest without using Event Injection mechanism meaning not required to
track the guest NMI and intercepting the IRET.
The presence of this feature is indicated via the CPUID function
0x8000000A_EDX[25].
automatic-ibrs :
The AMD Zen4 core supports a new feature called Automatic IBRS.
It is a "set-and-forget" feature that means that, unlike e.g.,
s/w-toggled SPEC_CTRL.IBRS, h/w manages its IBRS mitigation
resources automatically across CPL transitions.
The presence of this feature is indicated via the CPUID function
0x80000021_EAX[8].
The documention for the features are available in the links below.
a. Processor Programming Reference (PPR) for AMD Family 19h Model 01h,
Revision B1 Processors
b. AMD64 Architecture Programmer’s Manual Volumes 1–5 Publication No. Revision
40332 4.05 Date October 2022
Signed-off-by: Santosh Shukla <santosh.shukla@amd.com>
Signed-off-by: Kim Phillips <kim.phillips@amd.com>
Signed-off-by: Babu Moger <babu.moger@amd.com>
Link: https://www.amd.com/system/files/TechDocs/55898_B1_pub_0.50.zip
Link: https://www.amd.com/system/files/TechDocs/40332_4.05.pdf
Message-Id: <20230504205313.225073-7-babu.moger@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add the following feature bits for EPYC-Milan model and bump the version.
vaes : Vector VAES(ENC|DEC), VAES(ENC|DEC)LAST instruction support
vpclmulqdq : Vector VPCLMULQDQ instruction support
stibp-always-on : Single Thread Indirect Branch Prediction Mode has enhanced
performance and may be left Always on
amd-psfd : Predictive Store Forward Disable
no-nested-data-bp : Processor ignores nested data breakpoints
lfence-always-serializing : LFENCE instruction is always serializing
null-sel-clr-base : Null Selector Clears Base. When this bit is
set, a null segment load clears the segment base
These new features will be added in EPYC-Milan-v2. The "-cpu help" output
after the change will be.
x86 EPYC-Milan (alias configured by machine type)
x86 EPYC-Milan-v1 AMD EPYC-Milan Processor
x86 EPYC-Milan-v2 AMD EPYC-Milan Processor
The documentation for the features are available in the links below.
a. Processor Programming Reference (PPR) for AMD Family 19h Model 01h,
Revision B1 Processors
b. SECURITY ANALYSIS OF AMD PREDICTIVE STORE FORWARDING
c. AMD64 Architecture Programmer’s Manual Volumes 1–5 Publication No. Revision
40332 4.05 Date October 2022
Signed-off-by: Babu Moger <babu.moger@amd.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Link: https://www.amd.com/system/files/TechDocs/55898_B1_pub_0.50.zip
Link: https://www.amd.com/system/files/documents/security-analysis-predictive-store-forwarding.pdf
Link: https://www.amd.com/system/files/TechDocs/40332_4.05.pdf
Message-Id: <20230504205313.225073-6-babu.moger@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add the following feature bits.
no-nested-data-bp : Processor ignores nested data breakpoints.
lfence-always-serializing : LFENCE instruction is always serializing.
null-sel-cls-base : Null Selector Clears Base. When this bit is
set, a null segment load clears the segment base.
The documentation for the features are available in the links below.
a. Processor Programming Reference (PPR) for AMD Family 19h Model 01h,
Revision B1 Processors
b. AMD64 Architecture Programmer’s Manual Volumes 1–5 Publication No. Revision
40332 4.05 Date October 2022
Signed-off-by: Babu Moger <babu.moger@amd.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Link: https://www.amd.com/system/files/TechDocs/55898_B1_pub_0.50.zip
Link: https://www.amd.com/system/files/TechDocs/40332_4.05.pdf
Message-Id: <20230504205313.225073-5-babu.moger@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add the following feature bits.
amd-psfd : Predictive Store Forwarding Disable:
PSF is a hardware-based micro-architectural optimization
designed to improve the performance of code execution by
predicting address dependencies between loads and stores.
While SSBD (Speculative Store Bypass Disable) disables both
PSF and speculative store bypass, PSFD only disables PSF.
PSFD may be desirable for the software which is concerned
with the speculative behavior of PSF but desires a smaller
performance impact than setting SSBD.
Depends on the following kernel commit:
b73a54321ad8 ("KVM: x86: Expose Predictive Store Forwarding Disable")
stibp-always-on :
Single Thread Indirect Branch Prediction mode has enhanced
performance and may be left always on.
The documentation for the features are available in the links below.
a. Processor Programming Reference (PPR) for AMD Family 19h Model 01h,
Revision B1 Processors
b. SECURITY ANALYSIS OF AMD PREDICTIVE STORE FORWARDING
Signed-off-by: Babu Moger <babu.moger@amd.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Link: https://www.amd.com/system/files/documents/security-analysis-predictive-store-forwarding.pdf
Link: https://www.amd.com/system/files/TechDocs/55898_B1_pub_0.50.zip
Message-Id: <20230504205313.225073-4-babu.moger@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Introduce new EPYC cpu versions: EPYC-v4 and EPYC-Rome-v3.
The only difference vs. older models is an updated cache_info with
the 'complex_indexing' bit unset, since this bit is not currently
defined for AMD and may cause problems should it be used for
something else in the future. Setting this bit will also cause
CPUID validation failures when running SEV-SNP guests.
Signed-off-by: Michael Roth <michael.roth@amd.com>
Signed-off-by: Babu Moger <babu.moger@amd.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20230504205313.225073-3-babu.moger@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
New EPYC CPUs versions require small changes to their cache_info's.
Because current QEMU x86 CPU definition does not support versioned
cach_info, we would have to declare a new CPU type for each such case.
To avoid the dup work, add "cache_info" in X86CPUVersionDefinition",
to allow new cache_info pointers to be specified for a new CPU version.
Co-developed-by: Wei Huang <wei.huang2@amd.com>
Signed-off-by: Wei Huang <wei.huang2@amd.com>
Signed-off-by: Michael Roth <michael.roth@amd.com>
Signed-off-by: Babu Moger <babu.moger@amd.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20230504205313.225073-2-babu.moger@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Latest Intel platform Granite Rapids has introduced a new instruction -
PREFETCHIT0/1, which moves code to memory (cache) closer to the
processor depending on specific hints.
The bit definition:
CPUID.(EAX=7,ECX=1):EDX[bit 14]
Add CPUID definition for PREFETCHIT0/1.
Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-7-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
AVX-NE-CONVERT is a new set of instructions which can convert low
precision floating point like BF16/FP16 to high precision floating point
FP32, as well as convert FP32 elements to BF16. This instruction allows
the platform to have improved AI capabilities and better compatibility.
The bit definition:
CPUID.(EAX=7,ECX=1):EDX[bit 5]
Add CPUID definition for AVX-NE-CONVERT.
Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-6-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
AVX-VNNI-INT8 is a new set of instructions in the latest Intel platform
Sierra Forest, aims for the platform to have superior AI capabilities.
This instruction multiplies the individual bytes of two unsigned or
unsigned source operands, then adds and accumulates the results into the
destination dword element size operand.
The bit definition:
CPUID.(EAX=7,ECX=1):EDX[bit 4]
AVX-VNNI-INT8 is on a new feature bits leaf. Add a CPUID feature word
FEAT_7_1_EDX for this leaf.
Add CPUID definition for AVX-VNNI-INT8.
Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-5-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
AVX-IFMA is a new instruction in the latest Intel platform Sierra
Forest. This instruction packed multiplies unsigned 52-bit integers and
adds the low/high 52-bit products to Qword Accumulators.
The bit definition:
CPUID.(EAX=7,ECX=1):EAX[bit 23]
Add CPUID definition for AVX-IFMA.
Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-4-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Latest Intel platform Granite Rapids has introduced a new instruction -
AMX-FP16, which performs dot-products of two FP16 tiles and accumulates
the results into a packed single precision tile. AMX-FP16 adds FP16
capability and allows a FP16 GPU trained model to run faster without
loss of accuracy or added SW overhead.
The bit definition:
CPUID.(EAX=7,ECX=1):EAX[bit 21]
Add CPUID definition for AMX-FP16.
Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-3-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
CMPccXADD is a new set of instructions in the latest Intel platform
Sierra Forest. This new instruction set includes a semaphore operation
that can compare and add the operands if condition is met, which can
improve database performance.
The bit definition:
CPUID.(EAX=7,ECX=1):EAX[bit 7]
Add CPUID definition for CMPCCXADD.
Signed-off-by: Jiaxi Chen <jiaxi.chen@linux.intel.com>
Signed-off-by: Tao Su <tao1.su@linux.intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Message-Id: <20230303065913.1246327-2-tao1.su@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Update the setting of CPUID 0x8000001F EBX to clearly document the ranges
associated with fields being set.
Fixes: 6cb8f2a663 ("cpu/i386: populate CPUID 0x8000_001F when SEV is active")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <5822fd7d02b575121380e1f493a8f6d9eba2b11a.1664550870.git.thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
We want to get rid of the "#ifdef TARGET_X86_64" compile-time switch
in the long run, so we can drop the separate compilation of the
"qemu-system-i386" binary one day - but we then still need a way to
run a guest with max. CPU settings in 32-bit mode. So the "max" CPU
should determine its family/model/stepping settings according to the
"large mode" (LM) CPU feature bit during runtime, so that it is
possible to run "qemu-system-x86_64 -cpu max,lm=off" and still get
a sane family/model/stepping setting for the guest CPU.
To be able to check the LM bit, we have to move the code that sets
up these properties to a "realize" function, since the LM setting is
not available yet when the "instance_init" function is being called.
Message-Id: <20230306154311.476458-1-thuth@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
The previous patch wrongly replaced FEAT_XSAVE_XCR0_{LO|HI} with
FEAT_XSAVE_XSS_{LO|HI} in CPUID(EAX=12,ECX=1):{ECX,EDX}. As a result,
SGX enclaves only supported SSE and x87 feature (xfrm=0x3).
Fixes: 301e90675c ("target/i386: Enable support for XSAVES based features")
Signed-off-by: Yang Zhong <yang.zhong@linux.intel.com>
Reviewed-by: Yang Weijiang <weijiang.yang@intel.com>
Reviewed-by: Kai Huang <kai.huang@intel.com>
Message-Id: <20230406064041.420039-1-yang.zhong@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This had been pulled in via qemu/plugin.h from hw/core/cpu.h,
but that will be removed.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230310195252.210956-5-richard.henderson@linaro.org>
[AJB: add various additional cases shown by CI]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230315174331.2959-15-alex.bennee@linaro.org>
Reviewed-by: Emilio Cota <cota@braap.org>
Since commit a0e61807a3 ("qapi: Remove QMP events and commands from
user-mode builds") we don't generate the "qapi-commands-machine.h"
header in a user-emulation-only build.
Guard qmp_query_cpu_definitions() within CONFIG_USER_ONLY; move
x86_cpu_class_check_missing_features() closer since it is only used
by this QMP command handler.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230223155540.30370-3-philmd@linaro.org>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230227135202.9710-3-anjo@rev.ng>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Introduce support for emulating CPUID for Xen HVM guests. It doesn't make
sense to advertise the KVM leaves to a Xen guest, so do Xen unconditionally
when the xen-version machine property is set.
Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Obtain xen_version from KVM property, make it automatic]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>
The new CPU model mostly inherits features from Icelake-Server, while
adding new features:
- AMX (Advance Matrix eXtensions)
- Bus Lock Debug Exception
and new instructions:
- AVX VNNI (Vector Neural Network Instruction):
- VPDPBUS: Multiply and Add Unsigned and Signed Bytes
- VPDPBUSDS: Multiply and Add Unsigned and Signed Bytes with Saturation
- VPDPWSSD: Multiply and Add Signed Word Integers
- VPDPWSSDS: Multiply and Add Signed Integers with Saturation
- FP16: Replicates existing AVX512 computational SP (FP32) instructions
using FP16 instead of FP32 for ~2X performance gain
- SERIALIZE: Provide software with a simple way to force the processor to
complete all modifications, faster, allowed in all privilege levels and
not causing an unconditional VM exit
- TSX Suspend Load Address Tracking: Allows programmers to choose which
memory accesses do not need to be tracked in the TSX read set
- AVX512_BF16: Vector Neural Network Instructions supporting BFLOAT16
inputs and conversion instructions from IEEE single precision
- fast zero-length MOVSB (KVM doesn't support yet)
- fast short STOSB (KVM doesn't support yet)
- fast short CMPSB, SCASB (KVM doesn't support yet)
Features that may be added in future versions:
- CET (virtualization support hasn't been merged)
Signed-off-by: Wang, Lei <lei4.wang@intel.com>
Reviewed-by: Robert Hoo <robert.hu@linux.intel.com>
Message-Id: <20220812055751.14553-1-lei4.wang@intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
These are three more markers for string operation optimizations.
They can all be added to TCG, whose string operations are more or
less as fast as they can be for short lengths.
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fast short REP MOVS can be added to TCG, since a trivial translation
of string operation is a good option for short lengths.
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The new SGX Asynchronous Exit (AEX) notification mechanism (AEX-notify)
allows one enclave to receive a notification in the ERESUME after the
enclave exit due to an AEX. EDECCSSA is a new SGX user leaf function
(ENCLU[EDECCSSA]) to facilitate the AEX notification handling.
Whether the hardware supports to create enclave with AEX-notify support
is enumerated via CPUID.(EAX=0x12,ECX=0x1):EAX[10]. The new EDECCSSA
user leaf function is enumerated via CPUID.(EAX=0x12,ECX=0x0):EAX[11].
Add support to allow to expose the new SGX AEX-notify feature and the
new EDECCSSA user leaf function to KVM guest.
Link: https://lore.kernel.org/lkml/166760360549.4906.809756297092548496.tip-bot2@tip-bot2/
Link: https://lore.kernel.org/lkml/166760360934.4906.2427175408052308969.tip-bot2@tip-bot2/
Reviewed-by: Yang Zhong <yang.zhong@linux.intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
Message-Id: <20221109024834.172705-1-kai.huang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Convert the i386 CPU class to use 3-phase reset, so it doesn't
need to use device_class_set_parent_reset() any more.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Edgar E. Iglesias <edgar@zeroasic.com>
Reviewed-by: Taylor Simpson <tsimpson@quicinc.com>
Message-id: 20221124115023.2437291-7-peter.maydell@linaro.org
The has_FOO for pointer-valued FOO are redundant, except for arrays.
They are also a nuisance to work with. Recent commit "qapi: Start to
elide redundant has_FOO in generated C" provided the means to elide
them step by step. This is the step for qapi/machine*.json.
Said commit explains the transformation in more detail. The invariant
violations mentioned there do not occur here.
Cc: Eduardo Habkost <eduardo@habkost.net>
Cc: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
Cc: Yanan Wang <wangyanan55@huawei.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20221104160712.3005652-16-armbru@redhat.com>
The only issue with FMA instructions is that there are _a lot_ of them (30
opcodes, each of which comes in up to 4 versions depending on VEX.W and
VEX.L; a total of 96 possibilities). However, they can be implement with
only 6 helpers, two for scalar operations and four for packed operations.
(Scalar versions do not do any merging; they only affect the bottom 32
or 64 bits of the output operand. Therefore, there is no separate XMM
and YMM of the scalar helpers).
First, we can reduce the number of helpers to one third by passing four
operands (one output and three inputs); the reordering of which operands
go to the multiply and which go to the add is done in emit.c.
Second, the different instructions also dispatch to the same softfloat
function, so the flags for float32_muladd and float64_muladd are passed
in the helper as int arguments, with a little extra complication to
handle FMADDSUB and FMSUBADD.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
F16C only consists of two instructions, which are a bit peculiar
nevertheless.
First, they access only the low half of an YMM or XMM register for the
packed-half operand; the exact size still depends on the VEX.L flag.
This is similar to the existing avx_movx flag, but not exactly because
avx_movx is hardcoded to affect operand 2. To this end I added a "ph"
format name; it's possible to reuse this approach for the VPMOVSX and
VPMOVZX instructions, though that would also require adding two more
formats for the low-quarter and low-eighth of an operand.
Second, VCVTPS2PH is somewhat weird because it *stores* the result of
the instruction into memory rather than loading it.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Include AVX, AVX2 and VAES in the guest cpuid features supported by TCG.
Signed-off-by: Paul Brook <paul@nowt.org>
Message-Id: <20220424220204.2493824-40-paul@nowt.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The semantic difference between the deprecated device_legacy_reset()
function and the newer device_cold_reset() function is that the new
function resets both the device itself and any qbuses it owns,
whereas the legacy function resets just the device itself and nothing
else.
The x86_cpu_after_reset() function uses device_legacy_reset() to reset
the APIC; this is an APICCommonState and does not have any qbuses, so
for this purpose the two functions behave identically and we can stop
using the deprecated one.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20221013171926.1447899-1-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Resetting a guest that has Hyper-V VMBus support enabled triggers a QEMU
assertion failure:
hw/hyperv/hyperv.c:131: synic_reset: Assertion `QLIST_EMPTY(&synic->sint_routes)' failed.
This happens both on normal guest reboot or when using "system_reset" HMP
command.
The failing assertion was introduced by commit 64ddecc88b ("hyperv: SControl is optional to enable SynIc")
to catch dangling SINT routes on SynIC reset.
The root cause of this problem is that the SynIC itself is reset before
devices using SINT routes have chance to clean up these routes.
Since there seems to be no existing mechanism to force reset callbacks (or
methods) to be executed in specific order let's use a similar method that
is already used to reset another interrupt controller (APIC) after devices
have been reset - by invoking the SynIC reset from the machine reset
handler via a new x86_cpu_after_reset() function co-located with
the existing x86_cpu_reset() in target/i386/cpu.c.
Opportunistically move the APIC reset handler there, too.
Fixes: 64ddecc88b ("hyperv: SControl is optional to enable SynIc") # exposed the bug
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <cb57cee2e29b20d06f81dce054cbcea8b5d497e8.1664552976.git.maciej.szmigiero@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add support for saving/restoring extended save states when signals
are delivered. This allows using AVX, MPX or PKRU registers in
signal handlers.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
For the direct triple faults, i.e. hardware detected and KVM morphed
to VM-Exit, KVM will never lose them. But for triple faults sythesized
by KVM, e.g. the RSM path, if KVM exits to userspace before the request
is serviced, userspace could migrate the VM and lose the triple fault.
A new flag KVM_VCPUEVENT_VALID_TRIPLE_FAULT is defined to signal that
the event.triple_fault_pending field contains a valid state if the
KVM_CAP_X86_TRIPLE_FAULT_EVENT capability is enabled.
Acked-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Chenyi Qiang <chenyi.qiang@intel.com>
Message-Id: <20220929072014.20705-2-chenyi.qiang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When the user queries CPU models via QMP there is a 'deprecated' flag
present, however, this is not done for the CLI '-cpu help' command.
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20220707163720.1421716-5-berrange@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
When QEMU is started with '-cpu host,host-cache-info=on', it will
passthrough host's number of logical processors sharing cache and
number of processor cores in the physical package. QEMU already
fixes up the later to correctly reflect number of configured cores
for VM, however number of logical processors sharing cache is still
comes from host CPU, which confuses guest started with:
-machine q35,accel=kvm \
-cpu host,host-cache-info=on,l3-cache=off \
-smp 20,sockets=2,dies=1,cores=10,threads=1 \
-numa node,nodeid=0,memdev=ram-node0 \
-numa node,nodeid=1,memdev=ram-node1 \
-numa cpu,socket-id=0,node-id=0 \
-numa cpu,socket-id=1,node-id=1
on 2 socket Xeon 4210R host with 10 cores per socket
with CPUID[04H]:
...
--- cache 3 ---
cache type = unified cache (3)
cache level = 0x3 (3)
self-initializing cache level = true
fully associative cache = false
maximum IDs for CPUs sharing cache = 0x1f (31)
maximum IDs for cores in pkg = 0xf (15)
...
that doesn't match number of logical processors VM was
configured with and as result RHEL 9.0 guest complains:
sched: CPU #10's llc-sibling CPU #0 is not on the same node! [node: 1 != 0]. Ignoring dependency.
WARNING: CPU: 10 PID: 0 at arch/x86/kernel/smpboot.c:421 topology_sane.isra.0+0x67/0x80
...
Call Trace:
set_cpu_sibling_map+0x176/0x590
start_secondary+0x5b/0x150
secondary_startup_64_no_verify+0xc2/0xcb
Fix it by capping max number of logical processors to vcpus/socket
as it was configured, which fixes the issue.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=2088311
Message-Id: <20220524151020.2541698-3-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Accourding Intel's CPUID[EAX=04H] resulting bits 31 - 26 in EAX
should be:
"
**** The nearest power-of-2 integer that is not smaller than (1 + EAX[31:26]) is the number of unique
Core_IDs reserved for addressing different processor cores in a physical package. Core ID is a subset of
bits of the initial APIC ID.
"
ensure that values stored in EAX[31-26] always meets this condition.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20220524151020.2541698-2-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The previous patch used wrong count setting with index value, which got wrong
value from CPUID(EAX=12,ECX=0):EAX. So the SGX1 instruction can't be exposed
to VM and the SGX decice can't work in VM.
Fixes: d19d6ffa07 ("target/i386: introduce helper to access supported CPUID")
Signed-off-by: Yang Zhong <yang.zhong@intel.com>
Message-Id: <20220530131834.1222801-1-yang.zhong@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hyper-V TLFS allows for L0 and L1 hypervisors to collaborate on L2's
TLB flush hypercalls handling. With the correct setup, L2's TLB flush
hypercalls can be handled by L0 directly, without the need to exit to
L1.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20220525115949.1294004-6-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM kind of supported "extended GVA ranges" (up to 4095 additional GFNs
per hypercall) since the implementation of Hyper-V PV TLB flush feature
(Linux-4.18) as regardless of the request, full TLB flush was always
performed. "Extended GVA ranges for TLB flush hypercalls" feature bit
wasn't exposed then. Now, as KVM gains support for fine-grained TLB
flush handling, exposing this feature starts making sense.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20220525115949.1294004-5-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hyper-V specification allows to pass parameters for certain hypercalls
using XMM registers ("XMM Fast Hypercall Input"). When the feature is
in use, it allows for faster hypercalls processing as KVM can avoid
reading guest's memory.
KVM supports the feature since v5.14.
Rename HV_HYPERCALL_{PARAMS_XMM_AVAILABLE -> XMM_INPUT_AVAILABLE} to
comply with KVM.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20220525115949.1294004-4-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The newly introduced enlightenment allow L0 (KVM) and L1 (Hyper-V)
hypervisors to collaborate to avoid unnecessary updates to L2
MSR-Bitmap upon vmexits.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20220525115949.1294004-3-vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Since KVM commit 5f76f6f5ff96 ("KVM: nVMX: Do not expose MPX VMX controls when guest MPX disabled")
it is not possible to disable MPX on a "-cpu host" just by adding "-mpx"
there if the host CPU does indeed support MPX.
QEMU will fail to set MSR_IA32_VMX_TRUE_{EXIT,ENTRY}_CTLS MSRs in this case
and so trigger an assertion failure.
Instead, besides "-mpx" one has to explicitly add also
"-vmx-exit-clear-bndcfgs" and "-vmx-entry-load-bndcfgs" to QEMU command
line to make it work, which is a bit convoluted.
Make the MPX-related bits in FEAT_VMX_{EXIT,ENTRY}_CTLS dependent on MPX
being actually enabled so such workarounds are no longer necessary.
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <51aa2125c76363204cc23c27165e778097c33f0b.1653323077.git.maciej.szmigiero@oracle.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If CPUID.(EAX=07H, ECX=0):EDX[19] is set to 1, the processor
supports Architectural LBRs. In this case, CPUID leaf 01CH
indicates details of the Architectural LBRs capabilities.
XSAVE support for Architectural LBRs is enumerated in
CPUID.(EAX=0DH, ECX=0FH).
Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
Message-Id: <20220215195258.29149-9-weijiang.yang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
