Before changing default pauth algorithm, we need to make sure current
default one (QARMA5) can still be selected.
$ qemu-system-aarch64 -cpu max,pauth-qarma5=on ...
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20241219183211.3493974-2-pierrick.bouvier@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
helper.c includes some small TCG helper functions used for mostly
arithmetic instructions. These are TCG only and there's no need for
them to be in the large and unwieldy helper.c. Move them out to
their own source file in the tcg/ subdirectory, together with the
op_addsub.h multiply-included template header that they use.
Since we are moving op_addsub.h, we take the opportunity to
give it a name which matches our convention for files which
are not true header files but which are #included from other
C files: op_addsub.c.inc.
(Ironically, this means that helper.c no longer contains
any TCG helper function definitions at all.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20250110131211.2546314-1-peter.maydell@linaro.org
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
The correct usage is tracking and maintaining features in env->features[]
instead of manually set it in cpu_x86_cpuid().
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Link: https://lore.kernel.org/r/20241219110125.1266461-11-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Currently CPUID_HT is evaluated in cpu_x86_cpuid() each time. It's not a
correct usage of how feature bit is maintained and evaluated. The
expected practice is that features are tracked in env->features[] and
cpu_x86_cpuid() should be the consumer of env->features[].
Track CPUID_HT in env->features[FEAT_1_EDX] instead and evaluate it in
cpu's realizefn().
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Link: https://lore.kernel.org/r/20241219110125.1266461-10-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Now it changes to use env->topo_info.threads_per_core and doesn't depend
on qemu_init_vcpu() anymore. Put it together with other feature checks
before qemu_init_vcpu()
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Link: https://lore.kernel.org/r/20241219110125.1266461-8-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The name of nr_modules/nr_dies are ambiguous and they mislead people.
The purpose of them is to record and form the topology information. So
just maintain a X86CPUTopoInfo member in CPUX86State instead. Then
nr_modules and nr_dies can be dropped.
As the benefit, x86 can switch to use information in
CPUX86State::topo_info and get rid of the nr_cores and nr_threads in
CPUState. This helps remove the dependency on qemu_init_vcpu(), so that
x86 can get and use topology info earlier in x86_cpu_realizefn(); drop
the comment that highlighted the depedency.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Link: https://lore.kernel.org/r/20241219110125.1266461-7-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Introduce various helpers for getting the topology info of different
semantics. Using the helper is more self-explanatory.
Besides, the semantic of the helper will stay unchanged even when new
topology is added in the future. At that time, updating the
implementation of the helper without affecting the callers.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Link: https://lore.kernel.org/r/20241219110125.1266461-6-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Local variable cores_per_pkg is only used to calculate threads_per_pkg.
No need for it. Drop it and open-code it instead.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Link: https://lore.kernel.org/r/20241219110125.1266461-4-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
There are duplicated code to setup the value of MSR_CORE_THREAD_COUNT.
Extract a common function for it.
Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Link: https://lore.kernel.org/r/20241219110125.1266461-2-xiaoyao.li@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
kvm_install_msr_filters() uses KVM_MSR_FILTER_MAX_RANGES as the bound
when traversing msr_handlers[], while other places still compute the
size by ARRAY_SIZE(msr_handlers).
In fact, msr_handlers[] is an array with the fixed size
KVM_MSR_FILTER_MAX_RANGES, and this has to be true because
kvm_install_msr_filters copies from one array to the other.
For code consistency, assert that they match and use
ARRAY_SIZE(msr_handlers) everywehere.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Currently, there're following incorrect error handling cases in
kvm_arch_init():
* Missed to handle failure of kvm_get_supported_feature_msrs().
* Missed to return when kvm_vm_enable_disable_exits() fails.
* MSR filter related cases called exit() directly instead of returning
to kvm_init(). (The caller of kvm_arch_init() - kvm_init() - needs to
know if kvm_arch_init() fails in order to perform cleanup).
Fix the above cases.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Zide Chen <zide.chen@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-11-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
It is common practice to return a negative value (like -1) to indicate
an error, and other functions in kvm_arch_init() follow this style.
To avoid confusion (sometimes returned -1 indicates failure, and
sometimes -1, in a same function), return -1 when
kvm_msr_energy_thread_init() fails.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-10-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Before commit 0cc42e63bb54 ("kvm/i386: refactor kvm_arch_init and split
it into smaller functions"), error_report() attempts to print the error
code from kvm_filter_msr(). However, printing error code does not work
due to kvm_filter_msr() returns bool instead int.
0cc42e63bb54 fixed the error by removing error code printing, but this
lost useful error messages. Bring it back by making kvm_filter_msr()
return int.
This also makes the function call chain processing clearer, allowing for
better handling of error result propagation from kvm_filter_msr() to
kvm_arch_init(), preparing for the subsequent cleanup work of error
handling in kvm_arch_init().
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Zide Chen <zide.chen@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-9-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Update the comment to match the X86ConfidentialGuestClass
implementation.
Reported-by: Xiaoyao Li <xiaoyao.li@intel.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Pankaj Gupta <pankaj.gupta@amd.com>
Reviewed-by: Zide Chen <zide.chen@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-8-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The KVM_X86_DISABLE_EXITS_HTL typo has been fixed in commit
77d361b13c19 ("linux-headers: Update to kernel mainline commit
b357bf602").
Drop the related workaround.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Zide Chen <zide.chen@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-7-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
MSR_KVM_SYSTEM_TIME and MSR_KVM_WALL_CLOCK are attached with the (old)
kvmclock feature (KVM_FEATURE_CLOCKSOURCE).
So, just save/load them only when kvmclock (KVM_FEATURE_CLOCKSOURCE) is
enabled.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Zide Chen <zide.chen@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-5-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
These 2 MSRs have been already defined in kvm_para.h (standard-headers/
asm-x86/kvm_para.h).
Remove QEMU local definitions to avoid duplication.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
Reviewed-by: Zide Chen <zide.chen@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-4-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add feature definitions for KVM_CPUID_FEATURES in CPUID (
CPUID[4000_0001].EAX and CPUID[4000_0001].EDX), to get rid of lots of
offset calculations.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Zide Chen <zide.chen@intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-3-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In x86_cpu_filter_features(), if host doesn't support AVX10, the
configured avx10_version should be marked as filtered regardless of
whether prefix is NULL or not.
Check prefix before warn_report() instead of checking for
have_filtered_features.
Cc: qemu-stable@nongnu.org
Fixes: commit bccfb846fd52 ("target/i386: add AVX10 feature and AVX10 version property")
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Tao Su <tao1.su@linux.intel.com>
Link: https://lore.kernel.org/r/20241106030728.553238-2-zhao1.liu@intel.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Using a sextract or extract operation is only necessary if a
sign or zero extended value is needed. If not, a shift is
enough.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Because BT does not write back to the source operand, it can modify it to
ensure that one of the operands of TSTNE is a constant (after either gen_BT
or the optimizer's constant propagation). This produces better and more
optimizable TCG ops. For example, the sequence
movl $0x60013f, %ebx
btl %ecx, %ebx
becomes just
and_i32 tmp1,ecx,$0x1f dead: 1 2 pref=0xffff
shr_i32 tmp0,$0x60013f,tmp1 dead: 1 2 pref=0xffff
and_i32 tmp16,tmp0,$0x1 dead: 1 pref=0xbf80
On s390x, it can use four instructions to isolate bit 0 of 0x60013f >> (ecx & 31):
nilf %r12, 0x1f
lgfi %r11, 0x60013f
srlk %r12, %r11, 0(%r12)
nilf %r12, 1
Previously, it used five instructions to build 1 << (ecx & 31) and compute
TSTEQ, and also needed two more to construct the result of setcond:
nilf %r12, 0x1f
lghi %r11, 1
sllk %r12, %r11, 0(%r12)
lgfi %r9, 0x60013f
nrk %r0, %r12, %r9
lghi %r12, 0
locghilh %r12, 1
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
iFrom LoongArch Reference Manual pte width can be 64bit, 128bit
or more. Instead real hardware only supports 64bit pte width.
For 12bit pte, there is no detail definition for all 128bit
from manual.
Here only 64bit pte width is supported for simplicity, will add
this in later if real hw support it and there is definition for
all the bits from manual.
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Now that the last machine type that disabled cpu_model_allowed has
been removed, we can also remove the cpu_model_allowed flag itself
and all the related conditional code.
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Message-ID: <20250103144232.520383-9-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Only s390-ccw-virtio-2.6 and older used to set this switch to "off",
for newer machine types it is always enabled. Since we removed the
old machine types now, we can also remove the switch in the code and
assume that it is always enabled now.
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Message-ID: <20250103144232.520383-5-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Fixes: 644e3c5d812 ("missing vmx features for Skylake-Server and Cascadelake-Server")
Signed-off-by: Han Han <hhan@redhat.com>
Reviewed-by: Chenyi Qiang <chenyi.qiang@intel.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
(cherry picked from commit 93dcc9390e5ad0696ae7e9b7b3a5b08c2d1b6de6)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Hardcoded 32 bytes is used for vbsrl emulation check, there is
problem when options lsx=on,lasx=off is used for vbsrl.v instruction
in TCG mode. It injects LASX exception rather LSX exception.
Here actual operand size is used.
Cc: qemu-stable@nongnu.org
Fixes: df97f338076 ("target/loongarch: Implement xvreplve xvinsve0 xvpickve")
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
(cherry picked from commit d41989e7548397b469ec9c7be4cee699321a317e)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Fixes: 644e3c5d812 ("missing vmx features for Skylake-Server and Cascadelake-Server")
Signed-off-by: Han Han <hhan@redhat.com>
Reviewed-by: Chenyi Qiang <chenyi.qiang@intel.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Like LSX feature, add type OnOffAuto for LASX feature setting.
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Like LBT feature, add type OnOffAuto for LSX feature setting. Also
add LSX feature detection with new VM ioctl command, fallback to old
method if it is not supported.
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Hardcoded 32 bytes is used for vbsrl emulation check, there is
problem when options lsx=on,lasx=off is used for vbsrl.v instruction
in TCG mode. It injects LASX exception rather LSX exception.
Here actual operand size is used.
Cc: qemu-stable@nongnu.org
Fixes: df97f338076 ("target/loongarch: Implement xvreplve xvinsve0 xvpickve")
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Refer to the link below for a description of the vldi instructions:
https://jia.je/unofficial-loongarch-intrinsics-guide/lsx/misc/#synopsis_88
Fixed errors in vldi instruction implementation.
Signed-off-by: Guo Hongyu <guohongyu24@mails.ucas.ac.cn>
Tested-by: Xianglai Li <lixianglai@loongson.cn>
Signed-off-by: Xianglai Li <lixianglai@loongson.cn>
Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Convert all targets simultaneously, as the gen_intermediate_code
function disappears from the target. While there are possible
workarounds, they're larger than simply performing the conversion.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Initialize x with accumulated via direct assignment,
rather than multiplying by 1.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
No need to open-code 64x64->128-bit multiplication.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This structure, with bitfields, is incorrect for big-endian.
Use extract64 and deposit64 instead.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This structure, with bitfields, is incorrect for big-endian.
Use the existing float32_getexp_raw which uses extract32.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This massive macro is now only used once.
Expand it for use only by float64.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
There are multiple special cases for this instruction.
(1) The saturate to normal maximum instead of overflow to infinity is
handled by the new float_round_nearest_even_max rounding mode.
(2) The 0 * n + c special case is handled by the new
float_muladd_suppress_add_product_zero flag.
(3) The Inf - Inf -> 0 special case can be detected after the fact
by examining float_flag_invalid_isi.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This instruction has a special case that 0 * x + c returns c
without the normal sign folding that comes with 0 + -0.
Use the new float_muladd_suppress_add_product_zero to
describe this.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
There are no special cases for this instruction. Since hexagon
always uses default-nan mode, explicitly negating the first
input is unnecessary. Use float_muladd_negate_product instead.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
There are no special cases for this instruction.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
There are no special cases for this instruction.
Remove internal_mpyf as unused.
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Use the scalbn interface instead of float_muladd_halve_result.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Use the scalbn interface instead of float_muladd_halve_result.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
"Host Memory Backends" and "Memory devices" queue ("mem"):
- Fixup handling of virtio-mem unplug during system resets, as
preparation for s390x support (especially kdump in the Linux guest)
- virtio-mem support for s390x
-----BEGIN PGP SIGNATURE-----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=HDxr
-----END PGP SIGNATURE-----
Merge tag 'mem-2024-12-21' of https://github.com/davidhildenbrand/qemu into staging
Hi,
"Host Memory Backends" and "Memory devices" queue ("mem"):
- Fixup handling of virtio-mem unplug during system resets, as
preparation for s390x support (especially kdump in the Linux guest)
- virtio-mem support for s390x
# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEG9nKrXNcTDpGDfzKTd4Q9wD/g1oFAmdnFD4RHGRhdmlkQHJl
# ZGhhdC5jb20ACgkQTd4Q9wD/g1rWBBAAp7WkYaNAjRy1PgpjNZ3z1gUJc/vk+skJ
# xVgGodA8txrJOFpNrbTyfhrdLs2TV4oWDvB/zrZRRtuxvur3O1EhFd9k6EqXuydr
# 0FunvLvVJwRHfEZycjN4aacQMRH3CJw07OaTzexeSl5UR/6w5PRofwUK4HX7W/Ka
# arqomGa3OJrs1+WgkV0Qcn4vh9HLRVv3iNC2Xo4W1wOCr1Du9zSPn9oC7zOQ0EO4
# ZC//7QsdkNRjUX/yMXMkhlSXx3b/RmRg2DBrxo7BZXg27VwGu4uHxL4LRBZiB2A7
# V9MqFOcVKzPMkXKTRjrgZ0vXQx1MPJ6WprEihMzMpYU6DrpA7KN/l8Ca8H24B2ln
# h7+bmkDsHVVcWovE9ii/9cMRfws6uWXXg3KoA8RQ8IbX1tU02lblw2uHhXEzcoge
# npqp/Z5LAiKVMetEnNnLH5thjut5PAEjuqD00cmZAMy4DNngLX2bGSdzMeVBkDMa
# 78ehLGRplm3t7ibUfaZaMKe6UD9tFrcD6XKsvUTXXHNbYO8ynbx58WOxSZmY98zU
# n3JNQRqtXYjBVlH3Dqm47vOTZHgOzFv3raa8BmSLpcBDeTXCTcUIl20s77dGw/vT
# r5YNCMN7O4YPFKUoRK9604QTgw6qlYaRTQlJD09usprGqVylb6gQtfZZuZkYDMp8
# sEI77QHsePA=
# =HDxr
# -----END PGP SIGNATURE-----
# gpg: Signature made Sat 21 Dec 2024 14:17:18 EST
# gpg: using RSA key 1BD9CAAD735C4C3A460DFCCA4DDE10F700FF835A
# gpg: issuer "david@redhat.com"
# gpg: Good signature from "David Hildenbrand <david@redhat.com>" [unknown]
# gpg: aka "David Hildenbrand <davidhildenbrand@gmail.com>" [full]
# gpg: aka "David Hildenbrand <hildenbr@in.tum.de>" [unknown]
# gpg: WARNING: The key's User ID is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 1BD9 CAAD 735C 4C3A 460D FCCA 4DDE 10F7 00FF 835A
* tag 'mem-2024-12-21' of https://github.com/davidhildenbrand/qemu:
s390x: virtio-mem support
s390x/virtio-ccw: add support for virtio based memory devices
s390x: remember the maximum page size
s390x/pv: prepare for memory devices
s390x/s390-virtio-ccw: prepare for memory devices
s390x/s390-skeys: prepare for memory devices
s390x/s390-stattrib-kvm: prepare for memory devices and sparse memory layouts
s390x/s390-hypercall: introduce DIAG500 STORAGE_LIMIT
s390x: introduce s390_get_memory_limit()
s390x/s390-virtio-ccw: move setting the maximum guest size from sclp to machine code
s390x: rename s390-virtio-hcall* to s390-hypercall*
s390x/s390-virtio-hcall: prepare for more diag500 hypercalls
s390x/s390-virtio-hcall: remove hypercall registration mechanism
s390x/s390-virtio-ccw: don't crash on weird RAM sizes
virtio-mem: unplug memory only during system resets, not device resets
Conflicts:
- hw/s390x/s390-stattrib-kvm.c
sysemu/ -> system/ header rename conflict.
- hw/s390x/virtio-ccw-mem.c
Make Property array const and removed DEFINE_PROP_END_OF_LIST() to
conform to the latest conventions.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Since commit 5286c3662294 ("target/i386: properly reset TSC on reset")
QEMU writes the special value of "1" to each online vCPU TSC on VM reset
to reset it.
However parked vCPUs don't get that handling and due to that their TSCs
get desynchronized when the VM gets reset.
This in turn causes KVM to turn off PVCLOCK_TSC_STABLE_BIT in its exported
PV clock.
Note that KVM has no understanding of vCPU being currently parked.
Without PVCLOCK_TSC_STABLE_BIT the sched clock is marked unstable in
the guest's kvm_sched_clock_init().
This causes a performance regressions to show in some tests.
Fix this issue by writing the special value of "1" also to TSCs of parked
vCPUs on VM reset.
Reproducing the issue:
1) Boot a VM with "-smp 2,maxcpus=3" or similar
2) device_add host-x86_64-cpu,id=vcpu,node-id=0,socket-id=0,core-id=2,thread-id=0
3) Wait a few seconds
4) device_del vcpu
5) Inside the VM run:
# echo "t" >/proc/sysrq-trigger; dmesg | grep sched_clock_stable
Observe the sched_clock_stable() value is 1.
6) Reboot the VM
7) Once the VM boots once again run inside it:
# echo "t" >/proc/sysrq-trigger; dmesg | grep sched_clock_stable
Observe the sched_clock_stable() value is now 0.
Fixes: 5286c3662294 ("target/i386: properly reset TSC on reset")
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Link: https://lore.kernel.org/r/5a605a88e9a231386dc803c60f5fed9b48108139.1734014926.git.maciej.szmigiero@oracle.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 3f2a05b31ee9ce2ddb6c75a9bc3f5e7f7af9a76f)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
