Merge tag 'kvm-x86-2025.02.21' of https://github.com/kvm-x86/kvm-unit-tests into HEAD

KVM-Unit-Tests x86 changes:

- Expand the per-CPU data+stack area to 12KiB per CPU to reduce the
probability

Merge tag 'kvm-x86-2025.02.21' of https://github.com/kvm-x86/kvm-unit-tests into HEAD

KVM-Unit-Tests x86 changes:

- Expand the per-CPU data+stack area to 12KiB per CPU to reduce the
probability of tests overflowing their stack and clobbering pre-CPU
data.

- Add testcases for LA57 canonical checks.

- Add testcases for LAM.

- Add a smoke test to make sure KVM doesn't bleed split-lock #AC/#DB into
the guest.

- Fix many warts and bugs in the PMU test, and prepare it for PMU version 5
and beyond.

- Many misc fixes and cleanups.

show more ...

x86: Move SMP #defines from apic-defs.h to smp.h

Now that the __ASSEMBLY__ versus __ASSEMBLER_ mess is sorted out, move
the SMP related #defines from apic-defs.h to smp.h, and drop the comment
that

x86: Move SMP #defines from apic-defs.h to smp.h

Now that the __ASSEMBLY__ versus __ASSEMBLER_ mess is sorted out, move
the SMP related #defines from apic-defs.h to smp.h, and drop the comment
that explains the hackery.

Opportunistically make REALMODE_GDT_LOWMEM visible to assembly code as
well, and drop efistart64.S's local copy.

Link: https://lore.kernel.org/r/20250221233832.2251456-1-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>

show more ...

x86: replace segment selector magic number with macro definition

Add assembly check in desc.h, to replace segment selector
magic number with macro definition.

Signed-off-by: Hang SU <darcy.sh@antgr

x86: replace segment selector magic number with macro definition

Add assembly check in desc.h, to replace segment selector
magic number with macro definition.

Signed-off-by: Hang SU <darcy.sh@antgroup.com>
Link: https://lore.kernel.org/r/20250221225406.2228938-4-seanjc@google.com
[sean: fix KERNEL_CS vs. KERNEL_CS32 goof]
Signed-off-by: Sean Christopherson <seanjc@google.com>

show more ...

x86: Add a macro for the size of the per-CPU stack/data area

Add a macro to define the size of the per-CPU stack/data area so that it's
somewhat possible to make sense of the madness.

Link: https:/

x86: Add a macro for the size of the per-CPU stack/data area

Add a macro to define the size of the per-CPU stack/data area so that it's
somewhat possible to make sense of the madness.

Link: https://lore.kernel.org/r/20250215012032.1206409-3-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>

show more ...

x86: Make per-CPU stacks page-aligned

Align the per-CPU stacks, and by extension, the per-CPU data area to page
boundaries so that when things go sideways, it's at least somewhat obvious
that a test

x86: Make per-CPU stacks page-aligned

Align the per-CPU stacks, and by extension, the per-CPU data area to page
boundaries so that when things go sideways, it's at least somewhat obvious
that a test overran its stack. E.g. as is, stacks often start at the
*bottom* of a page, and so it looks like they're always broken because
they immediately split a page.

Link: https://lore.kernel.org/r/20250215012032.1206409-2-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>

show more ...

x86: Add a helper for the BSP's final init sequence common to all flavors

Add bsp_rest_init() to dedup bringing up APs and doing SMP initialization
across 32-bit, 64-bit, and EFI flavors of KVM-unit

x86: Add a helper for the BSP's final init sequence common to all flavors

Add bsp_rest_init() to dedup bringing up APs and doing SMP initialization
across 32-bit, 64-bit, and EFI flavors of KVM-unit-tests. The common
bucket will also be used in future to patches to init things that aren't
SMP related and thus don't fit in smp_init(), e.g. PMU setup.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20221102225110.3023543-15-seanjc@google.com

show more ...

Merge tag 'for_paolo' of https://github.com/sean-jc/kvm-unit-tests into HEAD

x86 fixes, cleanups, and new sub-tests:

- Bug fix for the VMX-preemption timer expiration test
- Refactor SVM tests

Merge tag 'for_paolo' of https://github.com/sean-jc/kvm-unit-tests into HEAD

x86 fixes, cleanups, and new sub-tests:

- Bug fix for the VMX-preemption timer expiration test
- Refactor SVM tests to split out NPT tests
- Add tests for MCE banks to MSR test
- Add SMP Support for x86 UEFI tests
- x86: nVMX: Add VMXON #UD test (and exception cleanup)
- PMU cleanup and related nVMX bug fixes

show more ...

x86: Rename ap_init() to bringup_aps()

Rename the helper that wakes and waits for APs to bringup_aps(), ap_init()
is terribly confusing because it's called from the BSP, not APs.

Signed-off-by: Sea

x86: Rename ap_init() to bringup_aps()

Rename the helper that wakes and waits for APs to bringup_aps(), ap_init()
is terribly confusing because it's called from the BSP, not APs.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220615232943.1465490-13-seanjc@google.com

show more ...

x86: Provide a common 64-bit AP entrypoint for EFI and non-EFI

ap_start64() currently serves as the 64-bit entrypoint for non-EFI
tests.

Having ap_start64() and save_id() written in asm prevents sh

x86: Provide a common 64-bit AP entrypoint for EFI and non-EFI

ap_start64() currently serves as the 64-bit entrypoint for non-EFI
tests.

Having ap_start64() and save_id() written in asm prevents sharing these
routines between EFI and non-EFI tests.

Rewrite them in C and use ap_start64 as the 64-bit entrypoint in the EFI
boot flow.

With this, EFI tests support -smp > 1. smptest.efi now passes.

Cc: Andrew Jones <drjones@redhat.com>
Cc: Marc Orr <marcorr@google.com>
Cc: Zixuan Wang <zxwang42@gmail.com>
Cc: Erdem Aktas <erdemaktas@google.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Thomas.Lendacky@amd.com
Cc: Joerg Roedel <jroedel@suse.de>
Cc: Borislav Petkov <bp@alien8.de>
Signed-off-by: Varad Gautam <varad.gautam@suse.com>
[sean: reset APIC before TSS setup, halt in AP wait loop, add comment]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220615232943.1465490-12-seanjc@google.com

show more ...

x86: Move 32-bit => 64-bit transition code to trampolines.S

Move the code for transitioning from unpaged 32-bit mode to paged 64-bit
mode to trampoline.S, it can be shared across EFI and non-EFI bui

x86: Move 32-bit => 64-bit transition code to trampolines.S

Move the code for transitioning from unpaged 32-bit mode to paged 64-bit
mode to trampoline.S, it can be shared across EFI and non-EFI builds.

Leave 5-level paging behind for the time being, EFI doesn't yet support
support disabling paging, which is required to get from 4-level to
5-level paging.

Signed-off-by: Varad Gautam <varad.gautam@suse.com>
[sean: move to trampolines.S instead of start32.S, reword changelog]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220615232943.1465490-10-seanjc@google.com

show more ...

x86: cstart64: Put APIC into xAPIC after loading TSS

Now that pre_boot_apic_id() works with either xAPIC or x2APIC, "reset"
the APIC after configuring loading the TSS. Previously, load_tss() =>
set

x86: cstart64: Put APIC into xAPIC after loading TSS

Now that pre_boot_apic_id() works with either xAPIC or x2APIC, "reset"
the APIC after configuring loading the TSS. Previously, load_tss() =>
setup_tss() needed to run after forcing the vCPU into xAPIC mode due to
pre_boot_apic_id() assuming xAPIC.

The order doesn't truly matter at this point, but loading the TSS first
will allow sharing code with the EFI boot flow, which "needs" to load the
TSS (more specifically, needs to configure GS.base) prior to forcing the
vCPU into xAPIC (and thus setting the per-vCPU APIC ops).

Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220725201336.2158604-3-seanjc@google.com

show more ...

x86: Move load_idt() to desc.c

This allows sharing IDT setup code between EFI (-fPIC) and
non-EFI builds.

Signed-off-by: Varad Gautam <varad.gautam@suse.com>
Signed-off-by: Sean Christopherson <sea

x86: Move load_idt() to desc.c

This allows sharing IDT setup code between EFI (-fPIC) and
non-EFI builds.

Signed-off-by: Varad Gautam <varad.gautam@suse.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220615232943.1465490-5-seanjc@google.com

show more ...

x86: Move ap_init() to smp.c

ap_init() copies the SIPI vector to lowmem, sends INIT/SIPI to APs
and waits on the APs to come up.

Port this routine to C from asm and move it to smp.c to allow sharin

x86: Move ap_init() to smp.c

ap_init() copies the SIPI vector to lowmem, sends INIT/SIPI to APs
and waits on the APs to come up.

Port this routine to C from asm and move it to smp.c to allow sharing
this functionality between the EFI (-fPIC) and non-EFI builds.

Call ap_init() from the EFI setup path to reset the APs to a known
location.

Signed-off-by: Varad Gautam <varad.gautam@suse.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220615232943.1465490-4-seanjc@google.com

show more ...

x86: Share realmode trampoline between i386 and x86_64

i386 and x86_64 each maintain their own copy of the realmode trampoline
(sipi_entry). Move the 16-bit SIPI vector and GDT to a new trampolines.

x86: Share realmode trampoline between i386 and x86_64

i386 and x86_64 each maintain their own copy of the realmode trampoline
(sipi_entry). Move the 16-bit SIPI vector and GDT to a new trampolines.S
to be shared by both. The common trampoline file will also be used to
shared 32-bit to 64-bit trampoline code.

Signed-off-by: Varad Gautam <varad.gautam@suse.com>
[sean: rename to trampolines.S to avoid naming conundrum]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220615232943.1465490-3-seanjc@google.com

show more ...

x86: Align incw instruction to avoid split lock

A split lock is any atomic operation whose operand crosses two cache
lines. Since the operand spans two cache lines and the operation must
be atomic,

x86: Align incw instruction to avoid split lock

A split lock is any atomic operation whose operand crosses two cache
lines. Since the operand spans two cache lines and the operation must
be atomic, the system locks the bus while the CPU accesses the two cache
lines. The bus lock operation is heavy weight and can cause
severe performance degradation.

Here's the log when run x86 test cases:
[ 3572.765921] x86/split lock detection: #AC: qemu-system-x86/24383
took a split_lock trap at address: 0x400306

Root caused 'cpu_online_count' spans two cache lines,
"lock incw cpu_online_count" instruction causes split lock.
'cpu_online_count' is the type of word(two bytes) and
therefore it needs to be aligned to 2 bytes to avoid split lock.

Signed-off-by: Junming Liu <junming.liu@intel.com>
Message-Id: <20220124111444.12548-1-junming.liu@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

x86: Look up the PTEs rather than assuming them

Rather than assuming which PTEs the SMEP test runs on, look them up to
ensure they are correct. If this test were to run on a different page
table (i

x86: Look up the PTEs rather than assuming them

Rather than assuming which PTEs the SMEP test runs on, look them up to
ensure they are correct. If this test were to run on a different page
table (ie: run in an L2 test) the wrong PTEs would be set. Switch to
looking up the PTEs to avoid this from happening.

Signed-off-by: Aaron Lewis <aaronlewis@google.com>
Message-Id: <20211110212001.3745914-11-aaronlewis@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

Merge branch 'gdt-idt-cleanup' into master

x86: Move 32-bit GDT and TSS to desc.c

Move the GDT and TSS data structures from x86/cstart.S to
lib/x86/desc.c, for consistency with the 64-bit version.

Signed-off-by: Paolo Bonzini <pbonzini@redh

x86: Move 32-bit GDT and TSS to desc.c

Move the GDT and TSS data structures from x86/cstart.S to
lib/x86/desc.c, for consistency with the 64-bit version.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

x86: Move 64-bit GDT and TSS to desc.c

Move the GDT and TSS data structures from x86/cstart64.S to
lib/x86/desc.c, so that the follow-up UEFI support commits can reuse
these definitions, without re-

x86: Move 64-bit GDT and TSS to desc.c

Move the GDT and TSS data structures from x86/cstart64.S to
lib/x86/desc.c, so that the follow-up UEFI support commits can reuse
these definitions, without re-defining them in UEFI's boot up assembly
code.

Signed-off-by: Zixuan Wang <zixuanwang@google.com>
Message-Id: <20211004204931.1537823-2-zxwang42@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

x86: get rid of ring0stacktop

The ring3 switch code relied on a special stack page that was used
for the ring0 stack during the ring3 part of the test. This special
stack page was used if an except

x86: get rid of ring0stacktop

The ring3 switch code relied on a special stack page that was used
for the ring0 stack during the ring3 part of the test. This special
stack page was used if an exception handler ran during the ring3 part
of the test.

This method is quite complex; it is easier to just use the same
stack for the "outer" part of the test and the exception handler.
To do so, store esp/rsp in the TSS just before doing the PUSH/IRET
sequence. On 64-bit, the TSS can also be used to restore rsp after
coming back from ring3.

Unifying the three copies of the ring switching code is left as an
exercise to the reader.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

x86: unify name of 32-bit and 64-bit GDT

There's no need to distinguish gdt32 and gdt64, since the same C functions
operate on both and selector numbers are mostly unified between 32-
and 64-bit ver

x86: unify name of 32-bit and 64-bit GDT

There's no need to distinguish gdt32 and gdt64, since the same C functions
operate on both and selector numbers are mostly unified between 32-
and 64-bit versions.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

x86: Move IDT to desc.c

Move the IDT data structures from x86/cstart.S and x86/cstart64.S to
lib/x86/desc.c, so that the follow-up UEFI support commits can reuse
these definitions, without re-defini

x86: Move IDT to desc.c

Move the IDT data structures from x86/cstart.S and x86/cstart64.S to
lib/x86/desc.c, so that the follow-up UEFI support commits can reuse
these definitions, without re-defining them in UEFI's boot up assembly
code.

Extracted by a patch by Zixuan Wang <zxwang42@gmail.com> and ported
to 32-bit too.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

replace tss_descr global with a function

tss_descr is declared as a struct descriptor_table_ptr but it is actualy
pointing to an _entry_ in the GDT. Also it is different per CPU, but
tss_descr does

replace tss_descr global with a function

tss_descr is declared as a struct descriptor_table_ptr but it is actualy
pointing to an _entry_ in the GDT. Also it is different per CPU, but
tss_descr does not recognize that. Fix both by reusing the code
(already present e.g. in the vmware_backdoors test) that extracts
the base from the GDT entry; and also provide a helper to retrieve
the limit, which is needed in vmx.c.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

svm: INIT and STARTUP ipi test

Init the vcpu and issue the STARTUP ipi to indicate the vcpu
should execute its startup routine.

Signed-off-by: Cathy Avery <cavery@redhat.com>
Message-Id: <202007171

svm: INIT and STARTUP ipi test

Init the vcpu and issue the STARTUP ipi to indicate the vcpu
should execute its startup routine.

Signed-off-by: Cathy Avery <cavery@redhat.com>
Message-Id: <20200717113422.19575-3-cavery@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...

x86: Replace instruction prefixes with spaces

There are three kinds of x86 prefix delimiters in GNU binutils:
'/', '\\' and a space.

The first works on Linux and few other platforms. The second on

x86: Replace instruction prefixes with spaces

There are three kinds of x86 prefix delimiters in GNU binutils:
'/', '\\' and a space.

The first works on Linux and few other platforms. The second one is
SVR-4 compatible and works on the generic elf target. The last kind is
universal and works everywhere, it's also used in the GAS manual [1].
Space delimiters fix the build errors on x86_64-elf binutils:

x86/cstart64.S:217: Error: invalid character '/' in mnemonic
x86/cstart64.S:313: Error: invalid character '/' in mnemonic

1. https://sourceware.org/binutils/docs/as/i386_002dPrefixes.html

Cc: Cameron Esfahani <dirty@apple.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200901085056.33391-3-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

show more ...