Split kvm_pgtable_stage2_destroy() into two:
- kvm_pgtable_stage2_destroy_range(), that performs the
page-table walk and free the entries over a range of addresses.
- kvm_pgtable_stage2_destroy_pgd(), that frees the PGD.
This refactoring enables subsequent patches to free large page-tables
in chunks, calling cond_resched() between each chunk, to yield the
CPU as necessary.
Existing callers of kvm_pgtable_stage2_destroy(), that probably cannot
take advantage of this (such as nVMHE), will continue to function as is.
Signed-off-by: Raghavendra Rao Ananta <rananta@google.com>
Suggested-by: Oliver Upton <oliver.upton@linux.dev>
Link: https://lore.kernel.org/r/20250820162242.2624752-2-rananta@google.com
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
vcpu_write_sys_reg()'s signature is not totally obvious, and it
is rather easy to write something that looks correct, except that...
Oh wait...
Swap addr and FAR_EL2 to restore some sanity in the nested SEA
department.
Fixes: 9aba641b9e ("KVM: arm64: nv: Respect exception routing rules for SEAs")
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250813163747.2591317-1-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
In pKVM, a race condition can occur if a guest updates its VBAR_EL1
register and, before a vCPU exit synchronizes this change, the
hypervisor needs to inject an undefined exception into a protected
guest.
In this scenario, the vCPU still holds the stale VBAR_EL1 value from
before the guest's update. When pKVM injects the exception, it ends up
using the stale value.
Explicitly read the live value of VBAR_EL1 from the guest and update the
vCPU value immediately before pending the exception. This ensures the
vCPU's value is the same as the guest's and that the exception will be
handled at the correct address upon resuming the guest.
Reported-by: Keir Fraser <keirf@google.com>
Signed-off-by: Fuad Tabba <tabba@google.com>
Link: https://lore.kernel.org/r/20250807120133.871892-3-tabba@google.com
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Since commit 17efc1acee ("arm64: Expose AIDR_EL1 via sysfs"), AIDR_EL1
is read early during boot. Therefore, a guest running as a protected VM
will fail to boot because when it attempts to access AIDR_EL1, access to
that register is restricted in pKVM for protected guests.
Similar to how MIDR_EL1 is handled by the host for protected VMs, let
the host handle accesses to AIDR_EL1 as well as REVIDR_EL1. However note
that, unlike MIDR_EL1, AIDR_EL1 and REVIDR_EL1 are trapped by
HCR_EL2.TID1. Therefore, explicitly mark them as handled by the host for
protected VMs. TID1 is always set in pKVM, because it needs to restrict
access to SMIDR_EL1, which is also trapped by that bit.
Reported-by: Will Deacon <will@kernel.org>
Signed-off-by: Fuad Tabba <tabba@google.com>
Link: https://lore.kernel.org/r/20250807120133.871892-2-tabba@google.com
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
The BUG_ON() macro adds a little bit of complexity over BUG(), and in
some cases this ends up confusing the compiler's control flow analysis
in a way that results in a warning. This one now shows up with clang-21:
arch/arm64/kvm/vgic/vgic-mmio.c:1094:3: error: variable 'len' is used uninitialized whenever 'if' condition is false [-Werror,-Wsometimes-uninitialized]
1094 | BUG_ON(1);
Change both instances of BUG_ON(1) to a plain BUG() in the arm64 kvm
code, to avoid the false-positive warning.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Link: https://lore.kernel.org/r/20250807072132.4170088-1-arnd@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
System register accesses redirected to the VNCR page can also generate
external aborts just like any other form of memory access. Route to
kvm_handle_guest_sea() for potential APEI handling, falling back to a
vSError if the kernel didn't handle the abort.
Take the opportunity to throw out the useless kvm_ras.h which provided a
helper with a single callsite...
Cc: Jiaqi Yan <jiaqiyan@google.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250729182342.3281742-1-oliver.upton@linux.dev
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Instead of checking for the ESR_EL2.VNCR bit being set (the only case
we should be here), we are actually testing random bits in ESR_EL2.DFSC.
13 obviously being a lucky number, it matches both permission and
translation fault status codes, which explains why we never saw it
failing. This was found by inspection, while reviewing a vaguely
related patch.
Whilst we're at it, turn the BUG_ON() into a WARN_ON_ONCE(), as
exploding here is just silly.
Fixes: 069a05e535 ("KVM: arm64: nv: Handle VNCR_EL2-triggered faults")
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Joey Gouly <joey.gouly@arm.com>
Link: https://lore.kernel.org/r/20250730101828.1168707-1-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
* kvm-arm64/vgic-v4-ctl:
: Userspace control of nASSGIcap, courtesy of Raghavendra Rao Ananta
:
: Allow userspace to decide if support for SGIs without an active state is
: advertised to the guest, allowing VMs from GICv3-only hardware to be
: migrated to to GICv4.1 capable machines.
Documentation: KVM: arm64: Describe VGICv3 registers writable pre-init
KVM: arm64: selftests: Add test for nASSGIcap attribute
KVM: arm64: vgic-v3: Allow userspace to write GICD_TYPER2.nASSGIcap
KVM: arm64: vgic-v3: Allow access to GICD_IIDR prior to initialization
KVM: arm64: vgic-v3: Consolidate MAINT_IRQ handling
KVM: arm64: Disambiguate support for vSGIs v. vLPIs
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
* kvm-arm64/el2-reg-visibility:
: Fixes to EL2 register visibility, courtesy of Marc Zyngier
:
: - Expose EL2 VGICv3 registers via the VGIC attributes accessor, not the
: KVM_{GET,SET}_ONE_REG ioctls
:
: - Condition visibility of FGT registers on the presence of FEAT_FGT in
: the VM
KVM: arm64: selftest: vgic-v3: Add basic GICv3 sysreg userspace access test
KVM: arm64: Enforce the sorting of the GICv3 system register table
KVM: arm64: Clarify the check for reset callback in check_sysreg_table()
KVM: arm64: vgic-v3: Fix ordering of ICH_HCR_EL2
KVM: arm64: Document registers exposed via KVM_DEV_ARM_VGIC_GRP_CPU_SYSREGS
KVM: arm64: selftests: get-reg-list: Add base EL2 registers
KVM: arm64: selftests: get-reg-list: Simplify feature dependency
KVM: arm64: Advertise FGT2 registers to userspace
KVM: arm64: Condition FGT registers on feature availability
KVM: arm64: Expose GICv3 EL2 registers via KVM_DEV_ARM_VGIC_GRP_CPU_SYSREGS
KVM: arm64: Let GICv3 save/restore honor visibility attribute
KVM: arm64: Define helper for ICH_VTR_EL2
KVM: arm64: Define constant value for ICC_SRE_EL2
KVM: arm64: Don't advertise ICH_*_EL2 registers through GET_ONE_REG
KVM: arm64: Make RVBAR_EL2 accesses UNDEF
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
* kvm-arm64/config-masks:
: More config-driven mask computation, courtesy of Marc Zyngier
:
: Converts more system registers to the config-driven computation of RESx
: masks based on the advertised feature set
KVM: arm64: Tighten the definition of FEAT_PMUv3p9
KVM: arm64: Convert MDCR_EL2 to config-driven sanitisation
KVM: arm64: Convert SCTLR_EL1 to config-driven sanitisation
KVM: arm64: Convert TCR2_EL2 to config-driven sanitisation
arm64: sysreg: Add THE/ASID2 controls to TCR2_ELx
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
* kvm-arm64/gcie-legacy:
: Support for GICv3 emulation on GICv5, courtesy of Sascha Bischoff
:
: FEAT_GCIE_LEGACY adds the necessary hardware for GICv5 systems to
: support the legacy GICv3 for VMs, including a backwards-compatible VGIC
: implementation that we all know and love.
:
: As a starting point for GICv5 enablement in KVM, enable + use the
: GICv3-compatible feature when running VMs on GICv5 hardware.
KVM: arm64: gic-v5: Probe for GICv5
KVM: arm64: gic-v5: Support GICv3 compat
arm64/sysreg: Add ICH_VCTLR_EL2
irqchip/gic-v5: Populate struct gic_kvm_info
irqchip/gic-v5: Skip deactivate for forwarded PPI interrupts
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
GICv5 initial host support
Add host kernel support for the new arm64 GICv5 architecture, which is
quite a departure from the previous ones.
Include support for the full gamut of the architecture (interrupt
routing and delivery to CPUs, wired interrupts, MSIs, and interrupt
translation).
* tag 'irqchip-gic-v5-host': (32 commits)
arm64: smp: Fix pNMI setup after GICv5 rework
arm64: Kconfig: Enable GICv5
docs: arm64: gic-v5: Document booting requirements for GICv5
irqchip/gic-v5: Add GICv5 IWB support
irqchip/gic-v5: Add GICv5 ITS support
irqchip/msi-lib: Add IRQ_DOMAIN_FLAG_FWNODE_PARENT handling
irqchip/gic-v3: Rename GICv3 ITS MSI parent
PCI/MSI: Add pci_msi_map_rid_ctlr_node() helper function
of/irq: Add of_msi_xlate() helper function
irqchip/gic-v5: Enable GICv5 SMP booting
irqchip/gic-v5: Add GICv5 LPI/IPI support
irqchip/gic-v5: Add GICv5 IRS/SPI support
irqchip/gic-v5: Add GICv5 PPI support
arm64: Add support for GICv5 GSB barriers
arm64: smp: Support non-SGIs for IPIs
arm64: cpucaps: Add GICv5 CPU interface (GCIE) capability
arm64: cpucaps: Rename GICv3 CPU interface capability
arm64: Disable GICv5 read/write/instruction traps
arm64/sysreg: Add ICH_HFGITR_EL2
arm64/sysreg: Add ICH_HFGWTR_EL2
...
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
* kvm-arm64/doublefault2: (33 commits)
: NV Support for FEAT_RAS + DoubleFault2
:
: Delegate the vSError context to the guest hypervisor when in a nested
: state, including registers related to ESR propagation. Additionally,
: catch up KVM's external abort infrastructure to the architecture,
: implementing the effects of FEAT_DoubleFault2.
:
: This has some impact on non-nested guests, as SErrors deemed unmasked at
: the time they're made pending are now immediately injected with an
: emulated exception entry rather than using the VSE bit.
KVM: arm64: Make RAS registers UNDEF when RAS isn't advertised
KVM: arm64: Filter out HCR_EL2 bits when running in hypervisor context
KVM: arm64: Check for SYSREGS_ON_CPU before accessing the CPU state
KVM: arm64: Commit exceptions from KVM_SET_VCPU_EVENTS immediately
KVM: arm64: selftests: Test ESR propagation for vSError injection
KVM: arm64: Populate ESR_ELx.EC for emulated SError injection
KVM: arm64: selftests: Catch up set_id_regs with the kernel
KVM: arm64: selftests: Add SCTLR2_EL1 to get-reg-list
KVM: arm64: selftests: Test SEAs are taken to SError vector when EASE=1
KVM: arm64: selftests: Add basic SError injection test
KVM: arm64: Don't retire MMIO instruction w/ pending (emulated) SError
KVM: arm64: Advertise support for FEAT_DoubleFault2
KVM: arm64: Advertise support for FEAT_SCTLR2
KVM: arm64: nv: Enable vSErrors when HCRX_EL2.TMEA is set
KVM: arm64: nv: Honor SError routing effects of SCTLR2_ELx.NMEA
KVM: arm64: nv: Take "masked" aborts to EL2 when HCRX_EL2.TMEA is set
KVM: arm64: Route SEAs to the SError vector when EASE is set
KVM: arm64: nv: Ensure Address size faults affect correct ESR
KVM: arm64: Factor out helper for selecting exception target EL
KVM: arm64: Describe SCTLR2_ELx RESx masks
...
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
* kvm-arm64/cacheable-pfnmap:
: Cacheable PFNMAP support at stage-2, courtesy of Ankit Agrawal
:
: For historical reasons, KVM only allows cacheable mappings at stage-2
: when a kernel alias exists in the direct map for the memory region. On
: hardware without FEAT_S2FWB, this is necessary as KVM must do cache
: maintenance to keep guest/host accesses coherent.
:
: This is unnecessarily restrictive on systems with FEAT_S2FWB and
: CTR_EL0.DIC, as KVM no longer needs to perform cache maintenance to
: maintain correctness.
:
: Allow cacheable mappings at stage-2 on supporting hardware when the
: corresponding VMA has cacheable memory attributes and advertise a
: capability to userspace such that a VMM can determine if a stage-2
: mapping can be established (e.g. VFIO device).
KVM: arm64: Expose new KVM cap for cacheable PFNMAP
KVM: arm64: Allow cacheable stage 2 mapping using VMA flags
KVM: arm64: Block cacheable PFNMAP mapping
KVM: arm64: Assume non-PFNMAP/MIXEDMAP VMAs can be mapped cacheable
KVM: arm64: Rename the device variable to s2_force_noncacheable
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
KVM unconditionally advertises GICD_TYPER2.nASSGIcap (which internally
implies vSGIs) on GICv4.1 systems. Allow userspace to change whether a
VM supports the feature. Only allow changes prior to VGIC initialization
as at that point vPEs need to be allocated for the VM.
For convenience, bundle support for vLPIs and vSGIs behind this feature,
allowing userspace to control vPE allocation for VMs in environments
that may be constrained on vPE IDs.
Signed-off-by: Raghavendra Rao Ananta <rananta@google.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250724062805.2658919-5-oliver.upton@linux.dev
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
KVM allows userspace to write GICD_IIDR for backwards-compatibility with
older kernels, where new implementation revisions have new features.
Unfortunately this is allowed to happen at runtime, and ripping features
out from underneath a running guest is a terrible idea.
While we can't do anything about the ABI, prepare for more ID-like
registers by allowing access to GICD_IIDR prior to VGIC initialization.
Hoist initializaiton of the default value to kvm_vgic_create() and
discard the incorrect comment that assumed userspace could access the
register before initialization (until now).
Subsequent changes will allow the VMM to further provision the GIC
feature set, e.g. the presence of nASSGIcap.
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250724062805.2658919-4-oliver.upton@linux.dev
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Consolidate the duplicated handling of the VGICv3 maintenance IRQ
attribute as a regular GICv3 attribute, as it is neither a register nor
a common attribute. As this is now handled separately from the VGIC
registers, the locking is relaxed to only acquire the intended
config_lock.
Reviewed-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250724062805.2658919-3-oliver.upton@linux.dev
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
vgic_supports_direct_msis() is a bit of a misnomer, as it returns true
if either vSGIs or vLPIs are supported. Pick it apart into a few
predicates and replace some open-coded checks for vSGIs, including an
opportunistic fix to always check if the CPUIF is capable of handling
vSGIs.
Reviewed-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250724062805.2658919-2-oliver.upton@linux.dev
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
check_sysreg_table() has a wonky 'is_32" parameter, which is really
an indication that we should enforce the presence of a reset helper.
Clean this up by naming the variable accordingly and inverting the
condition. Contrary to popular belief, system instructions don't
have a reset value (duh!), and therefore do not need to be checked
for reset (they escaped the check through luck...).
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Sebastian Ott <sebott@redhat.com>
Link: https://lore.kernel.org/r/20250718111154.104029-3-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
The R_QXXPC and R_NPBXC rules have some interesting (and pretty
sharp) corners when defining the behaviour of of WXN at S1:
- when S1 overlay is enabled, WXN applies to the overlay and
will remove W
- when S1 overlay is disabled, WXN applies to the base permissions
and will remove X.
Today, we lumb the two together in a way that doesn't really match
the rules, making things awkward to follow what is happening, in
particular when overlays are enabled.
Split these two rules over two distinct paths, which makes things
a lot easier to read and validate against the architecture rules.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250701151648.754785-3-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Some of the POE computation is a bit confused. Specifically, there
is an element of confusion between what wi->{e0,}poe an wr->{p,u}ov
actually represent.
- wi->{e0,}poe is an *input* to the walk, and indicates whether
POE is enabled at EL0 or EL{1,2}
- wr->{p,u}ov is a *result* of the walk, and indicates whether
overlays are enabled. Crutially, it is possible to have POE
enabled, and yet overlays disabled, while the converse isn't
true
What this all means is that once the base permissions have been
established, checking for wi->{e0,}poe makes little sense, because
the truth about overlays resides in wr->{p,u}ov. So constructs
checking for (wi->poe && wr->pov) only add perplexity.
Refactor compute_s1_overlay_permissions() and the way it is
called according to the above principles. Take the opportunity
to avoid reading registers that are not strictly required.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250701151648.754785-2-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
We currently always expose FEAT_RAS when available on the host.
As we are about to make this feature selectable from userspace,
check for it being present before emulating register accesses
as RAZ/WI, and inject an UNDEF otherwise.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250721101955.535159-4-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Most HCR_EL2 bits are not supposed to affect EL2 at all, but only
the guest. However, we gladly merge these bits with the host's
HCR_EL2 configuration, irrespective of entering L1 or L2.
This leads to some funky behaviour, such as L1 trying to inject
a virtual SError for L2, and getting a taste of its own medecine.
Not quite what the architecture anticipated.
In the end, the only bits that matter are those we have defined as
invariants, either because we've made them RESx (E2H, HCD...), or
that we actively refuse to merge because the mess with KVM's own
logic.
Use the sanitisation infrastructure to get the RES1 bits, and let
things rip in a safer way.
Fixes: 04ab519bb8 ("KVM: arm64: nv: Configure HCR_EL2 for FEAT_NV2")
Signed-off-by: Marc Zyngier <maz@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20250721101955.535159-3-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Mark Brown reports that since we commit to making exceptions
visible without the vcpu being loaded, the external abort selftest
fails.
Upon investigation, it turns out that the code that makes registers
affected by an exception visible to the guest is completely broken
on VHE, as we don't check whether the system registers are loaded
on the CPU at this point. We managed to get away with this so far,
but that's obviously as bad as it gets,
Add the required checksm and document the absolute need to check
for the SYSREGS_ON_CPU flag before calling into any of the
__vcpu_write_sys_reg_to_cpu()__vcpu_read_sys_reg_from_cpu() helpers.
Reported-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/18535df8-e647-4643-af9a-bb780af03a70@sirena.org.uk
Link: https://lore.kernel.org/r/20250720102229.179114-1-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
The current definition of FEAT_PMUv3p9 doesn't check for the lack
of an IMPDEF PMU, which is encoded as 0b1111, but considered unsigned.
Use the recently introduced helper to address the issue (which is
harmless, as KVM never advertises an IMPDEF PMU).
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250714115503.3334242-6-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
We never documented which GICv3 registers are available for save/restore
via the KVM_DEV_ARM_VGIC_GRP_CPU_SYSREGS interface.
Let's take the opportunity of adding the EL2 registers to document the whole
thing in one go.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20250714122634.3334816-12-maz@kernel.org
[ oliver: fix trailing whitespace ]
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
Describing the dependencies between registers and features is on
the masochistic side of things, with hard-coded values that would
be better taken from the existing description.
Add a couple of helpers to that effect, and repaint the dependency
array. More could be done to improve this test, but my interest is
wearing thin...
Signed-off-by: Marc Zyngier <maz@kernel.org>
Tested-by: Itaru Kitayama <itaru.kitayama@fujitsu.com>
Link: https://lore.kernel.org/r/20250714122634.3334816-10-maz@kernel.org
Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
