When KVM runs in non-protected VHE mode, there's no context synchronization event between __debug_switch_to_host() restoring the host debug registers and __kvm_vcpu_run() unmasking debug exceptions. Due to this, it's theoretically possible for the host to take an unexpected debug exception due to the stale guest configuration.

This cannot happen in NVHE/HVHE mode as debug exceptions are masked in the hyp code, and the exception return to the host will provide the necessary context synchronization before debug exceptions can be taken.

For now, avoid the problem by adding an ISB after VHE hyp code restores the host debug registers.

Signed-off-by: Mark Rutland mark.rutland@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Fuad Tabba tabba@google.com Cc: Marc Zyngier maz@kernel.org Cc: Mark Brown broonie@kernel.org Cc: Oliver Upton oliver.upton@linux.dev Cc: Will Deacon will@kernel.org Cc: stable@vger.kernel.org --- arch/arm64/kvm/hyp/include/hyp/debug-sr.h | 3 +++ 1 file changed, 3 insertions(+)

diff --git a/arch/arm64/kvm/hyp/include/hyp/debug-sr.h b/arch/arm64/kvm/hyp/include/hyp/debug-sr.h index 502a5b73ee70c..73881e1dc2679 100644 --- a/arch/arm64/kvm/hyp/include/hyp/debug-sr.h +++ b/arch/arm64/kvm/hyp/include/hyp/debug-sr.h @@ -167,6 +167,9 @@ static inline void __debug_switch_to_host_common(struct kvm_vcpu *vcpu)

__debug_save_state(guest_dbg, guest_ctxt); __debug_restore_state(host_dbg, host_ctxt); + + if (has_vhe()) + isb(); }

#endif /* __ARM64_KVM_HYP_DEBUG_SR_H__ */

The NVHE/HVHE and VHE modes have separate implementations of __activate_cptr_traps() and __deactivate_cptr_traps() in their respective switch.c files. There's some duplication of logic, and it's not currently possible to reuse this logic elsewhere.

Move the logic into the common switch.h header so that it can be reused, and de-duplicate the common logic.

This rework changes the way SVE traps are deactivated in VHE mode, aligning it with NVHE/HVHE modes:

* Before this patch, VHE's __deactivate_cptr_traps() would unconditionally enable SVE for host EL2 (but not EL0), regardless of whether the ARM64_SVE cpucap was set.

* After this patch, VHE's __deactivate_cptr_traps() will take the ARM64_SVE cpucap into account. When ARM64_SVE is not set, SVE will be trapped from EL2 and below.

The old and new behaviour are both benign:

* When ARM64_SVE is not set, the host will not touch SVE state, and will not reconfigure SVE traps. Host EL0 access to SVE will be trapped as expected.

* When ARM64_SVE is set, the host will configure EL0 SVE traps before returning to EL0 as part of reloading the EL0 FPSIMD/SVE/SME state.

Signed-off-by: Mark Rutland mark.rutland@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Fuad Tabba tabba@google.com Cc: Marc Zyngier maz@kernel.org Cc: Mark Brown broonie@kernel.org Cc: Oliver Upton oliver.upton@linux.dev Cc: Will Deacon will@kernel.org --- arch/arm64/kvm/hyp/include/hyp/switch.h | 130 ++++++++++++++++++++++++ arch/arm64/kvm/hyp/nvhe/switch.c | 59 ----------- arch/arm64/kvm/hyp/vhe/switch.c | 81 --------------- 3 files changed, 130 insertions(+), 140 deletions(-)

diff --git a/arch/arm64/kvm/hyp/include/hyp/switch.h b/arch/arm64/kvm/hyp/include/hyp/switch.h index aa5b561b92182..9b025276fcc8e 100644 --- a/arch/arm64/kvm/hyp/include/hyp/switch.h +++ b/arch/arm64/kvm/hyp/include/hyp/switch.h @@ -65,6 +65,136 @@ static inline void __activate_traps_fpsimd32(struct kvm_vcpu *vcpu) } }

+static inline void __activate_cptr_traps_nvhe(struct kvm_vcpu *vcpu) +{ + u64 val = CPTR_NVHE_EL2_RES1 | CPTR_EL2_TAM | CPTR_EL2_TTA; + + /* + * Always trap SME since it's not supported in KVM. + * TSM is RES1 if SME isn't implemented. + */ + val |= CPTR_EL2_TSM; + + if (!vcpu_has_sve(vcpu) || !guest_owns_fp_regs()) + val |= CPTR_EL2_TZ; + + if (!guest_owns_fp_regs()) + val |= CPTR_EL2_TFP; + + write_sysreg(val, cptr_el2); +} + +static inline void __activate_cptr_traps_vhe(struct kvm_vcpu *vcpu) +{ + /* + * With VHE (HCR.E2H == 1), accesses to CPACR_EL1 are routed to + * CPTR_EL2. In general, CPACR_EL1 has the same layout as CPTR_EL2, + * except for some missing controls, such as TAM. + * In this case, CPTR_EL2.TAM has the same position with or without + * VHE (HCR.E2H == 1) which allows us to use here the CPTR_EL2.TAM + * shift value for trapping the AMU accesses. + */ + u64 val = CPTR_EL2_TAM | CPACR_EL1_TTA; + u64 cptr; + + if (guest_owns_fp_regs()) { + val |= CPACR_EL1_FPEN; + if (vcpu_has_sve(vcpu)) + val |= CPACR_EL1_ZEN; + } + + if (!vcpu_has_nv(vcpu)) + goto write; + + /* + * The architecture is a bit crap (what a surprise): an EL2 guest + * writing to CPTR_EL2 via CPACR_EL1 can't set any of TCPAC or TTA, + * as they are RES0 in the guest's view. To work around it, trap the + * sucker using the very same bit it can't set... + */ + if (vcpu_el2_e2h_is_set(vcpu) && is_hyp_ctxt(vcpu)) + val |= CPTR_EL2_TCPAC; + + /* + * Layer the guest hypervisor's trap configuration on top of our own if + * we're in a nested context. + */ + if (is_hyp_ctxt(vcpu)) + goto write; + + cptr = vcpu_sanitised_cptr_el2(vcpu); + + /* + * Pay attention, there's some interesting detail here. + * + * The CPTR_EL2.xEN fields are 2 bits wide, although there are only two + * meaningful trap states when HCR_EL2.TGE = 0 (running a nested guest): + * + * - CPTR_EL2.xEN = x0, traps are enabled + * - CPTR_EL2.xEN = x1, traps are disabled + * + * In other words, bit[0] determines if guest accesses trap or not. In + * the interest of simplicity, clear the entire field if the guest + * hypervisor has traps enabled to dispel any illusion of something more + * complicated taking place. + */ + if (!(SYS_FIELD_GET(CPACR_EL1, FPEN, cptr) & BIT(0))) + val &= ~CPACR_EL1_FPEN; + if (!(SYS_FIELD_GET(CPACR_EL1, ZEN, cptr) & BIT(0))) + val &= ~CPACR_EL1_ZEN; + + if (kvm_has_feat(vcpu->kvm, ID_AA64MMFR3_EL1, S2POE, IMP)) + val |= cptr & CPACR_EL1_E0POE; + + val |= cptr & CPTR_EL2_TCPAC; + +write: + write_sysreg(val, cpacr_el1); +} + +static inline void __activate_cptr_traps(struct kvm_vcpu *vcpu) +{ + if (!guest_owns_fp_regs()) + __activate_traps_fpsimd32(vcpu); + + if (has_vhe() || has_hvhe()) + __activate_cptr_traps_vhe(vcpu); + else + __activate_cptr_traps_nvhe(vcpu); +} + +static inline void __deactivate_cptr_traps_nvhe(struct kvm_vcpu *vcpu) +{ + u64 val = CPTR_NVHE_EL2_RES1; + + if (!cpus_have_final_cap(ARM64_SVE)) + val |= CPTR_EL2_TZ; + if (!cpus_have_final_cap(ARM64_SME)) + val |= CPTR_EL2_TSM; + + write_sysreg(val, cptr_el2); +} + +static inline void __deactivate_cptr_traps_vhe(struct kvm_vcpu *vcpu) +{ + u64 val = CPACR_EL1_FPEN; + + if (cpus_have_final_cap(ARM64_SVE)) + val |= CPACR_EL1_ZEN; + if (cpus_have_final_cap(ARM64_SME)) + val |= CPACR_EL1_SMEN; + + write_sysreg(val, cpacr_el1); +} + +static inline void __deactivate_cptr_traps(struct kvm_vcpu *vcpu) +{ + if (has_vhe() || has_hvhe()) + __deactivate_cptr_traps_vhe(vcpu); + else + __deactivate_cptr_traps_nvhe(vcpu); +} + #define reg_to_fgt_masks(reg) \ ({ \ struct fgt_masks *m; \ diff --git a/arch/arm64/kvm/hyp/nvhe/switch.c b/arch/arm64/kvm/hyp/nvhe/switch.c index 73affe1333a49..0e752b515d0fd 100644 --- a/arch/arm64/kvm/hyp/nvhe/switch.c +++ b/arch/arm64/kvm/hyp/nvhe/switch.c @@ -47,65 +47,6 @@ struct fgt_masks hdfgwtr2_masks;

extern void kvm_nvhe_prepare_backtrace(unsigned long fp, unsigned long pc);

-static void __activate_cptr_traps(struct kvm_vcpu *vcpu) -{ - u64 val = CPTR_EL2_TAM; /* Same bit irrespective of E2H */ - - if (!guest_owns_fp_regs()) - __activate_traps_fpsimd32(vcpu); - - if (has_hvhe()) { - val |= CPACR_EL1_TTA; - - if (guest_owns_fp_regs()) { - val |= CPACR_EL1_FPEN; - if (vcpu_has_sve(vcpu)) - val |= CPACR_EL1_ZEN; - } - - write_sysreg(val, cpacr_el1); - } else { - val |= CPTR_EL2_TTA | CPTR_NVHE_EL2_RES1; - - /* - * Always trap SME since it's not supported in KVM. - * TSM is RES1 if SME isn't implemented. - */ - val |= CPTR_EL2_TSM; - - if (!vcpu_has_sve(vcpu) || !guest_owns_fp_regs()) - val |= CPTR_EL2_TZ; - - if (!guest_owns_fp_regs()) - val |= CPTR_EL2_TFP; - - write_sysreg(val, cptr_el2); - } -} - -static void __deactivate_cptr_traps(struct kvm_vcpu *vcpu) -{ - if (has_hvhe()) { - u64 val = CPACR_EL1_FPEN; - - if (cpus_have_final_cap(ARM64_SVE)) - val |= CPACR_EL1_ZEN; - if (cpus_have_final_cap(ARM64_SME)) - val |= CPACR_EL1_SMEN; - - write_sysreg(val, cpacr_el1); - } else { - u64 val = CPTR_NVHE_EL2_RES1; - - if (!cpus_have_final_cap(ARM64_SVE)) - val |= CPTR_EL2_TZ; - if (!cpus_have_final_cap(ARM64_SME)) - val |= CPTR_EL2_TSM; - - write_sysreg(val, cptr_el2); - } -} - static void __activate_traps(struct kvm_vcpu *vcpu) { ___activate_traps(vcpu, vcpu->arch.hcr_el2); diff --git a/arch/arm64/kvm/hyp/vhe/switch.c b/arch/arm64/kvm/hyp/vhe/switch.c index 20df44b49ceb8..32b4814eb2b7a 100644 --- a/arch/arm64/kvm/hyp/vhe/switch.c +++ b/arch/arm64/kvm/hyp/vhe/switch.c @@ -90,87 +90,6 @@ static u64 __compute_hcr(struct kvm_vcpu *vcpu) return hcr | (guest_hcr & ~NV_HCR_GUEST_EXCLUDE); }

-static void __activate_cptr_traps(struct kvm_vcpu *vcpu) -{ - u64 cptr; - - /* - * With VHE (HCR.E2H == 1), accesses to CPACR_EL1 are routed to - * CPTR_EL2. In general, CPACR_EL1 has the same layout as CPTR_EL2, - * except for some missing controls, such as TAM. - * In this case, CPTR_EL2.TAM has the same position with or without - * VHE (HCR.E2H == 1) which allows us to use here the CPTR_EL2.TAM - * shift value for trapping the AMU accesses. - */ - u64 val = CPACR_EL1_TTA | CPTR_EL2_TAM; - - if (guest_owns_fp_regs()) { - val |= CPACR_EL1_FPEN; - if (vcpu_has_sve(vcpu)) - val |= CPACR_EL1_ZEN; - } else { - __activate_traps_fpsimd32(vcpu); - } - - if (!vcpu_has_nv(vcpu)) - goto write; - - /* - * The architecture is a bit crap (what a surprise): an EL2 guest - * writing to CPTR_EL2 via CPACR_EL1 can't set any of TCPAC or TTA, - * as they are RES0 in the guest's view. To work around it, trap the - * sucker using the very same bit it can't set... - */ - if (vcpu_el2_e2h_is_set(vcpu) && is_hyp_ctxt(vcpu)) - val |= CPTR_EL2_TCPAC; - - /* - * Layer the guest hypervisor's trap configuration on top of our own if - * we're in a nested context. - */ - if (is_hyp_ctxt(vcpu)) - goto write; - - cptr = vcpu_sanitised_cptr_el2(vcpu); - - /* - * Pay attention, there's some interesting detail here. - * - * The CPTR_EL2.xEN fields are 2 bits wide, although there are only two - * meaningful trap states when HCR_EL2.TGE = 0 (running a nested guest): - * - * - CPTR_EL2.xEN = x0, traps are enabled - * - CPTR_EL2.xEN = x1, traps are disabled - * - * In other words, bit[0] determines if guest accesses trap or not. In - * the interest of simplicity, clear the entire field if the guest - * hypervisor has traps enabled to dispel any illusion of something more - * complicated taking place. - */ - if (!(SYS_FIELD_GET(CPACR_EL1, FPEN, cptr) & BIT(0))) - val &= ~CPACR_EL1_FPEN; - if (!(SYS_FIELD_GET(CPACR_EL1, ZEN, cptr) & BIT(0))) - val &= ~CPACR_EL1_ZEN; - - if (kvm_has_feat(vcpu->kvm, ID_AA64MMFR3_EL1, S2POE, IMP)) - val |= cptr & CPACR_EL1_E0POE; - - val |= cptr & CPTR_EL2_TCPAC; - -write: - write_sysreg(val, cpacr_el1); -} - -static void __deactivate_cptr_traps(struct kvm_vcpu *vcpu) -{ - u64 val = CPACR_EL1_FPEN | CPACR_EL1_ZEN_EL1EN; - - if (cpus_have_final_cap(ARM64_SME)) - val |= CPACR_EL1_SMEN_EL1EN; - - write_sysreg(val, cpacr_el1); -} - static void __activate_traps(struct kvm_vcpu *vcpu) { u64 val;

The hyp code FPSIMD/SVE/SME trap handling logic has some rather messy open-coded manipulation of CPTR/CPACR. This is benign for non-nested guests, but broken for nested guests, as the guest hypervisor's CPTR configuration is not taken into account.

Consider the case where L0 provides FPSIMD+SVE to an L1 guest hypervisor, and the L1 guest hypervisor only provides FPSIMD to an L2 guest (with L1 configuring CPTR/CPACR to trap SVE usage from L2). If the L2 guest triggers an FPSIMD trap to the L0 hypervisor, kvm_hyp_handle_fpsimd() will see that the vCPU supports FPSIMD+SVE, and will configure CPTR/CPACR to NOT trap FPSIMD+SVE before returning to the L2 guest. Consequently the L2 guest would be able to manipulate SVE state even though the L1 hypervisor had configured CPTR/CPACR to forbid this.

Clean this up, and fix the nested virt issue by always using __deactivate_cptr_traps() and __activate_cptr_traps() to manage the CPTR traps. This removes the need for the ad-hoc fixup in kvm_hyp_save_fpsimd_host(), and ensures that any guest hypervisor configuration of CPTR/CPACR is taken into account.

Signed-off-by: Mark Rutland mark.rutland@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Fuad Tabba tabba@google.com Cc: Marc Zyngier maz@kernel.org Cc: Mark Brown broonie@kernel.org Cc: Oliver Upton oliver.upton@linux.dev Cc: Will Deacon will@kernel.org --- arch/arm64/kvm/hyp/include/hyp/switch.h | 17 ++++++++--------- 1 file changed, 8 insertions(+), 9 deletions(-)

diff --git a/arch/arm64/kvm/hyp/include/hyp/switch.h b/arch/arm64/kvm/hyp/include/hyp/switch.h index 9b025276fcc8e..ce001a2b6421c 100644 --- a/arch/arm64/kvm/hyp/include/hyp/switch.h +++ b/arch/arm64/kvm/hyp/include/hyp/switch.h @@ -617,11 +617,6 @@ static void kvm_hyp_save_fpsimd_host(struct kvm_vcpu *vcpu) */ if (system_supports_sve()) { __hyp_sve_save_host(); - - /* Re-enable SVE traps if not supported for the guest vcpu. */ - if (!vcpu_has_sve(vcpu)) - cpacr_clear_set(CPACR_EL1_ZEN, 0); - } else { __fpsimd_save_state(host_data_ptr(host_ctxt.fp_regs)); } @@ -672,10 +667,7 @@ static inline bool kvm_hyp_handle_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code) /* Valid trap. Switch the context: */

/* First disable enough traps to allow us to update the registers */ - if (sve_guest || (is_protected_kvm_enabled() && system_supports_sve())) - cpacr_clear_set(0, CPACR_EL1_FPEN | CPACR_EL1_ZEN); - else - cpacr_clear_set(0, CPACR_EL1_FPEN); + __deactivate_cptr_traps(vcpu); isb();

/* Write out the host state if it's in the registers */ @@ -697,6 +689,13 @@ static inline bool kvm_hyp_handle_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code)

*host_data_ptr(fp_owner) = FP_STATE_GUEST_OWNED;

+ /* + * Re-enable traps necessary for the current state of the guest, e.g. + * those enabled by a guest hypervisor. The ERET to the guest will + * provide the necessary context synchronization. + */ + __activate_cptr_traps(vcpu); + return true; }

The VHE hyp code has recently gained a few ISBs. Simplify this to one unconditional ISB in __kvm_vcpu_run_vhe(), and remove the unnecessary ISB from the kvm_call_hyp_ret() macro.

While kvm_call_hyp_ret() is also used to invoke __vgic_v3_get_gic_config(), but no ISB is necessary in that case either.

For the moment, an ISB is left in kvm_call_hyp(), as there are many more users, and removign the ISB would require a more thorough audit.

Suggested-by: Marc Zyngier maz@kernel.org Signed-off-by: Mark Rutland mark.rutland@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Fuad Tabba tabba@google.com Cc: Marc Zyngier maz@kernel.org Cc: Mark Brown broonie@kernel.org Cc: Oliver Upton oliver.upton@linux.dev Cc: Will Deacon will@kernel.org --- arch/arm64/include/asm/kvm_host.h | 6 ++---- arch/arm64/kvm/hyp/include/hyp/debug-sr.h | 3 --- arch/arm64/kvm/hyp/vhe/switch.c | 25 +++++++++++------------ 3 files changed, 14 insertions(+), 20 deletions(-)

diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h index 382b382d14dac..2e3c152668eb8 100644 --- a/arch/arm64/include/asm/kvm_host.h +++ b/arch/arm64/include/asm/kvm_host.h @@ -1289,9 +1289,8 @@ void kvm_arm_resume_guest(struct kvm *kvm); })

/* - * The couple of isb() below are there to guarantee the same behaviour - * on VHE as on !VHE, where the eret to EL1 acts as a context - * synchronization event. + * The isb() below is there to guarantee the same behaviour on VHE as on !VHE, + * where the eret to EL1 acts as a context synchronization event. */ #define kvm_call_hyp(f, ...) \ do { \ @@ -1309,7 +1308,6 @@ void kvm_arm_resume_guest(struct kvm *kvm); \ if (has_vhe()) { \ ret = f(__VA_ARGS__); \ - isb(); \ } else { \ ret = kvm_call_hyp_nvhe(f, ##__VA_ARGS__); \ } \ diff --git a/arch/arm64/kvm/hyp/include/hyp/debug-sr.h b/arch/arm64/kvm/hyp/include/hyp/debug-sr.h index 73881e1dc2679..502a5b73ee70c 100644 --- a/arch/arm64/kvm/hyp/include/hyp/debug-sr.h +++ b/arch/arm64/kvm/hyp/include/hyp/debug-sr.h @@ -167,9 +167,6 @@ static inline void __debug_switch_to_host_common(struct kvm_vcpu *vcpu)

__debug_save_state(guest_dbg, guest_ctxt); __debug_restore_state(host_dbg, host_ctxt); - - if (has_vhe()) - isb(); }

#endif /* __ARM64_KVM_HYP_DEBUG_SR_H__ */ diff --git a/arch/arm64/kvm/hyp/vhe/switch.c b/arch/arm64/kvm/hyp/vhe/switch.c index 32b4814eb2b7a..477f1580ffeaa 100644 --- a/arch/arm64/kvm/hyp/vhe/switch.c +++ b/arch/arm64/kvm/hyp/vhe/switch.c @@ -558,10 +558,10 @@ static int __kvm_vcpu_run_vhe(struct kvm_vcpu *vcpu) host_ctxt = host_data_ptr(host_ctxt); guest_ctxt = &vcpu->arch.ctxt;

- sysreg_save_host_state_vhe(host_ctxt); - fpsimd_lazy_switch_to_guest(vcpu);

+ sysreg_save_host_state_vhe(host_ctxt); + /* * Note that ARM erratum 1165522 requires us to configure both stage 1 * and stage 2 translation for the guest context before we clear @@ -586,18 +586,23 @@ static int __kvm_vcpu_run_vhe(struct kvm_vcpu *vcpu)

__deactivate_traps(vcpu);

- /* Ensure CPTR trap deactivation has taken effect */ + sysreg_restore_host_state_vhe(host_ctxt); + + __debug_switch_to_host(vcpu); + + /* + * Ensure that all system register writes above have taken effect + * before returning to the host. In VHE mode, CPTR traps for + * FPSIMD/SVE/SME also apply to EL2, so FPSIMD/SVE/SME state must be + * manipulated after the ISB. + */ isb();

fpsimd_lazy_switch_to_host(vcpu);

- sysreg_restore_host_state_vhe(host_ctxt); - if (guest_owns_fp_regs()) __fpsimd_save_fpexc32(vcpu);

- __debug_switch_to_host(vcpu); - return exit_code; } NOKPROBE_SYMBOL(__kvm_vcpu_run_vhe); @@ -627,12 +632,6 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu) */ local_daif_restore(DAIF_PROCCTX_NOIRQ);

- /* - * When we exit from the guest we change a number of CPU configuration - * parameters, such as traps. We rely on the isb() in kvm_call_hyp*() - * to make sure these changes take effect before running the host or - * additional guests. - */ return ret; }