Re: [PATCH v14 26/44] arm64: RMI: Allow populating initial contents

On 14/07/2026 05:37, Kohei Enju wrote:
> On 05/13 14:17, Steven Price wrote:
>> The VMM needs to populate the realm with some data before starting (e.g.
>> a kernel and initrd). This is measured by the RMM and used as part of
>> the attestation later on.
>>
>> Signed-off-by: Steven Price <[email protected]>
>> ---
>> Changes since v13:
>>  * Rename realm_create_protected_data_page() to realm_data_map_init().
>> Changes since v12:
>>  * The ioctl now updates the structure with the amount populated rather
>>    than returning this through the ioctl return code.
>>  * Use the new RMM v2.0 range based RMI calls.
>>  * Adapt to upstream changes in kvm_gmem_populate().
>> Changes since v11:
>>  * The multiplex CAP is gone and there's a new ioctl which makes use of
>>    the generic kvm_gmem_populate() functionality.
>> Changes since v7:
>>  * Improve the error codes.
>>  * Other minor changes from review.
>> Changes since v6:
>>  * Handle host potentially having a larger page size than the RMM
>>    granule.
>>  * Drop historic "par" (protected address range) from
>>    populate_par_region() - it doesn't exist within the current
>>    architecture.
>>  * Add a cond_resched() call in kvm_populate_realm().
>> Changes since v5:
>>  * Refactor to use PFNs rather than tracking struct page in
>>    realm_create_protected_data_page().
>>  * Pull changes from a later patch (in the v5 series) for accessing
>>    pages from a guest memfd.
>>  * Do the populate in chunks to avoid holding locks for too long and
>>    triggering RCU stall warnings.
>> ---
>>  arch/arm64/include/asm/kvm_rmi.h |   4 ++
>>  arch/arm64/kvm/Kconfig           |   1 +
>>  arch/arm64/kvm/arm.c             |  13 ++++
>>  arch/arm64/kvm/rmi.c             | 106 +++++++++++++++++++++++++++++++
>>  4 files changed, 124 insertions(+)
>>
>> diff --git a/arch/arm64/include/asm/kvm_rmi.h b/arch/arm64/include/asm/kvm_rmi.h
>> index 007249a13dbc..a2b6bc412a22 100644
>> --- a/arch/arm64/include/asm/kvm_rmi.h
>> +++ b/arch/arm64/include/asm/kvm_rmi.h
>> @@ -88,6 +88,10 @@ int kvm_rec_enter(struct kvm_vcpu *vcpu);
>>  int kvm_rec_pre_enter(struct kvm_vcpu *vcpu);
>>  int handle_rec_exit(struct kvm_vcpu *vcpu, int rec_run_status);
>>  
>> +struct kvm_arm_rmi_populate;
>> +
>> +int kvm_arm_rmi_populate(struct kvm *kvm,
>> +			 struct kvm_arm_rmi_populate *arg);
>>  void kvm_realm_unmap_range(struct kvm *kvm,
>>  			   unsigned long ipa,
>>  			   unsigned long size,
>> diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
>> index 4e16719fda22..d0cd011cf672 100644
>> --- a/arch/arm64/kvm/Kconfig
>> +++ b/arch/arm64/kvm/Kconfig
>> @@ -38,6 +38,7 @@ menuconfig KVM
>>  	select GUEST_PERF_EVENTS if PERF_EVENTS
>>  	select KVM_GUEST_MEMFD
>>  	select KVM_GENERIC_MEMORY_ATTRIBUTES
>> +	select HAVE_KVM_ARCH_GMEM_POPULATE
>>  	help
>>  	  Support hosting virtualized guest machines.
>>  
>> diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
>> index ed88a203b892..073ba9181da9 100644
>> --- a/arch/arm64/kvm/arm.c
>> +++ b/arch/arm64/kvm/arm.c
>> @@ -2131,6 +2131,19 @@ int kvm_arch_vm_ioctl(struct file *filp, unsigned int ioctl, unsigned long arg)
>>  			return -EFAULT;
>>  		return kvm_vm_ioctl_get_reg_writable_masks(kvm, &range);
>>  	}
>> +	case KVM_ARM_RMI_POPULATE: {
>> +		struct kvm_arm_rmi_populate req;
>> +		int ret;
>> +
>> +		if (!kvm_is_realm(kvm))
>> +			return -ENXIO;
>> +		if (copy_from_user(&req, argp, sizeof(req)))
>> +			return -EFAULT;
>> +		ret = kvm_arm_rmi_populate(kvm, &req);
>> +		if (copy_to_user(argp, &req, sizeof(req)))
>> +			return -EFAULT;
>> +		return ret;
>> +	}
>>  	default:
>>  		return -EINVAL;
>>  	}
>> diff --git a/arch/arm64/kvm/rmi.c b/arch/arm64/kvm/rmi.c
>> index a89873a5eb77..209087bcf399 100644
>> --- a/arch/arm64/kvm/rmi.c
>> +++ b/arch/arm64/kvm/rmi.c
>> @@ -486,6 +486,75 @@ void kvm_realm_unmap_range(struct kvm *kvm, unsigned long start,
>>  		realm_unmap_private_range(kvm, start, end, may_block);
>>  }
>>  
>> +static int realm_data_map_init(struct kvm *kvm, unsigned long ipa,
>> +			       kvm_pfn_t dst_pfn, kvm_pfn_t src_pfn,
>> +			       unsigned long flags)
>> +{
>> +	struct realm *realm = &kvm->arch.realm;
>> +	phys_addr_t rd = virt_to_phys(realm->rd);
>> +	phys_addr_t dst_phys, src_phys;
>> +	int ret;
>> +
>> +	dst_phys = __pfn_to_phys(dst_pfn);
>> +	src_phys = __pfn_to_phys(src_pfn);
>> +
>> +	if (rmi_delegate_page(dst_phys))
>> +		return -ENXIO;
>> +
>> +	ret = rmi_rtt_data_map_init(rd, dst_phys, ipa, src_phys, flags);
>> +	if (RMI_RETURN_STATUS(ret) == RMI_ERROR_RTT) {
>> +		/* Create missing RTTs and retry */
>> +		int level = RMI_RETURN_INDEX(ret);
>> +
>> +		KVM_BUG_ON(level == KVM_PGTABLE_LAST_LEVEL, kvm);
>> +
>> +		ret = realm_create_rtt_levels(realm, ipa, level,
>> +					      KVM_PGTABLE_LAST_LEVEL, NULL);
>> +		if (!ret) {
>> +			ret = rmi_rtt_data_map_init(rd, dst_phys, ipa, src_phys,
>> +						    flags);
> 
> I think rmi_rtt_data_map_init() returns a raw RMI return value which is
> positive on failure. Shouldn't it be converted to -ENXIO or another
> Linux errno before returning it?
> 
> Otherwise, the positive error value can propagate through
> kvm_gmem_populate() and populate_region() to kvm_arm_rmi_populate().
> There, it would be treated as the number of pages populated, even though
> the operation did not populate any pages.

Good spot.

>> +		}
>> +	}
>> +
>> +	if (ret) {
>> +		if (WARN_ON(rmi_undelegate_page(dst_phys))) {
>> +			/* Undelegate failed, so we leak the page */
>> +			get_page(pfn_to_page(dst_pfn));
> 
> Since realm_create_rtt_levels() returns either 0 or a negative errno,
> how about preserving that error and converting any raw RMI erorr to
> -ENXIO here?
> 
> 		return ret < 0 ? ret : -ENXIO;

Makes sense, although this needs to be:

	return ret <= 0 ? ret : -ENXIO;

to handle the ret==0 case. I'll fix this in my next posting.

>> +		}
>> +	}
>> +
>> +	return ret;
>> +}
>> +
>> +static int populate_region_cb(struct kvm *kvm, gfn_t gfn, kvm_pfn_t pfn,
>> +			      struct page *src_page, void *opaque)
>> +{
>> +	unsigned long data_flags = *(unsigned long *)opaque;
>> +	phys_addr_t ipa = gfn_to_gpa(gfn);
>> +
>> +	if (!src_page)
>> +		return -EOPNOTSUPP;
>> +
>> +	return realm_data_map_init(kvm, ipa, pfn, page_to_pfn(src_page),
>> +				   data_flags);
>> +}
>> +
>> +static long populate_region(struct kvm *kvm,
>> +			    gfn_t base_gfn,
>> +			    unsigned long pages,
>> +			    u64 uaddr,
>> +			    unsigned long data_flags)
>> +{
>> +	long ret = 0;
>> +
>> +	mutex_lock(&kvm->slots_lock);
>> +	ret = kvm_gmem_populate(kvm, base_gfn, u64_to_user_ptr(uaddr), pages,
>> +				populate_region_cb, &data_flags);
>> +	mutex_unlock(&kvm->slots_lock);
>> +
>> +	return ret;
>> +}
>> +
>>  enum ripas_action {
>>  	RIPAS_INIT,
>>  	RIPAS_SET,
>> @@ -574,6 +643,43 @@ static int realm_ensure_created(struct kvm *kvm)
>>  	return -ENXIO;
>>  }
>>  
>> +int kvm_arm_rmi_populate(struct kvm *kvm,
>> +			 struct kvm_arm_rmi_populate *args)
>> +{
>> +	unsigned long data_flags = 0;
>> +	unsigned long ipa_start = args->base;
>> +	unsigned long ipa_end = ipa_start + args->size;
>> +	long pages_populated;
>> +	int ret;
>> +
>> +	if (args->reserved ||
>> +	    (args->flags & ~KVM_ARM_RMI_POPULATE_FLAGS_MEASURE) ||
>> +	    !IS_ALIGNED(ipa_start, PAGE_SIZE) ||
>> +	    !IS_ALIGNED(ipa_end, PAGE_SIZE) ||
>> +	    !IS_ALIGNED(args->source_uaddr, PAGE_SIZE))
>> +		return -EINVAL;
>> +
>> +	ret = realm_ensure_created(kvm);
>> +	if (ret)
>> +		return ret;
> 
> If I understand correctly, concurrent KVM_ARM_RMI_POPULATE calls can
> race with each other in realm_ensure_created(). 
> 
> realm_ensure_created() calls realm_create_rd() when kvm_realm_state() ==
> REALM_STATE_NONE. Since kvm->arch.config_lock is not held here, multiple
> threads could enter realm_create_rd() concurrently.
> 
> KVM_ARM_RMI_POPULATE can also race with KVM_RUN through
> kvm_activate_realm(). kvm_activate_realm() takes kvm->arch.config_lock
> while checking the Realm state, calling realm_ensure_created(), and
> updating the state, whereas KVM_ARM_RMI_POPULATE does not.
> 
> How about serializing Realm creation here as follows?
> 	scoped_guard(mutex, &kvm->arch.config_lock) {
> 		ret = realm_ensure_created(kvm);
> 		if (ret)
> 			return ret;
> 	}
> 
> Does that make sense?

Yes this is a bug I'd already caught and is due to the fixed in my next
posting - similar to what you suggest the config_lock is now held when
realm_ensure_created() is called (and I've put a lockdep_assert_held()
in realm_ensure_created).

Thanks,
Steve

>> +
>> +	if (args->flags & KVM_ARM_RMI_POPULATE_FLAGS_MEASURE)
>> +		data_flags |= RMI_MEASURE_CONTENT;
>> +
>> +	pages_populated = populate_region(kvm, gpa_to_gfn(ipa_start),
>> +					  args->size >> PAGE_SHIFT,
>> +					  args->source_uaddr, data_flags);
>> +
>> +	if (pages_populated < 0)
>> +		return pages_populated;
>> +
>> +	args->size -= pages_populated << PAGE_SHIFT;
>> +	args->source_uaddr += pages_populated << PAGE_SHIFT;
>> +	args->base += pages_populated << PAGE_SHIFT;
>> +
>> +	return 0;
>> +}
>> +
>>  static void kvm_complete_ripas_change(struct kvm_vcpu *vcpu)
>>  {
>>  	struct kvm *kvm = vcpu->kvm;
>> -- 
>> 2.43.0
>>
>>