Re: [PATCH 02/21] KVM: Allow page-sized MMU caches to be initialized with custom 64-bit values

[Binbin Wu <binbin.wu@linux.intel.com>]

On 3/5/2024 2:55 PM, Binbin Wu wrote:
>
>
> On 2/28/2024 7:20 AM, Paolo Bonzini wrote:
>> From: Sean Christopherson <seanjc@google.com>
>>
>> Add support to MMU caches for initializing a page with a custom 64-bit
>> value, e.g. to pre-fill an entire page table with non-zero PTE values.
>> The functionality will be used by x86 to support Intel's TDX, which 
>> needs
>> to set bit 63 in all non-present PTEs in order to prevent !PRESENT page
>> faults from getting reflected into the guest (Intel's EPT Violation #VE
>> architecture made the less than brilliant decision of having the per-PTE
>> behavior be opt-out instead of opt-in).
>>
>> Signed-off-by: Sean Christopherson <seanjc@google.com>
>> Signed-off-by: Isaku Yamahata <isaku.yamahata@intel.com>
>> Message-Id: 
>> <5919f685f109a1b0ebc6bd8fc4536ee94bcc172d.1705965635.git.isaku.yamahata@intel.com>
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> ---
>>   include/linux/kvm_types.h |  1 +
>>   virt/kvm/kvm_main.c       | 16 ++++++++++++++--
>>   2 files changed, 15 insertions(+), 2 deletions(-)
>
> Reviewed-by: Binbin Wu <binbin.wu@linux.intel.com>
>
>>
>> diff --git a/include/linux/kvm_types.h b/include/linux/kvm_types.h
>> index d93f6522b2c3..827ecc0b7e10 100644
>> --- a/include/linux/kvm_types.h
>> +++ b/include/linux/kvm_types.h
>> @@ -86,6 +86,7 @@ struct gfn_to_pfn_cache {
>>   struct kvm_mmu_memory_cache {
>>       gfp_t gfp_zero;
>>       gfp_t gfp_custom;
>> +    u64 init_value;
>>       struct kmem_cache *kmem_cache;
>>       int capacity;
>>       int nobjs;
>> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
>> index 9c99c9373a3e..c9828feb7a1c 100644
>> --- a/virt/kvm/kvm_main.c
>> +++ b/virt/kvm/kvm_main.c
>> @@ -401,12 +401,17 @@ static void kvm_flush_shadow_all(struct kvm *kvm)
>>   static inline void *mmu_memory_cache_alloc_obj(struct 
>> kvm_mmu_memory_cache *mc,
>>                              gfp_t gfp_flags)
>>   {
>> +    void *page;
>> +
>>       gfp_flags |= mc->gfp_zero;
>>         if (mc->kmem_cache)
>>           return kmem_cache_alloc(mc->kmem_cache, gfp_flags);
>> -    else
>> -        return (void *)__get_free_page(gfp_flags);
>> +
>> +    page = (void *)__get_free_page(gfp_flags);
>> +    if (page && mc->init_value)
>> +        memset64(page, mc->init_value, PAGE_SIZE / 
>> sizeof(mc->init_value));

Do we need a static_assert() to make sure mc->init_value is 64bit?

>> +    return page;
>>   }
>>     int __kvm_mmu_topup_memory_cache(struct kvm_mmu_memory_cache *mc, 
>> int capacity, int min)
>> @@ -421,6 +426,13 @@ int __kvm_mmu_topup_memory_cache(struct 
>> kvm_mmu_memory_cache *mc, int capacity,
>>           if (WARN_ON_ONCE(!capacity))
>>               return -EIO;
>>   +        /*
>> +         * Custom init values can be used only for page allocations,
>> +         * and obviously conflict with __GFP_ZERO.
>> +         */
>> +        if (WARN_ON_ONCE(mc->init_value && (mc->kmem_cache || 
>> mc->gfp_zero)))
>> +            return -EIO;
>> +
>>           mc->objects = kvmalloc_array(capacity, sizeof(void *), gfp);
>>           if (!mc->objects)
>>               return -ENOMEM;
>
>