Re: [RFCv2 04/13] x86/kvm: Use bounce buffers for KVM memory protection

From: Dave Hansen <dave.hansen@intel.com>
To: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	Andy Lutomirski <luto@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Sean Christopherson <seanjc@google.com>,
	Jim Mattson <jmattson@google.com>
Cc: David Rientjes <rientjes@google.com>,
	"Edgecombe, Rick P" <rick.p.edgecombe@intel.com>,
	"Kleen, Andi" <andi.kleen@intel.com>,
	"Yamahata, Isaku" <isaku.yamahata@intel.com>,
	Erdem Aktas <erdemaktas@google.com>,
	Steve Rutherford <srutherford@google.com>,
	Peter Gonda <pgonda@google.com>,
	David Hildenbrand <david@redhat.com>,
	x86@kernel.org, kvm@vger.kernel.org, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Subject: Re: [RFCv2 04/13] x86/kvm: Use bounce buffers for KVM memory protection
Date: Fri, 16 Apr 2021 09:21:33 -0700	[thread overview]
Message-ID: <2ea287f6-9e2f-1607-c9d6-8c985438f989@intel.com> (raw)
In-Reply-To: <20210416154106.23721-5-kirill.shutemov@linux.intel.com>

On 4/16/21 8:40 AM, Kirill A. Shutemov wrote:
> Mirror SEV, use SWIOTLB always if KVM memory protection is enabled.
...
>  arch/x86/mm/mem_encrypt.c          | 44 ---------------------------
>  arch/x86/mm/mem_encrypt_common.c   | 48 ++++++++++++++++++++++++++++++

The changelog need to at least mention what's going on here.  It doesn't
prepare me at all for having code move around.

> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> index d197b3beb904..c51d14db5620 100644
> --- a/arch/x86/Kconfig
> +++ b/arch/x86/Kconfig
> @@ -812,6 +812,7 @@ config KVM_GUEST
>  	select ARCH_CPUIDLE_HALTPOLL
>  	select X86_HV_CALLBACK_VECTOR
>  	select X86_MEM_ENCRYPT_COMMON
> +	select SWIOTLB
>  	default y
>  	help
>  	  This option enables various optimizations for running under the KVM

So, this feature is always compiled in with KVM.  Could you say a couple
of things about that?  Why did you decide not have a Kconfig option for it?

> diff --git a/arch/x86/include/asm/mem_encrypt.h b/arch/x86/include/asm/mem_encrypt.h
> index 31c4df123aa0..a748b30c2f23 100644
> --- a/arch/x86/include/asm/mem_encrypt.h
> +++ b/arch/x86/include/asm/mem_encrypt.h
> @@ -47,10 +47,8 @@ int __init early_set_memory_encrypted(unsigned long vaddr, unsigned long size);
>  
>  void __init mem_encrypt_free_decrypted_mem(void);
>  
> -/* Architecture __weak replacement functions */
> -void __init mem_encrypt_init(void);
> -
>  void __init sev_es_init_vc_handling(void);
> +
>  bool sme_active(void);
>  bool sev_active(void);
>  bool sev_es_active(void);
> @@ -91,6 +89,9 @@ static inline void mem_encrypt_free_decrypted_mem(void) { }
>  
>  #endif	/* CONFIG_AMD_MEM_ENCRYPT */
>  
> +/* Architecture __weak replacement functions */
> +void __init mem_encrypt_init(void);

FWIW, I'd rather have the code movement in separate patches from the
functional changes.

>  /*
>   * The __sme_pa() and __sme_pa_nodebug() macros are meant for use when
>   * writing to or comparing values from the cr3 register.  Having the
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index aed6034fcac1..ba179f5ca198 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -26,6 +26,7 @@
>  #include <linux/kprobes.h>
>  #include <linux/nmi.h>
>  #include <linux/swait.h>
> +#include <linux/swiotlb.h>
>  #include <asm/timer.h>
>  #include <asm/cpu.h>
>  #include <asm/traps.h>
> @@ -765,6 +766,7 @@ static void __init kvm_init_platform(void)
>  		pr_info("KVM memory protection enabled\n");
>  		mem_protected = true;
>  		setup_force_cpu_cap(X86_FEATURE_KVM_MEM_PROTECTED);
> +		swiotlb_force = SWIOTLB_FORCE;
>  	}
>  }
>  
> diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c
> index c2cfa5e7c152..814060a6ceb0 100644
> --- a/arch/x86/kernel/pci-swiotlb.c
> +++ b/arch/x86/kernel/pci-swiotlb.c
> @@ -13,6 +13,7 @@
>  #include <asm/dma.h>
>  #include <asm/xen/swiotlb-xen.h>
>  #include <asm/iommu_table.h>
> +#include <asm/kvm_para.h>
>  
>  int swiotlb __read_mostly;
>  
> @@ -49,7 +50,7 @@ int __init pci_swiotlb_detect_4gb(void)
>  	 * buffers are allocated and used for devices that do not support
>  	 * the addressing range required for the encryption mask.
>  	 */
> -	if (sme_active())
> +	if (sme_active() || kvm_mem_protected())
>  		swiotlb = 1;
>  
>  	return swiotlb;

While I don't doubt you got it right, it would be nice to also explain
in the changelog why you manipulate both 'swiotlb_force' and 'swiotlb'.

> diff --git a/arch/x86/mm/mem_encrypt.c b/arch/x86/mm/mem_encrypt.c
> index 9ca477b9b8ba..3478f20fb46f 100644
> --- a/arch/x86/mm/mem_encrypt.c
> +++ b/arch/x86/mm/mem_encrypt.c
> @@ -409,47 +409,3 @@ void __init mem_encrypt_free_decrypted_mem(void)
>  
>  	free_init_pages("unused decrypted", vaddr, vaddr_end);
>  }
> -
> -static void print_mem_encrypt_feature_info(void)
> -{
> -	pr_info("AMD Memory Encryption Features active:");
> -
> -	/* Secure Memory Encryption */
> -	if (sme_active()) {
> -		/*
> -		 * SME is mutually exclusive with any of the SEV
> -		 * features below.
> -		 */
> -		pr_cont(" SME\n");
> -		return;
> -	}
> -
> -	/* Secure Encrypted Virtualization */
> -	if (sev_active())
> -		pr_cont(" SEV");
> -
> -	/* Encrypted Register State */
> -	if (sev_es_active())
> -		pr_cont(" SEV-ES");
> -
> -	pr_cont("\n");
> -}
> -
> -/* Architecture __weak replacement functions */
> -void __init mem_encrypt_init(void)
> -{
> -	if (!sme_me_mask)
> -		return;
> -
> -	/* Call into SWIOTLB to update the SWIOTLB DMA buffers */
> -	swiotlb_update_mem_attributes();
> -
> -	/*
> -	 * With SEV, we need to unroll the rep string I/O instructions.
> -	 */
> -	if (sev_active())
> -		static_branch_enable(&sev_enable_key);
> -
> -	print_mem_encrypt_feature_info();
> -}
> -
> diff --git a/arch/x86/mm/mem_encrypt_common.c b/arch/x86/mm/mem_encrypt_common.c
> index 6bf0718bb72a..351b77361a5d 100644
> --- a/arch/x86/mm/mem_encrypt_common.c
> +++ b/arch/x86/mm/mem_encrypt_common.c
> @@ -11,6 +11,7 @@
>  #include <linux/mem_encrypt.h>
>  #include <linux/dma-direct.h>
>  #include <asm/kvm_para.h>
> +#include <asm/mem_encrypt.h>
>  
>  /* Override for DMA direct allocation check - ARCH_HAS_FORCE_DMA_UNENCRYPTED */
>  bool force_dma_unencrypted(struct device *dev)
> @@ -37,3 +38,50 @@ bool force_dma_unencrypted(struct device *dev)
>  
>  	return false;
>  }
> +
> +static void print_mem_encrypt_feature_info(void)
> +{
> +	if (kvm_mem_protected()) {
> +		pr_info("KVM memory protection enabled\n");
> +		return;
> +	}

I understand that they're touching similar areas of code, but I'm a bit
unnerved with memory protection being in all these "encryption"
functions and files.

I think some thoughtful renaming is in order.

> +	pr_info("AMD Memory Encryption Features active:");
> +
> +	/* Secure Memory Encryption */
> +	if (sme_active()) {
> +		/*
> +		 * SME is mutually exclusive with any of the SEV
> +		 * features below.
> +		 */
> +		pr_cont(" SME\n");
> +		return;
> +	}
> +
> +	/* Secure Encrypted Virtualization */
> +	if (sev_active())
> +		pr_cont(" SEV");
> +
> +	/* Encrypted Register State */
> +	if (sev_es_active())
> +		pr_cont(" SEV-ES");
> +
> +	pr_cont("\n");
> +}

This, for instance really shouldn't be in common code.  It should be in
an AMD-specific area.

> +void __init mem_encrypt_init(void)
> +{
> +	if (!sme_me_mask && !kvm_mem_protected())
> +		return;
> +
> +	/* Call into SWIOTLB to update the SWIOTLB DMA buffers */
> +	swiotlb_update_mem_attributes();
> +
> +	/*
> +	 * With SEV, we need to unroll the rep string I/O instructions.
> +	 */
> +	if (sev_active())
> +		static_branch_enable(&sev_enable_key);
> +
> +	print_mem_encrypt_feature_info();
> +}

This function is called like this:

>         /*
>          * This needs to be called before any devices perform DMA
>          * operations that might use the SWIOTLB bounce buffers. It will
>          * mark the bounce buffers as decrypted so that their usage will
>          * not cause "plain-text" data to be decrypted when accessed.
>          */
>         mem_encrypt_init();

So, maybe this should be x86_swiotlb_init() or something.  Then, move
the print_mem_encrypt_feature_info() elsewhere, probably back out to
mem_init().  Maybe even just call it print_arch_mem_features() or something.