linux-kernel - Re: [PATCH v3 11/15] crypto: x86/aes-kl - Support AES algorithm using Key Locker instructions

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 2 Dec 2021 15:21:22 +0100
From:   Peter Zijlstra <peterz@...radead.org>
To:     "Chang S. Bae" <chang.seok.bae@...el.com>
Cc:     tglx@...utronix.de, bp@...e.de, dave.hansen@...ux.intel.com,
        mingo@...nel.org, luto@...nel.org, x86@...nel.org,
        herbert@...dor.apana.org.au, linux-kernel@...r.kernel.org,
        linux-crypto@...r.kernel.org, dan.j.williams@...el.com,
        charishma1.gairuboyina@...el.com, kumar.n.dwarakanath@...el.com,
        lalithambika.krishnakumar@...el.com, ravi.v.shankar@...el.com
Subject: Re: [PATCH v3 11/15] crypto: x86/aes-kl - Support AES algorithm
 using Key Locker instructions

On Wed, Nov 24, 2021 at 12:06:56PM -0800, Chang S. Bae wrote:
> +	encodekey256 %eax, %eax

So this thing uses the fancy new keylocker instructions, however:

> diff --git a/crypto/Kconfig b/crypto/Kconfig
> index 285f82647d2b..784a04433549 100644
> --- a/crypto/Kconfig
> +++ b/crypto/Kconfig
> @@ -1113,6 +1113,50 @@ config CRYPTO_AES_NI_INTEL
>  	  ECB, CBC, LRW, XTS. The 64 bit version has additional
>  	  acceleration for CTR.
>  
> +config CRYPTO_AES_KL
> +	tristate "AES cipher algorithms (AES-KL)"
> +	depends on (LD_VERSION >= 23600) || (LLD_VERSION >= 120000)
> +	depends on DM_CRYPT
> +	select X86_KEYLOCKER
> +	select CRYPTO_AES_NI_INTEL


There is no dependency on the compiler actually supporting them..

config AS_HAS_KEYLOCKER
	def_bool $(as-instr,encodekey256)

	depends on AS_HAS_KEYLOCKER

Hmm?


> +
> +	help
> +	  Key Locker provides AES SIMD instructions (AES-KL) for secure
> +	  data encryption and decryption. While this new instruction
> +	  set is analogous to AES-NI, AES-KL supports to encode an AES
> +	  key to an encoded form ('key handle') and uses it to transform
> +	  data instead of accessing the AES key.
> +
> +	  The setkey() transforms an AES key to a key handle, then the AES
> +	  key is no longer needed for data transformation. A user may
> +	  displace their keys from possible exposition.
> +
> +	  This key encryption is done by the CPU-internal wrapping key. The
> +	  x86 core code loads a new random key at every boot time and
> +	  restores it from deep sleep states. This wrapping key support is
> +	  provided with X86_KEYLOCKER.
> +
> +	  AES-KL supports 128-/256-bit keys only. While giving a 192-bit
> +	  key does not return an error, as AES-NI is chosen to process it,
> +	  the claimed security property is not available with that.
> +
> +	  GNU binutils version 2.36 or above and LLVM version 12 or above
> +	  are assemblers that support AES-KL instructions.
> +
> +	  Bare metal disk encryption is the preferred use case. Make it
> +	  depend on DM_CRYPT.
> +
> +	  This selection enables an alternative crypto cipher for
> +	  cryptsetup, e.g. "capi:xts-aes-aeskl-plain", to use with dm-crypt
> +	  volumes. It trades off raw performance for reduced clear-text key
> +	  exposure and has an additional failure mode compared to AES-NI.
> +	  See Documentation/x86/keylocker.rst for more details. Key Locker
> +	  usage requires explicit opt-in at cryptsetup time. So, select it
> +	  if unsure.
> +
> +	  See also the CRYPTO_AES_NI_INTEL description for more about the
> +	  AES cipher algorithm.
> +
>  config CRYPTO_AES_SPARC64
>  	tristate "AES cipher algorithms (SPARC64)"
>  	depends on SPARC64
> -- 
> 2.17.1
>