From: Coly Li Date: Sun, 16 Aug 2020 17:22:26 +0000 Subject: Re: [PATCH RESEND] docs: update trusted-encrypted.rst Message-Id: <90ce83bf-3547-7ad4-ea6d-40e2e77e29dd@suse.de> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit List-Id: References: <20200815075143.47082-1-colyli@suse.de> <096636f4-a928-dd00-dba6-216651c3d63b@suse.de> <1597597699.8344.11.camel@linux.ibm.com> In-Reply-To: <1597597699.8344.11.camel@linux.ibm.com> To: jejb@linux.ibm.com Cc: Stefan Berger , keyrings@vger.kernel.org, linux-kernel@vger.kernel.org, Dan Williams , Jarkko Sakkinen , Mimi Zohar On 2020/8/17 01:08, James Bottomley wrote: > On Mon, 2020-08-17 at 01:01 +0800, Coly Li wrote: >> On 2020/8/17 00:06, Stefan Berger wrote: >>> On 8/15/20 3:51 AM, Coly Li wrote: > [...] >>>> Usage:: >>>> @@ -115,7 +114,7 @@ append 'keyhandle=0x81000001' to statements >>>> between quotes, such as >>> >>> >>> A note in this file states this: >>> >>> Note: When using a TPM 2.0 with a persistent key with handle >>> 0x81000001, append 'keyhandle=0x81000001' to statements between >>> quotes, such as "new 32 keyhandle=0x81000001". >>> >>> Now if someone was (still) interested in TPM 1.2 then the below >>> changes you are proposing wouldn't work for them. Maybe you should >>> adapt the note to state that these keyhandle=... should be removed >>> for the TPM 1.2 case. >>> >> >> I agree. Indeed I have no idea why number 0x81000001 is used, and I >> don't have practice experience with TPM 1.2. Now the purpose of this >> patch accomplished: experts response and confirm my guess :-) > > It was the conventional persistent value for the RSA 2048 version of > the primary storage seed. Originally the PC spec required the > manufacturer provision this on all TPM 2.0 based PC class systems. > Unfortunately in spite of it being in the Windows Hardware guide no > manufacturer ever did, meaning you either have to create it yourself or > do something different. Because of usability problems, every consumer > of TPM key function has opted to do something different, namely derive > the EC primary if no parent is specified. Aha, thanks for the hint :-) My motivation is for the NVDIMM security with TPM 2.0 chip on x86 server (Lenovo SR650). To automatically load a trusted key, I encounter the outdated command line in trusted-encrypted.rst. From your response, it seems 0x81000001 is still a working value that I can recommend to other people who want to encrypt/decrypt their NVDIMM banks. Coly Li