[PATCH v2 4/6] KEYS: trusted: Move tpm2_key_decode() to the TPM driver

[Jarkko Sakkinen]

On Wed May 22, 2024 at 1:45 AM EEST, Jarkko Sakkinen wrote:
> On Wed May 22, 2024 at 12:59 AM EEST, James Bottomley wrote:
> > On Tue, 2024-05-21 at 22:44 +0100, David Howells wrote:
> > > Jarkko Sakkinen <jarkko at kernel.org> wrote:
> > > 
> > > > On Tue May 21, 2024 at 9:18 PM EEST, James Bottomley wrote:
> > > > ...
> > > > You don't save a single byte of memory with any constant that
> > > > dictates the size requirements for multiple modules in two disjoint
> > > > subsystems.
> > > 
> > > I think James is just suggesting you replace your limit argument with
> > > a constant not that you always allocate that amount of memory.
> >
> > Exactly.  All we use it for is the -E2BIG check to ensure user space
> > isn't allowed to run away with loads of kernel memory.
>
> Not true.
>
> It did return -EINVAL. This patch changes it to -E2BIG.
>
> >
> > > What the limit should be, OTOH, is up for discussion, but PAGE_SIZE
> > > seems not unreasonable.
> >
> > A page is fine currently (MAX_BLOB_SIZE is 512).  However, it may be
> > too small for some of the complex policies when they're introduced. 
> > I'm not bothered about what it currently is, I just want it to be able
> > to be increased easily when the time comes.
>
> MAX_BLOB_SIZE would be used to cap key blob, not the policy.
>
> And you are ignoring it yourself too in the driver.

Obviously policy is part of the key blob i.e. expected value for that.

... but that does not reduce space requirements to rsa asymmetric keys.
It increases them but I think at this point 8192 is good starting point.
And it cap can be scaled later.

Being a parameter also allows to have even kernel-command line or sysfs
parameter and stuff like that. It is robust not a bad choice.

BR, Jarkko