On Fri May 15 20, Jarkko Sakkinen wrote:
On Thu, May 14, 2020 at 08:44:23PM -0700, James Bottomley wrote:
On Fri, 2020-05-15 at 05:22 +0300, Jarkko Sakkinen wrote:
> On Thu, 2020-05-14 at 17:31 +0300, Jarkko Sakkinen wrote:
> > I'm compiling now kernel with all series included.
> >
> > Kind of checking if I could just take the whole series. Let see.
> >
> > In all cases I want the style errors in 3/8 to be fixes with a
> > helper
> > but maybe better to hold before sending anything. Possibly that is
> > all
> > needed I'll just carve that patch myself.
> >
> > Please don't do anything for the moment.
>
> This is what I tried first (with the full series applied):
>
> #!/bin/sh
>
> die()
> {
> keyctl clear @u
> ./tpm2-flush --all-transient
> exit $1
> }
>
> KEYHANDLE=$(./tpm2-root-key || die 1)
> KEYID=$(keyctl add trusted kmk "new 32 keyhandle=$KEYHANDLE
> hash=sha256" @u || die 1)
>
> echo "$KEYID ($KEYHANDLE)"
>
> keyctl pipe $KEYID > blob.hex || die 1
> keyctl clear @u || die 1
>
> echo "Import key from blob"
>
> keyctl add trusted kmk "load `cat blob.hex` keyhandle=$KEYHANDLE" @u
> || die 1
>
> die 0
>
> Result:
>
> sudo ./keyctl-smoke.sh
> 566201053 (0x80000000)
> keyctl_read_alloc: Permission denied
I get keyctl_read_alloc -EPERM when I 'sudo su' and try to play with keyctl print.
If I 'sudo su -' and then try it works as expected. Also works for normal user.
Well, it's clearly failing in keyctl pipe
I do confess to never having tested a volatile primary, but I just did
so and it works for me. I will also add the keyhandle in the load
isn't necessary, because it should be in the blob, but there should
also be no harm (just tested).
However, I don't have keyctl_read_alloc in my tree, so it may be an
incompatibility with another patch set. What's your base and what
other patches do you have applied?
http://git.infradead.org/users/jjs/linux-tpmdd.git
Or exactly:
git://git.infradead.org/users/jjs/linux-tpmdd.git (master)
/Jarkko
