On Fri, Oct 04, 2019 at 11:24:34AM -0700, Jerry Snitselaar wrote: > On Fri Oct 04 19, James Bottomley wrote: > > On Fri, 2019-10-04 at 13:37 -0400, Mimi Zohar wrote: > > > On Fri, 2019-10-04 at 09:37 -0700, James Bottomley wrote: > > > > On Thu, 2019-10-03 at 21:51 +0300, Jarkko Sakkinen wrote: > > > > > As has been seen recently, binding the buffer allocation and > > > > > tpm_buf > > > > > together is sometimes far from optimal. > > > > > > > > Can you elaborate on this a bit more? I must have missed the > > > > discussion. > > > > > > Refer to e13cd21ffd50 ("tpm: Wrap the buffer from the caller to > > > tpm_buf in tpm_send()") for the details. > > > > Yes, I get that, but to my mind that calls for moving the > > tpm_init/destroy_buf into the callers of tpm_send (which, for the most > > part, already exist), which means there's no need to separate the buf > > and data lifetimes. > > > > James > > > > Sumit has been working on a patchset that does this. His patchset > converts both the asymmetric keys and trusted keys code to using the > tpm_buf manipulation functions.
And it is also in a shape that it can soon be merged (within few iterations at most). /Jarkko