TPM 2.0 issues a retry return code if the TPM cannot currently process
the command. Today's use case is an NV write that is rate limited to
prevent wear out. There could be others.
Architecturally, where should this be handled? Is it handled today?
Where was it handled for TPM 1.2?
- device driver
- kernel resource manager
- user space TSS
- user space application
My opinion is that the user space TSS should handle it, but I'm very
open to other opinions.
- If in the device driver, other applications would block
- The resource manager would be OK, but perhaps is complexity that we
don't want to add.
- It seems to be a burden on applications that the TSS could handle once.
