On Thu, Mar 27, 2025 at 01:57:48PM +0100, Michal Suchánek wrote: > This does not address the other problem that when the rimeout triggers > on something like LOAD CONTEXT then the kernel and the TPM don't agree > on what the state of the context is.
In the case of tpm2-sessions, and in-kernel callers, we could try to improve the situation in some of the edge cases. Out-of-my sleeve ideas: 1. E.g., tpm2_load_null() could potentially have retries instead of single trial. 2. Perhaps timeout in tpm_tis should be adaptive? It could be by defaault 2s and would grow on trials. > > Thanks > > Michal BR, Jarkko
