TPM event logs

When a Trusted Platform Module (TPM) Platform Configuration Register (PCR) Extend operation is performed, an event log entry is recorded in a TPM Event log file. This log file is used by external entities that depend on remote attestation and by host firmware during multi-node synchronization. The log files are used to reconstruct and validate the PCR values against known values. The event log files are not maintained by the TPM. Thus, the firmware must provide storage for the log files and provide interfaces to update the log files on PCR Extends and access the log files for attestation purposes.

Because the initial PCR Extend operations are performed by the Host Boot (HB) code, when Power Hypervisor (PHYP) is started, the event log information that is associated with the initial program load (IPL) time Extend operations are saved in the HB code. The HB code also communicates the relevant event log entries to PHYP through the host data area (HDAT) structure.

The PHYP maintains the TPM event log information in the physical TPM (pTPM) adjunct state. A maximum of 64 MB of storage area is allotted for each TPM log file. Preference is given to log entries that are created for concurrent firmware updates (also known as, unbounded log entries). Low and mid-range platforms have a single pTPM per node. Multinode enterprise platforms have another (redundant) pTPM per node.

If a TPM log buffer is full, additional PCR Extend operations to the TPM are allowed. The truncation of the log file is recorded and the attestation interfaces receive a flag that indicates that the delivered log files have been truncated.

At the time of initial program load (IPL), PCR Extend operations that have appropriate event log information are created. PCR Extend operations are also created for concurrent firmware updates. The log files include code measurements and configuration and platform history.

Current estimates on a first (cold) IPL are 50 event records per node on a single node system, and 200 event records per node on a four node system (at 128 B per event record, this rate is 25 KB per node per IPL).

The TPM event log information can be obtained through a resource dump. The event log files are NOT migrated with logical partitions because the log files are associated with the physical platform. Thus, the TPM history of the physical TPMs (pTPMs) might not be the same as the TPM history of the logical partitions.

TPM configuration settings that do not require a node TPM during a first (cold) or subsequent (warm) IPL do not require event log files. However, if the TPM Required option is set, PCR Extend operations and associated event log files must be maintained.