# PE/PCR4 verification fails: firmware emits a SHA-1-only TCG event log while the SHA-256 PCR bank is active

## Summary

On this machine, IncusOS fails very early under Secure Boot + TPM with a PCR4
mismatch. The root cause looks like the platform firmware records a **SHA-1-only**
TCG event log, even though the TPM's **SHA-256 PCR bank is active and populated**.
IncusOS's PE binary / PCR4 verification operates on SHA-256, so it has no SHA-256
event data to reconstruct the expected PCR4 from, and `computed != actual`.

## Environment

- **IncusOS build:** `IncusOS_202607060039` (image generated with the web
  customizer, written to a USB stick, installed on physical hardware)
- **Hardware:** `HP ProDesk 600 G2 SFF`
- **BIOS:** `Var.02.36`(latest)
- **TPM:** `Infineon SLB9670 TPM 2.0 ver. 7.63.3353.0`
- **Secure Boot:** enrolled via the HP BIOS "Clear Secure Boot keys" option
  (Setup Mode). IncusOS enrolled its own keys; the binary passes the firmware
  Secure Boot gate and then fails at IncusOS's *own* PE/PCR4 verification.

## Error shown on screen

```
IncusOS failed to verify PE binaries: ERROR: computed PCR4 (12771355e46cd47c71ed1721fd5319b383cca3a1f9fce3aa1c8cd3bd37af20d7) doesn't match actual value (de62c5b65745b270487978ceae097095d915ba971c36202036d9ad4097859046)
```

Both digests are 32 bytes (SHA-256), consistent with verification running against
the SHA-256 bank.

## Diagnosis (Ubuntu live session on the same hardware)

**1. The firmware event log contains SHA-1 digests only.** Every event carries a
single digest with `AlgorithmId: sha1`:

```
EventType: ...
DigestCount: 1
Digests:
- AlgorithmId: sha1
  Digest: "..."
```

There is not a single SHA-256 digest anywhere in the log. (The Spec ID / header at
the top of `eventlog.txt` shows the log's declared digest set — see attachment.)

**2. The SHA-1 log is internally faithful** — replaying it reproduces the live
SHA-1 PCRs exactly:

| PCR | tpm2_eventlog replay (sha1)                 | tpm2_pcrread sha1                          |
|-----|---------------------------------------------|--------------------------------------------|
| 4   | `06f109f35a2ef2494fd69c33e3a9a3749e7dca9e`  | `06F109F35A2EF2494FD69C33E3A9A3749E7DCA9E` |
| 7   | `afdf621a76da76a5915818bb1c3de5aa15fc7e01`  | `AFDF621A76DA76A5915818BB1C3DE5AA15FC7E01` |

Exact match — the log is not corrupt, it simply only covers the SHA-1 bank.

**3. The TPM's SHA-256 bank is active and populated** (values from the Ubuntu boot,
so they naturally differ from the IncusOS-boot PCR4 above — included only to show
the bank is in use):

```
tpm2_pcrread sha256:4,7
  4 : 0x6DC1146060584796A14D18B10DF39038E6125FCE324D01C47D9A0384C38C1304
  7 : 0x7D72C56F5D3F02DA45414DAD72D35CFFDD8E4410C91FC5D1D63CC04ADDE72C12
```

So: SHA-256 PCRs are being extended, but the firmware never writes SHA-256
digests into its event log. IncusOS therefore cannot rebuild the SHA-256 PCR4
from the log.

## Attachments

- `eventlog.txt` — `tpm2_eventlog` output (note the header declaring SHA-1)

## Notes

- Happy to capture the IncusOS-side log via
  `incus query /os/1.0/debug/secureboot/event-log` if useful, but the box can't
  boot IncusOS far enough with Secure Boot on. I can install the same build with
  Secure Boot off (TPM-only) to reach that endpoint if you want it, with the
  caveat that PCR7 / the measured chain differ with SB off.
- Possibly related to #813 / #824 / #909 (per-firmware PE/PCR verification
  handling), but the specific symptom here is a **SHA-1-only event log combined
  with an active SHA-256 PCR bank**.