initrd: add TPM DA lockout gating, counter auth fix, bad_auth tooling and documentation by tlaurion · Pull Request #2117 · linuxboot/heads

tlaurion · 2026-05-14T19:09:04Z

Summary

Fixes a TPM1 counter auth regression (PR #2068) where increment_tpm_counter was changed from hardcoded -pwdc '' (empty counter auth) to -pwdc "${tpm_passphrase:-}" (owner passphrase), while counters continued to be created with -pwdc "". This caused every increment to compute SHA1(owner_pass) against a counter created with SHA1(""), producing persistent TPM_AUTHFAIL.

Per TCG TPM Main Spec Part 3, TPM_CreateCounter uses owner auth (-pwdo) but TPM_IncrementCounter uses the counter's own authData, not the owner password. The correct design for Heads' rollback counter is empty auth.

The repeated auth failures (3 per boot) triggered TPM 1.2 dictionary-attack lockout (TPM_DEFEND_LOCK_RUNNING), which persisted through forceclear on some implementations.

Adds DA state diagnostics, preflight guard and testing tooling.

Changes

`initrd/bin/tpmr.sh` — auth fix + DA state + bad_auth

tpm1_counter_increment(): detect -pwdc '', call tpm directly (bypass _tpm_auth_retry)
tpm1_reset(): detect defend lock, cycle physical presence, retry takeown
tpm1_da_state(): TPM1 DA via TPM_CAP_DA_LOGIC, output DA: line
tpm2_da_state(): TPM2 DA via getcap properties-variable, unlock estimate
tpm1_bad_auth() / tpm2_bad_auth(): deliberate wrong-auth for testing
Add 'defend' and '0x98e|0x149' to auth detection patterns

`initrd/etc/functions.sh` — counter auth + DA preflight

check_tpm_counter(): -pwdc '' per TCG spec
increment_tpm_counter(): empty auth first, migration fallback with owner passphrase
DA preflight guard: DIE on lockout, WARN near threshold, DEBUG at each decision
Remove TPM1 owner-passphrase prompt from PR UX improvements: TPM reseal (HOTP/TOTP/DUK) adds integrity report; detects disk/tpm swap and guide user into action, add terminal colors and guidance! Reduced quiet noise #2068

Other

initrd/bin/tpm-reset.sh: TPM reset frontend
initrd/bin/oem-factory-reset.sh: -pwdc '' for consistency
doc/tpm.md: DA diagnosis, testing, escalation, physical presence

Copilot

Pull request overview

This PR improves Heads’ TPM error handling for TPM1 by correctly classifying tpmtotp “Defend lock running” output as an authorization-related failure (so it doesn’t immediately hard-fail), and adds a recovery path in tpm1_reset() to attempt clearing TPM1 defend-lock after forceclear by cycling physical presence.

Changes:

Extend auth-failure grep patterns to include defend (and unify inclusion of TPM2 auth hex codes in the shared retry helper).
Enhance tpm1_reset() to detect “defend lock” after takeown and retry after cycling physical presence.
Expand TPM documentation to describe tool selection, auth retry detection, and TPM1 defend-lock behavior.

Reviewed changes

Copilot reviewed 1 out of 2 changed files in this pull request and generated 3 comments.

File	Description
initrd/bin/tpmr.sh	Treat “defend lock” as auth-related and add TPM1 defend-lock recovery logic during reset.
doc/tpm.md	Document TPM toolchain selection and the updated auth retry / defend-lock behavior.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot

Pull request overview

Copilot reviewed 1 out of 2 changed files in this pull request and generated 2 comments.

Copilot

Pull request overview

Copilot reviewed 1 out of 2 changed files in this pull request and generated 1 comment.

Copilot

Pull request overview

Copilot reviewed 1 out of 2 changed files in this pull request and generated no new comments.

notgivenby · 2026-05-15T05:50:02Z

The fix did not work…will copy logs.

tlaurion · 2026-05-16T01:04:25Z

The fix did not work…will copy logs.

Found the bug and where the regression comes from. Damn that one was not easy. Pushing fix and updating the other pr

Copilot

Pull request overview

Copilot reviewed 2 out of 3 changed files in this pull request and generated 1 comment.

…iew fixes Add preflight dictionary attack (DA) lockout guard to increment_tpm_counter, querying da_state before every counter increment. DIE on active lockout, WARN when count nears threshold. Add tpm1_da_state and tpm2_da_state for unified DA state query: - TPM1: reads TPM_CAP_DA_LOGIC (0x19); actionDependValue>0 + state=1 = locked; DA: line timer=field maps to actionDependValue - TPM2: reads getcap properties-variable (no single DA query); count >= max_auth = locked; estimate = (counter - max_auth + 1) * interval; DA: line timer=present when locked, absent when clean - Both output machine-parsable DA: line for the preflight guard Fix da_timer sed pattern: use sed -n with /p so da_timer stays empty when DA: line has no timer= field (TPM2 count < threshold). Without -n, non-matching lines echoed the full line as da_timer. Add tpm1_bad_auth and tpm2_bad_auth for testing: - Uses NV index auth (-P) not owner auth (-C o -P) for TPM2 because NV auth failure produces TPM2_RC_AUTH_FAIL (0x98e) and increments LOCKOUT_COUNTER; owner auth may not increment on some TPM2 implementations - Intentionally wrong password TPM_DEFEND_LOCK_TEST_WRONG_PASSWORD - Shows DA state before and after; uses || true to survive set -e Add tpm-reset.sh as a TPM reset frontend via tpmr.sh wrapper. Add DA documentation in doc/tpm.md covering diagnosis, testing, escalation, TPM1 vs TPM2 DA parameter configurability. Review fixes: - Fix TPM2 property names: TPM_PT_ -> TPM2_PT_ prefix in doc/tpm.md - Remove misplaced design decision comment from tpm2_da_state (belongs in tpm2_bad_auth where it already exists) - Add DEBUG logging at every decision point across preflight guard and all DA state functions for runtime traceability - Document design decisions inline: timer logic, estimate formula, empty auth retry bypass, NV vs owner auth Signed-off-by: Thierry Laurion <insurgo@riseup.net>

Copilot

Pull request overview

Copilot reviewed 2 out of 5 changed files in this pull request and generated 3 comments.

- Fix TPM2 time remaining table entry: estimate is derived from LOCKOUT_COUNTER vs MAX_AUTH_FAIL times LOCKOUT_INTERVAL, not LOCKOUT_RECOVERY (which is the lockout-auth-blocked-after-failure timer, not the remaining-until-unlock) - Reword migration WARN: 'older Heads version' not 'older firmware' (the migration case is caused by previous Heads code, not platform firmware) - Remove fragile PR #2117 reference from preventing-future-lockouts section: describe the fix generically (restoring empty counter auth) so the doc is correct regardless of branch context Signed-off-by: Thierry Laurion <insurgo@riseup.net>

When TPM1 does not support TPM_CAP_DA_LOGIC (0x19), tpm getcapability may print raw TSS error text to stdout instead of returning empty. The empty-string check missed this because error text is non-empty. Fix: check that the output contains the 'State' field (expected for valid DA capability data) before echoing. If missing, return 'unavailable' and suppress the raw TSS garbage. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

TPM_CAP_DA_LOGIC (0x19) was added late in TPM 1.2 spec rev 103. Older Infineon TPMs (X230-era SLB9635/9645) and some Atmel chips do not implement it and return TPM_BAD_MODE (exit 44). Document this limitation in: - tpm1_da_state function comment: specific TPM models affected - increment_tpm_counter preflight guard: note that da_state may return unavailable on older TPM1 - doc/tpm.md: explain why some TPM1 hardware shows 'unavailable' Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…able When TPM_CAP_DA_LOGIC fails, also query TPM_CAP_VERSION_VAL (0x06) and log the TPM firmware version to debug.log. This helps identify which TPM1 chips lack rev 103+ DA capability support. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…iagnostics Log TPM2 firmware version (TPM2_PT_FIRMWARE_VERSION_1, decoded as major.minor) to debug.log in tpm2_da_state. Consistent with the TPM1 version probe already added to tpm1_da_state. This helps identify TPM chips and firmware revisions that may have limited or missing DA capability support. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…o 0x06) TPM_CAP_VERSION_VAL (0x1a) returns the full version info structure but, like TPM_CAP_DA_LOGIC (0x19), may be unsupported on older TPM1 chips. TPM_CAP_VERSION (0x06) returns a basic uint32 encoding and is more widely supported. Try 0x1a first, fall back to 0x06. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

DA: line always emits timer= field (with empty value when unavailable), not conditionally. Fix comment to match actual behavior. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…ry operator expected' The mount_usb() function used (...) to group the fallback assignment, creating a subshell. inside a subshell does not propagate, so USB_FAILED stayed unset when mount-usb.sh failed, producing the error at line 24. Fix: use { } brace grouping (no subshell) so USB_FAILED=1 is assigned in the current shell. Also fix the same pattern on line 27 (second mount attempt). Signed-off-by: Thierry Laurion <insurgo@riseup.net>

exit 1 inside { } brace grouping actually exits the script (unlike the original ( ) subshell where it was silently swallowed). Change to return 1 so aborting USB selection returns from mount_usb() gracefully instead of killing the entire script. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…ailable On X230-era TPM1 chips that lack TPM_CAP_DA_LOGIC, 'TPM DA state: unavailable' was too terse. Add user-facing explanation of why the query failed and how lockout can still be detected. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

… unavailable Extract revMajor/revMinor and VendorID from TPM_CAP_VERSION_VAL (0x1a) to report whether the TPM is pre-rev 103 (too old for CAP_DA_LOGIC) or a newer revision where the capability is still unexpectedly absent. User sees TPM spec revision and vendor alongside 'unavailable'. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

revMajor from TPM_CAP_VERSION_VAL is hex (e.g. 0x0D = 13). Convert to decimal before display and comparison, otherwise bash reads 0D as invalid octal and the -lt comparison silently fails. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…lity version requirements Reference TCG TPM Main Specification Part 2 rev 103 as the revision where TPM_CAP_DA_LOGIC was added. Reference TPM_CAP_VERSION_VAL (0x1a) in tpmtotp's getcapability.c as the source of spec version reporting. Update doc/tpm.md with same references. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

Skip TPM_CAP_DA_LOGIC entirely when TPM_CAP_VERSION_VAL reports revMajor < 103 (pre-rev 103 TPMs don't support it). This avoids the unnecessary TPM_BAD_MODE error and one TPM round trip on older TPMs like the STMicroelectronics rev 13 chip on X230. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…, skip migration fallback When da_state is unavailable (pre-rev 103 TPM1) and the counter increment itself fails, check the output for defend lock patterns before attempting the owner-passphrase migration fallback. On defend lock, DIE immediately with TPM reset guidance instead of confusing the user with an owner passphrase prompt. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…d path The preflight guard must run for both TPM1 and TPM2. TPM1 degrades gracefully (no DA: line → skip), TPM2 always works. Keep comments clear about the pre-rev 103 TPM1 limitation. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

When tpm1_da_state returns unavailable (pre-rev 103 TPM1), bad_auth now explains that the DA counter can't be read and the test proceeds by attempting the increment. User is told to repeat bad_auth until the increment itself fails to confirm lockout. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…n bad_auth guidance User-facing messages now say 'TPM 1.2 spec rev' instead of 'TPM spec rev' to avoid confusion with TPM 2.0. bad_auth guidance text rewritten to describe the limitation clearly without raw parenthetical formatting. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…03 TPM1 Per UX best practices: one line per state, no raw codes, explicit 'expected' for intentional failures. da_state shows one-line status with vendor/rev on second line. bad_auth shows action + outcome without redundant DA state dumps. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…er TPM1 Replace technical output ('TPM DA state: unavailable (TPM 1.2 pre-rev 103)') with plain language: 'TPM 1.2 too old to report DA lockout state.' bad_auth now reads: 'testing lockout by attempting increment...' 'Repeat until the increment fails with lockout.' Vendor/rev info moved to DEBUG log only. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

…nd pre-rev 103 fallback Document the TPM1 version-first probe (skip CAP_DA_LOGIC if rev < 103), the preflight guard defend lock fallback for pre-rev 103 TPMs, and the firmware version logging in tpm2_da_state. Update tpmr.sh function comments to reflect current behavior and non-developer UX patterns. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

- Only skip DA_LOGIC when revMajor is a valid hex value strictly < 103 - If revMajor is unparseable (format mismatch), try DA_LOGIC anyway - Use BusyBox-compatible basic regex (no \+ ERE operator) - If DA_LOGIC fails despite rev >= 103, show version in error message - Clearer DEBUG logging at each decision point Signed-off-by: Thierry Laurion <insurgo@riseup.net>

- Add da_state and bad_auth to the subcommand reference table - Fix misleading NVRAM index claim (0x3135106223 is not fixed for TPM2) - Fix bad_auth debug.log claim: increment output goes to console - Fix _tpm_auth_retry comment: 'owner auth' -> 'authorization' (handles counter auth too) Signed-off-by: Thierry Laurion <insurgo@riseup.net>

Copilot AI review requested due to automatic review settings May 14, 2026 19:09

Copilot started reviewing on behalf of tlaurion May 14, 2026 19:09 View session

Copilot AI reviewed May 14, 2026

View reviewed changes

Comment thread doc/tpm.md Outdated

Comment thread doc/tpm.md Outdated

Comment thread doc/tpm.md

tlaurion force-pushed the tpm1_fixes branch 2 times, most recently from b20be90 to a5a2ebb Compare May 14, 2026 19:17

tlaurion mentioned this pull request May 14, 2026

m900 tower port (rebased on tpm1_fixes with DA lockout gating) #2118

Open

tlaurion force-pushed the tpm1_fixes branch from 8d95198 to 072c089 Compare May 14, 2026 20:08

tlaurion requested a review from Copilot May 14, 2026 20:28

Copilot started reviewing on behalf of tlaurion May 14, 2026 20:29 View session

Copilot AI reviewed May 14, 2026

View reviewed changes

Comment thread doc/tpm.md Outdated

Comment thread doc/tpm.md Outdated

tlaurion force-pushed the tpm1_fixes branch from 072c089 to 091435a Compare May 14, 2026 20:35

tlaurion requested a review from Copilot May 14, 2026 20:40

Copilot started reviewing on behalf of tlaurion May 14, 2026 20:41 View session

Copilot AI reviewed May 14, 2026

View reviewed changes

Comment thread doc/tpm.md Outdated

tlaurion force-pushed the tpm1_fixes branch from 091435a to ea5cc85 Compare May 14, 2026 20:50

tlaurion requested a review from Copilot May 14, 2026 20:51

Copilot started reviewing on behalf of tlaurion May 14, 2026 20:52 View session

Copilot AI reviewed May 14, 2026

View reviewed changes

tlaurion force-pushed the tpm1_fixes branch from ea5cc85 to c20cd28 Compare May 16, 2026 01:05

tlaurion requested a review from Copilot May 16, 2026 01:08

tlaurion changed the title ~~initrd/bin/tpmr.sh: fix TPM1 auth failure detection and defend lock recovery~~ initrd: fix TPM1 counter auth regression and defend lock cascade failure May 16, 2026

Copilot started reviewing on behalf of tlaurion May 16, 2026 01:08 View session

tlaurion force-pushed the tpm1_fixes branch from c20cd28 to bf4ce3d Compare May 16, 2026 01:12

Copilot AI reviewed May 16, 2026

View reviewed changes

Comment thread initrd/etc/functions.sh

tlaurion force-pushed the tpm1_fixes branch from bf4ce3d to 8b497e5 Compare May 16, 2026 01:14

tlaurion requested a review from Copilot May 16, 2026 01:16

Copilot started reviewing on behalf of tlaurion May 16, 2026 01:17 View session

tlaurion force-pushed the tpm1_fixes branch from 8b497e5 to b3b3e8c Compare May 16, 2026 01:18

tlaurion requested a review from Copilot May 17, 2026 17:12

Copilot started reviewing on behalf of tlaurion May 17, 2026 17:12 View session

Copilot AI reviewed May 17, 2026

View reviewed changes

Comment thread initrd/etc/functions.sh Outdated

Comment thread doc/tpm.md Outdated

Comment thread doc/tpm.md Outdated

tlaurion changed the title ~~initrd: fix TPM1 counter auth regression and defend lock cascade failure~~ initrd: add TPM DA lockout gating, counter auth fix, bad_auth tooling and documentation May 17, 2026

tlaurion marked this pull request as draft May 17, 2026 18:35

tlaurion added 22 commits May 17, 2026 14:36

initrd/bin/tpmr.sh: fix timer= comment in tpm1_da_state

38c3112

DA: line always emits timer= field (with empty value when unavailable), not conditionally. Fix comment to match actual behavior. Signed-off-by: Thierry Laurion <insurgo@riseup.net>

initrd/bin/tpmr.sh: normalize TPM vendor/rev output format

5a6bb88

Signed-off-by: Thierry Laurion <insurgo@riseup.net>

Uh oh!

Conversation

tlaurion commented May 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

initrd/bin/tpmr.sh — auth fix + DA state + bad_auth

initrd/etc/functions.sh — counter auth + DA preflight

Other

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

notgivenby commented May 15, 2026

Uh oh!

tlaurion commented May 16, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

tlaurion commented May 14, 2026 •

edited

Loading

`initrd/bin/tpmr.sh` — auth fix + DA state + bad_auth

`initrd/etc/functions.sh` — counter auth + DA preflight