`ot_otp`: handle secret partition descrambling at OTP back-end load time #259

rivos-eblot · 2025-10-24T12:24:39Z

No description provided.

AlexJones0

Thanks for looking at this, it looks great to me! I'll base my other OTP PR (#258) on the changes in this PR.

You can ignore the CI failure in //sw/device/tests/crypto:aes_kwp_kat_functest_sim_qemu_sival_rom_ext, I need to find some time to look into why all the AES tests are recently flaky on Earlgrey regression CI - no QEMU change was made so some of the OpenTitanLib crypto refactoring must have started doing something differently. The errors are quite strange.

hw/opentitan/ot_otp_eg.c

hw/opentitan/ot_otp_dj.c

AlexJones0

Actually I've just checked and this is causing the otp_ctrl_descrambling_test to fail again, let me see if I can figure out what's going wrong.

AlexJones0 · 2025-10-24T23:18:10Z

I have something resembling a fix in my last commit here. At least for Earlgrey it gets the descrambling test passing. The commit message goes into more detail but this ensures that DAI reads from buffered partitions read from the buffer and do not erroneously calculate the ECC on bufferized data (which, if descrambled, will not have the same ECC as the scrambled data).

This is probably not entirely correct, as I haven't cross-referenced the RTL nor properly considered the implications on digests at all, but it is sufficient to at least get the descrambling datapath exercised by otp_ctrl_descrambling_test passing again after switching to descrambling during bufferization. Hopefully it is at least a useful starting point.

AlexJones0

Thanks, this LGTM (apart from the CI lint failure).

Also on the commit for fixing buffered DAI reads (542d0c0), this should be renamed to ot_otp instead of ot_otp_eg?

The CI regression run reports that the otp_ctrl_descrambling_test is passing, so this should be correct for secret partition descrambling.

hw/opentitan/ot_otp_dj.c

rivos-eblot · 2025-10-27T13:02:01Z

Should be ok, waiting for CI

AlexJones0

Thanks for all the work on this, it looks really nice to me.

If you use all the suggestions/fixes from the comments I've left and rebase my OTP PR (#258) on top, then I can see the the Earlgrey OTP tests that we have passing.

hw/opentitan/ot_otp_eg.c

hw/opentitan/ot_otp_dj.c

hw/opentitan/ot_otp_eg.c

Function naming was a bit confusing, rename it to match similar feature. Remove condition known to be always true. Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

…IZE} macros with ot_otp_eg_part_data_byte_{offset,size} static inline function. This aligns code with Darjeeling version. Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

hw/opentitan/ot_otp_dj.c

…ementation - Partially generated with autoreg.py and otptool.py Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

…igest management. - unscramble secret partitions at OTP load time - content of buffered partitions is stored, unscrambled, in partition buffers - all partition digest are loaded, ECC-corrected and stored in dedicated partition controller field, whether they are buffered or not - DAI read always access the OTP back-end data, never the buffered copy Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

Replace this legacy implementation based on partition "type", with the `integrity` partition descriptor to perform ECC. Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

AlexJones0

Thanks, this is much more closely aligned to the RTL.

With my PR rebased on top, all the OTP tests that I have locally are passing (when the QEMU support is merged I will try and enable them upstream so we get the checks in CI regressions).

rivos-eblot · 2025-10-28T13:00:09Z

Thanks, this is much more closely aligned to the RTL.

With my PR rebased on top, all the OTP tests that I have locally are passing (when the QEMU support is merged I will try and enable them upstream so we get the checks in CI regressions).

Thanks a lot for your help + review + diagnosis.

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch from d7559b3 to b4888f0 Compare October 24, 2025 13:27

AlexJones0 approved these changes Oct 24, 2025

View reviewed changes

hw/opentitan/ot_otp_eg.c Show resolved Hide resolved

hw/opentitan/ot_otp_dj.c Show resolved Hide resolved

AlexJones0 self-requested a review October 24, 2025 14:48

AlexJones0 reviewed Oct 24, 2025

View reviewed changes

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch from b4888f0 to de7b785 Compare October 24, 2025 15:13

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch 3 times, most recently from 2e770f4 to a04e37b Compare October 27, 2025 09:39

AlexJones0 reviewed Oct 27, 2025

View reviewed changes

hw/opentitan/ot_otp_dj.c Outdated Show resolved Hide resolved

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch from a04e37b to 94fd1b3 Compare October 27, 2025 13:01

rivos-eblot requested a review from luismarques October 27, 2025 13:54

AlexJones0 mentioned this pull request Oct 27, 2025

ot_otp: perform partition descrambling when loading tokens #254

Merged

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch 2 times, most recently from a379f03 to 598ac19 Compare October 27, 2025 20:22

rivos-eblot mentioned this pull request Oct 27, 2025

ot_otp: Add scrambling on DAI writes to secret partitions + REGWEN fixes #258

Merged

AlexJones0 reviewed Oct 27, 2025

View reviewed changes

rivos-eblot added 3 commits October 28, 2025 08:19

[ot] hw/opentitan: ot_otp: rename/simplify digest ECC check

2dcc90e

Function naming was a bit confusing, rename it to match similar feature. Remove condition known to be always true. Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

[ot] python/qemu: ot.top.descriptor: fix CASE_SUB range

23943d1

Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

[ot] hw/opentitan: ot_otp_eg: replace OT_OTP_PART_DATA_BYTE_{OFFSET,S…

49d6d22

…IZE} macros with ot_otp_eg_part_data_byte_{offset,size} static inline function. This aligns code with Darjeeling version. Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch from 598ac19 to aafab66 Compare October 28, 2025 07:40

AlexJones0 reviewed Oct 28, 2025

View reviewed changes

hw/opentitan/ot_otp_dj.c Outdated Show resolved Hide resolved

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch from aafab66 to 174c3a5 Compare October 28, 2025 10:31

rivos-eblot added 3 commits October 28, 2025 11:52

[ot] hw/opentitan: ot_otp_eg: update OTP to match Darjeeling new impl…

cb3b99b

…ementation - Partially generated with autoreg.py and otptool.py Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

[ot] hw/opentitan: ot_otp: cleanup partition index management

a8aab5c

Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch from 174c3a5 to cb32e1c Compare October 28, 2025 11:23

rivos-eblot added 2 commits October 28, 2025 12:32

[ot] hw/opentitan: ot_otp: use integrity descriptor bit

8333fbf

Replace this legacy implementation based on partition "type", with the `integrity` partition descriptor to perform ECC. Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

[ot] .gitlab-ci.d/opentitan: ot-smoke.yml: add Darjeeling smoke test

984e9bc

Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

[ot] .github/workflows: build_test.yaml: add Darjeeling smoke test

473089b

Signed-off-by: Emmanuel Blot <eblot@rivosinc.com>

rivos-eblot force-pushed the dev/ebl/otp_secret_scramble branch from cb32e1c to 473089b Compare October 28, 2025 11:32

AlexJones0 approved these changes Oct 28, 2025

View reviewed changes

rivos-eblot marked this pull request as ready for review October 28, 2025 12:57

rivos-eblot merged commit f35d78f into lowRISC:ot-9.2.0 Oct 28, 2025
10 checks passed

rivos-eblot deleted the dev/ebl/otp_secret_scramble branch October 28, 2025 13:01

ot_otp: handle secret partition descrambling at OTP back-end load time #259

ot_otp: handle secret partition descrambling at OTP back-end load time #259

Uh oh!

Conversation

rivos-eblot commented Oct 24, 2025

Uh oh!

AlexJones0 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

AlexJones0 left a comment

Choose a reason for hiding this comment

Uh oh!

AlexJones0 commented Oct 24, 2025

Uh oh!

AlexJones0 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

rivos-eblot commented Oct 27, 2025

Uh oh!

AlexJones0 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

AlexJones0 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

rivos-eblot commented Oct 28, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

`ot_otp`: handle secret partition descrambling at OTP back-end load time #259

`ot_otp`: handle secret partition descrambling at OTP back-end load time #259

AlexJones0 left a comment •

edited

Loading

AlexJones0 left a comment •

edited

Loading