)]}'
{
  "log": [
    {
      "commit": "5bb821ee5b1b88e2aa4414f175df26f840149811",
      "tree": "a75472c3da3eecb2bb5341558fd7a2ceeabadeee",
      "parents": [
        "cdb4fd23fe4a0df945fd85adff877895830c8f8d"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 17 16:19:30 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 17 16:39:55 2026 -0800"
      },
      "message": "hal/blocking, veer: Store the DPE signature through a reference\n\nSign::sign stores the signature in a caller-provided value and returns\nthe next handle, so that a large signature such as ML-DSA is not copied\nthrough the return value.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "cdb4fd23fe4a0df945fd85adff877895830c8f8d",
      "tree": "b116e78ea9369fe550c3f59676f1e73bb6256944",
      "parents": [
        "8804768fe1a59817b5623b3e16e0835231c10fbb"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 17 10:47:07 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 17 16:39:55 2026 -0800"
      },
      "message": "veer: Add a client for the Caliptra DPE Sign command\n\nThe driver sends the Sign command with INVOKE_DPE_ECC384 and implements\nthe Sign HAL trait. It takes the command and response layouts from the\ncaliptra-dpe crate.\n\nThe test signs a digest with the default context and checks the\nsignature with ECDSA384_SIGNATURE_VERIFY, using the public key from the\nDPE CertifyKey command. It passes on the emulator and on the VCK190\nFPGA.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "8804768fe1a59817b5623b3e16e0835231c10fbb",
      "tree": "a1776da07c6edaa513c9fa72edd8fdb18c32d399",
      "parents": [
        "5e6dfc7c6a5ede8e16f07a24fe0982e39cb77dd3"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 17 10:47:07 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 17 16:39:55 2026 -0800"
      },
      "message": "hal/blocking: Add DPE client traits\n\nSign is the trait for the DPE Sign command of a profile. P384Sha384 is\nthe profile with ECDSA P-384 and SHA-384. The other DPE commands have\nno trait yet.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "5e6dfc7c6a5ede8e16f07a24fe0982e39cb77dd3",
      "tree": "fa0e8d074b26647d8f45277b0b2133cb4b65659e",
      "parents": [
        "32fa56f153c46512fe209ff6f1e3f7b2b6d30693"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Mon Nov 16 22:01:58 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 17 10:19:32 2026 -0800"
      },
      "message": "mctp: remove the panic paths from the IPC client and the API stack\n\nIpcMctpClient borrowed its buffers with RefCell::borrow and indexed them with\nranges. Every borrow is now try_borrow, which returns an internal error if the\nbuffers are in use, and the slices use get. The copies use a loop without a\nlength check that can panic. The request and response buffers are borrowed as\nseparate fields, which removes the copy of the request.\n\nStackReqChannel::recv and StackListener::recv sliced the caller\u0027s buffer to the\nmessage\u0027s full length, which panicked for a message longer than the buffer. They\nnow cut the slice to what the buffer holds. payload_size still reports the full\nlength, so a caller can tell. Two tests cover it.\n\nencode_send writes its header and payload without indexing.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "32fa56f153c46512fe209ff6f1e3f7b2b6d30693",
      "tree": "f2abc3750469b44ef59b14e49b5911188074d325",
      "parents": [
        "22ed2237ef1d3bc71c5ca5e9556fb4fb209d9d75"
      ],
      "author": {
        "name": "Dave Roth",
        "email": "davidroth@google.com",
        "time": "Tue Nov 17 17:02:37 2026 +0000"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 17 09:44:35 2026 -0800"
      },
      "message": "Revert \"chore: Roll rules_rust to 0.75.0\"\n\nThis reverts commit 0d165d8a8e095b05809b3e9fb41049c539f8026b (#723).\n\nAlso moves `:linker_script` from `link_deps` to `deps` in\n`target/veer/tests/rng/BUILD.bazel` (added in 922d587 after #723) for\ncompatibility with `rules_rust` 0.70.0, and refreshes\n`MODULE.bazel.lock` to preserve `multi_process_app_entry.patch`.\n"
    },
    {
      "commit": "22ed2237ef1d3bc71c5ca5e9556fb4fb209d9d75",
      "tree": "b80b85483d6870279be7f96bc268b883e9abe634",
      "parents": [
        "ebb895896e98f335557c8fc4fe5370fc5a164f51"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 15:13:22 2026 -0800"
      },
      "committer": {
        "name": "Zach Halvorsen",
        "email": "zhalvorsen@google.com",
        "time": "Mon Nov 16 16:04:24 2026 -0800"
      },
      "message": "i3c: Remove the panic paths from the i3c_service app\n\nThe I3C server and client index their buffers with get. A Pigweed patch\ndoes the same in the entry point generated for a multi_process_app.\n\nWith these, the i3c_service image passes app_no_panics_test.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "ebb895896e98f335557c8fc4fe5370fc5a164f51",
      "tree": "536993d5d5c7ff0555bc95974a45a3dbbc8bd255",
      "parents": [
        "922d587ebbd2ba53a17de52ed614adf2cd0ef09c"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 15:01:40 2026 -0800"
      },
      "committer": {
        "name": "Zach Halvorsen",
        "email": "zhalvorsen@google.com",
        "time": "Mon Nov 16 16:04:24 2026 -0800"
      },
      "message": "target: Test that VeeR userspace apps cannot panic\n\nrust_binary_no_panics_test looks for the kernel\u0027s panic_is_possible\nsymbol, so it passes for code linked into a userspace app. That\nincludes the mailbox, SHA and RNG drivers.\n\napp_no_panics_test runs the toolchain\u0027s nm on each app ELF of a system\nimage and fails if the app links a core::panicking function.\n\nThe i3c_service image is left out. Its app can panic in the entry point\nPigweed generates for a multi_process_app and in i3c_server::dispatch.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "922d587ebbd2ba53a17de52ed614adf2cd0ef09c",
      "tree": "c5cb9721636431b714eb9c91a77ad7ff63a6ac45",
      "parents": [
        "d7efc0be75d2c73ad3ebf8a00c605678a2d4d896"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 14:31:14 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 15:50:11 2026 -0800"
      },
      "message": "veer: Add a random number generator that uses the Caliptra mailbox\n\nThe driver sends CM_RANDOM_GENERATE and implements the CryptoRng HAL\ntrait.\n\nThe test checks that every byte of a buffer is written, for several\nbuffer lengths, and that random bytes can be drawn in the middle of a\nSHA computation on the same mailbox. It passes on the emulator and on\nthe VCK190 FPGA.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "d7efc0be75d2c73ad3ebf8a00c605678a2d4d896",
      "tree": "59947fa6e1feecd8ae82e9670356acaed2b4a284",
      "parents": [
        "7c934852f0d663500272bae5af747b1c0552a15c"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 14:31:13 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 15:50:11 2026 -0800"
      },
      "message": "hal/blocking: Add random number generator traits\n\nCryptoRng fills a buffer with random bytes that are suitable for keys\nand nonces. The HAL has no trait for weaker generators. RandCore adapts\na CryptoRng to the rand_core TryRngCore and TryCryptoRng traits.\n\nReseeding is left out because not every generator supports it.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "7c934852f0d663500272bae5af747b1c0552a15c",
      "tree": "322f758dc2e6e1c5896e27c2c6c1963a5cc6aeb6",
      "parents": [
        "895a3600b48170cee6af2bfc9c8434f9f91567ad"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 14:21:38 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Mon Nov 16 15:50:11 2026 -0800"
      },
      "message": "veer: Let drivers share the Caliptra mailbox\n\nThe Mailbox methods take \u0026self, and the SHA driver holds a reference to\nthe mailbox instead of owning it. An SPDM responder needs several hash\ncomputations, a random number generator and a signer on the same\nmailbox.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "895a3600b48170cee6af2bfc9c8434f9f91567ad",
      "tree": "84ec2a579858af1be63cbe026edd488205d3cab1",
      "parents": [
        "45b088f3e972a347203c45960687843e3346feaa"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sat Nov 14 15:48:53 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 16 14:09:34 2026 -0800"
      },
      "message": "orchestrator/test: a pure device model for boot-sequence tests\n\nA managed device as seen from its reset line: hold reset and it is down,\nrelease it and one boot attempt starts. A script of Behaviours (Hang, Boots,\nBootsThenFails) says what each attempt does, indexed by how many times reset\nhas been released, so the device owns its behaviour and the system under test\nnever tells it how to act.\n\nIt is a pure state machine with injected time and no dependencies: no pins,\nno clock, no kernel, no heap. The delay counts from the release, a reset\nmid-boot restarts it, and asserting reset drops ready. Transitions are stamped\nwith the moment they were due, not the moment a late caller noticed, so tests\ncheck exact timing and an event loop can sleep until next_event_at. A small\nqueue keeps ready edges so a pulse between two polls is not lost; overflow is\ncounted, not hidden.\n\nIt lives beside the orchestrator\u0027s other test doubles in services/orchestrator/test\nand has a README documenting its seams, invariants and host-test target. All\narithmetic saturates and the script lookup avoids casts, per the review\nchecklist.\n\nNineteen host tests. Two deliberate breakages (boot one tick late, a reset\nthat forgets to record the drop) each fail several of them. ./pw presubmit\npasses.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "45b088f3e972a347203c45960687843e3346feaa",
      "tree": "1d3bbf47ee90f5a81351e526021f30e49d658f5f",
      "parents": [
        "62a7e7af17a9c5d08f1879649194941ffb4e7adf"
      ],
      "author": {
        "name": "dependabot[bot]",
        "email": "49699333+dependabot[bot]@users.noreply.github.com",
        "time": "Mon Nov 16 09:07:43 2026 +0000"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Mon Nov 16 16:38:30 2026 -0400"
      },
      "message": "ci(deps): bump actions/setup-python from 6.3.0 to 7.1.0\n\nBumps [actions/setup-python](https://github.com/actions/setup-python) from 6.3.0 to 7.1.0.\n- [Release notes](https://github.com/actions/setup-python/releases)\n- [Commits](https://github.com/actions/setup-python/compare/ece7cb06caefa5fff74198d8649806c4678c61a1...6d4fd22b20465ccc0f7354476588e8df7c9ca0e0)\n\n---\nupdated-dependencies:\n- dependency-name: actions/setup-python\n  dependency-version: 7.1.0\n  dependency-type: direct:production\n  update-type: version-update:semver-major\n...\n\nSigned-off-by: dependabot[bot] \u003csupport@github.com\u003e"
    },
    {
      "commit": "62a7e7af17a9c5d08f1879649194941ffb4e7adf",
      "tree": "02e101cd5602c93e11176e15035f840ae1c8a938",
      "parents": [
        "3646385c9599c859011cc2022b2e3a18724527e6"
      ],
      "author": {
        "name": "dependabot[bot]",
        "email": "49699333+dependabot[bot]@users.noreply.github.com",
        "time": "Mon Nov 16 09:07:34 2026 +0000"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Mon Nov 16 16:34:37 2026 -0400"
      },
      "message": "ci(deps): bump peter-evans/create-pull-request from 8.1.1 to 8.1.3\n\nBumps [peter-evans/create-pull-request](https://github.com/peter-evans/create-pull-request) from 8.1.1 to 8.1.3.\n- [Release notes](https://github.com/peter-evans/create-pull-request/releases)\n- [Commits](https://github.com/peter-evans/create-pull-request/compare/5f6978faf089d4d20b00c7766989d076bb2fc7f1...3ffccacd46053766916f087ce41d627c0b8cf2d1)\n\n---\nupdated-dependencies:\n- dependency-name: peter-evans/create-pull-request\n  dependency-version: 8.1.3\n  dependency-type: direct:production\n  update-type: version-update:semver-patch\n...\n\nSigned-off-by: dependabot[bot] \u003csupport@github.com\u003e"
    },
    {
      "commit": "3646385c9599c859011cc2022b2e3a18724527e6",
      "tree": "c9cf2aa140acc4c98b389c212dbc69a7ff268511",
      "parents": [
        "92750c8289f9ad182170c6980869b53d427c0927"
      ],
      "author": {
        "name": "dependabot[bot]",
        "email": "49699333+dependabot[bot]@users.noreply.github.com",
        "time": "Mon Nov 16 09:07:38 2026 +0000"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Mon Nov 16 16:33:05 2026 -0400"
      },
      "message": "ci(deps): bump bazel-contrib/setup-bazel from 0.19.1 to 0.19.2\n\nBumps [bazel-contrib/setup-bazel](https://github.com/bazel-contrib/setup-bazel) from 0.19.1 to 0.19.2.\n- [Release notes](https://github.com/bazel-contrib/setup-bazel/releases)\n- [Commits](https://github.com/bazel-contrib/setup-bazel/compare/104e3eea9447cdaa4b00d0fb46191a3985d26ab3...f34d6a05eae3c24acab7fa54587492daeaeffdab)\n\n---\nupdated-dependencies:\n- dependency-name: bazel-contrib/setup-bazel\n  dependency-version: 0.19.2\n  dependency-type: direct:production\n  update-type: version-update:semver-patch\n...\n\nSigned-off-by: dependabot[bot] \u003csupport@github.com\u003e"
    },
    {
      "commit": "92750c8289f9ad182170c6980869b53d427c0927",
      "tree": "9bebe3072b9829abac25d542226823e661b4a066",
      "parents": [
        "f0cdcba0ce6d14a07f928acf6bdd309ad5a9fd3b"
      ],
      "author": {
        "name": "dependabot[bot]",
        "email": "49699333+dependabot[bot]@users.noreply.github.com",
        "time": "Mon Nov 16 09:07:31 2026 +0000"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Mon Nov 16 16:22:55 2026 -0400"
      },
      "message": "ci(deps): bump actions/download-artifact from 4.3.0 to 8.0.2\n\nBumps [actions/download-artifact](https://github.com/actions/download-artifact) from 4.3.0 to 8.0.2.\n- [Release notes](https://github.com/actions/download-artifact/releases)\n- [Commits](https://github.com/actions/download-artifact/compare/d3f86a106a0bac45b974a628896c90dbdf5c8093...9000827ccba6bdab643e8b6fd33ac0654aef8333)\n\n---\nupdated-dependencies:\n- dependency-name: actions/download-artifact\n  dependency-version: 8.0.2\n  dependency-type: direct:production\n  update-type: version-update:semver-major\n...\n\nSigned-off-by: dependabot[bot] \u003csupport@github.com\u003e"
    },
    {
      "commit": "f0cdcba0ce6d14a07f928acf6bdd309ad5a9fd3b",
      "tree": "7cd6f2b42d2dec2012d0c6de0468bfe5d826d7fb",
      "parents": [
        "0d165d8a8e095b05809b3e9fb41049c539f8026b"
      ],
      "author": {
        "name": "dependabot[bot]",
        "email": "49699333+dependabot[bot]@users.noreply.github.com",
        "time": "Mon Nov 16 09:08:51 2026 +0000"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 16 12:02:53 2026 -0800"
      },
      "message": "ci(deps): bump actions/upload-artifact from 4.6.2 to 7.0.2\n\nBumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 4.6.2 to 7.0.2.\n- [Release notes](https://github.com/actions/upload-artifact/releases)\n- [Commits](https://github.com/actions/upload-artifact/compare/ea165f8d65b6e75b540449e92b4886f43607fa02...cf430e030ddbb5b0abf93d22962f4752f3646cd9)\n\n---\nupdated-dependencies:\n- dependency-name: actions/upload-artifact\n  dependency-version: 7.0.2\n  dependency-type: direct:production\n  update-type: version-update:semver-major\n...\n\nSigned-off-by: dependabot[bot] \u003csupport@github.com\u003e"
    },
    {
      "commit": "0d165d8a8e095b05809b3e9fb41049c539f8026b",
      "tree": "44ede39d5e404b43013e412d32a2fab50e63a63f",
      "parents": [
        "b38d346ed5459fc91a92d1eacdea2dc635e3ba25"
      ],
      "author": {
        "name": "Dave Roth",
        "email": "davidroth@google.com",
        "time": "Mon Nov 16 19:09:56 2026 +0000"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 16 12:01:40 2026 -0800"
      },
      "message": "chore: Roll rules_rust to 0.75.0\n\nRoll rules_rust and rules_rust_mdbook to 0.75.0:\n\n- Add cargo-bazel-lock.json lockfiles for the non-root caliptra_deps\n  crate_universe workspaces (rust_caliptra_crates and\n  rust_caliptra_crates_host), required by rules_rust \u003e\u003d 0.71.1.\n- Move target_linker_script C++ targets from rust_binary deps to\n  link_deps across //target/....\n- Keep protobuf at 36.2 and mark its test-only crate_universe extension\n  as dev_dependency \u003d True until @pigweed rolls protobuf \u003e\u003d 37.0.\n- Pass CFLAGS\u003d-Wno-undef to psm\u0027s build script inside rules_rust_mdbook.\n- Backport @pigweed\u0027s rules_rust 0.75.0 compatibility updates onto the\n  pinned pigweed commit until pwrev/445972 lands upstream.\n"
    },
    {
      "commit": "b38d346ed5459fc91a92d1eacdea2dc635e3ba25",
      "tree": "2cfdbe5fcad7f67781da8cd634d8414248f3b8fd",
      "parents": [
        "40b880b4013eb2a3c04263431fd281415e2ed95d"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 13 09:40:31 2026 -0800"
      },
      "committer": {
        "name": "Zach Halvorsen",
        "email": "zhalvorsen@google.com",
        "time": "Mon Nov 16 11:26:16 2026 -0800"
      },
      "message": "docs: Add a //:watch target to rebuild on file changes\n\n//:watch is an alias of Pigweed\u0027s pw_watch entry point for Bazel, and\nworkflows.json has a matching tool. `bazel run //:watch build //docs`\nand `./pw watch build //docs` rebuild the docs when a source file\nchanges.\n\nDocument the target in the README, the mdbook pages and AGENTS.md.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "40b880b4013eb2a3c04263431fd281415e2ed95d",
      "tree": "ba87d99c17d3754dfda27a19fb16cc1aeed495fc",
      "parents": [
        "b252d7db681fc8d660760f4ca658c1c53b347122"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 15 00:41:18 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 16 10:24:01 2026 -0800"
      },
      "message": "util/service: wait for the deadline when every channel is deferred\n\nThe kernel rejects object_wait on a WaitGroup with no members\n(InvalidArgument). A ServiceLoop whose channels were all deferred and which had\nno added source therefore failed its next call to `next`, so a server with a\nsingle deferred channel could not wait for its own deadline to complete the\nreply: its loop returned an error and the server exited.\n\n`next` now notices that no channel is serving and no source was added, and\nwaits on a parked channel\u0027s own handle for Signals::ERROR (never raised on a\nhandler) until the deadline, then returns Event::Deadline. With Instant::MAX\nnothing could ever wake the loop, so it returns FailedPrecondition instead of\nhanging. The new path runs exactly when the WaitGroup is empty, which is exactly\nwhen the old code failed, so a loop that worked before behaves the same.\n\nuserspace::time::sleep_until is not used: it waits on handle 0 assuming that is\nthe process, but in an app handle 0 is the first declared object.\n\nTests: a host test for the helper that decides this, and a QEMU regression test,\nutil_service/deferred_single_channel, with a one-channel server that defers each\nrequest for 50 ms and a client that makes two blocking calls. Without the fix\nthe server\u0027s `next` returns InvalidArgument, the server exits and the client\u0027s\ncall times out; with it the test passes. The source counter uses saturating_add.\n./pw presubmit passes.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "b252d7db681fc8d660760f4ca658c1c53b347122",
      "tree": "16bf66c175276e4c8a4cdde21b4b182da2cf9d82",
      "parents": [
        "eb90627b19f9def02dac69ae82632976b0f8a978"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 13 15:29:57 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Sun Nov 15 11:19:03 2026 -0800"
      },
      "message": "veer: Read the ROM\u0027s handoff table\n\nThe caliptra-mcu-sw ROM leaves a handoff table in the last 1 KiB of\nDCCM. It holds the handoff version, the source of the runtime image\nand the capabilities of the ROM.\n\nveer_handoff checks the table\u0027s marker and version and refers to the\ntable where it is in DCCM. The runtime image logs the table when it\nboots. A service can read the table through a 64-byte mapping at\n@HANDOFF_BASE@ in its system config.\n\nThe host tests pass. The on-device test and the runtime test pass on\nthe emulator and on the VCK190 FPGA.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "eb90627b19f9def02dac69ae82632976b0f8a978",
      "tree": "03a0b54c23280d29921eadaf14d8737164d8934a",
      "parents": [
        "d267cfd12dc1cbd0135a7f86541985c39ce7b1b6"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 10 14:51:50 2026 -0800"
      },
      "committer": {
        "name": "Zach Halvorsen",
        "email": "zhalvorsen@google.com",
        "time": "Fri Nov 13 16:10:26 2026 -0800"
      },
      "message": "ocp_recovery: run the recovery test on the emulator from the host\n\nThe test plays the BMC through the MCU\u0027s AXI view of the i3c-core. A\nsmall emulator patch lets the recovery FIFO be filled through the TTI TX\ndata port in streaming boot, and reports EMPTY and FULL from the FIFO\nlevel when Caliptra is not the reader.\n\nAssisted-by: Claude:claude-opus-5-5\n"
    },
    {
      "commit": "d267cfd12dc1cbd0135a7f86541985c39ce7b1b6",
      "tree": "2e3ee1899f10ee562ca2c88fe3c41eb074f83d3a",
      "parents": [
        "382f41060a941082e22a1e2496d3f61733ce673c"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 10 14:50:07 2026 -0800"
      },
      "committer": {
        "name": "Zach Halvorsen",
        "email": "zhalvorsen@google.com",
        "time": "Fri Nov 13 16:10:26 2026 -0800"
      },
      "message": "ocp_recovery: add an on-device recovery test with an FPGA agent\n\nThe firmware asks for recovery image 3, checks every word, waits for\nactivation and reports success. On the FPGA, caliptra-sw\u0027s BMC serves\nthe image over I3C. The emulator has no path from its BMC to the MCU\u0027s\nrecovery registers, so there is no emulator variant yet.\n\nAssisted-by: Claude:claude-opus-5-5\n"
    },
    {
      "commit": "382f41060a941082e22a1e2496d3f61733ce673c",
      "tree": "dde750dda4c1963e0ec2a0e44134736b7ecbbcae",
      "parents": [
        "65da5514b6f0fc5234f849a0e8763069a631dc85"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 10 14:18:47 2026 -0800"
      },
      "committer": {
        "name": "Zach Halvorsen",
        "email": "zhalvorsen@google.com",
        "time": "Fri Nov 13 16:10:26 2026 -0800"
      },
      "message": "ocp_recovery: add blocking HAL and VeeR i3c-core driver\n\nThe HAL runs the per-image OCP recovery sequence that Caliptra\u0027s\nDmaRecovery uses. The VeeR driver reads the i3c-core recovery FIFO one\nword at a time over MMIO, since the MCU has no DMA path to it.\n\nAssisted-by: Claude:claude-opus-5-5\n"
    },
    {
      "commit": "65da5514b6f0fc5234f849a0e8763069a631dc85",
      "tree": "0ec232836766a1259d0d05edbccd234bb40d114e",
      "parents": [
        "c5042c8fbee0ce22e455048b2b80ec9b2ccc14e0"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 13 14:04:22 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 13 15:10:05 2026 -0800"
      },
      "message": "veer: Add the runtime image\n\n//target/veer/runtime is the product image for the Caliptra MCU. It has\nthe kernel and no services.\n\nTarget::main sets up the hardware once the kernel has started, which is\nbefore any service would start. It stops the watchdog, because the ROM\nleaves it running.\n\nA host test checks that the watchdog is disabled after that. It passes\non the emulator and on the VCK190 FPGA.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "c5042c8fbee0ce22e455048b2b80ec9b2ccc14e0",
      "tree": "31399f6aaef6cc1f47264c9ea5ce3ecdaaa7daec",
      "parents": [
        "1b829740ceaf53e17cc979961a3d05cb30d3d7d7"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 13 14:04:22 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 13 15:10:05 2026 -0800"
      },
      "message": "veer: Add a driver for the MCU watchdog\n\nThe MCI has two watchdog timers. Timer 2 starts when timer 1 times out\nand raises an NMI when it times out itself. The caliptra-mcu-sw ROM\nleaves the watchdog running when it jumps to the runtime image.\n\nThe driver starts, restarts and stops the watchdog. It is generic over\nregister access so that host tests can run it against a model of the\nwatchdog RTL.\n\nThe host tests pass. The on-device test passes on the emulator and on\nthe VCK190 FPGA.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "1b829740ceaf53e17cc979961a3d05cb30d3d7d7",
      "tree": "da77dbdad76be1ce73d83f7595b0ecbe27e4115b",
      "parents": [
        "d9a10d4419ebc71040c6501c56daa842ca20c677"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Fri Nov 13 14:35:11 2026 -0400"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 13 11:22:07 2026 -0800"
      },
      "message": "docs: Add AGENTS.md\n\nAdd guidance for AI coding agents working in the repository. The file\ncarries only the CI gate commands (./pw presubmit, wildcard build tag\nfilters), a \"Before pushing\" checklist, and pointers into docs/src/,\nworkflows.json and .github/copilot-instructions.md, so each fact lives in\none place.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "d9a10d4419ebc71040c6501c56daa842ca20c677",
      "tree": "3d0910459bc87eb71c9143f8c5cc3d1179ac3ea0",
      "parents": [
        "25eb416a3a943df0b491b42e426106af370d1a70"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Thu Nov 12 00:13:18 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Fri Nov 13 11:37:36 2026 -0400"
      },
      "message": "docs: updated README.md and .gitignore\n\ndocs: add rust format-on-save to the VS Code settings example\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "25eb416a3a943df0b491b42e426106af370d1a70",
      "tree": "4486ea0993b9521431706857688da1984a3de6dd",
      "parents": [
        "69f7b911cf4100dcd00ba642348be78cc58cc56b"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Thu Nov 12 17:42:18 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Fri Nov 13 11:30:18 2026 -0400"
      },
      "message": "target/ast10x0: start the flash_service program mid-sector\n\nThe flash service does not report the program page, so the client\nassumed 256 bytes and started its page-crossing program at offset 250.\nA program page is a power of two smaller than the sector, so it divides\nthe middle of the sector. Start the program just before the middle and\nread back a fixed margin around the data, which crosses a page boundary\nfor any page size and removes the fixed page size from the test. Check\nthat the sector holds the read back and that the test offset is aligned\nto the sector.\n\nTested: flash_qemu_test and no_panics_test pass under\n--config\u003dvirt_ast10x0, flash_evb_test passes on the AST1060 EVB, and\n./pw format and ./pw presubmit pass.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\nSigned-off-by: Enmanuel Parache \u003cenmanuelparache@gmail.com\u003e\n"
    },
    {
      "commit": "69f7b911cf4100dcd00ba642348be78cc58cc56b",
      "tree": "5d04216cdb510f3cd0c880126c5f7b2b49f9929c",
      "parents": [
        "52ba4b06bab052b6cd6a6e424e2fcf2bfb9b22da"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Thu Nov 12 16:44:02 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Fri Nov 13 11:30:18 2026 -0400"
      },
      "message": "format: make presubmit happy\n"
    },
    {
      "commit": "52ba4b06bab052b6cd6a6e424e2fcf2bfb9b22da",
      "tree": "6053c0710ee8f8be97e6323bb68ff6ddd245d228",
      "parents": [
        "e857290ccffac73a331a6344cdd890c1dbdd8cec"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Thu Nov 12 16:22:31 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Fri Nov 13 11:30:18 2026 -0400"
      },
      "message": "target/ast10x0: derive flash test geometry from the device\n\nThe flash tests compared the geometry with fixed values: a 4 KiB sector,\na 64 MiB flash and a 256-byte program page. Take them from the flash\ninstead, which the backend discovers over SFDP, and check that they are\nconsistent: the erase sizes match the sector, the size is a multiple of\nthe sector and holds the test. tests/flash reads the program page from the\ndriver. The flash service does not report the program page, so\ntests/flash_service still assumes 256 bytes for its page crossing program.\n\nTested: flash_qemu_test and no_panics_test pass under\n--config\u003dvirt_ast10x0, and flash_evb_test passes on the AST1060 EVB for\nboth tests.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\nSigned-off-by: Enmanuel Parache \u003cenmanuelparache@gmail.com\u003e\n"
    },
    {
      "commit": "e857290ccffac73a331a6344cdd890c1dbdd8cec",
      "tree": "d16c9badfaea6d147824d3ecebb8c0e5e0060e02",
      "parents": [
        "29bb629b0b1e9d84f164784c471714bb5a899d11"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Thu Nov 12 14:13:44 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Fri Nov 13 11:30:18 2026 -0400"
      },
      "message": "target/ast10x0: restore flash after a failed flash_service check\n\nThe flash_service client returned from the first failed check without\nrestoring the sector it had saved, which left the EVB\u0027s flash erased or\nhalf-programmed. Run the checks in a function that returns the failure\nand restore the sector afterwards in every case, as tests/flash does.\n\nAlso fix comments in the test\u0027s BUILD.bazel and system.json5 that were\nwrong (FMC is modeled under QEMU, the images are 64 MiB, and the app\nflash layout) and derive the server buffer size from the driver\u0027s\nMAX_READ_SIZE.\n\nTested: flash_qemu_test and no_panics_test pass under\n--config\u003dvirt_ast10x0, and flash_evb_test passes on the AST1060 EVB.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\nSigned-off-by: Enmanuel Parache \u003cenmanuelparache@gmail.com\u003e\n"
    },
    {
      "commit": "29bb629b0b1e9d84f164784c471714bb5a899d11",
      "tree": "812ee481f092089a1be4b13272c16d7d35ac845a",
      "parents": [
        "3aa455380ee85f83ef23bae424791792723e4e6b"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Thu Nov 12 11:46:39 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Fri Nov 13 11:30:18 2026 -0400"
      },
      "message": "target/ast10x0: add flash and flash service tests\n\nAdd two ast10x0 images that test the FMC CS1 flash.\n\ntests/flash runs the flash HAL over Ast10x0FmcFlashDriver in the kernel,\nwithout IPC, like target/veer/tests/flash. tests/flash_service serves the\nflash from a server app through services/flash/server-runtime and drives\nit from a client app through services/flash/client-ipc over a kernel\nchannel, like target/veer/tests/flash_service. The tests erase, program\nacross a page boundary and read back, and check that bad requests are\nrejected. Both save and restore the sectors they use, so they leave the\nEVB\u0027s flash as it was.\n\nTested: flash_qemu_test and no_panics_test pass under\n--config\u003dvirt_ast10x0, and flash_evb_test passes on the AST1060 EVB\nunder --config\u003dk_ast1060_evb for both tests.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\nSigned-off-by: Enmanuel Parache \u003cenmanuelparache@gmail.com\u003e\n"
    },
    {
      "commit": "3aa455380ee85f83ef23bae424791792723e4e6b",
      "tree": "14fd56612fdd86ceeac720201a91a667b7c60e66",
      "parents": [
        "bd203c6839dde0634ca4e4af4eb32b6691371a0a"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 13:12:46 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer i3c driver: use div_ceil for the RX word count\n\nFixes the clippy::manual_div_ceil error CI reports now that the driver\nlibrary builds on the host (and so is linted).\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "bd203c6839dde0634ca4e4af4eb32b6691371a0a",
      "tree": "198453ee8c2f99ad82860ced747ffa1c223adbf8",
      "parents": [
        "215d49a074cee48d4d77eb839573f01eae93622f"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 12:37:59 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer i3c driver: access registers through an Mmio trait\n\nStore the register base as an address and build the pointer on each\naccess with with_exposed_provenance, as the flash driver does; new stays\na const fn. The driver is generic over Mmio so host tests can use a fake\nregister block, and the word packing becomes two small helpers.\n\nAdd i3c_test: register offsets, word packing tails, TX and IBI write\norder, interrupt enable bits, ResponseRead before InboundReady, the IBI\nretry bound, RX truncation and address decoding. Drop the veer-only\ntarget_compatible_with from the library so the test builds on the host.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "215d49a074cee48d4d77eb839573f01eae93622f",
      "tree": "10827d4e1ce445ce09414459211ac5b1acf621a8",
      "parents": [
        "dc4a07540d5d49bbce2dc5a19e2624bec1aae5f8"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 11:23:07 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "hal/blocking: convert DynamicAddress to u8 with From, not as_u8\n\nDynamicAddress already converts in from u8 with TryFrom, validating the 7-bit\nrange. Convert out with the matching impl From\u003cDynamicAddress\u003e for u8, the\nidiomatic spelling, instead of an as_u8 method; u8::from(addr) and .map(u8::from)\nwork wherever an Into\u003cu8\u003e is expected.\n\nas_u8 was a const fn, but nothing could use that: no const, static or const fn\nmentions DynamicAddress, and no value of the type can be built in a const\ncontext, since TryFrom is a trait method and the field is private. If a const\nconstructor is ever added, a const accessor can come back with it.\n\nSix call sites change (three HAL tests, three in the server\u0027s dispatch).\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "dc4a07540d5d49bbce2dc5a19e2624bec1aae5f8",
      "tree": "c977cd25dcee0781d0bfc3bd51fd5074070235fa",
      "parents": [
        "5a63d6b3b6c550a446ca6cc9fdbfdd723a25a413"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 10:04:40 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "i3c service: expose the target\u0027s device info\n\nA client of the service could read the target\u0027s dynamic address, but not its\ntransfer limits, so a consumer such as an MCTP binding had to assume a frame\nsize. The limits are the target\u0027s to report, and on some controllers they are\nruntime values the controller sets.\n\nAdd I3cOp::DeviceInfo, which returns a DeviceInfo: the dynamic and static\naddresses (0 meaning none; 0x00 is reserved on an I3C bus) and the most one\nSend and one received frame can carry, as two little-endian u16s, six bytes in\nall. The server builds it from I3cTarget::device_info and clamps both limits to\nthe frame ceiling (MAX_PAYLOAD), so a client never plans a frame the ring slots\nor the wire cannot hold. I3cClient gains device_info().\n\nThe new op takes the next free opcode, so existing clients and servers are\nunaffected.\n\nTests: the codec (round trip, zero address, wrong length), the server\u0027s dispatch\n(addresses and limits, an unassigned target, the clamp) and the client over the\nloopback. On target the i3c_service client now reads the device info after the\ndynamic address and fails unless the address is assigned and both limits are\n256, which covers the real driver on the emulator and, when run, the FPGA.\n\nPasses on the emulator; the FPGA image builds.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "5a63d6b3b6c550a446ca6cc9fdbfdd723a25a413",
      "tree": "81d1eb463ee1fb35597b4ca8bf8eece9ba9b438a",
      "parents": [
        "9ecf91f3587081c09c357469a8e0d304349114a7"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 10:01:53 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "hal/blocking: widen the I3C target facade\n\nThe I3cTarget facade was shaped by its one implementor, the Caliptra TTI. The\nonly other I3C target-mode code in the tree, the AST10x0 driver, shows what it\nlacked: the dynamic address is learned from the bus master asynchronously and\ncan be revoked; the maximum read and write lengths are runtime values the\ncontroller sets (SETMRL/SETMWL) and the receive path there is capped at 64\nbytes, not 256; and a Hot-Join request exists. caliptra-mcu-sw\u0027s own HIL has\nget_device_info and disable.\n\nAdd to the trait:\n - static_address and max_recv_len, next to dynamic_address and max_send_len,\n   and device_info, a provided method that assembles them into an\n   I3cDeviceInfo. The limits are documented as runtime values to re-read.\n - disable, the counterpart of enable.\n - request_hot_join, a provided method that reports \"not raised\" by default.\n - TargetEvent::AddressLost, for a revoked address. It and AddressAssigned are\n   documented as reported only by drivers that learn of them asynchronously; the\n   Caliptra ROM assigns the address before the driver runs.\n - The ordering rule for on_interrupt: when one interrupt carries several\n   events, report first the one that cannot be reported again (a clear-on-read\n   status such as a transmit completing) and leave the one that will fire again.\n   The Caliptra driver already follows it; it was unwritten.\n\nThe Caliptra driver implements static_address (from the standby-controller\ndevice-address register), max_recv_len (the 256-byte FIFO) and disable (clear\nthe TTI interrupt enables).\n\n//hal/blocking had no test target, so its unit tests, including the facade\u0027s,\nwere never built. Add blocking_test; all 34 tests pass.\n\nPasses on the emulator; the FPGA image builds.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "9ecf91f3587081c09c357469a8e0d304349114a7",
      "tree": "f1e9b2b0873a969fcc39c75245ebd5b85e6b79e1",
      "parents": [
        "5bf5b32690db8ae94a306f4022fc156967e9c9c5"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 09:44:07 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "i3c: make frame and ring limits a property of the target\n\nThe I3C service fixed two limits as crate constants that belong to the hardware\nor to a consumer.\n\nFrame size. MAX_PAYLOAD was 250, documented as the Caliptra i3c-core limit.\nThat core carries 256-byte private reads and writes (a 64-DWORD FIFO, matching\ncaliptra-mcu-sw\u0027s MAX_READ_WRITE_SIZE), so a 251 to 256 byte write from the\ncontroller was silently truncated. Make 256 the protocol ceiling in i3c_api, and\nadd I3cTarget::max_send_len so a target reports what it can stage in one\nresponse. dispatch refuses a Send over either with TooLong, before it reaches the\ndriver. The Caliptra driver reports 256. The receive side has no method: an\ninbound frame must fit the ceiling, which the trait now documents.\n\nRing depth. RX_RING was a constant justified by MCTP/PLDM fragment arithmetic.\nMake it a const parameter, Server\u003cT, const RING: usize \u003d RX_RING\u003e, with\nServer::with_ring for another depth; Server::new keeps the default of 6 so call\nsites are unchanged. LoopbackTransport and the runtime\u0027s run follow the\nparameter, and run infers it from the server it is given. The ring\u0027s docs no\nlonger describe one consumer\u0027s needs.\n\nTests: a send over the target\u0027s limit is TooLong and stages nothing; a send at\nthe ceiling is accepted; a full-size inbound frame arrives intact; a two-deep\nring overflows on the third frame and keeps the first two.\n\nPasses on the emulator; the FPGA image builds.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "5bf5b32690db8ae94a306f4022fc156967e9c9c5",
      "tree": "b1b3a1f7de93e311a5ee29b675a623278667cc64",
      "parents": [
        "50dd1dc9efe7c246efe930d08b0be72778a753c1"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 09:23:15 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "i3c api: use the shared util/transport Transport instead of a copy\n\nservices/i3c/api/src/transport.rs defined its own Transport trait and\nTransportError, a near copy of the I2C service\u0027s, with the same\ntransact(\u0026mut self, req, resp) -\u003e Result\u003cusize, _\u003e signature as the shared\nutil_transport crate that exists for exactly this.\n\nDelete the copy and re-export util_transport::{Transport, TransportError} from\ni3c_api. Everything that used i3c_api::Transport (I3cClient, IpcTransport,\nLoopbackTransport, the client tests) keeps compiling unchanged through the\nre-export. The shared TransportError has two more variants (WrongState,\nTooLarge) and is non_exhaustive; nothing here matches on its variants.\n\nThe Display text changes slightly (\"transport round-trip failed\" instead of\n\"i3c transport round-trip failed\"); nothing matches on it.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "50dd1dc9efe7c246efe930d08b0be72778a753c1",
      "tree": "8418d694ec947a8d05f78e1eab8877e58ba4d6f0",
      "parents": [
        "109fa77727128f3b7bb9109ce59aea7dbae8af66"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 09:21:40 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "i3c: allow the server runtime and client-ipc on every kernel target\n\ni3c_server_runtime and i3c_client_ipc were gated with the veer-only\nTARGET_COMPATIBLE_WITH, but nothing in either crate is specific to one SoC: the\nruntime uses hal/blocking, i3c_api, i3c_server, util/service and pw_kernel, and\nthe client uses userspace. Use KERNEL_TARGET_COMPATIBLE_WITH from //target:defs.bzl\ninstead, as the MCTP server runtime does, so they build for ast10x0 and\nearlgrey as well as veer.\n\nBuilding for another target does not make the service work there: no other\nSoC implements the I3cTarget facade yet. This only removes a gate that had no\ntechnical reason.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "109fa77727128f3b7bb9109ce59aea7dbae8af66",
      "tree": "2ab2fe6f8525f5514d112375c78d2738312a926a",
      "parents": [
        "1a12ee3d5a2e94efc638e98e4783ee46eb7b2fdc"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 09:07:29 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "mctp api: share the MCTP transport-header length\n\nThe 4-byte DSP0236 transport-header length was hard-coded in three places: two\ntest files (MCTP_HEADER_SIZE) and inline in the I2C sender (MCTP_I2C_MAXMTU +\n4); an I3C binding would be a fourth. mctp_estack::MctpHeader::LEN is the same\nvalue but is not re-exported, so it cannot be reused.\n\nAdd openprot_mctp_api::MCTP_HEADER_LEN (host-buildable, already a dependency of\nevery consumer) and point the existing copies at it.\n\nPorted from origin/mctp-i3c-fpga-bench (f59f0672).\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "1a12ee3d5a2e94efc638e98e4783ee46eb7b2fdc",
      "tree": "d05ad2d591f459e2975abdaf1c9b7a77c7762b7f",
      "parents": [
        "f812955b26e59bc90b1ae234a10c2f586e833135"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 01:11:37 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "hal/blocking: fix a broken intra-doc link in the I3C module docs\n\npub mod i3c_hardware has an outer doc comment in lib.rs as well as the inner\ndocs in i3c_hardware.rs. rustdoc merges them and resolves their links from the\nparent scope, so [`I3cTarget`] was not found and the doc build failed under\n-D warnings (build ci_tests: \"unresolved link to `I3cTarget`\"). Link it by its\nfull path. No other change.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "f812955b26e59bc90b1ae234a10c2f586e833135",
      "tree": "e5017ac744cc2d7b93337cc888da6bb3aaaae57a",
      "parents": [
        "466626d32340eb08ee725f30c8992fa491ad048b"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Thu Nov 12 00:36:31 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "i3c service: refuse a second Send while one is unread, as caliptra-mcu-sw does\n\nOnly one response is staged at a time. A Send while the previous one has not\nbeen read by the controller is answered with the new I3cStatus::Busy and\nstages nothing; the client retries. caliptra-mcu-sw\u0027s I3C driver gates\noutbound data the same way (transmit_read returns ErrorCode::BUSY), and\ndocuments why: staging more before the first is consumed can overrun the\ntarget\u0027s TX queue.\n\nThis replaces holding the second Send until the first was read, which could\nleave the client blocked forever if the controller never read it:\n\n - dispatch refuses the Send, so the rule is host-tested and also applies to\n   the in-process loopback transport.\n - The runtime no longer holds a request (no defer/complete), and a new\n   inbound frame no longer releases anything: a controller may write again\n   before it reads. ResponseRead is the only release.\n - The driver now reports ResponseRead first when it arrives together with an\n   inbound frame. It was dropped in that case, because the runtime used to\n   release on the inbound frame; without that it would have left the target\n   refusing sends forever. The inbound frame is reported by the interrupt that\n   fires again after the acknowledge (RxDescStat is level-triggered).\n\nOn target the client now retries a Busy send. The host test makes the\ncontroller read the first reply with an exact-length read, which is what\nraises ResponseRead and releases the retry; on the FPGA it requires the send\nto be refused first, so it fails if the hardware model read the reply on its\nown. The emulator reads every reply itself, so there the second send is\naccepted at once.\n\nTested on the emulator and on the VCK190.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "466626d32340eb08ee725f30c8992fa491ad048b",
      "tree": "17bcd74d9757e12e93256453f5ce0cf0be2a3c06",
      "parents": [
        "55fb1e912a55275e776223369be99b720cd498ac"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 17:02:18 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer: sort the I3C peripheral BUILD deps\n\nThe presubmit format check wants //third_party/caliptra:crate_tock-registers\nbefore the @rust_caliptra_crates dependencies. No other change.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "55fb1e912a55275e776223369be99b720cd498ac",
      "tree": "270f7bea4f1a75129c277a3ee6835642590158fa",
      "parents": [
        "e9c618e4bc1bd04f559ac2ebc06dbd4366b6e14a"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 16:29:57 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer tests: remove the unused i3c_host crate\n\nThe I3C tests that used its socket helpers (i3c_backpressure, i3c_irq_preempt,\ni3c_user_irq) were removed because they could not load, and the i3c_service\ntest now drives the I3C socket through caliptra-mcu-sw\u0027s BufferedStream. Nothing\ndepends on the crate any more.\n\nThe originals are in e9a830de.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "e9c618e4bc1bd04f559ac2ebc06dbd4366b6e14a",
      "tree": "96ac993ead5ddc62c85829051b5ac04db69d1e0d",
      "parents": [
        "5048b56851744a12504e40f3830cae42d98823ff"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 16:27:46 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "i3c service test: exercise the held-Send back-pressure path\n\nThe client now sends a second reply before the host has read the first. The\nserver holds it (defer) and the host test releases it with a second write\n(the InboundReady release). The server logs \"i3c: send held\" so the test can\nsee the hold happen.\n\nThe test server raises IBIs with a non-MCTP MDB (0x1f). After an MCTP IBI\n(0xAE) the hardware model reads the reply itself at once, which clears the\nback-pressure before a second send can be held. With 0x1f the FPGA model\nleaves it queued, so on the FPGA the test requires the send to be held and\nfails if it is not. The emulator reads every reply regardless of the MDB, so\nthere the second send is not held and the test carries on.\n\nNot covered: release on ResponseRead (a private read by the host).\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "5048b56851744a12504e40f3830cae42d98823ff",
      "tree": "7ee4ba92b2d298ef419c1f768172ee77e6c5b8e4",
      "parents": [
        "f7800eb315672f166c5ae86d7667617e9ee30e8c"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 16:27:46 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "caliptra_test: tell the test which interface it runs on\n\nSet CALIPTRA_INTERFACE to \"emulator\" or \"fpga\" in every generated test target,\nso a host test that has to behave differently on one of them can check it.\nA caller\u0027s own env still overrides it.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "f7800eb315672f166c5ae86d7667617e9ee30e8c",
      "tree": "6726a29d839de40277753466c2cfa22c02a1c48a",
      "parents": [
        "7ce372640f2aa4cfd1ad439e3154163d5aba788c"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 15:56:36 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "i3c server-runtime: build on util_service::ServiceLoop\n\nReplace the hand-rolled WaitGroup loop with a util_service::ServiceLoop,\nas the flash and MCTP server runtimes do.\n\n - The I3C IRQ is added to the loop as a wake source (add_source); the IRQ\n   branch is unchanged: on_interrupt, latch inbound frames, interrupt_ack,\n   and raise USER on the client channel.\n - Requests arrive as Event::Request, already read by the loop.\n - Outbound back-pressure uses defer/complete in place of wait_group_remove\n   and wait_group_add by hand. The held Send\u0027s bytes are still kept by the\n   runtime, since ServiceLoop does not store them.\n\nBehavior changes:\n\n - run() returns Result\u003c()\u003e instead of never returning, like the flash and\n   MCTP runtimes. Only setup and WaitGroup failures return; the test\u0027s\n   server process logs the error.\n - interrupt_ack acks the subscribed signals instead of pending \u0026 subscribed,\n   because next() does not report pending signals. Equivalent for the single\n   I3C IRQ signal.\n - Holding a second Send can fail, for example if the loop refuses to defer\n   the reply. The old code logged the error and left the client blocked\n   forever. Now the server replies to that client straight away with an\n   Internal error. The Send is dropped, not queued for transmit.\n\nThe held-Send path is not covered by any test on target: its test,\ni3c_backpressure, was removed earlier because it could not load.\n\nPasses on the emulator and on the VCK190.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "7ce372640f2aa4cfd1ad439e3154163d5aba788c",
      "tree": "a5c22fba67d8e2343264ea2dc3a93851ff3a0304",
      "parents": [
        "09aa258879485604b84608304c9886ecf4220a86"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 15:14:15 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer tests: remove the I3C emulator tests that cannot load\n\ni3c_backpressure, i3c_irq_preempt and i3c_user_irq drive the emulator through\ncaliptra_runner, which no longer exists in target/veer/tooling/caliptra_runner.bzl,\nso none of the three packages load and any wildcard over //target/veer/... fails\non them. Their host sides also depend on the emulator-process runner that the\nhardware-model tests replace.\n\nRemove them so the tree builds again. i3c_service is the I3C test that runs on\nboth the emulator and the VCK190. They can be brought back later as\ncaliptra_test targets; the originals are in e9a830de.\n\nAlso drop the comment in the i3c_service template that pointed at them.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "09aa258879485604b84608304c9886ecf4220a86",
      "tree": "1e0233e402554e2a11b1c22725e6af628fb8ea84",
      "parents": [
        "234b8ad359fb8cef0267da7ba1fe4cfa545496d0"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 15:11:36 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer tests: run the I3C service test through the hardware model, on emulator and FPGA\n\nThe I3C service test drove the emulator through caliptra_runner, which no\nlonger exists in target/veer/tooling/caliptra_runner.bzl, so the package did\nnot load. It also loaded rust_process from a .bzl that the pinned Pigweed does\nnot have (it is in rust_app.bzl).\n\nMove it onto the same tooling as the other VeeR tests:\n\n - system.json5 becomes system.json5.tmpl, rendered per platform by\n   veer_system_config: SRAM comes from @KERNEL_FLASH_START@ and\n   @KERNEL_RAM_START@, and the I3C register block from @I3C_BASE@.\n - The image is built with veer_system_image, which also defines the signed\n   i3c_service_image_image_emulator and _fpga targets.\n - The host side is a caliptra_test that drives the firmware through\n   caliptra_mcu_hw_model, following caliptra-mcu-sw\u0027s own\n   test_i3c_services: it boots with an I3C socket, waits for the firmware\u0027s\n   \"waiting for private write\" log line, sends one private write with\n   BufferedStream, and waits for the firmware to log that it received the\n   expected payload and exit with PASS. That generates\n   i3c_service_test_emulator and i3c_service_test_fpga.\n\nThe write reaches the real I3C target and its interrupt on the VCK190, and\nthe server wakes the client from that interrupt, so the test exercises the\nI3C receive interrupt path on hardware. Reading the staged reply back is not\ncovered: the driver does not append a PEC, which BufferedStream\u0027s private\nread requires.\n\nPasses on the emulator and on the VCK190 (with the VeeR claim patch).\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "234b8ad359fb8cef0267da7ba1fe4cfa545496d0",
      "tree": "436a246e974f3d7d6dc0dd993d86534126840894",
      "parents": [
        "24d5429696c309b2541eb1f70b835fdb07844ebf"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 15:09:47 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer: add an @I3C_BASE@ placeholder to the system config templates\n\nrender_system_config fills @KERNEL_FLASH_START@, @KERNEL_RAM_START@ and\n@MCI_BASE@ from caliptra-mcu-sw\u0027s EMULATOR_MEMORY_MAP / FPGA_MEMORY_MAP. Add\n@I3C_BASE@, taken from the same map\u0027s i3c_offset, so a template can map the\nI3C register block at the right address for either platform instead of\nhard-coding the emulator\u0027s 0x20004000.\n\nNothing uses it yet; the I3C service test template does in the next change.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "24d5429696c309b2541eb1f70b835fdb07844ebf",
      "tree": "91b97da6f78e0103c52ab2ab8fdebbce77e66eea",
      "parents": [
        "2f81a1b25db5ff710f97d0937d682ebf5e71abb7"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 15:08:15 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "pigweed: drop the MEICPCT write from the VeeR external-interrupt claim\n\npw_kernel\u0027s VeeR PIC handler claims an interrupt by writing MEICPCT (CSR\n0xBCA) and then reading MEIHAP. On the Caliptra FPGA\u0027s VeeR core the write\ntraps as an illegal instruction: running the I3C service test on the VCK190\ntook an exception (mcause 2, mtval 0xbca59073 \u003d csrw 0xbca, a1) right after\nthe I3C interrupt fired. caliptra-mcu-sw\u0027s own VeeR kernel never writes\nMEICPCT; it reads MEIHAP (0xFC8) directly, and that is what runs on the FPGA.\n\nAdd third_party/pigweed/veer_pic_claim.patch, wired into the pigweed\ngit_override after veer_pic_gateways.patch, to read MEIHAP directly. With it\nthe I3C service test passes on the VCK190 and on the emulator, and the\nemulator interrupts test still passes.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "2f81a1b25db5ff710f97d0937d682ebf5e71abb7",
      "tree": "d2618d002534dace7e6ddcb35d366b9eeaa96ed4",
      "parents": [
        "91cd3d65bf11c2ce3eec8aaac1e65b3ade78cc5e"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 10:50:41 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer tests: add I3C emulator tests\n\ni3c_host is the shared host-side I3C socket harness (wire protocol and PEC).\nThe firmware tests, each driven by a host test over the emulator\u0027s I3C socket:\n- i3c_service: the I3C client and server over real IPC.\n- i3c_backpressure: three staged responses, each Send held until the\n  controller reads the previous one.\n- i3c_user_irq: one inbound private write delivered to userspace.\n- i3c_irq_preempt: I3C interrupt delivery under preemption.\n\nThese use the emulator-only caliptra_runner path, not the --config\u003dfpga flow.\nNot yet built or run on this base.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "91cd3d65bf11c2ce3eec8aaac1e65b3ade78cc5e",
      "tree": "0d36e9d4712a05149103959326e54c5c56a33fd2",
      "parents": [
        "095042b472f16dbdf1fa9b3e5c1f0d0e045b9bef"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 10:50:41 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "veer: add the Caliptra-SS I3C target driver\n\nCaliptraI3cTarget implements I3cTarget for the Caliptra-SS i3c-core in target\nmode. The register base comes from caliptra-mcu-sw\u0027s memory map for the\nselected target type (emulator or FPGA).\n\nNotable behaviour:\n- Inbound frames are drained by RX descriptor queue depth, since RxDescStat can\n  drop while a descriptor is still queued on the FPGA.\n- ResponseRead is derived from TxDescComplete, which is only armed by\n  enable_response_read_events(); each interrupt status bit is handled\n  separately.\n- A failed IBI (non-zero LastIbiStatus) is raised again, up to 16 times.\n\nNot yet built on this base.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "095042b472f16dbdf1fa9b3e5c1f0d0e045b9bef",
      "tree": "41857fcf1c129b8df8428d3b35e64dfb3638ede7",
      "parents": [
        "0e5984b4c76805f3bed6e966bf4052386a2cd311"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 10:50:41 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "services/i3c: add the I3C service, client and IPC transport\n\n- api: the request/response wire format and the Transport seam.\n- server: an in-process I3C server that latches inbound frames into a ring and\n  stages outbound responses, with a loopback transport for host tests.\n- server-runtime: the pw_kernel loop that services the I3C interrupt and the\n  IPC channel, and holds a second Send until the controller has read the first\n  (back-pressure).\n- client and client-ipc: a typed client over the Transport, and the kernel IPC\n  transport.\n\nNot yet built on this base.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "0e5984b4c76805f3bed6e966bf4052386a2cd311",
      "tree": "39229bd9a384f9525e06f2bf175ad1a6d890d16d",
      "parents": [
        "3c0f676ec6dd9eaec61bab8d6ed0ff7e2bd4a34a"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 10:50:41 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "hal/blocking: add the I3C target hardware abstraction\n\nI3cTarget and its supporting types (DynamicAddress, I3cTargetConfig,\nTargetEvent) describe an I3C controller\u0027s target (secondary) mode: bring-up,\ninterrupt servicing, reading inbound frames and staging a response with an IBI.\nenable_response_read_events() is optional (default no-op) and arms the\nresponse-read event for runtimes that hold a second send until the first has\nbeen read.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "3c0f676ec6dd9eaec61bab8d6ed0ff7e2bd4a34a",
      "tree": "50376f32f71ce8e3d758854309558333c5891923",
      "parents": [
        "b5de3f90c250aea4f672ef00dbed9fdb0c909167"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Wed Nov 11 10:50:41 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Thu Nov 12 13:26:51 2026 -0800"
      },
      "message": "pigweed: make the VeeR PIC gateways level-triggered, active high\n\nAdd a local Pigweed patch, third_party/pigweed/veer_pic_gateways.patch, wired\ninto the pigweed git_override.\n\npw_kernel\u0027s VeeR PIC init sets each interrupt\u0027s priority but never writes its\ngateway control register (MeiGwCtrl), so the gateway type and polarity are\nwhatever the reset value, or a previous boot stage, left behind. The patch\nwrites MeiGwCtrlValue(0) for every interrupt at init, which selects\nlevel-triggered, active high, as caliptra-mcu-sw\u0027s PIC driver does explicitly.\nA no-op if the reset value already matches.\n\nNot yet checked against the pinned Pigweed commit: it was written against an\nolder pin.\n\nCo-Authored-By: Claude Sonnet 5.5 \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "b5de3f90c250aea4f672ef00dbed9fdb0c909167",
      "tree": "f7bc50d27c1003d10659e5dc649e0f5e7430dc0b",
      "parents": [
        "2cd5974e16e22d45537435fa7f616d643e71d905"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Thu Nov 12 12:33:51 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Thu Nov 12 12:45:28 2026 -0800"
      },
      "message": "veer: Check the mailbox lock and tie SHA sizes to caliptra-api\n\nAfter taking the mailbox lock, read the lock register again and return\nan error if it is not held, as caliptra-sw and caliptra-mcu-sw do.\n\nTake the response header size from MailboxRespHeader and check at\ncompile time that the SHA state has the size of CmShaInitResp.\n\nDocument that a request is always a whole number of words.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "2cd5974e16e22d45537435fa7f616d643e71d905",
      "tree": "02f9252da17f98ced8610534bfcbedc72f4cb327",
      "parents": [
        "596765783732cbc6d16c2f7199f9a7eb2930a721"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Wed Nov 11 15:16:12 2026 -0800"
      },
      "committer": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Thu Nov 12 12:45:28 2026 -0800"
      },
      "message": "veer: Add a SHA-2 driver that uses the Caliptra mailbox\n\nThe VeeR core has no hash hardware, so the driver sends the Caliptra\nCM_SHA commands over the mailbox. It implements the scoped and owned\ndigest HAL traits for SHA-384 and SHA-512. The CM_SHA commands do not\nsupport SHA-256.\n\nThe mailbox driver is a port of the mailbox code in caliptra-mcu-sw\nromtime. It handles requests and responses as 32-bit words and polls\nfor the response.\n\nThe test checks the digests of known messages and passes on the\nemulator and on the VCK190 FPGA. It is a userspace app because the\nstack of the kernel\u0027s bootstrap thread is too small for it.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "596765783732cbc6d16c2f7199f9a7eb2930a721",
      "tree": "5e655bbd1870f6fcb24639df7a5df9951f5fb4dd",
      "parents": [
        "68c1c9b2eb1861a428ebeebfcc84158851c3ed59"
      ],
      "author": {
        "name": "Evan Chuchen",
        "email": "evan_chuchen@jabil.com",
        "time": "Tue Oct 13 18:57:29 2026 +0800"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Thu Nov 12 09:05:52 2026 -0400"
      },
      "message": "target/ast10x0: add QEMU and hardware tests for the control responder\n\nAdd two ast10x0 images for services/mctp/control. The QEMU image runs\nin CI: fake_server_main stands in for the MCTP server and the bus\nowner, sends 11 requests, and checks every reply byte and the IPC\nheader the reply goes out under. The hardware image answers a BMC on\nI2C bus 0 at slave address 0x38 as a static endpoint at EID 8, and\nadds the PINCTRL_I2C0 pin group that bus 0 needs.\n\nTested: mctp_control_qemu_test and qemu_no_panics_test pass under\n--config\u003dvirt_ast10x0, and mctp_control_image builds under\n--config\u003dk_ast1060_evb.\n\nSigned-off-by: Evan Chuchen \u003cevan_chuchen@jabil.com\u003e\n"
    },
    {
      "commit": "68c1c9b2eb1861a428ebeebfcc84158851c3ed59",
      "tree": "880a8dff0247bd1a18f0941fb0c5399b0f927586",
      "parents": [
        "67ffab77d90303300bde54ffe2e9ae8d62806b2c"
      ],
      "author": {
        "name": "Evan Chuchen",
        "email": "evan_chuchen@jabil.com",
        "time": "Tue Oct 13 18:57:20 2026 +0800"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Thu Nov 12 09:05:52 2026 -0400"
      },
      "message": "services/mctp: add an MCTP control protocol responder\n\nNothing in the tree answered MCTP control (message type 0x00), so a bus\nowner could not assign this endpoint an EID or learn what it supports.\n\nservices/mctp/control answers Set Endpoint ID, Get Endpoint ID, Get\nMCTP Version Support and Get Message Type Support, and answers\nERROR_UNSUPPORTED_CMD for every other command. The crate holds no\ntransport: handle() turns request bytes into response bytes, and the\nEID lives in the caller\u0027s MCTP stack. The crate README records the\ndesign and the deferred commands.\n\nTested: bazelisk test //services/mctp:mctp_host_tests, 8 targets pass.\nSigned-off-by: Evan Chuchen \u003cevan_chuchen@jabil.com\u003e\n"
    },
    {
      "commit": "67ffab77d90303300bde54ffe2e9ae8d62806b2c",
      "tree": "d8f63a1cc9bb5113a065e170da1a2d2fb39c6e60",
      "parents": [
        "3f670bbfc40ad512de6eb717fa94a98aa6da3571"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 10 17:45:57 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 12:24:49 2026 -0800"
      },
      "message": "veer: Serve the VeeR flash driver through the flash service\n\nAdd an image with two apps. The flash server serves the primary flash\ncontroller of the emulator or the FPGA with SoftFlash and\nflash_server_runtime. The flash client erases, programs and reads back\nflash over IPC, and checks that the server rejects an out-of-bounds read\nand an unaligned erase.\n\nThe controller\u0027s address differs between the emulator and the FPGA and\nis not in caliptra-mcu-sw\u0027s memory maps, so render_system_config gets a\n@PRIMARY_FLASH_CTRL@ placeholder for it.\n\nThe emulator test passes. The FPGA image builds but has not been run on\na board.\n\nAssisted-by: Claude:claude-opus-5-5\n"
    },
    {
      "commit": "3f670bbfc40ad512de6eb717fa94a98aa6da3571",
      "tree": "9c1c437ab9f6491aa1b29a52371eba68ca431f32",
      "parents": [
        "2a5ef23b4a5d9d42950aa0a93da9055d827ca287"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Wed Nov 11 09:59:00 2026 -0400"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 10:56:15 2026 -0800"
      },
      "message": "added earlgrey to .bazelrc and update README.md\n"
    },
    {
      "commit": "2a5ef23b4a5d9d42950aa0a93da9055d827ca287",
      "tree": "a985408579a6ece67abab518b9cf87ec2246f9e0",
      "parents": [
        "a4b5c79042850c077280bcd4571690e1c4a0c329"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Wed Nov 04 14:04:05 2026 -0400"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 10:56:15 2026 -0800"
      },
      "message": "docs: add instructions to setup vscode for a better development experience\n"
    },
    {
      "commit": "a4b5c79042850c077280bcd4571690e1c4a0c329",
      "tree": "f80c993b62ff4f9854e6d917c9612b26b8ed934e",
      "parents": [
        "05a858197cd25e14d9b47d6b1757c705912fc43c"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Wed Nov 04 14:02:54 2026 -0400"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 10:56:15 2026 -0800"
      },
      "message": "development: add rust-analyzer to rust-toolchain.toml component\n"
    },
    {
      "commit": "05a858197cd25e14d9b47d6b1757c705912fc43c",
      "tree": "6667b7d6bd43242e1321bf08a797039bb0b57d4c",
      "parents": [
        "9fb347557fd2bf92e2801e26ef7e1cd5587b7ee9"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 08 18:54:45 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 08:43:22 2026 -0800"
      },
      "message": "services/flash/server-runtime: Serve one flash device per channel\n\nA controller with several chip selects needs several devices behind one\nserver process. Give each device its own channel and Server and run them\nthrough util_service\u0027s loop. The channel a request arrives on picks the\ndevice, so the wire protocol is unchanged and a client can only reach the\ndevices it was given a channel to.\n\nrun now returns an error if the WaitGroup fails instead of retrying the\nwait forever, and takes the WaitGroup handle and the channels in place of\na single channel handle.\n"
    },
    {
      "commit": "9fb347557fd2bf92e2801e26ef7e1cd5587b7ee9",
      "tree": "4961fdb6a45abb3d873d108002039e9d80ff4f80",
      "parents": [
        "ba0ef83a6a52ac05533a5c4637dbd15a5e1e7a8b"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 08 18:54:45 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 08:43:22 2026 -0800"
      },
      "message": "util/service: Let a caller see which channel a request came on\n\nA server that keeps one handler per channel, such as one flash device per\nchip select, has to pick the handler for the channel a request arrived\non. ChannelId only identified the channel to the loop. Add index(), the\nposition of the channel in the slice given to ServiceLoop::new.\n"
    },
    {
      "commit": "ba0ef83a6a52ac05533a5c4637dbd15a5e1e7a8b",
      "tree": "95c8c12c80258afb70143c29fcddb03f70d7ffd2",
      "parents": [
        "0c92fc4568e09637e98b7aa19cbb795e58ae477d"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 08 21:24:11 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 08:43:22 2026 -0800"
      },
      "message": "services/flash/server-runtime: Run the flash server on the kernel\n\nThe runtime reads a request frame from the channel, calls\nServer::handle_frame, and responds with the frame it leaves in the same\nbuffer. It also takes over the wait loop the target apps carried. It is\nthe only part of the server that does not build for the host.\n"
    },
    {
      "commit": "0c92fc4568e09637e98b7aa19cbb795e58ae477d",
      "tree": "38c48c3640cc017b46f9fbb5e24adffd307d5663",
      "parents": [
        "c16360669237c7e8f3d7c34e1f9791a1ef4bbf7e"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 08 21:24:11 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Wed Nov 11 08:43:22 2026 -0800"
      },
      "message": "services/flash/client-ipc: Carry the flash client over a kernel channel\n\nChannelTransport is a FlashTransport over one kernel channel. It passes\nthe pieces of a request and of a response to the kernel unchanged, so the\nclient still copies nothing. FlashIpcClient is FlashClient over it. This\nis the client that kernel apps use, and the only part of the client that\ndoes not build for the host.\n"
    },
    {
      "commit": "c16360669237c7e8f3d7c34e1f9791a1ef4bbf7e",
      "tree": "11125d08e6c9d217dd4817b72dc2ac30d346170d",
      "parents": [
        "d095ccbead648f3863ed64d1c72f97c6393e18f3"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 10 11:03:43 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 16:25:00 2026 -0800"
      },
      "message": "veer: Add a flash driver for the FPGA\u0027s flash controllers\n\nThe Caliptra MCU FPGA has a primary and a secondary flash controller.\nTheir registers are in the FPGA wrapper, and caliptra-sw\u0027s\nModelFpgaSubsystem, which runs on the board\u0027s ARM cores, does the work\nof the controllers. Add FpgaFlashCtrl so that SoftFlash supports them.\n\nThe driver has not been run on a board. A host test runs it against a\ncopy of the caliptra-sw host code, with the register layout compiled\nfrom caliptra-sw\u0027s fpga_regs.rs.\n\nAssisted-by: Claude:claude-fable-5-1\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "d095ccbead648f3863ed64d1c72f97c6393e18f3",
      "tree": "92dd673931db8ee60079364a7ce212e88798b0ad",
      "parents": [
        "567a1670aff97519c61ba2cd838399e100003a0d"
      ],
      "author": {
        "name": "Amit Kumar-Hermosillo",
        "email": "amitkh@google.com",
        "time": "Tue Nov 03 20:21:58 2026 +0000"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 16:24:16 2026 -0800"
      },
      "message": "[veer] document FPGA test tooling\n\nAdds target/veer/tooling/fpga/README.md covering how the FPGA test flow works\n(run_remote.sh / run_on_board.sh), board setup, running tests, writing\non-device and host-driven tests with the caliptra_test.bzl macros, updating\nthe bitstream, and troubleshooting. Points .bazelrc and\ndocs/src/architecture.md at it.\n"
    },
    {
      "commit": "567a1670aff97519c61ba2cd838399e100003a0d",
      "tree": "fcad3f3fa5597d28642408ab575cc9cea9fa4cbd",
      "parents": [
        "32a9227a7309c6542ef3c61c56e663a9bc211a76"
      ],
      "author": {
        "name": "Amit Kumar-Hermosillo",
        "email": "amitkh@google.com",
        "time": "Tue Oct 27 17:29:20 2026 +0000"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 16:24:16 2026 -0800"
      },
      "message": "[veer] add FPGA board bootstrap and remote test runner\n"
    },
    {
      "commit": "32a9227a7309c6542ef3c61c56e663a9bc211a76",
      "tree": "e2f314379e71c8c25e2136fff9869ddf67bb54db",
      "parents": [
        "be566399214a0bdee220a6c19c1c9c6a5fce6280"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 10 14:47:37 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 16:06:16 2026 -0800"
      },
      "message": "veer: Return NonZero from SoftFlash::erasable_sizes_bitmap\n\nFlashDriver::erasable_sizes_bitmap now returns NonZero\u003cu32\u003e. The VeeR\nflash driver merged just before that change and still returned u32, so\nmain does not build.\n\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "be566399214a0bdee220a6c19c1c9c6a5fce6280",
      "tree": "276a775747667cf11ee54f4d6162f38c936fc9e8",
      "parents": [
        "44aa3dd906c96b8c8cbd0d0f315d9b7ec97f3c41"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Fri Nov 06 13:39:15 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 14:31:31 2026 -0800"
      },
      "message": "Revising terminology\n\nTo answer the questions in #668:\n\n\u003e What does each of platform, target, SoC, board and product mean in this\n\u003e project? Write the definitions down once.\n\n**Product** could also be called system, board, or platform. Platform is\nproblematic since this is a \"platform\" root of trust, i.e., the thing\nthat we are protecting, not the system we\u0027re building firmware for.\n\n**Product class** is perhaps something helpful to define, as\n\"Caliptra MCU\" is really a very loose meta-architecture that\nhas many versions and implementations by vendors, but is useful\nfor certain drivers to be written for it.\n\n**Chip** could also be called SoC, CPU, processor, or MCU (though a common\nabbreviation for \"microcontroller\", this would be confusing as Caliptra\nMCU is a distinct class of products).\n\nDriver source code for hardware should generally be chip-specific, but not\nproduct-specific, except for final configuration details for a specific memory\nmap, address set, interrupt numbering, etc.\n\n\"Platform\" and \"target\" should not be used directly by OpenPRoT unless\nprefixed with another project for clarity, e.g., \"pigweed target\".\n\nThis will need to be reconciled with the rest of the docs and the openprot directory\nstructure. If we like this, I can come up with an issue to track migrating\nterminology and directories.\n\n\u003e Is a board part of the SoC support, or of the product that uses it?\n\nI think each physical board, which would have its own firmware, would be\na separate product. If we needed multiple boards (like the AST1060 EVB and\ntest harness), presumably they would want similar but different firmware\nso would have their own products, though they could reuse much of their\nconfiguration and code.\n\n\u003e What is platform/ for? Options: rename it to match its contents, move\n\u003e the software implementations elsewhere, or remove what is unused.\n\nI think we should merge `platform/` into `openprot/`, which is otherwise\nessentially unused? It seems like the \"core\" of what OpenPRoT is going\nto be is going to be the internal and external service traits\nand their implementations, along with the chip-specific drivers.\n\nI\u0027d propose consolidating traits under `openprot/`, perhaps:\n- `openprot/ipc` internal services, e.g., cryptography\n- `openprot/ext` external services, e.g., over MCTP\n\nThe existing `platform/` contents can put into this framework\nor archived, since a lot is more about multi-OS support, which\nwe don\u0027t really have as a goal anymore, from what I understand.\n(**Is this correct now, as the docs say otherwise?**)\n\n\u003e Should the top-level `target/` be renamed so it doesn\u0027t collide with\n\u003e pigweed\u0027s \"target\" and Bazel\u0027s?\n\nYes. I think we should move split it into `product/` and `chip/`.\nRust development also uses `target/`, so is often in people\u0027s\nignore files for Rust development.\n(In fact, this caused me a fair bit of difficulty when starting since I had\na global rust-analyzer ignore for \"target\".)\n\n\u003e What should happen to the unused mock hardware: wire its tests into\n\u003e Bazel, or archive it?\n\nArchive it. I would say to trim down the repository and revive the code\nif necessary when we need it.\n\nPartially addresses #668. We can split #668 into multiple sub-issues\nto track migrating these terms and reorganizing the directories and making\nthe rest of the documentation match this.\n"
    },
    {
      "commit": "44aa3dd906c96b8c8cbd0d0f315d9b7ec97f3c41",
      "tree": "f06e342fef7a79d9b892c0a13d52dccc685cc10b",
      "parents": [
        "cd9251cb74ff23ef5e84ac2e521a052de294140b"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Tue Nov 10 15:46:23 2026 -0500"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 14:30:31 2026 -0800"
      },
      "message": "flash: fix unaligned-read bounds, reject non-4K sector devices, and guard capacity overflow on 32-bit\n"
    },
    {
      "commit": "cd9251cb74ff23ef5e84ac2e521a052de294140b",
      "tree": "fa8bed91d645ab2dab44d7d4e26b7187d4b85c10",
      "parents": [
        "68af0d832facea7edd89e844d0df3cdedc474299"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Tue Nov 10 15:41:18 2026 -0500"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 14:30:31 2026 -0800"
      },
      "message": "flash: make program-window and erase-bitmap power-of-two/nonzero checks part of the FlashDriver interface\n"
    },
    {
      "commit": "68af0d832facea7edd89e844d0df3cdedc474299",
      "tree": "a60d78619986d1e1d6f9e4c2d11af125d179f430",
      "parents": [
        "c10ed4633d02e660ef6fd2f58ecf4a696ff5eb0f"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Tue Nov 10 11:21:29 2026 -0500"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 14:30:31 2026 -0800"
      },
      "message": "flash: replace runtime power-of-two check with compile-time proof at the source\n"
    },
    {
      "commit": "c10ed4633d02e660ef6fd2f58ecf4a696ff5eb0f",
      "tree": "5bd17b3c4a479f6c39262344815474dfa0fea866",
      "parents": [
        "79aca05f1746442e74dd5cd066f35d086ccae929"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Thu Nov 05 14:58:24 2026 -0500"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 14:30:31 2026 -0800"
      },
      "message": "ast10x0: bring the FMC flash backend to main from ocp-global-demo\n"
    },
    {
      "commit": "79aca05f1746442e74dd5cd066f35d086ccae929",
      "tree": "1d33ed64c83b525a0e09ba18a9d786bf9f0b9260",
      "parents": [
        "9a9575dfe0266cae7e57525502913dc93e02305b"
      ],
      "author": {
        "name": "Christopher Swenson",
        "email": "swenson@swenson.io",
        "time": "Tue Nov 10 10:37:46 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 14:28:22 2026 -0800"
      },
      "message": "veer: Add a flash driver for the emulator\u0027s flash controllers\n\nThe Caliptra MCU emulator has a primary and a secondary flash\ncontroller, implemented in software by DummyFlashCtrl in\ncaliptra-mcu-sw. This driver implements hal_flash_driver::FlashDriver\nfor them. It presents each controller as NOR flash with a 256-byte\nprogram window and 4 KiB erase blocks.\n\nThe tests are host unit tests, an on-device test that runs on the\nemulator, and a no_panics_test for the test image. A driver for the\nFPGA\u0027s flash controllers will follow in a separate change.\n\nAssisted-by: Claude:claude-fable-5-1\nAssisted-by: Claude:claude-fable-5-2\n"
    },
    {
      "commit": "9a9575dfe0266cae7e57525502913dc93e02305b",
      "tree": "dd31caecad2619877b9673a81a181bd0186da9dc",
      "parents": [
        "8f05c171baf8fee842f45842fc789b0b15b5befb"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Fri Nov 06 11:22:28 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 17:17:12 2026 -0400"
      },
      "message": "Define pigweed workflow to run flaky tests assesment\n"
    },
    {
      "commit": "8f05c171baf8fee842f45842fc789b0b15b5befb",
      "tree": "b49ab4ba475ab95670a77e93bf0a3dc3ea9f229c",
      "parents": [
        "3edd48da82afdfcc154c02e74c82c5f86e565947"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "eparache@amd.com",
        "time": "Tue Nov 10 16:00:47 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 16:55:06 2026 -0400"
      },
      "message": "presubmit: Tell users how to regenerate MODULE.bazel.lock\n\nDon\u0027t assume the regenerated lockfile is in the user\u0027s working tree (CI\nruns on a throwaway checkout). On failure, print the command that updates\nthe lockfile and say it must be committed.\n\nCo-Authored-By: Claude \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "3edd48da82afdfcc154c02e74c82c5f86e565947",
      "tree": "d276179d00fc0191fe8e1fb77493080adb6042f9",
      "parents": [
        "078a901c3a7bba4fea90367a24fd0568d8f569aa"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "eparache@amd.com",
        "time": "Tue Nov 10 15:55:48 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 16:55:06 2026 -0400"
      },
      "message": "presubmit: Compare regenerated lockfile to git instead of using error mode\n\n./pw and the other presubmit steps run Bazel in the default\n--lockfile_mode\u003dupdate, which rewrites a stale MODULE.bazel.lock before\nlockfile_check runs. `mod deps --lockfile_mode\u003derror` then validated the\nregenerated file, not the committed one, and could never fail. Regenerate the\nlockfile and fail if `git diff` shows a change.\n\nCo-Authored-By: Claude \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "078a901c3a7bba4fea90367a24fd0568d8f569aa",
      "tree": "014b7bd7d0ba64eea2eef84d365f3591cfabcb34",
      "parents": [
        "6a606fd8df56d435062c0cfb501c4f1deec2ada1"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "eparache@amd.com",
        "time": "Tue Nov 10 15:20:52 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 16:55:06 2026 -0400"
      },
      "message": "presubmit: Run the lockfile check from ./pw presubmit\n\nAddress review feedback on #674: move the `bazel mod deps\n--lockfile_mode\u003derror` check out of ci.yml and into workflows.json as a\n`lockfile_check` analyzer in the presubmit group, so it runs locally and in\nCI through the same entry point. The pw driver only runs bazelisk\nbuild/test/run, so a small pw_py_binary invokes the command.\n\nCo-Authored-By: Claude \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "6a606fd8df56d435062c0cfb501c4f1deec2ada1",
      "tree": "d5b18e91b4dc522574e7f496230b6ec62681cc7a",
      "parents": [
        "3c4c31c6d478ce89f876de2a3877504f4dddfa7f"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Tue Nov 10 14:47:18 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 16:55:06 2026 -0400"
      },
      "message": "deps: ran bazel mod deps --lockfile_mode\u003dupdate\n"
    },
    {
      "commit": "3c4c31c6d478ce89f876de2a3877504f4dddfa7f",
      "tree": "58fc35b933d94c743b7334e630d8a4b5114e1f2b",
      "parents": [
        "b9e2542d02feef49c8ba2b3e5025e8252268bdfc"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "eparache@amd.com",
        "time": "Tue Nov 10 14:21:20 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 16:55:06 2026 -0400"
      },
      "message": "ci: Publish release after provenance verification; check lockfile in presubmit\n\nAddress review feedback on #674:\n- Move release creation into a publish-release job that needs\n  verify-provenance, so a failed verification never leaves a published\n  release. build-firmware no longer needs contents: write.\n- Run `bazel mod deps --lockfile_mode\u003derror` in the CI presubmit job so\n  MODULE.bazel.lock drift is caught on every PR, not only at release time.\n\nCo-Authored-By: Claude \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "b9e2542d02feef49c8ba2b3e5025e8252268bdfc",
      "tree": "cff033d87d633e4afd391d98937279605deab0bc",
      "parents": [
        "b18c95b43885ea0606c866358e25de5a1b1fde99"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "eparache@amd.com",
        "time": "Fri Nov 06 14:48:03 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 16:55:06 2026 -0400"
      },
      "message": "ci: Make release cquery match the build configuration\n\ncquery inherited the `test` rc settings (--platform_suffix\u003dtest and\nenable_tests\u003dtrue), so it reported outputs under k8-fastbuild-test while the\nbuild wrote to k8-fastbuild. Reset both flags and stop hiding cquery errors.\n\nCo-Authored-By: Claude \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "b18c95b43885ea0606c866358e25de5a1b1fde99",
      "tree": "46631b2e9f3e5b2487f13d038a079165e80a9689",
      "parents": [
        "f06d76a3dcab0715f72670283d1a037e6d8e08fd"
      ],
      "author": {
        "name": "Parache",
        "email": "eparache@amd.com",
        "time": "Fri Nov 06 12:44:21 2026 -0400"
      },
      "committer": {
        "name": "Enmanuel Parache",
        "email": "EnmanuelParache@users.noreply.github.com",
        "time": "Tue Nov 10 16:55:06 2026 -0400"
      },
      "message": "ci: Add SLSA Build L2 provenance for release firmware\n\nAdd a release workflow that builds firmware on a GitHub-hosted runner with\na locked Bazel module graph and publishes signed build provenance via\nactions/attest-build-provenance, then re-verifies it. Harden existing\nworkflows by pinning all actions to commit SHAs and declaring\nleast-privilege permissions. Document verification and known gaps.\n\nCo-Authored-By: Claude \u003cnoreply@anthropic.com\u003e\n"
    },
    {
      "commit": "f06d76a3dcab0715f72670283d1a037e6d8e08fd",
      "tree": "4bd7ce3371cbeb417bc7ad6849bf4cebbf119bae",
      "parents": [
        "7496e0049c6073023ce8ee77261984e1d9f2aeda"
      ],
      "author": {
        "name": "Enmanuel Parache",
        "email": "enmanuelparache@gmail.com",
        "time": "Tue Nov 10 08:49:33 2026 -0400"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Tue Nov 10 08:37:07 2026 -0800"
      },
      "message": "cleanup: removed unused directory services/storage\n"
    },
    {
      "commit": "7496e0049c6073023ce8ee77261984e1d9f2aeda",
      "tree": "483e731e331902941e0c81c0e6bd5041c708a830",
      "parents": [
        "1929cb854a2652e89c6a94bbab8fabd2e42abbf4"
      ],
      "author": {
        "name": "Amit Kumar-Hermosillo",
        "email": "amitkh@google.com",
        "time": "Mon Oct 26 14:08:14 2026 +0000"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 09 14:45:42 2026 -0800"
      },
      "message": "[veer] unify emulator and FPGA tests on caliptra_mcu_hw_model\n\nCreate macros and rules to define Caliptra images and tests for both emulator\nand FPGA that transition and cross-compile firmware and tests\nappropriately for each platform.\n"
    },
    {
      "commit": "1929cb854a2652e89c6a94bbab8fabd2e42abbf4",
      "tree": "0011a1a8c293380ca1a2dde5ab677c3b3d5baf35",
      "parents": [
        "d2d49fcf4857b0d20f6f24cbe479caa4d651aba9"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Thu Nov 05 10:28:08 2026 -0500"
      },
      "committer": {
        "name": "JesseChambers",
        "email": "145486573+JesseMelon@users.noreply.github.com",
        "time": "Mon Nov 09 13:29:23 2026 -0800"
      },
      "message": "cortex-m-rt: enable the device feature\n\ncortex-m-rt emits __INTERRUPTS, a 240-entry array of DefaultHandler\npointers, into .vector_table.interrupts. pw_kernel\u0027s generated interrupt\ntable targets that same section. The linker concatenates both in input\norder, and only whichever lands first occupies the IRQ slots the core\nactually reads at vector_table_base + 0x40.\n\nNothing in either image chose the winner. The i2c_slave_rx_burst slave\nimage happened to place pw_kernel\u0027s table at 0x40 and worked; the master\nimage placed cortex-m-rt\u0027s filler there instead, pushing the real table\n960 bytes clear of where the core looks. Its first I2C interrupt\ndispatched to DefaultHandler and panicked.\n\nThe device feature is cortex-m-rt\u0027s own switch for \"an external crate\nsupplies the interrupt vector table\", which is exactly our arrangement.\nIt drops the filler, leaving pw_kernel\u0027s table as the only contributor so\nit lands at 0x40 deterministically in every image.\n"
    },
    {
      "commit": "d2d49fcf4857b0d20f6f24cbe479caa4d651aba9",
      "tree": "92bf6ce9d77e151e0c3948e9000b958a693acfd1",
      "parents": [
        "664ae35acbdc411196dc4ee3a2418b10c51ac5ad"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Thu Nov 05 10:27:18 2026 -0500"
      },
      "committer": {
        "name": "JesseChambers",
        "email": "145486573+JesseMelon@users.noreply.github.com",
        "time": "Mon Nov 09 13:29:23 2026 -0800"
      },
      "message": "i2c: build the burst test master as a kernel image\n\nDevice A drove the bus from a bare-metal register driver, which does not\nexercise the path real traffic takes. Rebuild it with the same shape as\nthe PLDM firmware-update agent: a kernel plus a DMA-mode i2c_server plus\nan IPC client, so a master write goes through write_dma_mode exactly as\nPLDM\u0027s does.\n\nThis is what surfaced the vector table and interrupt-ack defects fixed in\nthe neighbouring commits -- the previous image never registered an\ninterrupt object, so it could not hit either.\n"
    },
    {
      "commit": "664ae35acbdc411196dc4ee3a2418b10c51ac5ad",
      "tree": "b82d4022079d9e927bac7b67e20aa1a18ccda9a2",
      "parents": [
        "acb422f2f44adbb986e2dc8fe4f0300a826682f7"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Thu Nov 05 10:26:20 2026 -0500"
      },
      "committer": {
        "name": "JesseChambers",
        "email": "145486573+JesseMelon@users.noreply.github.com",
        "time": "Mon Nov 09 13:29:23 2026 -0800"
      },
      "message": "i2c: make Bus::irq optional\n\nThe server-runtime was written for slave buses, where the IRQ is the only\npath by which received bytes ever reach a client. A bus that only ever\nwrites as master has no such consumer -- its transfers complete by polling\nstatus registers -- so requiring it to name an IRQ handle states something\nuntrue about the bus.\n\nChange Bus::irq to Option\u003cu32\u003e and register with the wait group only when\nit is Some. Every existing call site passes Some(handle::I2C2_IRQ), so\nbehaviour is unchanged; this is the type admitting what the dispatch loop\nalready did.\n"
    },
    {
      "commit": "acb422f2f44adbb986e2dc8fe4f0300a826682f7",
      "tree": "f52e605f623e48b87376132235c41caf4048a000",
      "parents": [
        "f1af19fb7e4e4c132f3e4b9dd9a8e59d0284433e"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Thu Nov 05 10:25:55 2026 -0500"
      },
      "committer": {
        "name": "JesseChambers",
        "email": "145486573+JesseMelon@users.noreply.github.com",
        "time": "Mon Nov 09 13:29:23 2026 -0800"
      },
      "message": "i2c: clear the controller on every IRQ, not just armed buses\n\nA completed master write leaves bits set in I2CM14, and those bits hold\nthe controller\u0027s shared interrupt line asserted. handle_slave_interrupt()\nis the only code that clears them, and it was reached only via\ntry_next_slave_event() behind an `if bus.notif_enabled` guard.\n\ninterrupt_ack() ran unconditionally, and it re-enables the NVIC line. So\non a bus that never arms slave notification the hardware source was never\ncleared while the line was re-enabled, re-firing the interrupt forever.\n\nService the controller on every IRQ and gate only the latch and client\nwake on notif_enabled. Buses that do arm notification are unaffected --\nthe call already ran for them.\n"
    },
    {
      "commit": "f1af19fb7e4e4c132f3e4b9dd9a8e59d0284433e",
      "tree": "a8d9e0178d02f7b0f3efb183936168935d9489bc",
      "parents": [
        "dcf1e749fef8e94bd20c45191d31ad192b00a68c"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Tue Nov 03 16:22:35 2026 -0500"
      },
      "committer": {
        "name": "JesseChambers",
        "email": "145486573+JesseMelon@users.noreply.github.com",
        "time": "Mon Nov 09 13:29:23 2026 -0800"
      },
      "message": "i2c: fix slave clock-stretch gate and master STOP-settle race\n\nThe slave_read drain re-armed the receive path unconditionally on every\ncall, with no concept of whether the consumer had actually taken the\nprevious frame -- a frame arriving before the client drained the prior\none just overwrote it in the server\u0027s one-frame latch. Gate the re-arm\non a per-frame credit the consumer grants by draining, so the controller\nclock-stretches a slow client instead of losing data.\n\nwrite_buffer_mode returned as soon as PKT_DONE was observed, but the\npacket state machine (I2CM14[31:28]) was still reading STOP at that\nmoment -- the controller was still physically driving the stop\ncondition. The next write\u0027s command register landed on a busy\ncontroller and was silently dropped. Add a settle loop that waits for\nthe state machine to return to idle before returning.\n\nReading back what DMA wrote requires invalidating the CM4F\u0027s\nwrite-through cache first, which touches the SCU directly. Map scu_regs\ninto every i2c_server consumer of this driver, not just the one that\nexercised the path when this was written -- a config that lacked the\nmapping would MemoryManagement fault the moment a slave frame actually\narrived.\n"
    },
    {
      "commit": "dcf1e749fef8e94bd20c45191d31ad192b00a68c",
      "tree": "93f8d57431f8fb68920a35bae2c3460e03f80e42",
      "parents": [
        "3b1c8556ab6d5162fb833bbb0c471c80214efc72"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Thu Oct 29 12:57:42 2026 -0400"
      },
      "committer": {
        "name": "JesseChambers",
        "email": "145486573+JesseMelon@users.noreply.github.com",
        "time": "Mon Nov 09 13:29:23 2026 -0800"
      },
      "message": "i2c: fix i2c_slave_rx_burst server_main to use current bus-number i2c_backend API\n"
    },
    {
      "commit": "3b1c8556ab6d5162fb833bbb0c471c80214efc72",
      "tree": "9a2020fea20d602172f41581fead14eeff70d618",
      "parents": [
        "ef932877194475b654a0efcdf1e3aa5f42d78db5"
      ],
      "author": {
        "name": "JesseMelon",
        "email": "jessechambers2000@icloud.com",
        "time": "Fri Oct 09 17:27:39 2026 -0400"
      },
      "committer": {
        "name": "JesseChambers",
        "email": "145486573+JesseMelon@users.noreply.github.com",
        "time": "Mon Nov 09 13:29:23 2026 -0800"
      },
      "message": "standalone i2c burst test w readme\n"
    },
    {
      "commit": "ef932877194475b654a0efcdf1e3aa5f42d78db5",
      "tree": "e1b6fccdebf3cf1621a89737b538d5b55ac4f932",
      "parents": [
        "e06572b7f43c4d1a2963f351efa52779b6b09077"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Mon Nov 09 09:38:52 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 09 10:14:33 2026 -0800"
      },
      "message": "services/flash/server: Add the status length with saturating_add\n\nThe sum cannot overflow because the reply fits the buffer, but the\nreview checklist asks for checked arithmetic on every integer operation.\n"
    },
    {
      "commit": "e06572b7f43c4d1a2963f351efa52779b6b09077",
      "tree": "25533e90fc1d6908bcabc05825afdfd7150dbb98",
      "parents": [
        "8da76bf117f56f0a88245e23a2803779a246b276"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 08 21:53:05 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 09 10:14:33 2026 -0800"
      },
      "message": "services/flash/tests: Name the shared servers type\n\nClippy rejects the nested Rc\u003cRefCell\u003cVec\u003cServer\u003cRamFlash\u003e\u003e\u003e\u003e as a very\ncomplex type and failed the presubmit.\n"
    },
    {
      "commit": "8da76bf117f56f0a88245e23a2803779a246b276",
      "tree": "da10ce513179122f33a69de0af4937886431446f",
      "parents": [
        "53a2636d1884e7dbfceaaeb17696cb0c52db18a5"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 08 21:48:04 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 09 10:14:33 2026 -0800"
      },
      "message": "docs: Say plainly that the server validates client input\n\nSplit the rule into two points: treat everything that comes off the wire\nas untrusted input, third-party clients included, and validate it in the\nserver because the driver assumes valid inputs. Drop \"trust boundary\" and\n\"documented preconditions\".\n"
    },
    {
      "commit": "53a2636d1884e7dbfceaaeb17696cb0c52db18a5",
      "tree": "8341521b4d74059bc59af03406a2ccb5246e6b4b",
      "parents": [
        "5a6d48f0ebdde253c6d44adeddbefa436b52bd54"
      ],
      "author": {
        "name": "Anthony Rocha",
        "email": "anthony.rocha@amd.com",
        "time": "Sun Nov 08 19:03:23 2026 -0800"
      },
      "committer": {
        "name": "Anthony Rocha",
        "email": "116300062+rusty1968@users.noreply.github.com",
        "time": "Mon Nov 09 10:14:33 2026 -0800"
      },
      "message": "services/flash: Test one server with a device per channel\n\nPut two RAM devices of different sizes behind one server, with a channel\nto each, the way the runtime serves a controller\u0027s chip selects. Each\nchannel reports its own device\u0027s geometry, a write on one never reaches\nthe other, a range valid on the larger device is rejected on the smaller,\nand a channel with no device behind it gets no frame.\n"
    }
  ],
  "next": "5a6d48f0ebdde253c6d44adeddbefa436b52bd54"
}
