{"metadata": {"vulnerability": "CVE-2026-46728", "count": 20, "kinds": {"milestone": 5, "sighting": 6, "kev": 0, "other": 9}, "linked_omitted": 0}, "data": [{"timestamp": "2026-05-16T21:26:48.876Z", "date_only": false, "value": "2026-05-16T21:26:48.876Z", "kind": "milestone", "label": "Reserved", "source": "cvelistv5", "record": "CVE-2026-46728", "field": "cveMetadata.dateReserved"}, {"timestamp": "2026-05-16T21:26:49.527Z", "date_only": false, "value": "2026-05-16T21:26:49.527Z", "kind": "milestone", "label": "Published", "source": "cvelistv5", "record": "CVE-2026-46728", "field": "cveMetadata.datePublished", "metric_role": "disclosure", "metric_basis": "recorded_event"}, {"timestamp": "2026-05-16T22:16:00Z", "date_only": false, "value": "2026-05-16T22:16:00Z", "kind": "milestone", "label": "Published", "source": "osv_ubuntu \u00b7 UBUNTU-CVE-2026-46728", "record": "UBUNTU-CVE-2026-46728", "field": "published", "metric_role": "disclosure", "metric_basis": "recorded_event", "linked_record": true}, {"timestamp": "2026-05-16T22:16:13Z", "date_only": false, "value": "2026-05-16T22:16:13Z", "kind": "other", "label": "Nvd published at", "source": "github \u00b7 GHSA-5CV5-VM4Q-7WMX", "record": "GHSA-5CV5-VM4Q-7WMX", "field": "database_specific.nvd_published_at", "linked_record": true}, {"timestamp": "2026-05-16T22:16:13.317Z", "date_only": false, "value": "2026-05-16T22:16:13.317", "kind": "milestone", "label": "Published", "source": "fkie_nvd \u00b7 FKIE_CVE-2026-46728", "record": "FKIE_CVE-2026-46728", "field": "published", "metric_role": "disclosure", "metric_basis": "recorded_event", "linked_record": true}, {"timestamp": "2026-05-16T22:24:18.071Z", "date_only": false, "value": "2026-05-16T22:24:18.071Z", "kind": "other", "label": "Updated", "source": "cvelistv5 / mitre", "record": "CVE-2026-46728", "field": "containers.cna.providerMetadata.dateUpdated"}, {"timestamp": "2026-05-16T22:44:13.788Z", "date_only": false, "value": "2026-05-16T22:44:13.788107Z", "kind": "sighting", "label": "Seen", "source": "https://bsky.app/profile/cve.skyfleet.blue/post/3mlyw77gh532i", "record": "CVE-2026-46728", "field": "creation_timestamp", "uuid": "1ac750e8-daf2-4605-bc89-2dcf8067686e", "content": "CVE-2026-46728 - Das U-Boot FIT Signature Verification Bypass\nCVE ID : CVE-2026-46728\n \n Published : May 16, 2026, 9:26 p.m. | 59\u00a0minutes ago\n \n Description : Das U-Boot before 2026.04 allows FIT (Flat Image Tree) signature verification bypass because hashed-nodes is omitted f...", "url": "https://bsky.app/profile/cve.skyfleet.blue/post/3mlyw77gh532i"}, {"timestamp": "2026-05-16T23:01:29.001Z", "date_only": false, "value": "2026-05-16T23:01:29.001643Z", "kind": "sighting", "label": "Seen", "source": "https://bsky.app/profile/thehackerwire.bsky.social/post/3mlyx6zjxtm2e", "record": "CVE-2026-46728", "field": "creation_timestamp", "uuid": "20e896a7-44aa-46e1-865b-bad36552cd25", "content": "\ud83d\udfe0 CVE-2026-46728 - High (8.2)\n\nDas U-Boot before 2026.04 allows FIT (Flat Image Tree) signature verification bypass because hash...\n\nhttps://www.thehackerwire.com/vulnerability/CVE-2026-46728/\n\n#infosec #cybersecurity #CVE #vulnerability #security #patchstack", "url": "https://bsky.app/profile/thehackerwire.bsky.social/post/3mlyx6zjxtm2e"}, {"timestamp": "2026-05-17T00:00:39.391Z", "date_only": false, "value": "2026-05-17T00:00:39.391029Z", "kind": "sighting", "label": "Seen", "source": "https://bsky.app/profile/offseq.bsky.social/post/3mlz2lgkzjx2j", "record": "cve-2026-46728", "field": "creation_timestamp", "uuid": "433fb6be-213e-40b0-b93e-9595c8fa4e08", "content": "\ud83d\udea8 HIGH: CVE-2026-46728 in denx U-Boot &lt;2026.04 lets local privileged users bypass FIT signature checks. Restrict privileged access &amp; monitor vendor updates. https://radar.offseq.com/threat/cve-2026-46728-cwe-346-origin-validation-error-in--fd4fda20 #OffSeq #Uboot #FirmwareSecurity", "url": "https://bsky.app/profile/offseq.bsky.social/post/3mlz2lgkzjx2j"}, {"timestamp": "2026-05-17T00:00:51.686Z", "date_only": false, "value": "2026-05-17T00:00:51.686324Z", "kind": "sighting", "label": "Seen", "source": "https://infosec.exchange/users/offseq/statuses/116586973497847352", "record": "cve-2026-46728", "field": "creation_timestamp", "uuid": "fdac47ce-3d59-40f5-b2ab-35ad80f42384", "content": "\ud83d\udd12 CVE-2026-46728 (HIGH, CVSS 8.2) in denx U-Boot &lt;2026.04 lets local privileged attackers bypass FIT signature checks due to hashed-nodes omission. Restrict local access &amp; watch for patches. https://radar.offseq.com/threat/cve-2026-46728-cwe-346-origin-validation-error-in--fd4fda20 #OffSeq #Uboot #Infosec #Firmware", "url": "https://infosec.exchange/users/offseq/statuses/116586973497847352"}, {"timestamp": "2026-05-17T00:31:38Z", "date_only": false, "value": "2026-05-17T00:31:38Z", "kind": "milestone", "label": "Published", "source": "github \u00b7 GHSA-5CV5-VM4Q-7WMX", "record": "GHSA-5CV5-VM4Q-7WMX", "field": "published", "metric_role": "disclosure", "metric_basis": "recorded_event", "linked_record": true}, {"timestamp": "2026-05-17T00:31:38Z", "date_only": false, "value": "2026-05-17T00:31:38Z", "kind": "other", "label": "Updated", "source": "github \u00b7 GHSA-5CV5-VM4Q-7WMX", "record": "GHSA-5CV5-VM4Q-7WMX", "field": "modified", "linked_record": true}, {"timestamp": "2026-05-17T21:50:57.544Z", "date_only": false, "value": "2026-05-17T21:50:57.544510Z", "kind": "sighting", "label": "Seen", "source": "https://bsky.app/profile/postac001.bsky.social/post/3mm3dshrdb32v", "record": "CVE-2026-46728", "field": "creation_timestamp", "uuid": "eda11c50-e097-490b-a416-79accccb5942", "content": "U-Boot 2026.04\u672a\u6e80\u3067\u3001FIT\u7f72\u540d\u691c\u8a3c\u30d0\u30a4\u30d1\u30b9\u306e\u53ef\u80fd\u6027\u3002\u30cf\u30c3\u30b7\u30e5\u30ce\u30fc\u30c9\u304c\u30cf\u30c3\u30b7\u30e5\u304b\u3089\u7701\u7565\u3055\u308c\u3001\u653b\u6483\u8005\u306f\u4e0d\u6b63\u306a\u30a4\u30e1\u30fc\u30b8\u306e\u5b9f\u884c\u304c\u53ef\u80fd\u306b\u306a\u308b\u3002\nCVE-2026-46728 CVSS 8.2 | HIGH", "url": "https://bsky.app/profile/postac001.bsky.social/post/3mm3dshrdb32v"}, {"timestamp": "2026-05-18T12:53:04.053Z", "date_only": false, "value": "2026-05-18T12:53:04.053259Z", "kind": "other", "label": "Recorded", "source": "cvelistv5 / CISA-ADP", "record": "CVE-2026-46728", "field": "containers.adp.0.metrics.0.other.content.timestamp"}, {"timestamp": "2026-05-18T12:53:04.053Z", "date_only": false, "value": "2026-05-18T12:53:04.053259Z", "kind": "other", "label": "Recorded", "source": "fkie_nvd \u00b7 FKIE_CVE-2026-46728", "record": "FKIE_CVE-2026-46728", "field": "metrics.ssvcV203.0.ssvcData.timestamp", "linked_record": true}, {"timestamp": "2026-05-18T12:53:13.033Z", "date_only": false, "value": "2026-05-18T12:53:13.033Z", "kind": "other", "label": "Updated", "source": "cvelistv5", "record": "CVE-2026-46728", "field": "cveMetadata.dateUpdated"}, {"timestamp": "2026-05-18T12:53:13.033Z", "date_only": false, "value": "2026-05-18T12:53:13.033Z", "kind": "other", "label": "Updated", "source": "cvelistv5 / CISA-ADP", "record": "CVE-2026-46728", "field": "containers.adp.0.providerMetadata.dateUpdated"}, {"timestamp": "2026-10-21T20:11:16Z", "date_only": false, "value": "2026-10-21T20:11:16Z", "kind": "other", "label": "Updated", "source": "osv_ubuntu \u00b7 UBUNTU-CVE-2026-46728", "record": "UBUNTU-CVE-2026-46728", "field": "modified", "linked_record": true}, {"timestamp": "2026-10-26T13:10:00.140Z", "date_only": false, "value": "2026-10-26T13:10:00.140", "kind": "other", "label": "Updated", "source": "fkie_nvd \u00b7 FKIE_CVE-2026-46728", "record": "FKIE_CVE-2026-46728", "field": "lastModified", "linked_record": true}, {"timestamp": "2026-11-16T01:48:35Z", "date_only": false, "value": "2026-11-16T01:48:35.000000Z", "kind": "sighting", "label": "Seen", "source": "https://gist.github.com/superflyer11/d08089fc65ab3c3f74b0f4ade0240a47", "record": "CVE-2026-46728", "field": "creation_timestamp", "uuid": "9fc161ee-d7ed-44a7-9699-1ffd196a5bc6", "content": "# Echo Dot 5th gen (`cannoli`) \u2014 boot chain analysis and the missing key artifacts\n\nStatic analysis of the Echo Dot 5 (C2N6L4, `cannoli`, MediaTek MT8519) boot\nchain, done entirely from public OTA images. No device was modified, and there is\nno exploit here.\n\nTwo results that answer questions currently open in this thread:\n\n1. **The `key-fit` / `key-da` artifacts are not in any OTA, by design.** They are\n   injected into the U-Boot control DTB at runtime by the Preloader, which is not\n   shipped in `payload.bin`.\n2. **`cannoli` is unpatched for CVE-2026-46728**, confirmed at the disassembly\n   level, and its recovery configurations boot a `ramdisk-1` that no signature\n   covers.\n\n---\n\n## 1. Getting firmware\n\nFTVDB indexes Echo firmware with the original Amazon CloudFront URLs:\n\n- `https://ftvdb.com/echo/firmware/com.amazon.cannoli.android.os/` \u2014 7 full\n  records, 2023-11-09 \u2192 2026-10-27.\n\nThe record analysed here:\n\n```\nFire OS 6574.1 (NS65741/8149)   build 13222532484\nhttps://d1s31zyz7dcc2d.cloudfront.net/2026/9/29/8e21a1df-3243-418e-9866-ecaa42164024/update-kindle-cannoli-NS65741_user_8149_0013222532484.bin\nsize 130,569,604 B   md5 e3a63824225f84eaab30e88bf3d5dfec\n```\n\nNewest record is Fire OS 6.5.7.5 (NS6575/8349), build `13323246980`.\n\n## 2. Container\n\nA plain ZIP (`PK\\x03\\x04`, comment `signed by SignApk`) holding an **uncompressed\n`payload.bin`** which is stock Android `update_engine`:\n\n```\nmagic \"CrAU\", major version 2\nmanifest 5131 B, metadata signature 264 B, data at 0x152b\nblock_size 4096\n```\n\nOTA properties: `product=cannoli`,\n`Amazon/cannoli/cannoli:6.0/NS65741/8149N:user/amz-p,release-keys`,\n`key_type=release-keys`, `binary_type=full`. Android 6.0 based Fire OS 6.\n\nPartitions and their compression:\n\n| partition | size | ops | compression |\n|---|---|---|---|\n| `system` | 131,088,384 | 63 | REPLACE / BZ / XZ |\n| `boot` | 10,067,968 | 5 | REPLACE / XZ |\n| `uboot` | 991,232 | 1 | XZ |\n| `tee1`, `tee2` | 647,168 | 1 each | XZ |\n| `scp` | 1,028,096 | 1 | XZ |\n| `mcupm` | 122,880 | 1 | XZ |\n| `brhgptlk_0..3` | 249,856 | 1 each | XZ |\n| `reserve_0..2` | 4,096 | 1 each | BZ |\n\n**No `preloader` and no key database.**\n\nExtraction: my `dump_payload.py` parses the manifest by hand, walks operations,\nand verifies each partition against the SHA-256 in `PartitionInfo`. All 14\nverified.\n\n## 3. Boot chain\n\n`U-Boot 2019.04-g71f44a240c`, built **2026-04-07**, `board=mt8519`. Every\npartition is a FIT image \u2014 verification is FIT hashes and RSA throughout, not\nAVB at this level.\n\n`uboot.img` contains two FDTs: the FIT at `0x0` (one `kernel-1`, `compression =\nnone`, load/entry `0x44e00000`) and the U-Boot control DTB at `0xee42c`.\n\nFIT signature node:\n\n```\nalgo                  = \"sha256,rsa2048\"\npadding               = \"pss\"\nkey-name-hint         = \"fit\"\nanti-rollback-version = &lt;0x02&gt;\nhashed-nodes          = \"/\", \"/configurations/conf-1\",\n                        \"/images/kernel-1\", \"/images/kernel-1/hash-1\"\nsigner-version        = \"2021.07\"\n```\n\n`boot.img` is signed by a **different** key, `key-name-hint =\n\"fireos-production\"`, `signer-version = \"2020.01\"`. It carries one kernel, one\nramdisk and five device trees for hardware revisions (HVT, EVT aluminium/plastic,\nDVT aluminium/plastic \u2014 DVT blobs also used for PVT), across **13\nconfigurations** with `-recovery` variants.\n\nStorage is **SPI-NAND with ubifs**, not eMMC; `boot`, `uboot` and `misc` are MTD.\n`/system` is `squashfs ro wait,verify,slotselect`.\n\n## 4. Where the keys actually are\n\nU-Boot's control DTB `/signature` subtree contains exactly one node, and it is a\nstub:\n\n```dts\nsignature {\n    key-dev {\n        required = \"conf\";\n        algo = \"sha256,rsa2048\";\n        key-name-hint = \"dev\";\n    };\n};\n```\n\nNo `rsa,modulus`, no `rsa,exponent`. The property parser at `0x44e87354` reads:\n\n```\n0x44eb9fa0  \"key-fireos-production\"\n0x44eba025  \"key-fireos-engineering\"\n0x44eb9fb6  \"rsa,r-squared\"\n0x44eb9fc4  \"rsa,modulus\"\n0x44eb9fd0  \"rsa,exponent\"      (rejected if &lt;= 7 bytes: cmp w0,#7 ; b.hi)\n0x44eb9fdd  \"rsa,n0-inverse\"\n0x44eb9ff1  \"rsa,num-bits\"\n```\n\nand returns `-0xe` when the modulus is absent.\n\nSo U-Boot looks keys up **by name**, and `cannoli`'s OTA provides neither\n`key-fireos-production` nor `key-fireos-engineering` \u2014 only the `key-dev` stub\nthat this parser would reject. **The real key nodes must be injected into the\ncontrol DTB before U-Boot's first verification.** The Preloader is the only stage\npositioned to do that, and it is in the NAND boot region, absent from every\npublic OTA.\n\nThat is the concrete answer to \"where are `key-fit` / `key-da`\": not missing from\nthis OTA by accident, and not obtainable from any other OTA either. Also worth\nknowing: `key-fireos-engineering` exists in retail bootloader code alongside the\nproduction key, so a second signing identity is wired in. Which one is accepted\nis decided by state the Preloader sets.\n\n## 5. CVE-2026-46728 is unpatched here\n\nCVE-2026-46728, CWE-346, CVSS 3.1 8.8: *\"U-Boot before 2026.04 allows FIT\nsignature verification bypass because hashed-nodes is omitted from a hash.\"*\nFix: `2092322b31cc`, reported by Apple SEAR. The commit message is explicit:\n\n&gt; The hashed-nodes property in a FIT signature node lists which FDT paths are\n&gt; included in the signature hash. **It is intended as a hint so should not be\n&gt; used for verification.** Add a function to build the node list from scratch by\n&gt; iterating the configuration's image references.\n\nPre-2026.04 U-Boot trusts the FIT's own `hashed-nodes` property \u2014 a property\ninside the image being verified.\n\n`cannoli`'s verifier is at `0x44e2fb00`. String xrefs were recovered by decoding\nADRP/ADD pairs on the flat image (`xref.py`):\n\n```\n0x44e2fb90  adrp x2, #0x44ec2000\n0x44e2fb9c  add  x2, x2, #0x1ac        x2 = \"hashed-nodes\"\n0x44e2fb94  mov  x0, x20               x0 = fit     (image under test)\n0x44e2fb98  mov  w1, w22               w1 = noffset\n0x44e2fba8  bl   #0x44e83e10           fdt_getprop(fit, noffset, \"hashed-nodes\", &amp;len)\n0x44e2fbc4   count NUL-separated paths\n0x44e2fbe4  \"Can't get hashed-nodes property\"\n...\n0x44e2fdec  ldr x5, [x0, #0x20] ; blr x5   info-&gt;verify(...)\n```\n\nEvery error string in the function is the stock pre-fix set:\n\n```\nCan't get hashed-nodes property\nhashed-nodes property must be null-terminated\nNumber of hashed nodes exceeds maximum          (cmp w26, #0x64 = 100)\nSelected config not in hashed nodes\nFailed to hash configuration\nNo data to hash\nToo many hash regions\nVerification failed\n```\n\nThe three strings the fix adds \u2014 `Failed to build hash node list`,\n`No hash nodes in image '%s'`, `No images in config '%s'` \u2014 are absent. A patched\nbuild would not contain the `fdt_getprop(..., \"hashed-nodes\", ...)` call at all.\n\n**Verdict: unpatched.**\n\n### Why it matters on this image specifically\n\nIn all six `-recovery` configurations, the config references a ramdisk that the\nsignature never covers:\n\n```\ncannoli-hvt-recovery\n   config image refs: kernel -&gt; kernel-1, fdt -&gt; fdt-hvt, ramdisk -&gt; ramdisk-1\n      kernel    covered_by_signature = True\n      fdt       covered_by_signature = True\n      ramdisk   covered_by_signature = False       (sign-images = \"fdt kernel\")\n```\n\nThe ramdisk is the initramfs: `init` (814,812 B), `init.rc`,\n`init.recovery.mt8519.rc`, `init.fosflags.sh`, `sbin/multi_init`, `sbin/adbd`,\n`file_contexts.bin`. That is root in early userspace, and it is the stage that\nwrites `/proc/idme/*` flags \u2014 the same flags U-Boot names as\n`FOS_FLAGS_DM_VERITY_OFF`. The kernel honours the resulting state:\n`device-mapper: android-verity: Allow invalid metadata when unlocked`.\n\nSo the integrity primitive is: swap the ramdisk in a recovery config, and pre-fix\nU-Boot verifies kernel and fdt, accepts, and boots it.\n\n**It is still not an unlock.** What is missing is a write primitive into `boot`.\n\n## 6. Where a write primitive is not\n\n- `swdl`, `usbdl`, `download agent`, `bootctrl`: **absent** from U-Boot. The\n  MediaTek download agent is a Preloader feature, so BROM/DA mode needs hardware.\n- `fastboot`: present as a string, but **zero** ADRP/ADD references \u2014 no code\n  path builds it.\n- In-U-Boot write machinery is the BCB/`misc` flow (`0x44e01000`, `0x44e010b0`,\n  `0x44e01110`) plus the MTD/UBI layer, behind `amzn_block_commands`.\n- The OTA path is closed: SignApk-signed zip, per-partition SHA-256 in the\n  payload manifest, both verified.\n\n## 7. Negative result worth recording\n\n`init.fosflags.sh` defines `FOS_FLAGS_ADB_AUTH_DISABLE = 0x20` and points at\n`/proc/idme/usr_flags` with the comment *\"User controllable flags (privilege not\nrequired to set)\"*. That looks alarming and is not a path:\n\n| item | U-Boot | kernel |\n|---|---|---|\n| `fos_flags` | referenced (`0x44e02f38`) | **absent** |\n| `dev_flags` | referenced (`0x44e427a8`) | `/idme/dev_flags` |\n| `usr_flags` | **zero xrefs** | **absent** |\n\nThe kernel's whole idme surface is `/idme`, `alscal`, `alsparams`, `board_id`,\n`bootmode`, `dev_flags`, `productid2`, `sensorcal`. The script is shared Amazon\nboilerplate across Echo products; it declares `usrflagsfile` and never uses it.\n\n## 8. Reproducing any of this\n\n```\n7z e -y update-kindle-cannoli-*.bin payload.bin\n./dump_payload.py payload.bin -o ../extract        # verifies all SHA-256\n./fit_extract.py ../extract/boot.img --summary\n./fit_extract.py ../extract/uboot.img -o ../extract/uboot\npython3 -m venv /tmp/csvenv &amp;&amp; /tmp/csvenv/bin/pip install capstone\n/tmp/csvenv/bin/python xref.py ../extract/uboot/kernel-1.bin \\\n    --base 0x44e00000 --strings \"hashed-nodes\"\n```\n\nThe U-Boot payload loads at `0x44e00000`, so every VA quoted here is\n`0x44e00000 + file offset`.\n\nTools: `dump_payload.py` (CrAU extractor with per-partition verification),\n`fit_extract.py` (FIT parser \u2014 summary, extraction, signature coverage audit),\n`xref.py` (ADRP/ADD xref recovery and disassembly for flat ARM64 images, with a\npickle cache).\n\n## 9. What would move this forward\n\n1. **A Preloader image** \u2014 factory/service dump or a device-side read. With it,\n   the key database and the engineering-key gate become answerable.\n2. **A write primitive into `boot`** \u2014 nothing in the shipped images provides\n   one.\n3. **Older `cannoli` OTAs diffed** \u2014 an engineering-key-signed build may have\n   shipped by accident at some point; a diff would show it.\n\n## Credits\n\n- **R0rt1z2 (Roger O.)** \u2014 the MediaTek BROM exploits and bootloader payloads that\n  make any Echo work possible; none of the surrounding ecosystem exists without\n  them.\n- **Simon Glass / Apple SEAR** \u2014 CVE-2026-46728 analysis and the fix.\n- **Ighor July / FTVDB** \u2014 firmware indexing with original source URLs.\n- EchoLocal, TECHO5 and echo-dot-assist for mapping what these devices can be\n  usefully repurposed into.\n\n---\n\n# Short version for each venue\n\n## For the XDA thread\n\n&gt; Answering the \"where are the signed DA/FIT artifacts\" question with a negative\n&gt; result plus one new finding.\n&gt;\n&gt; **The key artifacts are not in any OTA, by design.** I pulled\n&gt; `com.amazon.cannoli.android.os` Fire OS 6574.1 (NS65741/8149, md5\n&gt; `e3a63824225f84eaab30e88bf3d5dfec`) from the Amazon CloudFront URL in the FTVDB\n&gt; record. It is a stock Android A/B `CrAU` payload \u2014 no proprietary container \u2014\n&gt; and it ships `uboot`, `boot`, both `tee` slots, `scp`, `mcupm` and four\n&gt; `brhgptlk` slots, all FIT images signed `sha256,rsa2048` with PSS.\n&gt;\n&gt; What it does **not** ship is the Preloader or any key material. U-Boot's\n&gt; control DTB contains only a stub `key-dev` node with no `rsa,modulus`; the\n&gt; parser at `0x44e87354` rejects that. U-Boot resolves keys **by name**, and the\n&gt; names compiled into it are `key-fireos-production` and `key-fireos-engineering`\n&gt; \u2014 neither present in the OTA. So the key nodes have to be injected into the\n&gt; control DTB before U-Boot's first verification, and the only stage that can do\n&gt; that is the Preloader. That is why hunting OTAs for `key-fit` won't produce\n&gt; them.\n&gt;\n&gt; Second finding: **cannoli is unpatched for CVE-2026-46728.** The verifier at\n&gt; `0x44e2fb00` still calls `fdt_getprop(fit, noffset, \"hashed-nodes\", &amp;len)` and\n&gt; hashes exactly the nodes that property names \u2014 the pre-fix behaviour Apple SEAR\n&gt; reported. None of the three strings the 2026.04 fix adds are present. On top of\n&gt; that, every `-recovery` config references `ramdisk-1` and `sign-images` is only\n&gt; `\"fdt kernel\"`, so the initramfs is outside the signature.\n&gt;\n&gt; It is not an unlock: there is no write primitive into `boot` in any shipped\n&gt; image \u2014 `swdl`/`usbdl`/download-agent/bootctrl are all absent from U-Boot, and\n&gt; fastboot has zero code path. Full notes, tooling and reproduction steps in the\n&gt; write-up. Happy to answer questions or hand the extraction scripts to whoever\n&gt; wants them.\n\n## For the EchoLocal discussion\n\n&gt; Relevant to the `cannoli` / MT8519 blocker.\n&gt;\n&gt; Short version: the signed artifacts you're looking for are not in the OTA, and\n&gt; won't be. The payload ships `uboot`, `boot`, `tee1`/`tee2`, `scp`, `mcupm` and\n&gt; `brhgptlk_0..3`, all FIT images, but no Preloader and no key material. U-Boot\n&gt; carries only a stub `key-dev` node without `rsa,modulus`; the parser at\n&gt; `0x44e87354` requires it. The key names compiled in are\n&gt; `key-fireos-production` and `key-fireos-engineering`, so keys are resolved by\n&gt; name and must be injected into the control DTB at runtime \u2014 Preloader's job,\n&gt; and it's in the NAND boot region.\n&gt;\n&gt; Also: cannoli is unpatched for CVE-2026-46728, and the `-recovery` configs boot\n&gt; a `ramdisk-1` that no signature covers (`sign-images = \"fdt kernel\"`). Full\n&gt; disassembly evidence and reproduction steps in the write-up.\n&gt;\n&gt; One negative result that might save someone time: the `FOS_FLAGS_*` and\n&gt; `/proc/idme/usr_flags` path in `init.fosflags.sh` is dead on cannoli \u2014 the\n&gt; kernel exposes no `fos_flags` or `usr_flags` idme node, and `usr_flags` has zero\n&gt; references in U-Boot. It's shared boilerplate across Echo products.\n\n", "url": "https://gist.github.com/superflyer11/d08089fc65ab3c3f74b0f4ade0240a47"}], "metrics": {"as_of": "2026-11-16T13:36:20.065Z", "anchors": {"origin": {"timestamp": "2026-05-16T21:26:49.527Z", "date_only": false, "value": "2026-05-16T21:26:49.527Z", "kind": "milestone", "label": "Published", "source": "cvelistv5", "record": "CVE-2026-46728", "field": "cveMetadata.datePublished", "metric_role": "disclosure", "metric_basis": "recorded_event"}, "discovery": null, "report": null, "acknowledgement": null, "disclosure": {"timestamp": "2026-05-16T21:26:49.527Z", "date_only": false, "value": "2026-05-16T21:26:49.527Z", "kind": "milestone", "label": "Published", "source": "cvelistv5", "record": "CVE-2026-46728", "field": "cveMetadata.datePublished", "metric_role": "disclosure", "metric_basis": "recorded_event"}, "exploit_publication": null, "exploitation": null, "patch": null}, "vendor_response_time": {"available": false, "missing": ["start", "end"]}, "time_to_exploit": {"available": false, "missing": ["end"]}, "disclosure_to_exploit": {"available": false, "missing": ["end"]}, "disclosure_to_exploit_publication": {"available": false, "missing": ["end"]}, "time_to_patch": {"available": false, "missing": ["start", "end"]}, "total_time_to_patch": {"available": false, "missing": ["start", "end"]}, "disclosure_to_patch": {"available": false, "missing": ["end"]}, "time_unpatched": {"available": true, "seconds": 15869370.538679, "hours": 4408.158482966389, "days": 183.67327012359954, "min_seconds": 15869370.538679, "max_seconds": 15869370.538679, "approximate": false, "ongoing": true, "order_known": true, "batch_listing": false, "end_before_start": false, "start": {"timestamp": "2026-05-16T21:26:49.527Z", "date_only": false, "value": "2026-05-16T21:26:49.527Z", "kind": "milestone", "label": "Published", "source": "cvelistv5", "record": "CVE-2026-46728", "field": "cveMetadata.datePublished", "metric_role": "disclosure", "metric_basis": "recorded_event"}, "end": {"timestamp": "2026-11-16T13:36:20.065Z", "date_only": false, "label": "As of", "source": "Calculation clock"}}, "exploitation_to_patch": {"available": false, "missing": ["start"]}, "exploit_publication_to_patch": {"available": false, "missing": ["start"]}}}
