CVE-2026-98143

Description

In the Linux kernel, the following vulnerability has been resolved: accel: ethosu: Don't read the U65 rounding mode as a storage mode Bits 15:14 of NPU_SET_{IFM,OFM}_PRECISION select the activation storage mode on U85 only. On U65 the same field holds the rounding mode, and the command stream parser has read it as a storage mode since the driver was added. That went unnoticed while unknown values fell through the switch, but now that they are rejected, every U65 command stream that asks for natural rounding (2) fails CMDSTREAM_BO_CREATE with -EINVAL. Mesa emits it for average pooling, concatenation, split, unpack, strided slice, LUT and argmax, which is 72 failures of the Teflon test suite on an i.MX93. Truncating rounding (1) is misread as well: it picks the two-tile address path and computes a bogus feature map size from tile bases the command stream never set. Read the field as a storage mode only on the hardware where it is one.

Overview

State: PUBLISHED
Assigner (CNA): Linux
CVSS severity: HIGH
CVSS score: 7.8 / 10
CVSS vector: CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Effective score: 7.8 / 10 HIGH source: CNA overview
CWE(s): —
Reserved: 2026-09-25
Published: 2026-09-25 10:36 UTC
Last updated: 2026-09-25 14:42 UTC
Source: https://raw.githubusercontent.com/CVEProject/cvelistV5/main/cves/2026/98xxx/CVE-2026-98143.json
Linked Threat: CVE-2026-98143 — accel: ethosu: Don't read the U65 rounding mode as a storage mode

NVD triage scoring NVD CVE 2.0

Layer NVD adds on top of the CNA's CVE record — published / last-modified timestamps, exploitability / impact subscores, and the FIRST.org EPSS probability that this CVE will be exploited in the wild in the next 30 days.

NVD published: 2026-09-25 11:17:45 UTC
NVD last modified: 2026-09-25 15:18:06 UTC
NVD CVSS v3.1: 7.8 / 10 HIGH source: 416baaa9-dc9f-4396-8d5f-8c081fb06d67
NVD CVSS vector: CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Exploitability subscore: 1.8 / 10
Impact subscore: 5.9 / 10

European Union Vulnerability Database ENISA EUVD

ENISA's official EU repository for curated vulnerability intelligence. Carries a separate identifier (EUVD-YYYY-NNNN) and frequently exposes an earlier-published description + CVSS than NVD does.

EUVD ID: EUVD-2026-86974
Assigner: Linux
Published: Sep 25, 2026, 10:36:18 AM
Updated: Sep 25, 2026, 2:42:11 PM
EUVD base score (CVSS 3.1): 7.8 / 10
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
EUVD-reported EPSS: 0.1100
Vendors: Linux
Products: Linux (5a5e9c0228e613f0ef2a58b9782d7c0ea8f1e58b <6b08adbda8ea797849e3654ce12cb3856ce6051a)

Linux (6.19)

Linux (patch: 0)

Linux (patch: 7.2.7)

Linux (patch: 7.3-rc2)

Linux (5a5e9c0228e613f0ef2a58b9782d7c0ea8f1e58b <db9deec5a345abc538d081fb221dc0b00a9695bd)
Aliases: GHSA-m67f-86jg-mhm5

ENISA description: In the Linux kernel, the following vulnerability has been resolved: accel: ethosu: Don't read the U65 rounding mode as a storage mode Bits 15:14 of NPU_SET_{IFM,OFM}_PRECISION select the activation storage mode on U85 only. On U65 the same field holds the rounding mode, and the command stream parser has read it as a storage mode since the driver was added. That went unnoticed while unknown values fell through the switch, but now that they are rejected, every U65 command stream that asks for natural rounding (2) fails CMDSTREAM_BO_CREATE with -EINVAL. Mesa emits it for average pooling, concatenation, split, unpack, strided slice, LUT and argmax, which is 72 failures of the Teflon test suite on an i.MX93. Truncating rounding (1) is misread as well: it picks the two-tile address path and computes a bogus feature map size from tile bases the command stream never set. Read the field as a storage mode only on the hardware where it is one.

EUVD references (2)

Affected products (2)

Vendor	Product	Versions	Platforms
Linux	Linux	`5a5e9c0228e613f0ef2a58b9782d7c0ea8f1e58b` (affected), `5a5e9c0228e613f0ef2a58b9782d7c0ea8f1e58b` (affected)	—
Linux	Linux	`6.19` (affected), `0` (unaffected), `7.2.7` (unaffected), `7.3-rc2` (unaffected)	—

Vendor references (2)

References embedded in the original CVE record by the assigning CNA.

Web references (0)

DuckDuckGo results ranked by threat-intel / vendor advisory domains. Generated by the 🔎 Find references (web) button above — same flow as the Remediations search.

No web references attached yet.

NVD-tagged references (2)

Reference list NVD curates from the CNA record, vendor advisories, and third-party reports. The tag chips below are NVD's analyst-assigned categories.

https://git.kernel.org/stable/c/6b08adbda8ea797849e3654ce12cb3856ce6051a 416baaa9-dc9f-4396-8d5f-8c081fb06d67
https://git.kernel.org/stable/c/db9deec5a345abc538d081fb221dc0b00a9695bd 416baaa9-dc9f-4396-8d5f-8c081fb06d67

Remediations (10)

Remediations are stored against the linked Threat row; the list below is deduplicated across both pages.

Security Update Guide - Microsoft Security Response Center

web:msrc.microsoft.com

Security Update Guide - Microsoft Security Response Center

2026-09-27 22:54 UTC
Cisco Advance Notification for Publication of September 16, 2026 ...

web:sec.cloudapps.cisco.com

On September 16, 2026 , the Cisco Product Security Incident Response Team (PSIRT) published the advisories that are listed in the following tables. To remediate these vulnerabilities, Cisco strongly recommends that customers upgrade to the fixed software that is indicated in the advisories. For more information about changes in Cisco PSIRT vulnerability disclosure, see Strengthening the ...

2026-09-27 22:54 UTC
Microsoft Patch Tracker (September 2026): Exploit-Ranked | Senserva

web:senserva.com

Every Microsoft security patch (KB) and the CVEs it fixes, with severity, CVSS, and CISA KEV status. Cross-linked to the CVE reference.

2026-09-27 22:54 UTC
[Security Advisory] September 2026 CVE Advisory Notification

web:support.servicenow.com

What you need to know: On September 24, 2026 , ServiceNow issued CVE - 2026 -86857, CVE - 2026 -86858, CVE - 2026 -13016, CVE - 2026 -86859, and CVE - 2026 -86860. Each of these security issues was identified through internal security testing, customer security assessments, or reports submitted through ServiceNow's responsible disclosure and bug bounty programs and was remediated independently. For security ...

2026-09-27 22:54 UTC
CVE-2026-98143 | Ubuntu

web:ubuntu.com

Ubuntu is an open source software operating system that runs from the desktop, to the cloud, to all your internet connected things.

2026-09-27 22:54 UTC
What Is CVE Remediation in 2026? - aikido.dev

web:www.aikido.dev

CVE remediation is fixing known flaws in the software you run. Why upgrading often fails, what remediation actually involves, and how backporting fixes it.

2026-09-27 22:54 UTC
Remediate IOS-XE Security Advisory - Aug 2026 and Related ... - Cisco

web:www.cisco.com

This document provides upgrade guidance to mitigate IOS-XE Security Hardening Advisory - Aug 2026 and related critical bugs

2026-09-27 22:54 UTC
The 8 Best Automated Vulnerability Remediation Tools in 2026

web:www.openhands.dev

Compare 8 automated vulnerability remediation tools for 2026 on fixes, prioritization, integrations, and deployment control.

2026-09-27 22:54 UTC
Oracle Critical Patch Update Advisory - July 2026

web:www.oracle.com

This Critical Patch Update contains 1448 new security patches across the product families listed below. Please note that a My Oracle Support (MOS) note summarizing the content of this Critical Patch Update and other Oracle Software Security Assurance activities is located at July 2026 Critical Patch Update: Executive Summary and Analysis.

2026-09-27 22:54 UTC
Oracle Critical Security Patch Update Advisory - June 2026

web:www.oracle.com

Additional CVEs addressed are: The patch for CVE - 2026 -34481 also addresses CVE - 2026 -34477, CVE - 2026 -34478, CVE - 2026 -34479, and CVE - 2026 -34480. Oracle Fusion Middleware Risk Matrix This Critical Security Patch Update contains 106 new security patches for Oracle Fusion Middleware. 53 of these vulnerabilities may be remotely exploitable without authentication, i.e., may be exploited over a ...

2026-09-27 22:54 UTC

AI Forensic Analysis

Only Available for Registered Users. Sign in to view.

Raw JSON

The full cvelistV5 record. Download as CVE-2026-98143.json.

{
  "containers": {
    "cna": {
      "affected": [
        {
          "defaultStatus": "unaffected",
          "product": "Linux",
          "programFiles": [
            "drivers/accel/ethosu/ethosu_gem.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "lessThan": "6b08adbda8ea797849e3654ce12cb3856ce6051a",
              "status": "affected",
              "version": "5a5e9c0228e613f0ef2a58b9782d7c0ea8f1e58b",
              "versionType": "git"
            },
            {
              "lessThan": "db9deec5a345abc538d081fb221dc0b00a9695bd",
              "status": "affected",
              "version": "5a5e9c0228e613f0ef2a58b9782d7c0ea8f1e58b",
              "versionType": "git"
            }
          ]
        },
        {
          "defaultStatus": "affected",
          "product": "Linux",
          "programFiles": [
            "drivers/accel/ethosu/ethosu_gem.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "status": "affected",
              "version": "6.19"
            },
            {
              "lessThan": "6.19",
              "status": "unaffected",
              "version": "0",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "7.2.*",
              "status": "unaffected",
              "version": "7.2.7",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "*",
              "status": "unaffected",
              "version": "7.3-rc2",
              "versionType": "original_commit_for_fix"
            }
          ]
        }
      ],
      "cpeApplicability": [
        {
          "nodes": [
            {
              "cpeMatch": [
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "7.2.7",
                  "versionStartIncluding": "6.19",
                  "vulnerable": true
                },
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "7.3-rc2",
                  "versionStartIncluding": "6.19",
                  "vulnerable": true
                }
              ],
              "negate": false,
              "operator": "OR"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "In the Linux kernel, the following vulnerability has been resolved:\n\naccel: ethosu: Don't read the U65 rounding mode as a storage mode\n\nBits 15:14 of NPU_SET_{IFM,OFM}_PRECISION select the activation storage\nmode on U85 only. On U65 the same field holds the rounding mode, and the\ncommand stream parser has read it as a storage mode since the driver was\nadded.\n\nThat went unnoticed while unknown values fell through the switch, but\nnow that they are rejected, every U65 command stream that asks for\nnatural rounding (2) fails CMDSTREAM_BO_CREATE with -EINVAL. Mesa emits\nit for average pooling, concatenation, split, unpack, strided slice, LUT\nand argmax, which is 72 failures of the Teflon test suite on an i.MX93.\nTruncating rounding (1) is misread as well: it picks the two-tile\naddress path and computes a bogus feature map size from tile bases the\ncommand stream never set.\n\nRead the field as a storage mode only on the hardware where it is one."
        }
      ],
      "metrics": [
        {
          "cvssV3_1": {
            "baseScore": 7.8,
            "baseSeverity": "HIGH",
            "vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
            "version": "3.1"
          },
          "scenarios": [
            {
              "lang": "en",
              "value": "AV:L - The malformed data is an NPU command stream a local process passes through DRM_IOCTL_ETHOSU_CMDSTREAM_BO_CREATE on /dev/accel/accelN, which is validated by ethosu_gem_cmdstream_copy_and_validate() -> calc_sizes() -> feat_matrix_length(), then run via DRM_IOCTL_ETHOSU_SUBMIT. No network or radio transport carries it.\nAC:L - The attacker writes the whole command stream: rounding bits 15:14 = 1 in NPU_SET_OFM_PRECISION, x past WIDTH0_M1, and an arbitrary NPU_SET_OFM_BASE1. The validator then checks only base[0] while the U65 hardware uses base[1]. The result is deterministic and needs no race.\nPR:L - The ethosu ioctls CMDSTREAM_BO_CREATE and SUBMIT are registered with flags 0, with no DRM_ROOT_ONLY or capability check. Only permission to open the accel node is needed, which an ordinary user in the render/video group on a U65 board (e.g. i.MX93) holds.\nUI:N - The attacker's own process builds and submits the command stream. No other user has to do anything.\nS:U - The broken check is the kernel's own command-stream validator, and the NPU DMA it lets through corrupts the host kernel's own memory. That is the same security authority, i.e. a normal local privilege escalation, not a guest or sandbox escape.\nC:H - With rounding mode 1 on U65, feat_matrix_length() uses the storage-1 tile path and never bounds IFM/IFM2 base[1]/base[3]. The NPU can therefore DMA-read memory past the region BO at an offset the attacker chooses, and the result comes back through the attacker's mmapped OFM BO.\nI:H - The same mismatch on NPU_SET_OFM_PRECISION leaves OFM base[1]/base[3] unchecked, so region_size[] understates what the hardware writes. The ethosu_job.c region_size > gem->size check passes while the NPU writes attacker-computed output outside the BO, into kernel memory.\nA:H - Uncontrolled NPU DMA writes outside the BO can corrupt kernel data and panic the system. On unaffected command streams, the misread also makes every Mesa U65 job with natural rounding fail with -EINVAL."
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-09-25T14:42:11.786Z",
        "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
        "shortName": "Linux"
      },
      "references": [
        {
          "url": "https://git.kernel.org/stable/c/6b08adbda8ea797849e3654ce12cb3856ce6051a"
        },
        {
          "url": "https://git.kernel.org/stable/c/db9deec5a345abc538d081fb221dc0b00a9695bd"
        }
      ],
      "title": "accel: ethosu: Don't read the U65 rounding mode as a storage mode",
      "x_generator": {
        "engine": "bippy-1.2.0"
      }
    }
  },
  "cveMetadata": {
    "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
    "assignerShortName": "Linux",
    "cveId": "CVE-2026-98143",
    "datePublished": "2026-09-25T10:36:18.470Z",
    "dateReserved": "2026-09-25T10:25:14.319Z",
    "dateUpdated": "2026-09-25T14:42:11.786Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2"
}