{
  "schema_version": "1.6.1",
  "id": "RPI-CVE-2026-72148",
  "modified": "2026-10-09T11:39:44Z",
  "upstream": [
    "CVE-2026-72148"
  ],
  "published": "2026-08-15T06:04:33Z",
  "summary": "dmaengine: dw-edma: Add spinlock to protect DONE_INT_MASK and ABORT_INT_MASK",
  "details": "In the Linux kernel, the following vulnerability has been resolved:\n\ndmaengine: dw-edma: Add spinlock to protect DONE_INT_MASK and ABORT_INT_MASK\n\nThe DONE_INT_MASK and ABORT_INT_MASK registers are shared by all DMA\nchannels, and modifying them requires a read-modify-write sequence.\nBecause this operation is not atomic, concurrent calls to\ndw_edma_v0_core_start() can introduce race conditions if two channels\nupdate these registers simultaneously.\n\nAdd a spinlock to serialize access to these registers and prevent race\nconditions.\n\n[den: update dw_edma.lock comment]",
  "severity": [
    {
      "type": "CVSS_V3",
      "score": "CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:C/C:L/I:L/A:H"
    }
  ],
  "affected": [
    {
      "package": {
        "ecosystem": "Raspberry Pi OS:12",
        "name": "linux"
      },
      "versions": [
        "1:6.1.47-1+rpt4"
      ],
      "ecosystem_specific": {
        "binaries": [
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-2712",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v6",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v7",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v7l",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v8",
            "binary_version": "1:6.1.47-1+rpt4"
          }
        ]
      },
      "database_specific": {
        "per_version": {
          "1:6.1.47-1+rpt4": {
            "linux_commit": "655fc658a15ae7a6f37103754adb39ba52a9a14e",
            "upstream_version": "6.1.47",
            "assessment": "commit ancestry (strak rule)",
            "upstream_fixed": "6.1.178",
            "compiled_filter": false
          }
        }
      }
    }
  ],
  "references": [
    {
      "type": "ADVISORY",
      "url": "https://www.cve.org/CVERecord?id=CVE-2026-72148"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/3989b4775bc2cdbdb4ddc4b1d2420a82b40913f2"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/f60c7463d44fbc1585d247d0bd6976f7a1099472"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/2247cc25a91fb1b5b86586ed55fdd5b725a7477c"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/3ee0f478bb29b4ee892b178179a9a76ddd194149"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/21a9834f56d6249aaa6ca7c2d8c182d66c48c3e1"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/ddbc4a8a4fe296f1fa2e59f7d176fc7c773df640"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/1553ca96e9df158d8f37137cf4bf5fb0dc981d94"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/8ffba0171c6bbce5f093c6dba5a02c0805b31203"
    }
  ],
  "database_specific": {
    "source": "https://git.kernel.org/pub/scm/linux/security/vulns.git",
    "upstream_modified": "2026-08-23T12:46:35Z",
    "severity": {
      "source": "kernel CNA",
      "cvss_v3_score": 8.8,
      "cvss_v3_severity": "High",
      "scenario": "AV:A - On PCIe endpoint SoCs (R-Car Gen4, DWC EP) the dw-edma v0 engine is driven when a physically connected PCIe root host initiates endpoint DMA (e.g. pci-epf-mhi); that host peer needs no network path to the victim, fitting Adjacent attack vector.\nAC:L - The race needs only two eDMA channels starting their first chunk concurrently; an attacker controls both sides by issuing parallel DMA requests or timing concurrent completions, with no luck-dependent layout or rare config.\nPR:N - In the highest-impact PCIe EP deployment a malicious or compromised upstream host triggers concurrent endpoint DMA through BAR/MHI traffic without any local Linux credentials, shell, or CAP_* on the vulnerable endpoint system.\nUI:N - Exploitation is fully automated once the attacker can submit concurrent DMA operations; no victim user action such as mounting filesystems or opening files is required.\nS:C - Corrupted shared DONE/ABORT masks and linked-list error enables on an endpoint can disrupt PCIe DMA to/from upstream host memory, impacting resources outside the endpoint kernel security authority across the PCIe boundary.\nC:L - Lost interrupt unmasking can leave host-visible DMA buffers partially updated or stall reads, enabling limited cross-boundary disclosure of in-flight or stale data rather than arbitrary kernel memory read primitives.\nI:L - Register corruption can disable linked-list error detection and desynchronize DMA completion handling, causing limited unintended modification of host or device DMA buffers even though addresses are normally driver-controlled.\nA:H - A lost DONE/ABORT unmask leaves channels permanently BUSY with completions never delivered, wedging dmaengine clients and endpoint services until reboot, constituting a severe availability failure."
    }
  }
}