cve-2024-26757
Vulnerability from cvelistv5
Published
2024-04-03 17:00
Modified
2024-08-02 00:14
Severity
Summary
md: Don't ignore read-only array in md_check_recovery()
References
SourceURLTags
416baaa9-dc9f-4396-8d5f-8c081fb06d67https://git.kernel.org/stable/c/2ea169c5a0b1134d573d07fc27a16f327ad0e7d3
416baaa9-dc9f-4396-8d5f-8c081fb06d67https://git.kernel.org/stable/c/55a48ad2db64737f7ffc0407634218cc6e4c513b
Impacted products
VendorProduct
LinuxLinux
LinuxLinux
Show details on NVD website

To clipboard 

{
  "containers": {
    "adp": [
      {
        "metrics": [
          {
            "other": {
              "content": {
                "id": "CVE-2024-26757",
                "options": [
                  {
                    "Exploitation": "none"
                  },
                  {
                    "Automatable": "no"
                  },
                  {
                    "Technical Impact": "partial"
                  }
                ],
                "role": "CISA Coordinator",
                "timestamp": "2024-04-03T19:35:42.965370Z",
                "version": "2.0.3"
              },
              "type": "ssvc"
            }
          }
        ],
        "providerMetadata": {
          "dateUpdated": "2024-06-04T17:48:28.099Z",
          "orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
          "shortName": "CISA-ADP"
        },
        "title": "CISA ADP Vulnrichment"
      },
      {
        "providerMetadata": {
          "dateUpdated": "2024-08-02T00:14:13.354Z",
          "orgId": "af854a3a-2127-422b-91ae-364da2661108",
          "shortName": "CVE"
        },
        "references": [
          {
            "tags": [
              "x_transferred"
            ],
            "url": "https://git.kernel.org/stable/c/2ea169c5a0b1134d573d07fc27a16f327ad0e7d3"
          },
          {
            "tags": [
              "x_transferred"
            ],
            "url": "https://git.kernel.org/stable/c/55a48ad2db64737f7ffc0407634218cc6e4c513b"
          }
        ],
        "title": "CVE Program Container"
      }
    ],
    "cna": {
      "affected": [
        {
          "defaultStatus": "unaffected",
          "product": "Linux",
          "programFiles": [
            "drivers/md/md.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "lessThan": "2ea169c5a0b1",
              "status": "affected",
              "version": "ecbfb9f118bc",
              "versionType": "git"
            },
            {
              "lessThan": "55a48ad2db64",
              "status": "affected",
              "version": "ecbfb9f118bc",
              "versionType": "git"
            }
          ]
        },
        {
          "defaultStatus": "affected",
          "product": "Linux",
          "programFiles": [
            "drivers/md/md.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "status": "affected",
              "version": "4.8"
            },
            {
              "lessThan": "4.8",
              "status": "unaffected",
              "version": "0",
              "versionType": "custom"
            },
            {
              "lessThanOrEqual": "6.7.*",
              "status": "unaffected",
              "version": "6.7.7",
              "versionType": "custom"
            },
            {
              "lessThanOrEqual": "*",
              "status": "unaffected",
              "version": "6.8",
              "versionType": "original_commit_for_fix"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nmd: Don\u0027t ignore read-only array in md_check_recovery()\n\nUsually if the array is not read-write, md_check_recovery() won\u0027t\nregister new sync_thread in the first place. And if the array is\nread-write and sync_thread is registered, md_set_readonly() will\nunregister sync_thread before setting the array read-only. md/raid\nfollow this behavior hence there is no problem.\n\nAfter commit f52f5c71f3d4 (\"md: fix stopping sync thread\"), following\nhang can be triggered by test shell/integrity-caching.sh:\n\n1) array is read-only. dm-raid update super block:\nrs_update_sbs\n ro = mddev-\u003ero\n mddev-\u003ero = 0\n  -\u003e set array read-write\n md_update_sb\n\n2) register new sync thread concurrently.\n\n3) dm-raid set array back to read-only:\nrs_update_sbs\n mddev-\u003ero = ro\n\n4) stop the array:\nraid_dtr\n md_stop\n  stop_sync_thread\n    set_bit(MD_RECOVERY_INTR, \u0026mddev-\u003erecovery);\n    md_wakeup_thread_directly(mddev-\u003esync_thread);\n    wait_event(..., !test_bit(MD_RECOVERY_RUNNING, \u0026mddev-\u003erecovery))\n\n5) sync thread done:\n md_do_sync\n set_bit(MD_RECOVERY_DONE, \u0026mddev-\u003erecovery);\n md_wakeup_thread(mddev-\u003ethread);\n\n6) daemon thread can\u0027t unregister sync thread:\n md_check_recovery\n  if (!md_is_rdwr(mddev) \u0026\u0026\n      !test_bit(MD_RECOVERY_NEEDED, \u0026mddev-\u003erecovery))\n   return;\n  -\u003e -\u003e MD_RECOVERY_RUNNING can\u0027t be cleared, hence step 4 hang;\n\nThe root cause is that dm-raid manipulate \u0027mddev-\u003ero\u0027 by itself,\nhowever, dm-raid really should stop sync thread before setting the\narray read-only. Unfortunately, I need to read more code before I\ncan refacter the handler of \u0027mddev-\u003ero\u0027 in dm-raid, hence let\u0027s fix\nthe problem the easy way for now to prevent dm-raid regression."
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2024-05-29T05:22:11.706Z",
        "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
        "shortName": "Linux"
      },
      "references": [
        {
          "url": "https://git.kernel.org/stable/c/2ea169c5a0b1134d573d07fc27a16f327ad0e7d3"
        },
        {
          "url": "https://git.kernel.org/stable/c/55a48ad2db64737f7ffc0407634218cc6e4c513b"
        }
      ],
      "title": "md: Don\u0027t ignore read-only array in md_check_recovery()",
      "x_generator": {
        "engine": "bippy-a5840b7849dd"
      }
    }
  },
  "cveMetadata": {
    "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
    "assignerShortName": "Linux",
    "cveId": "CVE-2024-26757",
    "datePublished": "2024-04-03T17:00:41.633Z",
    "dateReserved": "2024-02-19T14:20:24.170Z",
    "dateUpdated": "2024-08-02T00:14:13.354Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.1",
  "meta": {
    "nvd": "{\"cve\":{\"id\":\"CVE-2024-26757\",\"sourceIdentifier\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\",\"published\":\"2024-04-03T17:15:52.207\",\"lastModified\":\"2024-04-03T17:24:18.150\",\"vulnStatus\":\"Awaiting Analysis\",\"descriptions\":[{\"lang\":\"en\",\"value\":\"In the Linux kernel, the following vulnerability has been resolved:\\n\\nmd: Don\u0027t ignore read-only array in md_check_recovery()\\n\\nUsually if the array is not read-write, md_check_recovery() won\u0027t\\nregister new sync_thread in the first place. And if the array is\\nread-write and sync_thread is registered, md_set_readonly() will\\nunregister sync_thread before setting the array read-only. md/raid\\nfollow this behavior hence there is no problem.\\n\\nAfter commit f52f5c71f3d4 (\\\"md: fix stopping sync thread\\\"), following\\nhang can be triggered by test shell/integrity-caching.sh:\\n\\n1) array is read-only. dm-raid update super block:\\nrs_update_sbs\\n ro = mddev-\u003ero\\n mddev-\u003ero = 0\\n  -\u003e set array read-write\\n md_update_sb\\n\\n2) register new sync thread concurrently.\\n\\n3) dm-raid set array back to read-only:\\nrs_update_sbs\\n mddev-\u003ero = ro\\n\\n4) stop the array:\\nraid_dtr\\n md_stop\\n  stop_sync_thread\\n    set_bit(MD_RECOVERY_INTR, \u0026mddev-\u003erecovery);\\n    md_wakeup_thread_directly(mddev-\u003esync_thread);\\n    wait_event(..., !test_bit(MD_RECOVERY_RUNNING, \u0026mddev-\u003erecovery))\\n\\n5) sync thread done:\\n md_do_sync\\n set_bit(MD_RECOVERY_DONE, \u0026mddev-\u003erecovery);\\n md_wakeup_thread(mddev-\u003ethread);\\n\\n6) daemon thread can\u0027t unregister sync thread:\\n md_check_recovery\\n  if (!md_is_rdwr(mddev) \u0026\u0026\\n      !test_bit(MD_RECOVERY_NEEDED, \u0026mddev-\u003erecovery))\\n   return;\\n  -\u003e -\u003e MD_RECOVERY_RUNNING can\u0027t be cleared, hence step 4 hang;\\n\\nThe root cause is that dm-raid manipulate \u0027mddev-\u003ero\u0027 by itself,\\nhowever, dm-raid really should stop sync thread before setting the\\narray read-only. Unfortunately, I need to read more code before I\\ncan refacter the handler of \u0027mddev-\u003ero\u0027 in dm-raid, hence let\u0027s fix\\nthe problem the easy way for now to prevent dm-raid regression.\"},{\"lang\":\"es\",\"value\":\"En el kernel de Linux, se resolvi\u00f3 la siguiente vulnerabilidad: md: No ignorar la matriz de solo lectura en md_check_recovery() Generalmente, si la matriz no es de lectura y escritura, md_check_recovery() no registrar\u00e1 un nuevo sync_thread en primer lugar. Y si la matriz es de lectura y escritura y sync_thread est\u00e1 registrado, md_set_readonly() cancelar\u00e1 el registro de sync_thread antes de configurar la matriz como de solo lectura. md/raid sigue este comportamiento, por lo que no hay problema. Despu\u00e9s de commit f52f5c71f3d4 (\\\"md: arreglar la detenci\u00f3n del hilo de sincronizaci\u00f3n\\\"), el siguiente bloqueo puede activarse mediante test shell/integrity-caching.sh: 1) la matriz es de solo lectura. Superbloque de actualizaci\u00f3n dm-raid: rs_update_sbs ro = mddev-\u0026gt;ro mddev-\u0026gt;ro = 0 -\u0026gt; establecer matriz de lectura y escritura md_update_sb 2) registrar un nuevo hilo de sincronizaci\u00f3n simult\u00e1neamente. 3) dm-raid vuelve a configurar la matriz en solo lectura: rs_update_sbs mddev-\u0026gt;ro = ro 4) detiene la matriz: raid_dtr md_stop stop_sync_thread set_bit(MD_RECOVERY_INTR, \u0026amp;mddev-\u0026gt;recovery); md_wakeup_thread_directly(mddev-\u0026gt;sync_thread); wait_event(..., !test_bit(MD_RECOVERY_RUNNING, \u0026amp;mddev-\u0026gt;recovery)) 5) hilo de sincronizaci\u00f3n realizado: md_do_sync set_bit(MD_RECOVERY_DONE, \u0026amp;mddev-\u0026gt;recovery); md_wakeup_thread(mddev-\u0026gt;thread); 6) el hilo del demonio no puede cancelar el registro del hilo de sincronizaci\u00f3n: md_check_recovery if (!md_is_rdwr(mddev) \u0026amp;\u0026amp; !test_bit(MD_RECOVERY_NEEDED, \u0026amp;mddev-\u0026gt;recovery)) return; -\u0026gt; -\u0026gt; MD_RECOVERY_RUNNING no se puede borrar, por lo tanto el paso 4 se bloquea; La causa principal es que dm-raid manipula \u0027mddev-\u0026gt;ro\u0027 por s\u00ed solo; sin embargo, dm-raid realmente deber\u00eda detener el hilo de sincronizaci\u00f3n antes de configurar la matriz como de solo lectura. Desafortunadamente, necesito leer m\u00e1s c\u00f3digo antes de poder refactorizar el controlador de \u0027mddev-\u0026gt;ro\u0027 en dm-raid, por lo tanto, solucionemos el problema de la manera m\u00e1s f\u00e1cil por ahora para evitar la regresi\u00f3n de dm-raid.\"}],\"metrics\":{},\"references\":[{\"url\":\"https://git.kernel.org/stable/c/2ea169c5a0b1134d573d07fc27a16f327ad0e7d3\",\"source\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\"},{\"url\":\"https://git.kernel.org/stable/c/55a48ad2db64737f7ffc0407634218cc6e4c513b\",\"source\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\"}]}}"
  }
}


Log in or create an account to share your comment.




Tags
Taxonomy of the tags.