{"dataType":"CVE_RECORD","dataVersion":"5.2","cveMetadata":{"cveId":"CVE-2026-72034","assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","state":"PUBLISHED","assignerShortName":"Linux","dateReserved":"2026-08-09T03:40:39.901Z","datePublished":"2026-08-15T05:51:55.739Z","dateUpdated":"2026-08-17T05:39:37.172Z"},"containers":{"cna":{"providerMetadata":{"orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux","dateUpdated":"2026-08-17T05:39:37.172Z"},"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nfhandle: reject detached mounts in capable_wrt_mount()\n\nThe recent fhandle RCU fix moved the mount namespace capability check\ninto capable_wrt_mount(), so a non-NULL mnt_namespace survives the\nns_capable() dereference. The helper still assumes the later\nREAD_ONCE(mount->mnt_ns) must be non-NULL because may_decode_fh()\nchecked is_mounted() first.\n\nThat assumption is not stable. A detached mount from\nopen_tree(..., OPEN_TREE_CLONE) can be dissolved on fput while\nopen_by_handle_at() is between those checks, and umount_tree() can\nclear mount->mnt_ns. If the helper observes NULL, it dereferences\nmnt_ns->user_ns and panics.\n\nReturn false when the RCU read observes a detached mount. This keeps\nthe relaxed permission path conservative: a mount no longer attached\nto a namespace cannot authorize open_by_handle_at() access."}],"metrics":[{"cvssV3_1":{"version":"3.1","vectorString":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","baseScore":7.8,"baseSeverity":"HIGH"},"scenarios":[{"lang":"en","value":"AV:L - Exploitation requires local syscalls only: open_tree(OPEN_TREE_CLONE) to obtain a detached mount fd, open_by_handle_at() to enter may_decode_fh()/capable_wrt_mount(), and close()/fput to dissolve the mount; no network, adjacent, or physical interface reaches this VFS fhandle path.\nAC:L - The attacker controls both sides of the TOCTOU race by concurrently calling open_by_handle_at() on a detached mount fd while another thread closes that fd (or a dup) to trigger dissolve_on_fput()/umount_tree() clearing mount->mnt_ns, making success depend on attacker timing rather than external state.\nPR:L - The relaxed permission path requires CAP_SYS_ADMIN in the current mount namespace (may_mount/open_tree) and CAP_DAC_READ_SEARCH in the current user namespace while lacking init_user_ns CAP_DAC_READ_SEARCH; these are obtainable by an unprivileged local user via user/mount namespaces (unshare -Urn), not real init-namespace root.\nUI:N - No victim action is required; the attacker creates the detached mount, supplies the file handle, and drives the concurrent open_by_handle_at()/close race entirely through their own syscalls and threads.\nS:U - Impact is confined to kernel memory/availability within the same security authority (local/container-triggered kernel panic or namespace lifetime corruption); it does not cross a VM escape, sandbox, or IOMMU boundary per kernel CVSS scope guidance.\nC:H - The affected window (since 620c266f3949) includes unsynchronized mount::mnt_ns access that can UAF freed mnt_namespace memory before the RCU fix, and later NULL/stale namespace dereferences; per kernel guidance UAF and mount-namespace lifetime races enable arbitrary kernel memory disclosure.\nI:H - Concurrent umount_tree() clearing mount->mnt_ns while capable_wrt_mount() dereferences mnt_ns->user_ns is a mount-namespace lifetime corruption bug; UAF in the same code path can be leveraged for heap corruption and arbitrary kernel writes/code execution, not merely a benign error return.\nA:H - Observing NULL or freed mnt_ns during ns_capable(mnt_ns->user_ns, CAP_SYS_ADMIN) causes a kernel oops/panic; the race is repeatable by reopening detached mounts, enabling reliable host DoS from containerized tenants on multi-tenant cloud nodes."}]}],"affected":[{"product":"Linux","vendor":"Linux","defaultStatus":"unaffected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/fhandle.c"],"versions":[{"version":"620c266f394932e5decc4b34683a75dfc59dc2f4","lessThan":"15104234c267ebe04b9f9a73e5c2179cc265ff60","status":"affected","versionType":"git"},{"version":"620c266f394932e5decc4b34683a75dfc59dc2f4","lessThan":"6c52226072a3c61337b1eec1799bb987748884fa","status":"affected","versionType":"git"},{"version":"620c266f394932e5decc4b34683a75dfc59dc2f4","lessThan":"6ee183d89261bf1d1cf9f06d80a40dab8f36ee55","status":"affected","versionType":"git"},{"version":"620c266f394932e5decc4b34683a75dfc59dc2f4","lessThan":"6c732471740bc2ac9b0946134f9f551dc75f4369","status":"affected","versionType":"git"}]},{"product":"Linux","vendor":"Linux","defaultStatus":"affected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/fhandle.c"],"versions":[{"version":"6.11","status":"affected"},{"version":"0","lessThan":"6.11","status":"unaffected","versionType":"semver"},{"version":"6.12.97","lessThanOrEqual":"6.12.*","status":"unaffected","versionType":"semver"},{"version":"6.18.40","lessThanOrEqual":"6.18.*","status":"unaffected","versionType":"semver"},{"version":"7.1.5","lessThanOrEqual":"7.1.*","status":"unaffected","versionType":"semver"},{"version":"7.2","lessThanOrEqual":"*","status":"unaffected","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"operator":"OR","negate":false,"cpeMatch":[{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.11","versionEndExcluding":"6.12.97"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.11","versionEndExcluding":"6.18.40"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.11","versionEndExcluding":"7.1.5"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.11","versionEndExcluding":"7.2"}]}]}],"references":[{"url":"https://git.kernel.org/stable/c/15104234c267ebe04b9f9a73e5c2179cc265ff60"},{"url":"https://git.kernel.org/stable/c/6c52226072a3c61337b1eec1799bb987748884fa"},{"url":"https://git.kernel.org/stable/c/6ee183d89261bf1d1cf9f06d80a40dab8f36ee55"},{"url":"https://git.kernel.org/stable/c/6c732471740bc2ac9b0946134f9f551dc75f4369"}],"title":"fhandle: reject detached mounts in capable_wrt_mount()","x_generator":{"engine":"bippy-1.2.0"}}}}