{"id":"CVE-2026-80527","state":"PUBLISHED","title":"ceph: fix hanging __ceph_get_caps() with stale mds_wanted","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nceph: fix hanging __ceph_get_caps() with stale mds_wanted\n\nA reader can hang forever in __ceph_get_caps() when the client no\nlonger holds `FILE_RD`, but local cap state still says that the\ncapability is already wanted (via `mds_wanted`).\n\nOne way to trigger this is through MDS cap revocation.  If another\nclient performs a conflicting operation, the MDS can revoke `FILE_RD`\nfrom the reader; the next read then has to reacquire `FILE_RD`.  If\nthe cap update that should request `FILE_RD` never reaches the MDS\nafter `cap->mds_wanted` was raised, the reader is left holding only\nnon-file caps while local `mds_wanted` still includes the file read\ncaps.\n\nIn that state, try_get_cap_refs() sees `need <= mds_wanted` and\nreturns 0, so __ceph_get_caps() just waits on `i_cap_wq`.  If the cap\nupdate that was supposed to request `FILE_RD never reaches the MDS\nafter `cap->mds_wanted was` raised, no further request is sent and the\nwaiter can sleep indefinitely until unrelated cap traffic happens to\nwake it up.\n\nThe ordering issue is that `cap->mds_wanted` is updated in\n__prep_cap() before the `CEPH_MSG_CLIENT_CAPS message` is actually\nqueued for send.  That makes one field serve two different meanings at\nonce: what this client wants, and what the client believes the MDS\nalready knows it wants.\n\nA proper fix would be to split those states and track whether a cap\nupdate is actually in flight or has been observed by the MDS.\nHowever, simply moving the `cap->mds_wanted assignment` later would\nnot be sufficient: queueing the message in the messenger does not\nguarantee that the MDS processed that specific wanted set, and\nreconnect or message loss can still invalidate that assumption.\nFixing that properly would require a larger rework of the cap state\nmachine.\n\nTo allow simpler backports to stable kernels, this patch implements a\nsimpler workaround:\n\n- stop waiting forever in __ceph_get_caps(); after a bounded wait,\n  fall back to the renew path\n\n- make ceph_renew_caps() issue a synchronous `OPEN` request whenever\n  the inode still does not actually hold the wanted caps, instead of\n  only calling ceph_check_caps()\n\nThe extra issued-vs-wanted check in ceph_renew_caps() is necessary\nbecause the previous test only checked whether the inode still had any\nreal caps at all.  That is not enough after revocation: the client can\nstill hold something like `pLs` and yet be missing `FILE_RD`\ncompletely.  In that case, falling back to ceph_check_caps() is not\nsufficient, because it still trusts `cap->mds_wanted` and may resend\nnothing.  By requiring `(issued & wanted) == wanted` before taking the\nasynchronous path, the code only uses ceph_check_caps() when the\n`wanted caps` are already actually issued.  Otherwise, it sends the\nsynchronous `OPEN` renew.\n\nThis preserves the existing asynchronous fast path when the wanted\ncaps are already issued, avoids changing cap-state semantics, and\nfixes the hang by guaranteeing that a stalled waiter eventually\nretries through a path that does not rely on the stale `mds_wanted`\nstate.\n\n[ idryomov: move CEPH_GET_CAPS_WAIT_TIMEOUT from libceph.h to\n  mds_client.h, formatting ]","publishedAt":"2026-08-26T14:37:06.178Z","updatedAt":"2026-10-16T19:26:24.678Z","severity":{"score":7.5,"label":"High","version":"3.1","vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","source":"Linux"},"assigner":"Linux","products":["Linux Linux"],"affected":[{"vendor":"Linux","product":"Linux","defaultStatus":"unaffected","versions":[{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"b5661524c5a45085a866864ca9b8ae2513dfd67a","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"5e84bc6f67e19fdd192d8b215de728acbfc12572","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"5fedf279a1ea369d39c8b06dd4547cdc576065d0","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"e05c315b4da0c16ea800ee4b2cb6c617f586d1b5","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"fcce1b3be6d286aa80831e730289f4c062053ae6","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"a3bc6b3e9ef3f5f5cb85a902a30a090c7931127c","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"9e55fe24c548ad3163903eb58bb002d28d32a630","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"50958bb928bad3bdba9e5d1b7ff4bbadcf6951e6","versionType":"git"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/ceph/caps.c","fs/ceph/file.c","fs/ceph/mds_client.h"]},{"vendor":"Linux","product":"Linux","defaultStatus":"affected","versions":[{"version":"5.8","status":"affected"},{"version":"0","status":"unaffected","lessThan":"5.8","versionType":"semver"},{"version":"5.10.266","status":"unaffected","lessThanOrEqual":"5.10.*","versionType":"semver"},{"version":"5.15.217","status":"unaffected","lessThanOrEqual":"5.15.*","versionType":"semver"},{"version":"6.1.184","status":"unaffected","lessThanOrEqual":"6.1.*","versionType":"semver"},{"version":"6.6.153","status":"unaffected","lessThanOrEqual":"6.6.*","versionType":"semver"},{"version":"6.12.105","status":"unaffected","lessThanOrEqual":"6.12.*","versionType":"semver"},{"version":"6.18.46","status":"unaffected","lessThanOrEqual":"6.18.*","versionType":"semver"},{"version":"7.1.10","status":"unaffected","lessThanOrEqual":"7.1.*","versionType":"semver"},{"version":"7.2","status":"unaffected","lessThanOrEqual":"*","versionType":"original_commit_for_fix"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/ceph/caps.c","fs/ceph/file.c","fs/ceph/mds_client.h"]}],"references":["https://git.kernel.org/stable/c/b5661524c5a45085a866864ca9b8ae2513dfd67a","https://git.kernel.org/stable/c/5e84bc6f67e19fdd192d8b215de728acbfc12572","https://git.kernel.org/stable/c/5fedf279a1ea369d39c8b06dd4547cdc576065d0","https://git.kernel.org/stable/c/e05c315b4da0c16ea800ee4b2cb6c617f586d1b5","https://git.kernel.org/stable/c/fcce1b3be6d286aa80831e730289f4c062053ae6","https://git.kernel.org/stable/c/a3bc6b3e9ef3f5f5cb85a902a30a090c7931127c","https://git.kernel.org/stable/c/9e55fe24c548ad3163903eb58bb002d28d32a630","https://git.kernel.org/stable/c/50958bb928bad3bdba9e5d1b7ff4bbadcf6951e6"],"weaknesses":[],"ingestedAt":"2026-10-29T20:13:07.220Z","source":{"repository":"CVEProject/cvelistV5","path":"cves/2026/80xxx/CVE-2026-80527.json","commit":"556c9d3a8cd17e218bb8e452ba7cd619a384ff47"},"raw":{"dataType":"CVE_RECORD","dataVersion":"5.2","cveMetadata":{"cveId":"CVE-2026-80527","state":"PUBLISHED","datePublished":"2026-08-26T14:37:06.178Z","dateUpdated":"2026-10-16T19:26:24.678Z","assignerShortName":"Linux","assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","dateReserved":"2026-08-26T14:34:25.764Z"},"containers":{"cna":{"title":"ceph: fix hanging __ceph_get_caps() with stale mds_wanted","descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nceph: fix hanging __ceph_get_caps() with stale mds_wanted\n\nA reader can hang forever in __ceph_get_caps() when the client no\nlonger holds `FILE_RD`, but local cap state still says that the\ncapability is already wanted (via `mds_wanted`).\n\nOne way to trigger this is through MDS cap revocation.  If another\nclient performs a conflicting operation, the MDS can revoke `FILE_RD`\nfrom the reader; the next read then has to reacquire `FILE_RD`.  If\nthe cap update that should request `FILE_RD` never reaches the MDS\nafter `cap->mds_wanted` was raised, the reader is left holding only\nnon-file caps while local `mds_wanted` still includes the file read\ncaps.\n\nIn that state, try_get_cap_refs() sees `need <= mds_wanted` and\nreturns 0, so __ceph_get_caps() just waits on `i_cap_wq`.  If the cap\nupdate that was supposed to request `FILE_RD never reaches the MDS\nafter `cap->mds_wanted was` raised, no further request is sent and the\nwaiter can sleep indefinitely until unrelated cap traffic happens to\nwake it up.\n\nThe ordering issue is that `cap->mds_wanted` is updated in\n__prep_cap() before the `CEPH_MSG_CLIENT_CAPS message` is actually\nqueued for send.  That makes one field serve two different meanings at\nonce: what this client wants, and what the client believes the MDS\nalready knows it wants.\n\nA proper fix would be to split those states and track whether a cap\nupdate is actually in flight or has been observed by the MDS.\nHowever, simply moving the `cap->mds_wanted assignment` later would\nnot be sufficient: queueing the message in the messenger does not\nguarantee that the MDS processed that specific wanted set, and\nreconnect or message loss can still invalidate that assumption.\nFixing that properly would require a larger rework of the cap state\nmachine.\n\nTo allow simpler backports to stable kernels, this patch implements a\nsimpler workaround:\n\n- stop waiting forever in __ceph_get_caps(); after a bounded wait,\n  fall back to the renew path\n\n- make ceph_renew_caps() issue a synchronous `OPEN` request whenever\n  the inode still does not actually hold the wanted caps, instead of\n  only calling ceph_check_caps()\n\nThe extra issued-vs-wanted check in ceph_renew_caps() is necessary\nbecause the previous test only checked whether the inode still had any\nreal caps at all.  That is not enough after revocation: the client can\nstill hold something like `pLs` and yet be missing `FILE_RD`\ncompletely.  In that case, falling back to ceph_check_caps() is not\nsufficient, because it still trusts `cap->mds_wanted` and may resend\nnothing.  By requiring `(issued & wanted) == wanted` before taking the\nasynchronous path, the code only uses ceph_check_caps() when the\n`wanted caps` are already actually issued.  Otherwise, it sends the\nsynchronous `OPEN` renew.\n\nThis preserves the existing asynchronous fast path when the wanted\ncaps are already issued, avoids changing cap-state semantics, and\nfixes the hang by guaranteeing that a stalled waiter eventually\nretries through a path that does not rely on the stale `mds_wanted`\nstate.\n\n[ idryomov: move CEPH_GET_CAPS_WAIT_TIMEOUT from libceph.h to\n  mds_client.h, formatting ]"}],"affected":[{"vendor":"Linux","product":"Linux","defaultStatus":"unaffected","versions":[{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"b5661524c5a45085a866864ca9b8ae2513dfd67a","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"5e84bc6f67e19fdd192d8b215de728acbfc12572","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"5fedf279a1ea369d39c8b06dd4547cdc576065d0","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"e05c315b4da0c16ea800ee4b2cb6c617f586d1b5","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"fcce1b3be6d286aa80831e730289f4c062053ae6","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"a3bc6b3e9ef3f5f5cb85a902a30a090c7931127c","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"9e55fe24c548ad3163903eb58bb002d28d32a630","versionType":"git"},{"version":"0a454bdd501ad1aa30bb72e9581efa338ad6ce5c","status":"affected","lessThan":"50958bb928bad3bdba9e5d1b7ff4bbadcf6951e6","versionType":"git"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/ceph/caps.c","fs/ceph/file.c","fs/ceph/mds_client.h"]},{"vendor":"Linux","product":"Linux","defaultStatus":"affected","versions":[{"version":"5.8","status":"affected"},{"version":"0","status":"unaffected","lessThan":"5.8","versionType":"semver"},{"version":"5.10.266","status":"unaffected","lessThanOrEqual":"5.10.*","versionType":"semver"},{"version":"5.15.217","status":"unaffected","lessThanOrEqual":"5.15.*","versionType":"semver"},{"version":"6.1.184","status":"unaffected","lessThanOrEqual":"6.1.*","versionType":"semver"},{"version":"6.6.153","status":"unaffected","lessThanOrEqual":"6.6.*","versionType":"semver"},{"version":"6.12.105","status":"unaffected","lessThanOrEqual":"6.12.*","versionType":"semver"},{"version":"6.18.46","status":"unaffected","lessThanOrEqual":"6.18.*","versionType":"semver"},{"version":"7.1.10","status":"unaffected","lessThanOrEqual":"7.1.*","versionType":"semver"},{"version":"7.2","status":"unaffected","lessThanOrEqual":"*","versionType":"original_commit_for_fix"}],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["fs/ceph/caps.c","fs/ceph/file.c","fs/ceph/mds_client.h"]}],"references":[{"url":"https://git.kernel.org/stable/c/b5661524c5a45085a866864ca9b8ae2513dfd67a"},{"url":"https://git.kernel.org/stable/c/5e84bc6f67e19fdd192d8b215de728acbfc12572"},{"url":"https://git.kernel.org/stable/c/5fedf279a1ea369d39c8b06dd4547cdc576065d0"},{"url":"https://git.kernel.org/stable/c/e05c315b4da0c16ea800ee4b2cb6c617f586d1b5"},{"url":"https://git.kernel.org/stable/c/fcce1b3be6d286aa80831e730289f4c062053ae6"},{"url":"https://git.kernel.org/stable/c/a3bc6b3e9ef3f5f5cb85a902a30a090c7931127c"},{"url":"https://git.kernel.org/stable/c/9e55fe24c548ad3163903eb58bb002d28d32a630"},{"url":"https://git.kernel.org/stable/c/50958bb928bad3bdba9e5d1b7ff4bbadcf6951e6"}],"metrics":[{"cvssV3_1":{"version":"3.1","vectorString":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","baseScore":7.5,"baseSeverity":"HIGH"},"scenarios":[{"lang":"en","value":"AV:N - On nodes with cephfs mounted (common in Ceph/Rook/Kubernetes), a remote Ceph client can perform conflicting writes so the MDS revokes FILE_RD via CEPH_MSG_CLIENT_CAPS over the network; the victim kernel then hits the stale mds_wanted hang on the next VFS read/mmap/ioctl path into __ceph_get_caps().\nAC:L - The fix commit documents MDS cap revocation from another client as a reliable trigger; an attacker controls both sides by holding a second Ceph client session and driving conflicting I/O and reconnect/message-loss timing, without depending on uncontrollable victim heap layout or rare kernel configs.\nPR:N - Exploitation requires no Linux account, capability, or init-namespace root on the victim host; any remote (or co-tenant) Ceph client with write access to the shared file can induce revocation and the hang while victims only perform normal reads on an already-mounted cephfs export.\nUI:N - No interactive victim action is needed beyond routine automated reads on cephfs-backed workloads; the attacker does not require the victim to mount media, click a link, or perform a one-off administrative step at exploitation time.\nS:U - Impact is confined to threads blocked in __ceph_get_caps() on the same host kernel security domain; it does not cross VM, container, or IOMMU boundaries or grant elevated privileges beyond availability loss on the affected client.\nC:N - This is a capability wait-loop logic error with no out-of-bounds access, use-after-free, or information disclosure; stale mds_wanted only prevents cap reacquisition and does not read or leak kernel or file data to an attacker.\nI:N - The bug does not modify inode data, kernel memory, or capabilities; it causes indefinite sleeping in cap acquisition with no path to arbitrary write, metadata corruption, or code execution.\nA:H - Before the fix, __ceph_get_caps() could wait forever on i_cap_wq when mds_wanted falsely indicated FILE_RD was already requested but not issued, hanging the reader thread (read/write/mmap/getattr) until unrelated cap traffic; repeated exploitation can deny service to cephfs-backed applications."}]}],"providerMetadata":{"shortName":"Linux","orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","dateUpdated":"2026-08-27T05:01:22.018Z"},"cpeApplicability":[{"nodes":[{"operator":"OR","negate":false,"cpeMatch":[{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"5.10.266"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"5.15.217"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"6.1.184"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"6.6.153"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"6.12.105"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"6.18.46"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"7.1.10"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"5.8","versionEndExcluding":"7.2"}]}]}],"x_generator":{"engine":"bippy-1.2.0"}},"adp":[{"title":"CISA ADP Vulnrichment","metrics":[{"other":{"type":"ssvc","content":{"id":"CVE-2026-80527","role":"CISA Coordinator","options":[{"Exploitation":"none"},{"Automatable":"yes"},{"Technical Impact":"partial"}],"version":"2.0.3","timestamp":"2026-10-16T19:16:34.310764Z"}}}],"providerMetadata":{"shortName":"CISA-ADP","orgId":"134c704f-9b21-4f2e-91b3-4a467353bcc0","dateUpdated":"2026-10-16T19:26:24.678Z"}}]}}}