| { |
| "containers": { |
| "cna": { |
| "providerMetadata": { |
| "orgId": "f4215fc3-5b6b-47ff-a258-f7189bd81038" |
| }, |
| "descriptions": [ |
| { |
| "lang": "en", |
| "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nmm/thp: fix deferred split unqueue naming and locking\n\nRecent changes are putting more pressure on THP deferred split queues:\nunder load revealing long-standing races, causing list_del corruptions,\n\"Bad page state\"s and worse (I keep BUGs in both of those, so usually\ndon't get to see how badly they end up without). The relevant recent\nchanges being 6.8's mTHP, 6.10's mTHP swapout, and 6.12's mTHP swapin,\nimproved swap allocation, and underused THP splitting.\n\nBefore fixing locking: rename misleading folio_undo_large_rmappable(),\nwhich does not undo large_rmappable, to folio_unqueue_deferred_split(),\nwhich is what it does. But that and its out-of-line __callee are mm\ninternals of very limited usability: add comment and WARN_ON_ONCEs to\ncheck usage; and return a bool to say if a deferred split was unqueued,\nwhich can then be used in WARN_ON_ONCEs around safety checks (sparing\ncallers the arcane conditionals in __folio_unqueue_deferred_split()).\n\nJust omit the folio_unqueue_deferred_split() from free_unref_folios(), all\nof whose callers now call it beforehand (and if any forget then bad_page()\nwill tell) - except for its caller put_pages_list(), which itself no\nlonger has any callers (and will be deleted separately).\n\nSwapout: mem_cgroup_swapout() has been resetting folio->memcg_data 0\nwithout checking and unqueueing a THP folio from deferred split list;\nwhich is unfortunate, since the split_queue_lock depends on the memcg\n(when memcg is enabled); so swapout has been unqueueing such THPs later,\nwhen freeing the folio, using the pgdat's lock instead: potentially\ncorrupting the memcg's list. __remove_mapping() has frozen refcount to 0\nhere, so no problem with calling folio_unqueue_deferred_split() before\nresetting memcg_data.\n\nThat goes back to 5.4 commit 87eaceb3faa5 (\"mm: thp: make deferred split\nshrinker memcg aware\"): which included a check on swapcache before adding\nto deferred queue, but no check on deferred queue before adding THP to\nswapcache. That worked fine with the usual sequence of events in reclaim\n(though there were a couple of rare ways in which a THP on deferred queue\ncould have been swapped out), but 6.12 commit dafff3f4c850 (\"mm: split\nunderused THPs\") avoids splitting underused THPs in reclaim, which makes\nswapcache THPs on deferred queue commonplace.\n\nKeep the check on swapcache before adding to deferred queue? Yes: it is\nno longer essential, but preserves the existing behaviour, and is likely\nto be a worthwhile optimization (vmstat showed much more traffic on the\nqueue under swapping load if the check was removed); update its comment.\n\nMemcg-v1 move (deprecated): mem_cgroup_move_account() has been changing\nfolio->memcg_data without checking and unqueueing a THP folio from the\ndeferred list, sometimes corrupting \"from\" memcg's list, like swapout. \nRefcount is non-zero here, so folio_unqueue_deferred_split() can only be\nused in a WARN_ON_ONCE to validate the fix, which must be done earlier:\nmem_cgroup_move_charge_pte_range() first try to split the THP (splitting\nof course unqueues), or skip it if that fails. Not ideal, but moving\ncharge has been requested, and khugepaged should repair the THP later:\nnobody wants new custom unqueueing code just for this deprecated case.\n\nThe 87eaceb3faa5 commit did have the code to move from one deferred list\nto another (but was not conscious of its unsafety while refcount non-0);\nbut that was removed by 5.6 commit fac0516b5534 (\"mm: thp: don't need care\ndeferred split queue in memcg charge move path\"), which argued that the\nexistence of a PMD mapping guarantees that the THP cannot be on a deferred\nlist. As above, false in rare cases, and now commonly false.\n\nBackport to 6.11 should be straightforward. Earlier backports must take\ncare that other _deferred_list fixes and dependencies are included. There\nis not a strong case for backports, but they can fix cornercases." |
| } |
| ], |
| "affected": [ |
| { |
| "product": "Linux", |
| "vendor": "Linux", |
| "defaultStatus": "unaffected", |
| "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", |
| "programFiles": [ |
| "mm/huge_memory.c", |
| "mm/internal.h", |
| "mm/memcontrol-v1.c", |
| "mm/memcontrol.c", |
| "mm/migrate.c", |
| "mm/page_alloc.c", |
| "mm/swap.c", |
| "mm/vmscan.c" |
| ], |
| "versions": [ |
| { |
| "version": "87eaceb3faa59b9b4d940ec9554ce251325d83fe", |
| "lessThan": "fc4951c3e3358dd82ea508e893695b916c813f17", |
| "status": "affected", |
| "versionType": "git" |
| }, |
| { |
| "version": "87eaceb3faa59b9b4d940ec9554ce251325d83fe", |
| "lessThan": "afb1352d06b1b6b2cfd1f901c766a430c87078b3", |
| "status": "affected", |
| "versionType": "git" |
| }, |
| { |
| "version": "87eaceb3faa59b9b4d940ec9554ce251325d83fe", |
| "lessThan": "f8f931bba0f92052cf842b7e30917b1afcc77d5a", |
| "status": "affected", |
| "versionType": "git" |
| } |
| ] |
| }, |
| { |
| "product": "Linux", |
| "vendor": "Linux", |
| "defaultStatus": "affected", |
| "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", |
| "programFiles": [ |
| "mm/huge_memory.c", |
| "mm/internal.h", |
| "mm/memcontrol-v1.c", |
| "mm/memcontrol.c", |
| "mm/migrate.c", |
| "mm/page_alloc.c", |
| "mm/swap.c", |
| "mm/vmscan.c" |
| ], |
| "versions": [ |
| { |
| "version": "5.4", |
| "status": "affected" |
| }, |
| { |
| "version": "0", |
| "lessThan": "5.4", |
| "status": "unaffected", |
| "versionType": "semver" |
| }, |
| { |
| "version": "6.6.62", |
| "lessThanOrEqual": "6.6.*", |
| "status": "unaffected", |
| "versionType": "semver" |
| }, |
| { |
| "version": "6.11.8", |
| "lessThanOrEqual": "6.11.*", |
| "status": "unaffected", |
| "versionType": "semver" |
| }, |
| { |
| "version": "6.12", |
| "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": "5.4", |
| "versionEndExcluding": "6.6.62" |
| }, |
| { |
| "vulnerable": true, |
| "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", |
| "versionStartIncluding": "5.4", |
| "versionEndExcluding": "6.11.8" |
| }, |
| { |
| "vulnerable": true, |
| "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", |
| "versionStartIncluding": "5.4", |
| "versionEndExcluding": "6.12" |
| } |
| ] |
| } |
| ] |
| } |
| ], |
| "references": [ |
| { |
| "url": "https://git.kernel.org/stable/c/fc4951c3e3358dd82ea508e893695b916c813f17" |
| }, |
| { |
| "url": "https://git.kernel.org/stable/c/afb1352d06b1b6b2cfd1f901c766a430c87078b3" |
| }, |
| { |
| "url": "https://git.kernel.org/stable/c/f8f931bba0f92052cf842b7e30917b1afcc77d5a" |
| } |
| ], |
| "title": "mm/thp: fix deferred split unqueue naming and locking", |
| "x_generator": { |
| "engine": "bippy-1.2.0" |
| } |
| } |
| }, |
| "cveMetadata": { |
| "assignerOrgId": "f4215fc3-5b6b-47ff-a258-f7189bd81038", |
| "cveID": "CVE-2024-53079", |
| "requesterUserId": "gregkh@kernel.org", |
| "serial": "1", |
| "state": "PUBLISHED" |
| }, |
| "dataType": "CVE_RECORD", |
| "dataVersion": "5.0" |
| } |
