{"id":"CVE-2026-98225","title":"In the Linux kernel, the following vulnerability has been resolved:\n\nmm/shrinker: fix bogus set_shrinker_bit() with cgroup.memory=nokmem\n\nWith cgroup.memory=nokmem, shrinker_memcg_alloc() bails out early and\nnever allocates an id, so shr…","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\nmm/shrinker: fix bogus set_shrinker_bit() with cgroup.memory=nokmem\n\nWith cgroup.memory=nokmem, shrinker_memcg_alloc() bails out early and\nnever allocates an id, so shr…","severity":"none","vendor":"Linux","product":"Linux","affected":["Linux >= fafaeceb89a5e2e856ff04c2cacb6cae4a2ecb67 < 469c12a69ab1842a9090b6c074ffedbfd9584f9c","Linux >= fafaeceb89a5e2e856ff04c2cacb6cae4a2ecb67 < 932cfb25e7ce98d1f93895671ec186a3087e4f80","Linux 7.2"],"published":"2026-10-06","updated":"2026-10-06","sourceUpdated":"2026-10-06T09:18:09.587","source":"NVD","sourceUrl":"https://nvd.nist.gov/vuln/detail/CVE-2026-98225","references":[{"url":"https://git.kernel.org/stable/c/469c12a69ab1842a9090b6c074ffedbfd9584f9c","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/932cfb25e7ce98d1f93895671ec186a3087e4f80","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"}],"tags":["nvd","cve.org"],"ingestedAt":"2026-10-06T08:50:17.438Z","slug":"CVE-2026-98225","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\nmm/shrinker: fix bogus set_shrinker_bit() with cgroup.memory=nokmem\n\nWith cgroup.memory=nokmem, shrinker_memcg_alloc() bails out early and\nnever allocates an id, so shrinker->id keeps the 0 it got from the\nkzalloc() in shrinker_alloc().  __list_lru_init() then copies that 0 into\nlru->shrinker_id, where it looks like a valid bit index.\n\nNothing calls expand_shrinker_info() on nokmem either, so shrinker_nr_max\nstays 0 and every memcg ends up with an empty map (map_nr_max == 0).\n\ndeferred_split_folio() hands a real memcg to __list_lru_add() regardless\nof whether the lru is memcg aware, so the first THP queued in a cgroup\ndoes set_shrinker_bit(memcg, nid, 0) and trips the bounds check:\n\nWARNING: mm/shrinker.c:212 at set_shrinker_bit+0x7d/0x90, CPU#126\nCall Trace:\n <TASK>\n deferred_split_folio+0x18c/0x220\n map_anon_folio_pmd_nopf+0xdd/0x130\n map_anon_folio_pmd_pf+0x14/0xb0\n do_huge_pmd_anonymous_page+0x1a1/0x620\n __handle_mm_fault+0xea9/0x10d0\n handle_mm_fault+0xe5/0x320\n do_user_addr_fault+0x1cc/0x870\n exc_page_fault+0x81/0x1b0\n asm_exc_page_fault+0x27/0x30\n </TASK>\n\nHarmless, the WARN_ON_ONCE() is what keeps the out of bounds unit[] read\nfrom happening, but the id should not look valid in the first place. \nClear it before returning.\n\nTwo other spots could paper over this: drop the id in __list_lru_init()\nwhen nokmem turns memcg_aware off, or make deferred_split_folio() pass\nNULL like list_lru_add_obj() does.  Both leave shrinker->id lying around\nfor the next caller, so fix it where the id is handed out.\n\n## Remediation\n\nRefer to the linked advisories for vendor-supplied fixes and affected version ranges.","depth":"sunlit","depthScore":3,"depthScoreParts":{"impact":2.8,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}