{"cve":"CVE-2026-64130","nvd":{"published":"2026-07-19T16:17:54.723","modified":"2026-08-13T15:10:33.113","status":"Analyzed","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nmm/page_alloc: fix initialization of tags of the huge zero folio with init_on_free\n\n__GFP_ZEROTAGS semantics are currently a bit weird, but effectively this\nflag is only ever set alongside __GFP_ZERO and __GFP_SKIP_KASAN.\n\nIf we run with init_on_free, we will zero out pages during\n__free_pages_prepare(), to skip zeroing on the allocation path.\n\nHowever, when allocating with __GFP_ZEROTAG set, post_alloc_hook() will\nconsequently not only skip clearing page content, but also skip clearing\ntag memory.\n\nNot clearing tags through __GFP_ZEROTAGS is irrelevant for most pages that\nwill get mapped to user space through set_pte_at() later: set_pte_at() and\nfriends will detect that the tags have not been initialized yet\n(PG_mte_tagged not set), and initialize them.\n\nHowever, for the huge zero folio, which will be mapped through a PMD\nmarked as special, this initialization will not be performed, ending up\nexposing whatever tags were still set for the pages.\n\nThe docs (Documentation/arch/arm64/memory-tagging-extension.rst) state\nthat allocation tags are set to 0 when a page is first mapped to user\nspace.  That no longer holds with the huge zero folio when init_on_free is\nenabled.\n\nFix it by decoupling __GFP_ZEROTAGS from __GFP_ZERO, passing to\ntag_clear_highpages() whether we want to also clear page content.\n\nInvert the meaning of the tag_clear_highpages() return value to have\nclearer semantics.\n\nReproduced with the huge zero folio by modifying the check_buffer_fill\narm64/mte selftest to use a 2 MiB area, after making sure that pages have\na non-0 tag set when freeing (note that, during boot, we will not actually\ninitialize tags, but only set KASAN_TAG_KERNEL in the page flags).\n\n\t$ ./check_buffer_fill\n\t1..20\n\t...\n\tnot ok 17 Check initial tags with private mapping, sync error mode and mmap memory\n\tnot ok 18 Check initial tags with private mapping, sync error mode and mmap/mprotect memory\n\t...\n\nThis code needs more cleanups; we'll tackle that next, like\ndecoupling __GFP_ZEROTAGS from __GFP_SKIP_KASAN.\n\n[akpm@linux-foundation.org: s/__GPF_ZERO/__GFP_ZERO/, per David]","cvss_v4":null,"cvss_v3":{"version":"3.1","vector":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H","score":5.5},"cwes":["CWE-908"],"references":["https://git.kernel.org/stable/c/2f2aec5120b93a8f8b52dc50cdc60dbb8aec72f6","https://git.kernel.org/stable/c/6a288a4ddb4a994490505ab5f41c445f8e6b6467","https://git.kernel.org/stable/c/738d18f1da3513d17b6f7bf30146cc4ac2480ffd"],"products":["linux:linux_kernel"],"source":"nvd","synced_at":"2026-10-19T22:45:36.895656543+00:00"},"epss":{"cve":"CVE-2026-64130","score":0.0016,"percentile":0.05168,"date":"2026-11-01"},"kev":null,"osv":[],"attribution":["This product uses the NVD API but is not endorsed or certified by the NVD.","EPSS scores: FIRST Exploit Prediction Scoring System, https://www.first.org/epss","Known exploited vulnerabilities: CISA Known Exploited Vulnerabilities Catalog, https://www.cisa.gov/known-exploited-vulnerabilities-catalog","Package advisories: OSV (https://osv.dev), from the GitHub Advisory Database, PyPA, Go, RustSec and other ecosystem databases, under CC-BY 4.0 or CC0","CVE records: the CVE Program's CVE List (https://www.cve.org), under its terms of use (https://www.cve.org/Legal/TermsOfUse). CVE is a registered trademark of The MITRE Corporation; this product is not affiliated with the CVE Program."]}