{"id":"CVE-2026-31397","title":"In the Linux kernel, the following vulnerability has been resolved:\n\nmm/huge_memory: fix use of NULL folio in move_pages_huge_pmd()\n\nmove_pages_huge_pmd() handles UFFDIO_MOVE for both normal THPs and huge\nzero pages","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\nmm/huge_memory: fix use of NULL folio in move_pages_huge_pmd()\n\nmove_pages_huge_pmd() handles UFFDIO_MOVE for both normal THPs and huge\nzero pages.  For the huge zero p…","severity":"high","cvss":7.8,"cvssVector":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","cwe":["CWE-476"],"vendor":"linux","product":"linux_kernel","affected":["linux_kernel >= 6.16, < 6.18.20","linux_kernel >= 6.19, < 6.19.10","linux_kernel = 7.0"],"patched":["linux_kernel 6.19.10"],"published":"2026-04-03","updated":"2026-07-24","source":"NVD","sourceUrl":"https://nvd.nist.gov/vuln/detail/CVE-2026-31397","references":[{"url":"https://git.kernel.org/stable/c/e3133d0986dc5a231d5419167dbac65312b28b41","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/f3caaee0f9e489fd2282d4ce45791dc8aed2da62","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/fae654083bfa409bb2244f390232e2be47f05bfc","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"}],"tags":["nvd"],"epss":0.00119,"epssPercentile":0.01988,"ingestedAt":"2026-07-25T22:05:04.182Z","slug":"CVE-2026-31397","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\nmm/huge_memory: fix use of NULL folio in move_pages_huge_pmd()\n\nmove_pages_huge_pmd() handles UFFDIO_MOVE for both normal THPs and huge\nzero pages.  For the huge zero page path, src_folio is explicitly set to\nNULL, and is used as a sentinel to skip folio operations like lock and\nrmap.\n\nIn the huge zero page branch, src_folio is NULL, so folio_mk_pmd(NULL,\npgprot) passes NULL through folio_pfn() and page_to_pfn().  With\nSPARSEMEM_VMEMMAP this silently produces a bogus PFN, installing a PMD\npointing to non-existent physical memory.  On other memory models it is a\nNULL dereference.\n\nUse page_folio(src_page) to obtain the valid huge zero folio from the\npage, which was obtained from pmd_page() and remains valid throughout.\n\nAfter commit d82d09e48219 (\"mm/huge_memory: mark PMD mappings of the huge\nzero folio special\"), moved huge zero PMDs must remain special so\nvm_normal_page_pmd() continues to treat them as special mappings.\n\nmove_pages_huge_pmd() currently reconstructs the destination PMD in the\nhuge zero page branch, which drops PMD state such as pmd_special() on\narchitectures with CONFIG_ARCH_HAS_PTE_SPECIAL.  As a result,\nvm_normal_page_pmd() can treat the moved huge zero PMD as a normal page\nand corrupt its refcount.\n\nInstead of reconstructing the PMD from the folio, derive the destination\nentry from src_pmdval after pmdp_huge_clear_flush(), then handle the PMD\nmetadata the same way move_huge_pmd() does for moved entries by marking it\nsoft-dirty and clearing uffd-wp.\n\n## Affected\n\n- `linux_kernel >= 6.16, < 6.18.20`\n- `linux_kernel >= 6.19, < 6.19.10`\n- `linux_kernel = 7.0`\n\n## Remediation\n\nUpgrade past the affected range:\n\n- `linux_kernel 6.19.10`","depth":"twilight","depthScore":43,"depthScoreParts":{"impact":42.9,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}