CVE-2026-98918

In the Linux kernel, the following vulnerability has been resolved:

mm/damon/ops-common: use a page-aligned address in damon ptep mkold()

damon va prepare access check() picks a random byte address within the region and stores it in r->sampling addr. damon va mkold() passes it into a page table walk, which hands it to damon ptep mkold() as the address of the page to sample:

damon va mkold(mm, r->sampling addr) damon va walk page range(mm, addr, addr + 1) damon mkold pmd entry() damon ptep mkold(pte, vma, addr) ptep test and clear young(vma, addr, pte) mmu notifier clear young(mm, addr, addr + PAGE SIZE)

For arm64, before commit 6f0e1142173a ("arm64: mm: support batch clearing of the young flag for large folios"), the contpte helper walked exactly CONT PTES entries from the aligned-down page table pointer and used @addr only to pass down to each entry, so an unaligned value was harmless:

ptep = contpte align down(ptep);
addr = ALIGN DOWN(addr, CONT PTE SIZE);
for (i = 0; i < CONT PTES; i++, ptep++, addr += PAGE SIZE)

Now the range to walk is derived from @addr instead: end = addr + nr * PAGE SIZE, rounded up to CONT PTE SIZE. For a sample in the last page of a contpte block, the sub-page offset puts end just past the block boundary, so the round-up lands a whole block further and the walk clears PTE AF in CONT PTES entries beyond the sampled block. For the last block in a page table page, those entries are past the end of that page, so the walk writes into the page that follows.

Triggered by the full 7.1/7.2 kernel selftest suite on arm64 (EC2 c/m6g.4xlarge). The kernel sometimes crashes at or shortly after the DAMON test.

What the overrun does depends on the page that happens to follow the page table, so there is no single signature. If that page is read-only, the write faults in the sampling path itself:

Unable to handle kernel write to read-only memory at virtual address ffff0003c5d2d000 FSC = 0x0f: level 3 permission fault CM = 0, WnR = 1, TnD = 0, TagAccess = 0 CPU: 10 UID: 0 PID: 3487 Comm: kdamond.2 pc : contpte test and clear young ptes+0x70/0xc0 lr : damon ptep mkold+0x1e8/0x1f8 Call trace: contpte test and clear young ptes+0x70/0xc0 (P) damon mkold pmd entry+0x150/0x170 walk pmd range+0x110/0x2b0 walk pud range+0x10c/0x208 walk pgd range+0x134/0x258 walk page range+0x98/0x1b0 walk page range vma unsafe+0x90/0x148 walk page range vma+0x28/0x40 damon va walk page range+0x114/0x2b8 damon va prepare access checks+0xec/0x1a8 kdamond fn+0x534/0x770 kthread+0x128/0x138 ret from fork+0x10/0x20

Otherwise the page is writable, the PTE AF clearing succeeds silently and the damage only surfaces later, in whatever happened to own the page, so the backtrace is unrelated to DAMON and differs between runs.

Pass a page-aligned address to the ptep test and clear young() call in damon ptep mkold(), which is the only place DAMON can reach contpte test and clear young ptes() from. Nothing else sees the aligned address, and r->sampling addr itself is left as is, so the sampling and region bookkeeping semantics are unchanged.