CVE-2026-99648: mm/vma: call ->may_split() before a partial merge
In the Linux kernel, the following vulnerability has been resolved:
mm/vma: call ->maysplit() before a partial merge
A driver can use the ->maysplit() hook to forbid a VMA boundary at certain addresses.
A device-dax device is a character device that gives raw access to a range of memory through mmap(). device-dax refuses any address that is not a multiple of the device's alignment through vmops->maysplit.
vmamergeexistingrange() can move boundaries, but does not call vmops->maysplit.
If a range at the edge of a VMA gets the attributes of the adjacent VMA, the function moves only that range over to the neighbour: one VMA shrinks and the other grows. This is a partial merge, and vmamodify() tries it before any split.
Take a device-dax VMA that lies next to another VMA of the same file. The user calls mprotect() on the one page of the first VMA that touches the second VMA, and asks for the protection that the second VMA already has. The result is a partial merge. It puts the boundary inside a 1 GiB mapping, which is a single leaf entry at the PUD level.
As part of the merge, the kernel looks for a huge PMD to split at the new boundary. That code does not expect a PUD leaf:
vmamergeexistingrange() commitmerge() vmaadjusttranshuge() splithugepmdaddress() mmfindpmd() treats the PUD leaf as a pointer to a PMD table, so the "table" is the device memory that the leaf maps splithugepmd() reads a PMD that the user wrote to the device memory folioput() on the folio that this PMD names
Bad. While the code must at some point be taught to split PUD entries as well, the issue here is that we move VMA boundaries such that they are incompatible with the driver.
Fix this by calling ->maysplit() on the address that a partial merge would move the boundary to, and by abandoning the merge if the driver refuses. vmamodify() then tries the split, which fails with -EINVAL.
Factor out existing split handling into checksplit() and reuse it.
Fix generated by an LLM and sent by Chris Mason in private.
[ david: improve description; cleanup the code ]
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 6.18.64 - Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 7.3.5 - Upgrade
Upgrade
debian/linuxto a version that resolves this vulnerability.Fixed in 6.1.176-1Fixed in 6.1.192-2Fixed in 6.12.107-1Fixed in 6.12.117-1Fixed in 7.3.5-1
Event History
Frequently Asked Questions
Which deployments are realistically exposed?
Systems using device-dax mappings are the relevant exposure. The issue requires a device-dax VMA to be adjacent to another VMA of the same file, with an alignment constraint enforced through the VMA may_split hook.
What operation is needed to trigger the vulnerable path?
A user must call mprotect() on the page at the edge of the first VMA, requesting protections already used by the adjacent VMA. This can cause a partial merge that moves the VMA boundary without first consulting may_split().