KernelScan.io

HIGH Introduced in 5.13

swiotlb BouncePage OOB

CVE-2026-98254

CVSS 7.8 / 10.0 NVD

CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

KernelScan AI7.5HIGH

01

In the Linux kernel, the following vulnerability has been resolved: swiotlb: use the adjusted address for the highmem page lookup swiotlb_bounce() reads the page frame number from the slot's recorded orig_addr, then advances orig_addr by tlb_offset to reach the address the caller asked about. The highmem branch mixes the two: the offset within the page comes from the adjusted address, the page from the value before it. Once the adjustment crosses a page boundary the pair no longer describes one location, and the whole copy lands one page below the intended one for a positive tlb_offset, one above for a negative one. DMA_FROM_DEVICE writes the device data over the wrong page and leaves the intended one stale, DMA_TO_DEVICE feeds the device from a page the mapping may not cover. Partial syncs through dma_sync_single_range_for_*() are what make tlb_offset non-zero. The branch test is picked the same way, so a slot recorded in lowmem can be adjusted into highmem and the lowmem path then hands a highmem address to phys_to_virt(). Take both from orig_addr once it is final and keep pfn in the branch that uses it. PhysHighMem() asks the question straight from the address, as dma-debug already does.

02

Engine v0.7.0

Risk summary

A bug in the kernel's software I/O TLB (swiotlb) DMA bounce buffer code causes data to be copied to or from the wrong physical page when a partial DMA sync crosses a page boundary. This results in memory corruption — device data overwrites an unintended kernel page, or the device receives stale data from the wrong page. Systems with swiotlb active (common in virtualized environments and with limited-DMA devices) are at risk; any local user performing I/O on an affected device can trigger it.

Affectedkernel/dma/swiotlb.c (DMA subsystem - Software I/O TLB)

Vulnerability analysis

When the swiotlb bounce buffer performs a partial DMA sync with a non-zero offset, it computes the target page from the original address before applying the offset, but takes the offset within the page from the address after applying it. If the offset crosses a page boundary, these two values no longer describe the same location, and the entire DMA copy lands on the wrong physical page — one page below or above the intended target. This corrupts kernel memory for incoming DMA and feeds the device stale or unrelated data for outgoing DMA. The fix computes the page frame number from the final adjusted address so that both the page and the page-offset refer to the same location, and it also uses the adjusted address for the highmem branch test. The vulnerable path is reachable through normal device I/O on any system where swiotlb is active and a driver performs partial syncs, which is common in virtualized environments and on hardware with limited DMA addressing capability.

03

BranchIntroducedFixed inPatch commit
5.105.10.475.116e53b4d6afbd
5.125.12.145.13aa4709813b29
6.125.136.12.1120219b72f5c20
6.185.136.18.54b7d7914a9ae3
7.25.137.2.8—
mainline5.137.3—
cip/6.125.136.12.112-cip336e53b4d6afbd

04

State of the newest release of each Civil Infrastructure Platform SLTS tree, from CIP's own fix records.

CIP treeState
cip/6.12Fixed in 6.12.112-cip33
cip/6.1Affected — no CIP fix yet (latest 6.1.190-cip62)
cip/6.1-rtAffected — no CIP fix yet (latest 6.1.189-cip61-rt32)
cip/5.10Affected — no CIP fix yet (latest 5.10.272-cip79)
cip/5.10-rtAffected — no CIP fix yet (latest 5.10.272-cip79-rt35)