Look up another CVE · open CVE-2026-74417 in the interactive tracker
CVE-2026-74417
drm/radeon: fix integer overflow in radeon_align_pitch()
Published 2026-08-15 (2 months ago) by the Linux kernel CNA. Subsystem drivers/gpu. Debian urgency: not yet assigned.
What to do, per Debian release
Debian's own status for CVE-2026-74417 in every release this site tracks, and what each one means in practice. Every status below is the one the Debian Security Team publishes for this CVE. source
| Release | Status | Version | What this means |
|---|---|---|---|
| Debian 12bookworm | vulnerable | 6.1.190-1 | No fix published anywhere yet, neither in Debian nor upstream for the series bookworm tracks. Watch the Debian tracker page for this CVE. Advisories for bookworm have historically landed a median of 23 days after publication. |
| Debian 13trixie | fixed | 6.12.100-1 | Update the kernel and reboot. Fixed in 6.12.100-1. A running kernel keeps the old code until the machine restarts. |
| Debian 14forky | fixed | 7.1.5-1 | Update the kernel and reboot. Fixed in 7.1.5-1. A running kernel keeps the old code until the machine restarts. |
| Debian unstablesid | fixed | 7.1.5-1 | Update the kernel and reboot. Fixed in 7.1.5-1. A running kernel keeps the old code until the machine restarts. |
The vulnerability
What the bug is
radeon_align_pitch() has the same kind of overflow issue as the old amdgpu helper: both the alignment round-up add and the final 'aligned * cpp' calculation can overflow signed int. If that wraps, radeon_mode_dumb_create() can end up returning an invalid pitch or creating a zero-sized dumb buffer. Fix this by using check_add_overflow() for the alignment round-up and check_mul_overflow() for the final pitch calculation, returning 0 on overflow. Also reject zero pitch and size in radeon_mode_dumb_create(). Found via AST-based call-graph analysis using sqry.
Where it lives
The fix touches these files. If the subsystem is one you do not use, the practical exposure is lower, though the package is still the vulnerable one.
drivers/gpu/drm/radeon/radeon_gem.c
- What this is
- RADEON and AMDGPU DRM DRIVERSMAINTAINERS section matched on drivers/gpu/drm/radeon/ · amd-gfx@lists.freedesktop.org · Supported
- Kernel module
- radeon
Check it on the machine
grep -w CONFIG_DRM_RADEON /boot/config-$(uname -r)Reads the booted kernel's own configuration, on the machine and nowhere else. =y is built in, =m is a module, and an is not set line means the code is not there.
lsmod | grep -w radeonPrints a line if that module is loaded right now, and nothing if it is not. One that is not loaded can still be loaded later by anything that needs it.
Show whether Debian builds this, per release
Debian builds this, but not on every flavour. At least one release below builds the code on some of its kernel flavours and not on others, so the answer depends on the architecture and flavour installed.
- bookwormMnot built on amd64/cloud-amd64, arm64/cloud-arm64, armhf/armmp, armhf/armmp-lpae, armhf/rt/armmp
- trixieMnot built on amd64/cloud-amd64, arm64/cloud-arm64, armel/rpi, armhf/armmp, armhf/armmp-lpae, armhf/rt/armmp, s390x/s390x
- forky, sidMnot built on amd64/cloud-amd64, arm64/cloud-arm64, armhf/armmp, armhf/armmp-lpae, s390x/s390x
- M
- a module where it is built at all; some flavours do not build it
- Build switch
- CONFIG_DRM_RADEON
From the configuration Debian publishes for every kernel flavour it builds. It covers only the architectures Debian ships a linux-config package for, and nothing about a kernel you built yourself. Debian's own status above stays authoritative about the source package.MAINTAINERS kbuild Makefiles Debian linux-config packages
Signals
- Published
- 2026-08-15 (2 months ago)source
- CVSS
- 7.8 high
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:Hsource recompute it - Attack vector
- local (AV:L)source
- EPSS
- 0.18%, higher than 6.6% of all CVEssource
- CISA KEV
- not listedthe KEV catalogue
- Debian urgency
- not yet assignedsource
- Subsystem
- drivers/gpu
Upstream fixes
The versions the CNA records as introducing and fixing this, and the commit for each. source
| Introduced | Fixed in | Commit |
|---|---|---|
| 2.6.39 | 6.6.145 | b7b44937c548 |
| 2.6.39 | 6.12.97 | 415bb9893e24 |
| 2.6.39 | 6.18.40 | d9dfa176899d |
| 2.6.39 | 7.1.5 | dfc7b5b55994 |
| 2.6.39 | 7.2 | ce3b24eb3ee8 |
Why the CNA scored it this way
Quoted from the kernel CNA's own scoring file, word for word. A highlighted metric carries the most severe value CVSS v3.1 defines for it, which is what pushes the score up. source
- AV:LAttack vector, local
- The bug is reached only via DRM_IOCTL_MODE_CREATE_DUMB on a local primary DRM device node (/dev/dri/card*); radeon_drm_ioctl() dispatches to radeon_mode_dumb_create() -> radeon_align_pitch(), with no network, Bluetooth, or remote packet path to this legacy radeon KMS driver.
- AC:LAttack complexity, low
- An attacker fully controls width, height, and bpp in the ioctl and can pick values (e.g. width=0x3FFFFFC0, bpp=32) that pass generic drm_mode_create_dumb() checks yet wrap signed-int pitch math, deterministically yielding zero or invalid pitch without races or uncontrollable layout.
- PR:LPrivileges required, low
- Exploitation requires opening the primary DRM node (render clients are blocked because CREATE_DUMB lacks DRM_RENDER_ALLOW); typical access is via the video group or logind session ACLs, i.e. a low-privileged local account, not root/CAP_SYS_ADMIN in the init namespace.
- UI:NUser interaction, none
- The attacker triggers the overflow from their own process by issuing CREATE_DUMB with crafted dimensions; no victim must mount filesystems, click links, or perform any separate action.
- S:UScope, unchanged
- Impact is in-kernel GEM/TTM memory corruption and privilege escalation within the host kernel security boundary, not a KVM guest-to-host escape, IOMMU bypass, or other cross-authority scope change.
- C:HConfidentiality impact, high
- Signed-int overflow can return pitch 0 or a bogus u32 pitch while drm_mode_create_dumb() already validated a much larger stride, creating undersized VRAM/GTT GEM objects whose metadata mismatches actual backing, enabling out-of-bounds kernel reads of adjacent allocations.
- I:HIntegrity impact, high
- A zero-sized or undersized dumb buffer paired with attacker-chosen width/height lets subsequent mmap, map_dumb, and GPU scanout paths treat the object as larger than allocated, corrupting adjacent kernel heap/VRAM objects and enabling control-flow hijack primitives.
- A:HAvailability impact, high
- Pre-fix code allocates zero-byte GEM objects on overflow instead of failing cleanly, and malformed pitch/size metadata can provoke kernel oopses, GPU faults, or repeatable denial of service when the BO is mapped or used for display.