Imagination Technologies vulnerabilities and threats

CVE-2026-45197 is a low-severity time-of-check to time-of-use (TOCTOU) race condition vulnerability in Imagination Technologies Graphics DDK. It affects kernel software running inside a Guest VM that interacts with the GPU Firmware. The flaw allows improper commands to be posted to the GPU Firmware, potentially causing reads or writes outside the Guest's virtualized GPU memory due to a TOCTOU bug in memory access validation.

Software installed and run as a non-privileged user may conduct improper GPU driver IOCTL calls to create an allocation scenario that when freed would cause double free and kernel heap corruption. Scenario caused by fabricating a specific combination of flags on the allocation interface that would cause an incorrect double free event when freed.

CVE-2026-45202 is a medium severity vulnerability in Imagination Technologies Graphics DDK involving a double free condition. Non-privileged users can trigger GPU system calls that lead to GPU memory leaks and potential kernel heap corruption. The issue arises from improper handling of memory free paths that do not maintain state data for upgraded higher order allocations, causing double free or memory leak events.

CVE-2026-45201 is a high-severity vulnerability in Imagination Technologies Graphics DDK where improper validation of a specified log2 page size during GPU system calls can lead to out-of-bounds (OOB) read and/or write. This occurs when software running as a non-privileged user passes an invalid log2 page size, causing 4K pages to be treated as higher order pages and enabling memory access beyond intended boundaries.

CVE-2026-45199 is a high-severity vulnerability in Imagination Technologies Graphics DDK where kernel software running inside a Guest VM can send improper commands to the GPU firmware. This triggers out-of-bounds writes outside the Guest's virtualized GPU memory, potentially allowing privilege escalation.

CVE-2026-49746 is a high-severity vulnerability in Imagination Technologies Graphics DDK involving improper validation of array indices. This flaw allows software running as a non-privileged user to perform out-of-bounds (OOB) read accesses to kernel memory via GPU system calls. In some cases, it can lead to use-after-free (UAF) conditions on arbitrary GPU memory pages, potentially causing denial of service or information disclosure.

CVE-2026-45204 is a medium severity vulnerability in Imagination Technologies Graphics DDK where a null pointer dereference occurs in a kernel thread error path. This can be triggered by improper GPU system calls from software running as a non-privileged user, leading to kernel exceptions, platform instability, and denial of service.

CVE-2026-45198 is a high severity vulnerability in Imagination Technologies Graphics DDK where GPU firmware may use pointers from non-secure memory controlled by the Rich Execution Environment (REE). This allows an attacker with REE kernel control to modify pointer values, potentially corrupting GPU firmware data. The flaw involves untrusted pointer dereference and trust boundary violation between secure and non-secure memory regions.

An integer overflow when calculating physical offsets for sparse PMRs may result in 32-bit truncation of address computations for PMRs larger than 4 GB. This can lead to incorrect GPU MMU mappings and may allow a non-privileged user to trigger access to unintended physical memory, resulting in memory corruption or information disclosure.

Kernel software installed and running inside a Guest VM may post improper commands to the GPU Firmware to trigger a write of data outside the Guest's virtualised GPU memory. Software installed and run under a Guest VM can send commands to the GPU which result in out of bounds memory accesses. These can be used to escalate privileges.

Kernel software installed and running inside a Guest VM may post improper commands to the GPU Firmware to trigger a write of data outside the Guest's virtualised GPU memory. Out of bounds accesses triggered by malware introduced to a Guest KMD could allow privilege escalation which escapes virtualization boundaries.

Software installed and run as a non-privileged user may conduct improper GPU system calls to manipulate the lifetimes of synchronisation objects in the kernel, leading to read/write UAFs. During workload submission involving a fence exported by the GPU driver, the reference count of the underlying synchronisation primitive is not properly incremented. This can be exploited, by destroying the exported fence and prematurely release the underlying primitive, resulting in a potential use-after-free condition.

Kernel software installed and running inside a Host VM may post improper commands to the GPU Firmware to trigger a memory write outside the permitted range of memory for the host kernel. A TOCTOU bug existed where a malicious driver could modify values in memory after firmware validation but before use.

Kernel software installed and running inside a Host VM may post improper commands to the GPU Firmware to trigger a GPU register access which can lead to privilege escalation.

Software installed and run as a non-privileged user may conduct a sequence of improper GPU system calls causing use after free, which helps in facilitating unprivileged memory access from a shader code. Triggering failure path in the MMU mapping logic by a malicious code could lead to incomplete cleanup of an internal driver state, allowing for future unauthorized access to the contents of the physical memory.

Software installed and run as a non-privileged user may cause OOB kernel memory reads or writes through GPU API calls. When indexing pages larger than 4kB in the page freeing logic of the sparse memory implementation, incorrect buffer indexing leads to OOB access.

Software installed and run as a non-privileged user may conduct improper GPU system calls to cause an integer overflow and map two GPU virtual addresses to the same physical address. One of these virutal mappings can be freed along with the physical page, allowing for a read/write UAF via the second mapping The second virtual mapping references a physical address that has been freed after the first virtual mapping has been freed. This allows the physical memory to be allocated (for example) by another process and read/written to.

Kernel software installed and running inside a Host VM may post improper commands to the GPU Firmware to trigger a memory read or write outside the permitted range of memory for the host kernel. Addresses passed to the GPU Firmware can be used by the Firmware for more privileged memory accesses than are permitted by the system.

A web page that contains unusual GPU shader code is loaded into the GPU compiler process and can trigger a write out-of-bounds write crash in the GPU shader compiler library. On certain platforms, when the compiler process has system privileges this could enable further exploits on the device. An edge case using a very small value in GPU shader code can cause a segmentation fault in the GPU shader compiler due to am out-of-bounds write.

Software installed and run as a non-privileged user may conduct improper GPU system calls to cause mismanagement of resources creating a write use after free scenario. A shared resource (memory page) managed by a CPU thread of control (driver) and accessed by a GPU thread of control (Firmware) can cause a write UAF when the CPU thread frees the resource before the GPU FW has finished accessing it.