CVE-2026-68249 drm/amdgpu/sdma5.0: replace BUG_ON() with WARN_ON()

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/sdma5.0: replace BUGON with WARNON There's no need to crash the kernel for these cases. cherry picked from commit 8d144a0eb09537055841af48c9e7c2d4cd48e84d...

EUVD-2026-55347

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/gfx11: replace BUGON with WARNON There's no need to crash the kernel for these cases. cherry picked from commit daa62107452d2451787c4248ca38fa2d1a0cbefd...

CVE-2026-68246 drm/amdgpu/gfx11: replace BUG_ON() with WARN_ON()

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/gfx11: replace BUGON with WARNON There's no need to crash the kernel for these cases. cherry picked from commit daa62107452d2451787c4248ca38fa2d1a0cbefd...

EUVD-2026-55346

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: fix lifetime issue of amdgpuvmgettaskinfopasid The vm pointer returned from amdgpuvmgetvmfrompasid is only valid while the lock is still being held. Once xaunlockirqrestore is called and returned, the pointer is no...

CVE-2026-68245 drm/amdgpu: fix lifetime issue of amdgpu_vm_get_task_info_pasid()

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: fix lifetime issue of amdgpuvmgettaskinfopasid The vm pointer returned from amdgpuvmgetvmfrompasid is only valid while the lock is still being held. Once xaunlockirqrestore is called and returned, the pointer is no...

CVE-2026-68245

CVE-2026-68245 affects the Linux kernel’s DRM/amdgpu component. The issue is a use-after-unlock: a vm pointer retrieved via amdgpu_vm_get_vm_from_pasid() is used after xa_unlock_irqrestore releases the lock, leaving the pointer vulnerable to modification. The documented fix removes the dependency...

CVE-2026-68245 drm/amdgpu: fix lifetime issue of amdgpu_vm_get_task_info_pasid()

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: fix lifetime issue of amdgpuvmgettaskinfopasid The vm pointer returned from amdgpuvmgetvmfrompasid is only valid while the lock is still being held. Once xaunlockirqrestore is called and returned, the pointer is no...

CVE-2026-68238

In the Linux kernel, the drm/amdgpu path released the VFCT ACPI table reference via acpi_get_table() without a matching acpi_put_table(), causing a leak on each call. The fix routes all exit paths through a common acpi_put_table() after copying the VBIOS image with kmemdup(), ensuring the VFCT ta...

CVE-2026-68238 drm/amdgpu: Release VFCT ACPI table reference

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: Release VFCT ACPI table reference amdgpuacpivfctbios fetches the VFCT table with acpigettable but never releases it. acpigettable takes a reference on the table incrementing its validationcount and mapping it on the 0...

CVE-2026-68238 drm/amdgpu: Release VFCT ACPI table reference

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: Release VFCT ACPI table reference amdgpuacpivfctbios fetches the VFCT table with acpigettable but never releases it. acpigettable takes a reference on the table incrementing its validationcount and mapping it on the 0...

CVE-2026-68237 drm/amdgpu/userq: fix indefinite fence wait during GPU reset

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/userq: fix indefinite fence wait during GPU reset prereset only force-completes fences of MAPPED queues. A queue in any other state e.g. mid-eviction keeps its lastfence pending; after a GPU reset that fence never...

CVE-2026-68237

CVE-2026-68237 resolves an indefinite fence wait in drm/amdgpu/userq during a GPU reset. The pre_reset path only force-completes fences of MAPPED queues, leaving fences for queues in states like mid-eviction pending; after a reset, those fences may never signal, causing eviction/suspend workers a...

CVE-2026-68237 drm/amdgpu/userq: fix indefinite fence wait during GPU reset

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/userq: fix indefinite fence wait during GPU reset prereset only force-completes fences of MAPPED queues. A queue in any other state e.g. mid-eviction keeps its lastfence pending; after a GPU reset that fence never...

CVE-2026-68234 drm/amdgpu: fix bo->pin leaking in amdgpu_bo_create_reserved

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: fix bo-pin leaking in amdgpubocreatereserved amdgpubocreatereserved only allocates a new BO when boptr struct amdgpubo boptr as input parameter is NULL, it simply skips creation when boptr is non-NULL. But it...

EUVD-2026-55335

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: fix bo-pin leaking in amdgpubocreatereserved amdgpubocreatereserved only allocates a new BO when boptr struct amdgpubo boptr as input parameter is NULL, it simply skips creation when boptr is non-NULL. But it...

CVE-2026-68234

The CVE describes a bug in drm/amdgpu where amdgpu_bo_create_reserved pins and reserves a BO even when a non-NULL pointer is passed, leading to pin_count increments on repeated uses and eventual BO leaks because teardown unpins only once. The fix makes pinning occur only at creation, so repeated ...

CVE-2026-68234 drm/amdgpu: fix bo->pin leaking in amdgpu_bo_create_reserved

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: fix bo-pin leaking in amdgpubocreatereserved amdgpubocreatereserved only allocates a new BO when boptr struct amdgpubo boptr as input parameter is NULL, it simply skips creation when boptr is non-NULL. But it...

CVE-2026-68115 drm/amdgpu/gfx10: replace BUG_ON() with WARN_ON()

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/gfx10: replace BUGON with WARNON There's no need to crash the kernel for these cases. cherry picked from commit ac6f00beb658239bced4aaed9efbb04a35348d48...

CVE-2026-68115

The CVE-2026-68115 entry relates to the Linux kernel DRM/AMDGPU gfx10 driver. The issue was that BUG_ON() caused a kernel crash and has been replaced with WARN_ON(), so the code no longer forces a crash in these cases. This is a code fix (cherry-picked from commit ac6f00beb658239bced4aaed9efbb04a...

EUVD-2026-55496

In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/gfx10: replace BUGON with WARNON There's no need to crash the kernel for these cases. cherry picked from commit ac6f00beb658239bced4aaed9efbb04a35348d48...