Live › vulnerability › CVE-2026-98520
Go to dashboardThe amdgpu KFD driver cached the address space of the first /dev/kfd opener, which becomes a dangling pointer when that container exits, and later openers' mappings were not zapped on GPU reset. The result is a kernel hang on GPU reset under SRIOV with multiple containers, and stale doorbell and MMIO mappings left reachable by guest userspace.
Summary written from the sources below. Check the originals before acting.
First seen by InfraStack 2026-10-23 21:57 UTC. All times on this page are UTC.
Systems running the Linux kernel amdgpu driver with KFD (ROCm compute) under SRIOV virtualization with multiple containers or guests sharing the GPU. Systems without AMD GPUs using SRIOV, or not using KFD, are unlikely to be affected; the sources do not say more.
No product ranges from the sources yet.
Each row is the first time that source carried this signal. The analyst’s rows are its own findings.
In the Linux kernel, the following vulnerability has been resolved:
drm/amdkfd: fix use-after-free and multi-container gap in kfd_dev_mapping
kfd_dev_mapping caches the address_space of the first /dev/kfd opener so that the GPU reset path can call unmap_mapping_range() to zap all userspace mappings of doorbell and MMIO ranges. This design has two bugs that both manifest under SRIOV with multiple containers:
Use-after-free / rwsem deadlock. The cached pointer refers to an inode owned by the first opener's container. When that container exits and its inode is released, kfd_dev_mapping becomes a dangling pointer. A subsequent GPU reset dereferences it inside unmap_mapping_range(), which takes i_mmap_rwsem on the freed inode, causing a hard hang observable as an uninterruptible rwsem wait.
Multi-container gap. Only the first opener's address_space is cached; VMAs created by later openers live in a different address_space and are never reached by unmap_mapping_range(). After a GPU reset those stale mappings keep doorbell and MMIO pages accessible to guest userspace with no GPU behind them, risking PCIe transaction timeouts and NMI panics.
Fix both bugs with the same approach used by DRM core (drm_drv.c): create a private pseudo-filesystem at module init time and allocate one anonymous inode from it. In kfd_open() redirect every opener's file->f_mapping to that inode's address_space. The inode is module-owned, lives exactly as long as the amdgpu module, and collects VMAs from all openers in one address_space. A single unmap_mapping_range() call in the reset path then correctly reaches every container's mappings with no dangling pointer risk.
The hang manifests as an NMI backtrace on the GPU reset workqueue stuck spinning in rwsem_down_read_slowpath() with a corrupted i_mmap_rwsem:
Workqueue: amdgpu-reset-dev xgpu_ai_mailbox_flr_work [amdgpu] Call Trace: <TASK> kvm_wait+0x1f/0x40 __pv_queued_spin_lock_slowpath+0x31d/0x3a0 _raw_spin_lock_irq+0x51/0x80 rwsem_down_read_slowpath+0xb3/0x550 down_read+0x48/0xd0 unmap_mapping_range+0x71/0x140 kfd_dev_unmap_mapping_range+0x5b/0x140 [amdgpu] amdgpu_amdkfd_clear_kfd_mapping+0xd8/0x190 [amdgpu] amdgpu_device_gpu_recover+0x232/0x450 [amdgpu] xgpu_ai_mailbox_flr_work+0xb5/0xc0 [amdgpu] process_one_work+0x18e/0x3e0 worker_thread+0x2e3/0x420 kthread+0x10a/0x230
(cherry picked from commit 1128b4a52de1572e87431de837fd9850cb99542c)
This page as .md · .json · in your agent: get_signal("CVE-2026-98520")
NVD data is provided by NIST and is public domain. Posts quoted belong to their authors.