CVE-2026-99450
vfio/nvgrace-gpu: Use a core-only open_device in fallback mode
What happened?
In the Linux kernel, the following vulnerability has been resolved:
vfio/nvgrace-gpu: Use a core-only open_device in fallback mode
When the ACPI memory properties are missing or the C2C link is down, the device is exposed through nvgrace_gpu_pci_core_ops. That table opens through nvgrace_gpu_open_device() but closes through vfio_pci_core_close_device(), so open and close are not symmetric.
nvgrace_gpu_open_device() registers nvdev->usemem for poison handling unconditionally. In fallback mode usemem.memlength is zero, and register_pfn_address_space() inserts a node covering [0, ULONG_MAX]. Nothing removes it: a second open fails with -EBUSY, and once the device is unbound the node outlives the nvdev that contains it. The next open after a rebind reports:
BUG: KASAN: slab-use-after-free in interval_tree_iter_first+0x209/0x310 Read of size 8 at addr ffff88800894aa30 by task nvgrace_repro/203 Call Trace: <TASK> dump_stack_lvl+0x5d/0x80 print_report+0x153/0x4c6 kasan_report+0xe4/0x1a0 interval_tree_iter_first+0x209/0x310 register_pfn_address_space+0x97/0x100 nvgrace_gpu_open_device+0x21c/0x730 vfio_df_open+0x213/0x520 vfio_df_ioctl_bind_iommufd+0x3e8/0x6e0 vfio_device_fops_unl_ioctl+0x782/0x1570 __x64_sys_ioctl+0x13c/0x1c0 do_syscall_64+0x106/0x630 entry_SYSCALL_64_after_hwframe+0x77/0x7f </TASK>
Allocated by task 188: ... _vfio_alloc_device+0x3a/0x500 nvgrace_gpu_probe+0xcb/0x660 local_pci_probe+0xcf/0x170 ...
Freed by task 188: ... kfree+0x20d/0x600 device_release+0xcf/0x250 kobject_put+0x152/0x2f0 pci_device_remove+0xb0/0x1c0 device_release_driver_internal+0x38d/0x540 unbind_store+0xde/0x100 ...
Give the fallback table its own open_device that only calls vfio_pci_core_enable() and vfio_pci_core_finish_enable(), the counterpart of vfio_pci_core_close_device(). Fallback mode then never touches the nvgrace memory regions, as Alex suggested.
This should be safe because the fallback table uses the vfio-pci core read, write and mmap handlers, so nothing there uses the state nvgrace_gpu_open_device() sets up. The one visible difference is that a fallback open no longer fails when BAR0 cannot be mapped, which matches vfio-pci.
Affected software and versions
| Software | Version information | Fixed version |
|---|---|---|
| LinuxLinux | affected: e5f19b619fa0b691ccb537d72240bd20eb72087c to < a2942b3abb5bddd1bac418abd61fb5f0616ab897; affected: e5f19b619fa0b691ccb537d72240bd20eb72087c to < 9c48b994b66d7b585698f9aaee8ff559a7755c0a | See vendor advisory |
| LinuxLinux | affected: 7.0; unaffected: 0 to < 7.0; unaffected: 7.3.2 to ≤ 7.3.*; unaffected: 7.4-rc1 to ≤ * | See vendor advisory |
Version information comes from the source record and can include unaffected products and exceptions. Distribution packages may contain backported fixes: an older upstream version alone does not prove a server is affected.
Is it being exploited?
This CVE was not listed in the CISA catalogue at our last retrieval. This does not establish that exploitation is absent.
How to respond
- Confirm the product, installed package version, and any configuration requirements in the vendor advisory.
- Follow the vendor’s remediation or workaround guidance in the references below. Structured fixed-version data is unavailable here; the publisher’s description may include a fix or workaround.
- Restart affected services if required, then verify the running version. Known exploitation may warrant investigation as well as patching.