CVE-2025-37754

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/i915/huc: Fix fence not released on early probe errors<br /> <br /> HuC delayed loading fence, introduced with commit 27536e03271da<br /> ("drm/i915/huc: track delayed HuC load with a fence"), is registered with<br /> object tracker early on driver probe but unregistered only from driver<br /> remove, which is not called on early probe errors. Since its memory is<br /> allocated under devres, then released anyway, it may happen to be<br /> allocated again to the fence and reused on future driver probes, resulting<br /> in kernel warnings that taint the kernel:<br /> <br /> [309.731371] ------------[ cut here ]------------<br /> [309.731373] ODEBUG: init destroyed (active state 0) object: ffff88813d7dd2e0 object type: i915_sw_fence hint: sw_fence_dummy_notify+0x0/0x20 [i915]<br /> [309.731575] WARNING: CPU: 2 PID: 3161 at lib/debugobjects.c:612 debug_print_object+0x93/0xf0<br /> ...<br /> [309.731693] CPU: 2 UID: 0 PID: 3161 Comm: i915_module_loa Tainted: G U 6.14.0-CI_DRM_16362-gf0fd77956987+ #1<br /> ...<br /> [309.731700] RIP: 0010:debug_print_object+0x93/0xf0<br /> ...<br /> [309.731728] Call Trace:<br /> [309.731730] <br /> ...<br /> [309.731949] __debug_object_init+0x17b/0x1c0<br /> [309.731957] debug_object_init+0x34/0x50<br /> [309.732126] __i915_sw_fence_init+0x34/0x60 [i915]<br /> [309.732256] intel_huc_init_early+0x4b/0x1d0 [i915]<br /> [309.732468] intel_uc_init_early+0x61/0x680 [i915]<br /> [309.732667] intel_gt_common_init_early+0x105/0x130 [i915]<br /> [309.732804] intel_root_gt_init_early+0x63/0x80 [i915]<br /> [309.732938] i915_driver_probe+0x1fa/0xeb0 [i915]<br /> [309.733075] i915_pci_probe+0xe6/0x220 [i915]<br /> [309.733198] local_pci_probe+0x44/0xb0<br /> [309.733203] pci_device_probe+0xf4/0x270<br /> [309.733209] really_probe+0xee/0x3c0<br /> [309.733215] __driver_probe_device+0x8c/0x180<br /> [309.733219] driver_probe_device+0x24/0xd0<br /> [309.733223] __driver_attach+0x10f/0x220<br /> [309.733230] bus_for_each_dev+0x7d/0xe0<br /> [309.733236] driver_attach+0x1e/0x30<br /> [309.733239] bus_add_driver+0x151/0x290<br /> [309.733244] driver_register+0x5e/0x130<br /> [309.733247] __pci_register_driver+0x7d/0x90<br /> [309.733251] i915_pci_register_driver+0x23/0x30 [i915]<br /> [309.733413] i915_init+0x34/0x120 [i915]<br /> [309.733655] do_one_initcall+0x62/0x3f0<br /> [309.733667] do_init_module+0x97/0x2a0<br /> [309.733671] load_module+0x25ff/0x2890<br /> [309.733688] init_module_from_file+0x97/0xe0<br /> [309.733701] idempotent_init_module+0x118/0x330<br /> [309.733711] __x64_sys_finit_module+0x77/0x100<br /> [309.733715] x64_sys_call+0x1f37/0x2650<br /> [309.733719] do_syscall_64+0x91/0x180<br /> [309.733763] entry_SYSCALL_64_after_hwframe+0x76/0x7e<br /> [309.733792] <br /> ...<br /> [309.733806] ---[ end trace 0000000000000000 ]---<br /> <br /> That scenario is most easily reproducible with<br /> igt@i915_module_load@reload-with-fault-injection.<br /> <br /> Fix the issue by moving the cleanup step to driver release path.<br /> <br /> (cherry picked from commit 795dbde92fe5c6996a02a5b579481de73035e7bf)