CVE-2023-53628

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdgpu: drop gfx_v11_0_cp_ecc_error_irq_funcs<br /> <br /> The gfx.cp_ecc_error_irq is retired in gfx11. In gfx_v11_0_hw_fini still<br /> use amdgpu_irq_put to disable this interrupt, which caused the call trace<br /> in this function.<br /> <br /> [ 102.873958] Call Trace:<br /> [ 102.873959] <br /> [ 102.873961] gfx_v11_0_hw_fini+0x23/0x1e0 [amdgpu]<br /> [ 102.874019] gfx_v11_0_suspend+0xe/0x20 [amdgpu]<br /> [ 102.874072] amdgpu_device_ip_suspend_phase2+0x240/0x460 [amdgpu]<br /> [ 102.874122] amdgpu_device_ip_suspend+0x3d/0x80 [amdgpu]<br /> [ 102.874172] amdgpu_device_pre_asic_reset+0xd9/0x490 [amdgpu]<br /> [ 102.874223] amdgpu_device_gpu_recover.cold+0x548/0xce6 [amdgpu]<br /> [ 102.874321] amdgpu_debugfs_reset_work+0x4c/0x70 [amdgpu]<br /> [ 102.874375] process_one_work+0x21f/0x3f0<br /> [ 102.874377] worker_thread+0x200/0x3e0<br /> [ 102.874378] ? process_one_work+0x3f0/0x3f0<br /> [ 102.874379] kthread+0xfd/0x130<br /> [ 102.874380] ? kthread_complete_and_exit+0x20/0x20<br /> [ 102.874381] ret_from_fork+0x22/0x30<br /> <br /> v2:<br /> - Handle umc and gfx ras cases in separated patch<br /> - Retired the gfx_v11_0_cp_ecc_error_irq_funcs in gfx11<br /> <br /> v3:<br /> - Improve the subject and code comments<br /> - Add judgment on gfx11 in the function of amdgpu_gfx_ras_late_init<br /> <br /> v4:<br /> - Drop the define of CP_ME1_PIPE_INST_ADDR_INTERVAL and<br /> SET_ECC_ME_PIPE_STATE which using in gfx_v11_0_set_cp_ecc_error_state<br /> - Check cp_ecc_error_irq.funcs rather than ip version for a more<br /> sustainable life<br /> <br /> v5:<br /> - Simplify judgment conditions