Hi Howard, FYI, the error/warning still remains. tree: https://android.googlesource.com/kernel/common android12-5.10-lts head: a7c8ce84600703f8b5e77ca13b192b20068408e3 commit: 26afe6712de0fad3a5de2b372286d029f8ff3231 [5373/14863] FROMLIST: usb: host: export symbols for xhci hooks usage config: arm-randconfig-r036-20210928 (attached as .config) compiler: arm-linux-gnueabi-gcc (GCC) 11.2.0 reproduce (this is a W=1 build): wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross git remote add android-common https://android.googlesource.com/kernel/common git fetch --no-tags android-common android12-5.10-lts git checkout 26afe6712de0fad3a5de2b372286d029f8ff3231 # save the attached .config to linux build tree COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross ARCH=arm If you fix the issue, kindly add following tag as appropriate Reported-by: kernel test robot <lkp(a)intel.com&gt; All warnings (new ones prefixed by >>): 2900 | int xhci_handle_event(struct xhci_hcd *xhci) | ^~~~~~~~~~~~~~~~~ 2976 | void xhci_update_erst_dequeue(struct xhci_hcd *xhci, | ^~~~~~~~~~~~~~~~~~~~~~~~ vim +/xhci_handle_event +2900 drivers/usb/host/xhci-ring.c 2893 2894 /* 2895 * This function handles all OS-owned events on the event ring. It may drop 2896 * xhci->lock between event processing (e.g. to pass up port status changes). 2897 * Returns >0 for "possibly more events to process" (caller should call again), 2898 * otherwise 0 if done. In future, <0 returns should indicate error code. 2899 */ 2901 { 2902 union xhci_trb *event; 2903 int update_ptrs = 1; 2904 u32 trb_type; 2905 int ret; 2906 2907 /* Event ring hasn't been allocated yet. */ 2908 if (!xhci->event_ring || !xhci->event_ring->dequeue) { 2909 xhci_err(xhci, "ERROR event ring not ready\n"); 2910 return -ENOMEM; 2911 } 2912 2913 event = xhci->event_ring->dequeue; 2914 /* Does the HC or OS own the TRB? */ 2915 if ((le32_to_cpu(event->event_cmd.flags) & TRB_CYCLE) != 2916 xhci->event_ring->cycle_state) 2917 return 0; 2918 2919 trace_xhci_handle_event(xhci->event_ring, &event->generic); 2920 2921 /* 2922 * Barrier between reading the TRB_CYCLE (valid) flag above and any 2923 * speculative reads of the event's flags/data below. 2924 */ 2925 rmb(); 2926 trb_type = TRB_FIELD_TO_TYPE(le32_to_cpu(event->event_cmd.flags)); 2927 /* FIXME: Handle more event types. */ 2928 2929 switch (trb_type) { 2930 case TRB_COMPLETION: 2931 handle_cmd_completion(xhci, &event->event_cmd); 2932 break; 2933 case TRB_PORT_STATUS: 2934 handle_port_status(xhci, event); 2935 update_ptrs = 0; 2936 break; 2937 case TRB_TRANSFER: 2938 ret = handle_tx_event(xhci, &event->trans_event); 2939 if (ret >= 0) 2940 update_ptrs = 0; 2941 break; 2942 case TRB_DEV_NOTE: 2943 handle_device_notification(xhci, event); 2944 break; 2945 default: 2946 if (trb_type >= TRB_VENDOR_DEFINED_LOW) 2947 handle_vendor_event(xhci, event, trb_type); 2948 else 2949 xhci_warn(xhci, "ERROR unknown event type %d\n", trb_type); 2950 } 2951 /* Any of the above functions may drop and re-acquire the lock, so check 2952 * to make sure a watchdog timer didn't mark the host as non-responsive. 2953 */ 2954 if (xhci->xhc_state & XHCI_STATE_DYING) { 2955 xhci_dbg(xhci, "xHCI host dying, returning from " 2956 "event handler.\n"); 2957 return 0; 2958 } 2959 2960 if (update_ptrs) 2961 /* Update SW event ring dequeue pointer */ 2962 inc_deq(xhci, xhci->event_ring); 2963 2964 /* Are there more items on the event ring? Caller will call us again to 2965 * check. 2966 */ 2967 return 1; 2968 } 2969 EXPORT_SYMBOL_GPL(xhci_handle_event); 2970 2971 /* 2972 * Update Event Ring Dequeue Pointer: 2973 * - When all events have finished 2974 * - To avoid "Event Ring Full Error" condition 2975 */ 2977 union xhci_trb *event_ring_deq) 2978 { 2979 u64 temp_64; 2980 dma_addr_t deq; 2981 2982 temp_64 = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue); 2983 /* If necessary, update the HW's version of the event ring deq ptr. */ 2984 if (event_ring_deq != xhci->event_ring->dequeue) { 2985 deq = xhci_trb_virt_to_dma(xhci->event_ring->deq_seg, 2986 xhci->event_ring->dequeue); 2987 if (deq == 0) 2988 xhci_warn(xhci, "WARN something wrong with SW event ring dequeue ptr\n"); 2989 /* 2990 * Per 4.9.4, Software writes to the ERDP register shall 2991 * always advance the Event Ring Dequeue Pointer value. 2992 */ 2993 if ((temp_64 & (u64) ~ERST_PTR_MASK) == 2994 ((u64) deq & (u64) ~ERST_PTR_MASK)) 2995 return; 2996 2997 /* Update HC event ring dequeue pointer */ 2998 temp_64 &= ERST_PTR_MASK; 2999 temp_64 |= ((u64) deq & (u64) ~ERST_PTR_MASK); 3000 } 3001 3002 /* Clear the event handler busy flag (RW1C) */ 3003 temp_64 |= ERST_EHB; 3004 xhci_write_64(xhci, temp_64, &xhci->ir_set->erst_dequeue); 3005 } 3006 EXPORT_SYMBOL_GPL(xhci_update_erst_dequeue); 3007 --- 0-DAY CI Kernel Test Service, Intel Corporation https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org