tree: https://gitee.com/openeuler/kernel.git openEuler-1.0-LTS head: ba07135cf74b1424e2f7f0c60e59c5b206ab320c commit: 699b7dcfd1ed2f3b930ad3c4712ac9d685591e4a [1743/1743] nvme-pci: shutdown on timeout during deletion config: x86_64-buildonly-randconfig-2004-20250802 (https://download.01.org/0day-ci/archive/20250804/202508042248.qt1mRSJK-lkp@i...) compiler: gcc-11 (Debian 11.3.0-12) 11.3.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250804/202508042248.qt1mRSJK-lkp@i...) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp@intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202508042248.qt1mRSJK-lkp@intel.com/ All warnings (new ones prefixed by >>): In file included from drivers/nvme/host/pci.c:25: include/linux/module.h:140:14: warning: 'cleanup_module' specifies less restrictive attribute than its target 'nvme_exit': 'cold' [-Wmissing-attributes] 140 | void cleanup_module(void) __attribute__((alias(#exitfn))); | ^~~~~~~~~~~~~~ drivers/nvme/host/pci.c:2768:1: note: in expansion of macro 'module_exit' 2768 | module_exit(nvme_exit); | ^~~~~~~~~~~ drivers/nvme/host/pci.c:2757:20: note: 'cleanup_module' target declared here 2757 | static void __exit nvme_exit(void) | ^~~~~~~~~ In file included from drivers/nvme/host/pci.c:25: include/linux/module.h:134:13: warning: 'init_module' specifies less restrictive attribute than its target 'nvme_init': 'cold' [-Wmissing-attributes] 134 | int init_module(void) __attribute__((alias(#initfn))); | ^~~~~~~~~~~ drivers/nvme/host/pci.c:2767:1: note: in expansion of macro 'module_init' 2767 | module_init(nvme_init); | ^~~~~~~~~~~ drivers/nvme/host/pci.c:2752:19: note: 'init_module' target declared here 2752 | static int __init nvme_init(void) | ^~~~~~~~~ drivers/nvme/host/pci.c: In function 'nvme_timeout':

...

drivers/nvme/host/pci.c:1174:26: warning: this statement may fall through [-Wimplicit-fallthrough=] 1174 | shutdown = true; | ~~~~~~~~~^~~~~~ drivers/nvme/host/pci.c:1175:9: note: here 1175 | case NVME_CTRL_CONNECTING: | ^~~~ drivers/nvme/host/pci.c:1261: warning: Function parameter or member 'nvmeq' not described in 'nvme_suspend_queue'

vim +1174 drivers/nvme/host/pci.c 1128 1129 static enum blk_eh_timer_return nvme_timeout(struct request *req, bool reserved) 1130 { 1131 struct nvme_iod *iod = blk_mq_rq_to_pdu(req); 1132 struct nvme_queue *nvmeq = iod->nvmeq; 1133 struct nvme_dev *dev = nvmeq->dev; 1134 struct request *abort_req; 1135 struct nvme_command cmd; 1136 bool shutdown = false; 1137 u32 csts = readl(dev->bar + NVME_REG_CSTS); 1138 1139 /* If PCI error recovery process is happening, we cannot reset or 1140 * the recovery mechanism will surely fail. 1141 */ 1142 mb(); 1143 if (pci_channel_offline(to_pci_dev(dev->dev))) 1144 return BLK_EH_RESET_TIMER; 1145 1146 /* 1147 * Reset immediately if the controller is failed 1148 */ 1149 if (nvme_should_reset(dev, csts)) { 1150 nvme_warn_reset(dev, csts); 1151 nvme_dev_disable(dev, false); 1152 nvme_reset_ctrl(&dev->ctrl); 1153 return BLK_EH_DONE; 1154 } 1155 1156 /* 1157 * Did we miss an interrupt? 1158 */ 1159 if (__nvme_poll(nvmeq, req->tag)) { 1160 dev_warn(dev->ctrl.device, 1161 "I/O %d QID %d timeout, completion polled\n", 1162 req->tag, nvmeq->qid); 1163 return BLK_EH_DONE; 1164 } 1165 1166 /* 1167 * Shutdown immediately if controller times out while starting. The 1168 * reset work will see the pci device disabled when it gets the forced 1169 * cancellation error. All outstanding requests are completed on 1170 * shutdown, so we return BLK_EH_DONE. 1171 */ 1172 switch (dev->ctrl.state) { 1173 case NVME_CTRL_DELETING:

1174 shutdown = true; 1175 case NVME_CTRL_CONNECTING: 1176 case NVME_CTRL_RESETTING: 1177 dev_warn_ratelimited(dev->ctrl.device, 1178 "I/O %d QID %d timeout, disable controller\n", 1179 req->tag, nvmeq->qid); 1180 nvme_dev_disable(dev, shutdown); 1181 nvme_req(req)->flags |= NVME_REQ_CANCELLED; 1182 return BLK_EH_DONE; 1183 default: 1184 break; 1185 } 1186 1187 /* 1188 * Shutdown the controller immediately and schedule a reset if the 1189 * command was already aborted once before and still hasn't been 1190 * returned to the driver, or if this is the admin queue. 1191 */ 1192 if (!nvmeq->qid || iod->aborted) { 1193 dev_warn(dev->ctrl.device, 1194 "I/O %d QID %d timeout, reset controller\n", 1195 req->tag, nvmeq->qid); 1196 nvme_dev_disable(dev, false); 1197 nvme_reset_ctrl(&dev->ctrl); 1198 1199 nvme_req(req)->flags |= NVME_REQ_CANCELLED; 1200 return BLK_EH_DONE; 1201 } 1202 1203 if (atomic_dec_return(&dev->ctrl.abort_limit) < 0) { 1204 atomic_inc(&dev->ctrl.abort_limit); 1205 return BLK_EH_RESET_TIMER; 1206 } 1207 iod->aborted = 1; 1208 1209 memset(&cmd, 0, sizeof(cmd)); 1210 cmd.abort.opcode = nvme_admin_abort_cmd; 1211 cmd.abort.cid = req->tag; 1212 cmd.abort.sqid = cpu_to_le16(nvmeq->qid); 1213 1214 dev_warn(nvmeq->dev->ctrl.device, 1215 "I/O %d QID %d timeout, aborting\n", 1216 req->tag, nvmeq->qid); 1217 1218 abort_req = nvme_alloc_request(dev->ctrl.admin_q, &cmd, 1219 BLK_MQ_REQ_NOWAIT, NVME_QID_ANY); 1220 if (IS_ERR(abort_req)) { 1221 atomic_inc(&dev->ctrl.abort_limit); 1222 return BLK_EH_RESET_TIMER; 1223 } 1224 1225 abort_req->timeout = ADMIN_TIMEOUT; 1226 abort_req->end_io_data = NULL; 1227 blk_execute_rq_nowait(abort_req->q, NULL, abort_req, 0, abort_endio); 1228 1229 /* 1230 * The aborted req will be completed on receiving the abort req. 1231 * We enable the timer again. If hit twice, it'll cause a device reset, 1232 * as the device then is in a faulty state. 1233 */ 1234 return BLK_EH_RESET_TIMER; 1235 } 1236

-- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki