When the controller is reconnecting, the host fails I/O and admin commands as the host cannot reach the controller. ns scanning may revalidate namespaces during that period and it is wrong to remove namespaces due to these failures as we may hang (see 205da2434301). One command that may fail is nvme_identify_ns_descs. Since we return success due to having ns identify descriptor list optional, we continue to compare ns identifiers in nvme_revalidate_disk, obviously fail and return -ENODEV to nvme_validate_ns, which will remove the namespace. Exactly what we don't want to happen.
Unified · Split
| @@ -1110,7 +1110,7 @@ static int nvme_identify_ns_descs(struct nvme_ctrl *ctrl, unsigned nsid, · 1109 unmodified lines | |||
| 1110 | * Don't treat an error as fatal, as we potentially already | 1110 | * Don't treat an error as fatal, as we potentially already |
| 1111 | * have a NGUID or EUI-64. | 1111 | * have a NGUID or EUI-64. |
| 1112 | */ | 1112 | */ |
| 1113 | if (status > 0) | 1113 | if (status > 0 && !(status & NVME_SC_DNR)) |
| 1114 | status = 0; | 1114 | status = 0; |
| 1115 | goto free_data; | 1115 | goto free_data; |
| 1116 | } | 1116 | } |