On Tue, Nov 10, 2020 at 05:41:15PM -0800, Sagi Grimberg wrote: > >>> - if (id->nsattr & NVME_NS_ATTR_RO) >>> + if (id->nsattr & NVME_NS_ATTR_RO || test_bit(NVME_NS_FORCE_RO, &ns->flags)) >>> set_disk_ro(disk, true); >> >> If the FORCE_RO flag is set, the disk is set to read-only. If that flag >> is later cleared, nothing clears the disk's read-only setting. > > Yea, that is true also for the non-force case, but before it broke > BLKROSET so I reverted that. We can use this FORCE_RO for BLKROSET as > well I think... Let me prioritize the hard r/o setting. mkp actually has an older patch that did just that which I'll resurrect.
