On 12 Dec 2017, at 9:38, Christoph Hellwig wrote: > On Fri, Dec 08, 2017 at 12:52:58PM -0500, Benjamin Coddington wrote: >> If a SCSI device already has an exclusive access registrants only >> reservation, the device's size cannot be known by the client until after >> registration. We should therefore delay setting the block device's length >> until after the registration and revalidation of the disk. > > Table 13 in sbc4r14.pdf clearly indicates READ CAPACITY is allowed even > for a not registered I_T nexus. What kind of setup do you see this > problem with? It does indeed, I should have checked. So it seems this patch should be unnecessary, however I think if we do end up with d->len = 0, bad things happen later. My memory is fuzzy on this, I'll have to re-test it. The SCSI devices are presented by the TCM driver on a 4.11 era kernel. I don't know that code at all, but with quick look it seems that __target_execute_cmd() wants to check for reservations for all commands. I think this patch should be dropped, and we can try to fix this on the TCM side. Ben -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
