Douglas, >> What exact regression causes this for real users and no just people >> playing around with scsi_debug? > > Sorry, you are regressing something that has been in place for over 20 > years and required by SPC (1 through 5) standards. The onus should > not be on me to prove that regression is not safe. It should be the > other way around (i.e. for you to prove that it is safe). Christoph's question is valid: Ignoring obsolete command sets that we no longer support, what is the real world use case for variable length CDBs larger than 32 bytes? Which devices require them, and do we want to support those devices? That fact that a SCSI spec permits something is not the same as saying we must support it. "Required by SPC" is news to me. -- Martin K. Petersen Oracle Linux Engineering
