cctech Digest, Vol 2, Issue 12

[Craig Ruff]

On Dec 12, 2014, at 11:00 AM, Chuck Guzis wrote;
> Subject: SCSI tape question
> Now, when I come to a filenark, I'm expecting that the filemark bit (bit 
> 7)  will be set in the sense key.  It doesn't happen, the result just 
> comes back as 0 bytes read and a normal (NO_SENSE) sense result.  The 
> additional sense code and sense qualifier bytes are also 0.

I wrote the SCSI tape drivers for the Unix based version of the NCAR Mass Storage System (MSS) Storage Manager servers, we moved a few petabytes of data through them. It made use of the SCSI generic interfaces of Irix, Solaris and Linux, but we couldn't trust the default OS tape drive driver behaviors would do the right thing for an archive.

You should be able to detect file mark indications in either the variable length or fixed block read modes.  When the tape (or drive controller buffer position for those that perform buffering) is positioned at the file mark, you will get a check condition response to the issuance of the read command, then you (or the SCSI generic driver) issue the read sense command (immediately after the read, no other intervening commands!) and the file mark bit will be set in the sense buffer.  For the generic SCSI devices, you can typically tell the driver to perform an auto sense for check condition responses. Just make sure you have properly told the driver where to put the sense data. I suppose a drive could have a manufacturer specific mode that could hide file marks, but I've never seen one.  We did make use of manufacturer specific procedures to place the read head beyond the detected EOT mark so we could recover tapes that had been accidentally overwritten in the middle of recorded data.