This message is from the T13 list server.
> "Mcgrath, Jim" <[EMAIL PROTECTED]> 12/06/01 05:52PM > I don't know of any case since SCSI-2 > where you need to allow for an odd number of returned bytes > (except possibly unblocked data in tapes > (I'm not an expert there) - certainly no control information). > This change happened ... Ahem. You mean to say this change BEGAN then? Scsi-2 did not forbid such transfers - did it? I don't think it even bothered to deprecate odd byte transfers? I know I don't remember reading any such "should not", but it's been a long time. I've met a number of vehement committee folk who argue to delete "should not"s from any spec whenever we fail to convert them into "shall"s, on the grounds that "should not"s discourage implementation enough to create de facto "shall not"s over time. I've never understood why that's a Bad Thing. Options are evil, aye. I'd go further: as far as I know, anyone who designs both host and device can design both to comply with Scsi and yet still avoid transferring anything lengths other than a multiple of 8 bytes, not just lengths other than a multiple of 2 bytes. They may have to add an optional changeable mode page to round up the all-pages available mode select/sense lengths. > anyone who designs both host and device Even in isolation, device folk can at least arrange for no "available length" to be anything but a multiple of N bytes, so that they never encourage a host to move anything but a multiple of N. We can't do better than a multiple of 4 until/unless we persuade Windows to stop issuing x 12 0 0 0 24 0 i.e. Inquiry for x24 (36) bytes. I hear non-multiples of 4 have been seen broken on Ide Dma motherboards already. Should be fun to watch what happens when non-multiples of 8 break. > Device designers alone can at least arrange Well, designers of NEW devices. Designers of old devices have to ask questions like, if I round up my x73 bytes of Inquiry to make x80 bytes available, who will I find who allocated just x73 bytes on a stack because they thought they knew that was all I had. > x73 bytes of Inquiry If you know a better place than Inquiry to put the c*pyright notice, please tell me. > This change happened ... > when SCSI first introduced wide SCSI, > which meant that in wide SCSI mode > (a purely physical mode > invisible to the upper layers of software) > you can only transfer an even number of bytes. I understand this to mean you can only clock across an even number of bytes. > SCSI did allow for an odd byte indicator > (IGNORE WIDE RESIDUE message), > but I personally have never seen it being used Ditto. Did the low end market ever even learn to support wide Scsi? Pat LaVarre Subscribe/Unsubscribe instructions can be found at www.t13.org.
