On Sat, 2012-10-06 at 23:11 -0700, Nicholas A. Bellinger wrote: > On Sat, 2012-10-06 at 21:49 -0400, Christoph Hellwig wrote: > > Currenly all non-pscsi bakcneds report their standards version as > > SPC 2 via ->get_device_rev. > > No, the proper on-the-wire bits to signal SPC-3 compliance are already > being returned by virtual backend drivers within standard INQUIRY > payload data. > > Notice the comment: > > root@tifa:/usr/src/target-pending.git# grep SCSI_SPC_2 drivers/target/*.c > drivers/target/target_core_file.c: return SCSI_SPC_2; /* Returns SPC-3 in Initiator Data */ > drivers/target/target_core_iblock.c: return SCSI_SPC_2; /* Returns SPC-3 in Initiator Data */ > drivers/target/target_core_rd.c: return SCSI_SPC_2; /* Returns SPC-3 in Initiator Data */ That's causing confusion, I think! > > In addition to putting it into the > > inquirty data in spc_emulate_inquiry_std we also use it in two > > places to check on the version of the persistent reservation and > > alua support. What is the benefit of supporting the old-style SCSI 2 > > reservations and ALUA support when they can't be used at all with > > the virtual backends, and can only be used in the corner case emulated > > ALUA/PR support for pscsi? > > > > It's the include/scsi/scsi.h SCSI_3 + SCSI_SPC_* version definition > names that are incorrect: > > #define SCSI_UNKNOWN 0 > #define SCSI_1 1 > #define SCSI_1_CCS 2 > #define SCSI_2 3 > #define SCSI_3 4 /* SPC */ > #define SCSI_SPC_2 5 > #define SCSI_SPC_3 6 > > from spc4r30 section 6.4.2 Standard INQUIRY data, Table 142 -- Version: > > 00h The device server does not claim conformance to any standard. > 01h to 02h Obsolete > 03h The device server complies to ANSI INCITS 301-1997 (a withdrawn standard). > 04h The device server complies to ANSI INCITS 351-2001 (SPC-2). > 05h The device server complies to ANSI INCITS 408-2005 (SPC-3). > 06h The device server complies to this standard. > > How about the following patch to fix the long-standing incorrect name > usage of these three scsi.h defines..? Because it's not incorrect. Your confusion is that the SCSI_ constants should match the inquiry data ... they shouldn't. Look where we set scsi_level in scsi_scan.c:708-713. We have to fit an extra identifier for SCSI_1_CCS so the scsi_level doesn't actually match the inquiry data; it's incremented by 1 for SCSI_3 and above. SCSI_3 does exist, by the way, it was defined in the first version of SPC and there are some devices which conform to it. James -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
