>From your first two messages, I gather that it might not be safe to rely on all LUNs in a SCSI target having the same PDT. Fair enough. On Tue, 21 Feb 2006, Pat LaVarre wrote: > Working from comments, not yet grep, I see only three distinct Groups > of flags: > > a) Mysteries: > > >> 19 #define BLIST_NOSTARTONADD 0x1000 /* do not do automatic > >> start on add */ > >> 9 #define BLIST_BORKEN 0x004 /* Flag for broken > >> handshaking */ > >> 10 #define BLIST_KEY 0x008 /* unlock by special > >> command */ > >> 28 #define BLIST_NO_ULD_ATTACH 0x100000 /* device is actually > >> for RAID config */ > >> 29 #define BLIST_SELECT_NO_ATN 0x200000 /* select without ATN */ > > > The jargon in BLIST_SELECT_NO_ATN is specific to Spi transport of Scsi. > > b) Scsi Simplifications: > > >> 12 #define BLIST_NOTQ 0x020 /* Buggy Tagged Command > >> Queuing */ > >> 15 #define BLIST_ISROM 0x100 /* Treat as (removable) > >> CD-ROM */ > >> 17 #define BLIST_INQUIRY_36 0x400 /* override additional > >> length field */ > >> 18 #define BLIST_INQUIRY_58 0x800 /* ... for broken > >> inquiry responses */ > >> 20 #define BLIST_MS_SKIP_PAGE_08 0x2000 /* do not send ms page > >> 0x08 */ > >> 21 #define BLIST_MS_SKIP_PAGE_3F 0x4000 /* do not send ms page > >> 0x3f */ > >> 22 #define BLIST_USE_10_BYTE_MS 0x8000 /* use 10 byte ms > >> before 6 byte ms */ > >> 23 #define BLIST_MS_192_BYTES_FOR_3F 0x10000 /* 192 byte ms > >> page 0x3f request */ > >> 27 #define BLIST_NOT_LOCKABLE 0x80000 /* don't use PREVENT- > >> ALLOW commands */ > > NOTQ is for all requests All requests to the LUN or to the target? Are tagged requests for different LUNs on the same target handled separately? For example, could there be simultaneous outstanding requests having the same tag value because they are for separate LUNs? > ISROM and _INQUIRY redefine Scsi op x12 Inquiry > _MS redefine Scsi op x1A Mode Sense 6 and op x5A Mode Sense 10 > NOT_LOCKABLE redefine op x1E Prevent/ Allow Medium Removal. > > All these I imagine we might have to store per Lun. Agreed. > c) Lun Counting Techniques: > > >> 6 #define BLIST_NOLUN 0x001 /* Only scan LUN 0 */ > >> 7 #define BLIST_FORCELUN 0x002 /* Known to have LUNs, > >> force scanning, deprecated: Use max_luns=N */ > >> 11 #define BLIST_SINGLELUN 0x010 /* Do not use LUNs in > >> parallel */ > >> 13 #define BLIST_SPARSELUN 0x040 /* Non consecutive LUN > >> numbering */ > >> 14 #define BLIST_MAX5LUN 0x080 /* Avoid LUNS >= 5 */ > >> 16 #define BLIST_LARGELUN 0x200 /* LUNs past 7 on a > >> SCSI-2 device */ > >> 24 #define BLIST_REPORTLUN2 0x20000 /* try REPORT_LUNS even > >> for SCSI-2 devs (if HBA supports more than 8 LUNs) */ > >> 26 #define BLIST_NOREPORTLUN 0x40000 /* don't try > >> REPORT_LUNS scan (SCSI-3 devs) */ > > All those I imagine we can store per device, not per Lun, by your > original logic: "any flag that influences how LUN detection should be > performed" "should apply to entire targets". That seems pretty clear. Thanks for the help. Alan Stern - : 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
