RFC: Add I/O class enum values to <scsi/srp.h>

Does anyone have an objection to me merging the trivial patch below through my git tree? This will be used by the IB SRP initiator to work with SilverStorm targets, which still implement rev. 10 of the SRP spec. I could just make these values private to the IB initiator, but I figured that things directly from the SRP spec belong in <scsi/srp.h> rather than in a particular driver's private header.

Thanks,
Roland

diff-tree a13ac0e9f99636a043d197f3349a67303ce4a701 (from bb61dd1fbf59f2291295986bed1f99b48f513fa4)

Author: Ramachandra K <[email protected]>
Date:   Mon Jun 5 12:13:52 2006 -0700

[SCSI] srp.h: Add I/O Class values

Add enum values for I/O Class values from rev. 10 and rev. 16a SRP drafts. The values are used to detect targets that implement obsolete revisions of SRP, so that the initiator can use the old format for port identifier when connecting to them.

index 637f77e..ad178fa 100644
@@ -87,6 +87,11 @@ enum srp_login_rej_reason {
 	SRP_LOGIN_REJ_CHANNEL_LIMIT_REACHED		= 0x00010006
 };
 
+enum {
+	SRP_REV10_IB_IO_CLASS	= 0xff00,
+	SRP_REV16A_IB_IO_CLASS	= 0x0100
+};
+
 struct srp_direct_buf {
 	__be64	va;
 	__be32	key;

No objection here ... but if you do, it will entangle our git trees even more nastily, since the srp.h file is created in the scsi-misc-2.6 tree.

James

James> No objection here ... but if you do, it will entangle our James> git trees even more nastily, since the srp.h file is James> created in the scsi-misc-2.6 tree.

No, I think we're OK. srp.h is already in Linus's tree (it went in as part of the original IB SRP initiator merge), and scsi-misc doesn't have any changes after ec448a0a36 (which is already upstream) in it. So putting the IO Class change in my tree actually reduces the dependency between our trees, since I can put the IB SRP changes in my tree without worrying about you merging the srp.h change first.

- R.