I made a bad assumption. Turns out if the target is on 4.7.0, I can't use SRP under an circumstances - even with the initiator booted from a hard drive. So, the 4.7.0 issue is separate than the 4.6.4 10-30 second delay from linux trying to make its own SRP connection while the iPXE SRP connection is still up. After I upgraded my target to 4.7.0, the first thing I tested was SRP booting, and when I saw it fail, I didn't ever test if my hard drive installed initiator could still use SRP. I just reverted target back to 4.6.4. Doing a git bisect now. (Stumbled into my bad assumption using SRP volumes for the kernel bisect, while target happened to be 4.7.0!) Have it narrowed down between commits (good) 9903fd1 - Thu Jun 23 12:22:33 2016 -0400 - Doug Ledford : Merge branches '4.7-rc-misc', 'hfi1-fixes', 'i40iw-rc-fixes' and 'mellanox-rc-fixes' into k.o/for-4.7-rc (bad) 30f56e3 - Tue Jul 19 16:44:11 2016 -0700 - Eugenia Emantayev : net/mlx4_en: Move filters cleanup to a proper location Reverting 30f56e3 doesnt' fix it. There's no comits between there to drivers/infiniband/{core/sq_query.c, hw/mlx4/, ulp/srp, ulp/srpt} or drivers/net/mellanox/mlx4, so it will be interesting to see what's causing it. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html