On Mon, 2016-02-29 at 07:05 +0100, Christoph Hellwig wrote: > On Sun, Feb 28, 2016 at 04:37:40PM -0800, Bart Van Assche wrote: > > On 02/28/16 12:43, Nicholas A. Bellinger wrote: > > > Anyways, I'll sit down this week and figure out what's going on with > > > Doug's active I/O shutdown regression. > > > > The crash occurs in the core_tpg_del_initiator_node_acl() function > > and a call to that function has been added recently in > > target_fabric_nacl_base_release(). I think it was added through the > > following patch: > > That patch just moved the call from the .fabric_drop_nodeacl instances > (in the SRPT case srpt_drop_nodeacl) to the caller in > target_fabric_nacl_base_release. I've not re-produced with v4.5-rc, but IIRC pre-commit 59fae4de usage of ib_create_cq() w/ srpt_compl_thread() -> kthread_stop(ch->thread) in srpt_destroy_ch_ib() did play a role wrt IB CQ active I/O shutdown completion in original code. Btw, is original ib_create_cq() usage a incompatible with chain RDMA READ/WRITE requests, or was that an extra improvement..? -- 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
