On Sun, 2016-03-06 at 15:18 +0100, Christoph Hellwig wrote: > On Sat, Mar 05, 2016 at 02:58:40PM -0800, Nicholas A. Bellinger wrote: > > For v4.5, we need to identify the ib_srpt active I/O session shutdown > > regression vs. chain RDMA READ/WRITE requests, > > As far as I can tell they have not been identified as a regression - > Doug just added new testing. And a relation to chained RDMA > READ/WRITE seems unlikely from all I know about the workings of the > driver. srpt_destroy_ch_ib() originally used kthread_stop(ch->thread) to wait until ib_qp->comp_handler() callbacks finished before ib_destroy_qp(), and before subsequent srpt_release_channel_work() to free srpt_rdma_ch associated resources. Is ib_destroy_qp() + ib_free_cq() in srpt_destroy_ch_ib() supposed to be waiting for outstanding ib_wr descriptors before release of *ch now..? -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
