The two main improvements in this series are: 1. Break RPC-over-RDMA connections after an RPC timeout. This gives the client's CM an opportunity to perform server and network path rediscovery before retrying the timed-out RPC. This design was selected because it is simple, and does not make any changes to the normal RPC Call send hot path. Also, note the logic already in xprt_rdma_send_request() that breaks the connection anyway, just before sending a retransmit. Internal testing at Oracle has demonstrated this logic is effective. 2. Support unloading the driver of the underlying device. Full support for the DEVICE_REMOVAL CM upcall is implemented in the client-side RPC-over-RDMA consumer. Devesh's workaround is reverted, since it is now no longer necessary. In addition, support is added for restoring transport operation when a new driver is subsequently loaded or when another device is already available with connectivity to the NFS server. Hopefully this work can be the basis for device hotplug, suspend/resume with NFS/RDMA mounts, and handling device failover. Available in the "nfs-rdma-for-4.12" topic branch of this git repo: git://git.linux-nfs.org/projects/cel/cel-2.6.git Or for browsing: http://git.linux-nfs.org/?p=cel/cel-2.6.git;a=log;h=refs/heads/nfs-rdma-for-4.12 Changes since v1: - Rebased on v4.11-rc4 + recent non-rdma fixes - Reordered so bug fixes come first - Three small new clean ups --- Chuck Lever (13): xprtrdma: Cancel refresh worker during buffer shutdown sunrpc: Export xprt_force_disconnect() xprtrdma: Detect unreachable NFS/RDMA servers more reliably xprtrdma: Refactor rpcrdma_ia_open() xprtrdma: Use same device when mapping or syncing DMA buffers xprtrdma: Support unplugging an HCA from under an NFS mount xprtrdma: Refactor rpcrdma_ep_connect xprtrdma: Restore transport after device removal xprtrdma: Revert commit d0f36c46deea xprtrdma: Annotate receive workqueue xprtrdma: Squelch ENOBUFS warnings sunrpc: Fix xdr_init_decode_pages() documenting comment xprtrdma: Remove rpcrdma_buffer::rb_pool net/sunrpc/xdr.c | 2 net/sunrpc/xprt.c | 1 net/sunrpc/xprtrdma/rpc_rdma.c | 12 + net/sunrpc/xprtrdma/transport.c | 57 ++++++- net/sunrpc/xprtrdma/verbs.c | 323 ++++++++++++++++++++++++++------------- net/sunrpc/xprtrdma/xprt_rdma.h | 22 ++- 6 files changed, 295 insertions(+), 122 deletions(-) -- Chuck Lever -- 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
