blob: 30065a28628cdaeb07ab814d2f21c7c274f17565 [file] [log] [blame]
Andrew Scullb4b6d4a2019-01-02 15:54:55 +00001// SPDX-License-Identifier: GPL-2.0
2/*
3 * Copyright (c) 2015, 2017 Oracle. All rights reserved.
4 * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved.
5 */
6
7/* Lightweight memory registration using Fast Registration Work
8 * Requests (FRWR).
9 *
David Brazdil0f672f62019-12-10 10:32:29 +000010 * FRWR features ordered asynchronous registration and invalidation
11 * of arbitrarily-sized memory regions. This is the fastest and safest
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000012 * but most complex memory registration mode.
13 */
14
15/* Normal operation
16 *
David Brazdil0f672f62019-12-10 10:32:29 +000017 * A Memory Region is prepared for RDMA Read or Write using a FAST_REG
18 * Work Request (frwr_map). When the RDMA operation is finished, this
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000019 * Memory Region is invalidated using a LOCAL_INV Work Request
David Brazdil0f672f62019-12-10 10:32:29 +000020 * (frwr_unmap_async and frwr_unmap_sync).
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000021 *
David Brazdil0f672f62019-12-10 10:32:29 +000022 * Typically FAST_REG Work Requests are not signaled, and neither are
23 * RDMA Send Work Requests (with the exception of signaling occasionally
24 * to prevent provider work queue overflows). This greatly reduces HCA
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000025 * interrupt workload.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000026 */
27
28/* Transport recovery
29 *
David Brazdil0f672f62019-12-10 10:32:29 +000030 * frwr_map and frwr_unmap_* cannot run at the same time the transport
31 * connect worker is running. The connect worker holds the transport
32 * send lock, just as ->send_request does. This prevents frwr_map and
33 * the connect worker from running concurrently. When a connection is
34 * closed, the Receive completion queue is drained before the allowing
35 * the connect worker to get control. This prevents frwr_unmap and the
36 * connect worker from running concurrently.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000037 *
David Brazdil0f672f62019-12-10 10:32:29 +000038 * When the underlying transport disconnects, MRs that are in flight
39 * are flushed and are likely unusable. Thus all flushed MRs are
40 * destroyed. New MRs are created on demand.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000041 */
42
43#include <linux/sunrpc/rpc_rdma.h>
44#include <linux/sunrpc/svc_rdma.h>
45
46#include "xprt_rdma.h"
47#include <trace/events/rpcrdma.h>
48
49#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
50# define RPCDBG_FACILITY RPCDBG_TRANS
51#endif
52
David Brazdil0f672f62019-12-10 10:32:29 +000053/**
54 * frwr_is_supported - Check if device supports FRWR
55 * @device: interface adapter to check
56 *
57 * Returns true if device supports FRWR, otherwise false
58 */
59bool frwr_is_supported(struct ib_device *device)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000060{
David Brazdil0f672f62019-12-10 10:32:29 +000061 struct ib_device_attr *attrs = &device->attrs;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000062
63 if (!(attrs->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS))
64 goto out_not_supported;
65 if (attrs->max_fast_reg_page_list_len == 0)
66 goto out_not_supported;
67 return true;
68
69out_not_supported:
70 pr_info("rpcrdma: 'frwr' mode is not supported by device %s\n",
David Brazdil0f672f62019-12-10 10:32:29 +000071 device->name);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000072 return false;
73}
74
David Brazdil0f672f62019-12-10 10:32:29 +000075/**
76 * frwr_release_mr - Destroy one MR
77 * @mr: MR allocated by frwr_init_mr
78 *
79 */
80void frwr_release_mr(struct rpcrdma_mr *mr)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000081{
82 int rc;
83
84 rc = ib_dereg_mr(mr->frwr.fr_mr);
85 if (rc)
David Brazdil0f672f62019-12-10 10:32:29 +000086 trace_xprtrdma_frwr_dereg(mr, rc);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000087 kfree(mr->mr_sg);
88 kfree(mr);
89}
90
David Brazdil0f672f62019-12-10 10:32:29 +000091static void frwr_mr_recycle(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr *mr)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000092{
David Brazdil0f672f62019-12-10 10:32:29 +000093 trace_xprtrdma_mr_recycle(mr);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000094
David Brazdil0f672f62019-12-10 10:32:29 +000095 if (mr->mr_dir != DMA_NONE) {
96 trace_xprtrdma_mr_unmap(mr);
97 ib_dma_unmap_sg(r_xprt->rx_ia.ri_id->device,
98 mr->mr_sg, mr->mr_nents, mr->mr_dir);
99 mr->mr_dir = DMA_NONE;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000100 }
101
David Brazdil0f672f62019-12-10 10:32:29 +0000102 spin_lock(&r_xprt->rx_buf.rb_lock);
103 list_del(&mr->mr_all);
104 r_xprt->rx_stats.mrs_recycled++;
105 spin_unlock(&r_xprt->rx_buf.rb_lock);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000106
David Brazdil0f672f62019-12-10 10:32:29 +0000107 frwr_release_mr(mr);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000108}
109
David Brazdil0f672f62019-12-10 10:32:29 +0000110/* MRs are dynamically allocated, so simply clean up and release the MR.
111 * A replacement MR will subsequently be allocated on demand.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000112 */
113static void
David Brazdil0f672f62019-12-10 10:32:29 +0000114frwr_mr_recycle_worker(struct work_struct *work)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000115{
David Brazdil0f672f62019-12-10 10:32:29 +0000116 struct rpcrdma_mr *mr = container_of(work, struct rpcrdma_mr,
117 mr_recycle);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000118
David Brazdil0f672f62019-12-10 10:32:29 +0000119 frwr_mr_recycle(mr->mr_xprt, mr);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000120}
121
David Brazdil0f672f62019-12-10 10:32:29 +0000122/* frwr_recycle - Discard MRs
123 * @req: request to reset
124 *
125 * Used after a reconnect. These MRs could be in flight, we can't
126 * tell. Safe thing to do is release them.
127 */
128void frwr_recycle(struct rpcrdma_req *req)
129{
130 struct rpcrdma_mr *mr;
131
132 while ((mr = rpcrdma_mr_pop(&req->rl_registered)))
133 frwr_mr_recycle(mr->mr_xprt, mr);
134}
135
136/* frwr_reset - Place MRs back on the free list
137 * @req: request to reset
138 *
139 * Used after a failed marshal. For FRWR, this means the MRs
140 * don't have to be fully released and recreated.
141 *
142 * NB: This is safe only as long as none of @req's MRs are
143 * involved with an ongoing asynchronous FAST_REG or LOCAL_INV
144 * Work Request.
145 */
146void frwr_reset(struct rpcrdma_req *req)
147{
148 struct rpcrdma_mr *mr;
149
150 while ((mr = rpcrdma_mr_pop(&req->rl_registered)))
151 rpcrdma_mr_put(mr);
152}
153
154/**
155 * frwr_init_mr - Initialize one MR
156 * @ia: interface adapter
157 * @mr: generic MR to prepare for FRWR
158 *
159 * Returns zero if successful. Otherwise a negative errno
160 * is returned.
161 */
162int frwr_init_mr(struct rpcrdma_ia *ia, struct rpcrdma_mr *mr)
163{
164 unsigned int depth = ia->ri_max_frwr_depth;
165 struct scatterlist *sg;
166 struct ib_mr *frmr;
167 int rc;
168
169 /* NB: ib_alloc_mr and device drivers typically allocate
170 * memory with GFP_KERNEL.
171 */
172 frmr = ib_alloc_mr(ia->ri_pd, ia->ri_mrtype, depth);
173 if (IS_ERR(frmr))
174 goto out_mr_err;
175
176 sg = kcalloc(depth, sizeof(*sg), GFP_NOFS);
177 if (!sg)
178 goto out_list_err;
179
180 mr->frwr.fr_mr = frmr;
181 mr->mr_dir = DMA_NONE;
182 INIT_LIST_HEAD(&mr->mr_list);
183 INIT_WORK(&mr->mr_recycle, frwr_mr_recycle_worker);
184 init_completion(&mr->frwr.fr_linv_done);
185
186 sg_init_table(sg, depth);
187 mr->mr_sg = sg;
188 return 0;
189
190out_mr_err:
191 rc = PTR_ERR(frmr);
192 trace_xprtrdma_frwr_alloc(mr, rc);
193 return rc;
194
195out_list_err:
196 ib_dereg_mr(frmr);
197 return -ENOMEM;
198}
199
200/**
201 * frwr_open - Prepare an endpoint for use with FRWR
202 * @ia: interface adapter this endpoint will use
203 * @ep: endpoint to prepare
204 *
205 * On success, sets:
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000206 * ep->rep_attr.cap.max_send_wr
207 * ep->rep_attr.cap.max_recv_wr
David Brazdil0f672f62019-12-10 10:32:29 +0000208 * ep->rep_max_requests
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000209 * ia->ri_max_segs
210 *
211 * And these FRWR-related fields:
212 * ia->ri_max_frwr_depth
213 * ia->ri_mrtype
David Brazdil0f672f62019-12-10 10:32:29 +0000214 *
215 * On failure, a negative errno is returned.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000216 */
David Brazdil0f672f62019-12-10 10:32:29 +0000217int frwr_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000218{
David Brazdil0f672f62019-12-10 10:32:29 +0000219 struct ib_device_attr *attrs = &ia->ri_id->device->attrs;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000220 int max_qp_wr, depth, delta;
221
222 ia->ri_mrtype = IB_MR_TYPE_MEM_REG;
223 if (attrs->device_cap_flags & IB_DEVICE_SG_GAPS_REG)
224 ia->ri_mrtype = IB_MR_TYPE_SG_GAPS;
225
David Brazdil0f672f62019-12-10 10:32:29 +0000226 /* Quirk: Some devices advertise a large max_fast_reg_page_list_len
227 * capability, but perform optimally when the MRs are not larger
228 * than a page.
229 */
230 if (attrs->max_sge_rd > 1)
231 ia->ri_max_frwr_depth = attrs->max_sge_rd;
232 else
233 ia->ri_max_frwr_depth = attrs->max_fast_reg_page_list_len;
234 if (ia->ri_max_frwr_depth > RPCRDMA_MAX_DATA_SEGS)
235 ia->ri_max_frwr_depth = RPCRDMA_MAX_DATA_SEGS;
236 dprintk("RPC: %s: max FR page list depth = %u\n",
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000237 __func__, ia->ri_max_frwr_depth);
238
239 /* Add room for frwr register and invalidate WRs.
240 * 1. FRWR reg WR for head
241 * 2. FRWR invalidate WR for head
242 * 3. N FRWR reg WRs for pagelist
243 * 4. N FRWR invalidate WRs for pagelist
244 * 5. FRWR reg WR for tail
245 * 6. FRWR invalidate WR for tail
246 * 7. The RDMA_SEND WR
247 */
248 depth = 7;
249
250 /* Calculate N if the device max FRWR depth is smaller than
251 * RPCRDMA_MAX_DATA_SEGS.
252 */
253 if (ia->ri_max_frwr_depth < RPCRDMA_MAX_DATA_SEGS) {
254 delta = RPCRDMA_MAX_DATA_SEGS - ia->ri_max_frwr_depth;
255 do {
256 depth += 2; /* FRWR reg + invalidate */
257 delta -= ia->ri_max_frwr_depth;
258 } while (delta > 0);
259 }
260
David Brazdil0f672f62019-12-10 10:32:29 +0000261 max_qp_wr = ia->ri_id->device->attrs.max_qp_wr;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000262 max_qp_wr -= RPCRDMA_BACKWARD_WRS;
263 max_qp_wr -= 1;
264 if (max_qp_wr < RPCRDMA_MIN_SLOT_TABLE)
265 return -ENOMEM;
David Brazdil0f672f62019-12-10 10:32:29 +0000266 if (ep->rep_max_requests > max_qp_wr)
267 ep->rep_max_requests = max_qp_wr;
268 ep->rep_attr.cap.max_send_wr = ep->rep_max_requests * depth;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000269 if (ep->rep_attr.cap.max_send_wr > max_qp_wr) {
David Brazdil0f672f62019-12-10 10:32:29 +0000270 ep->rep_max_requests = max_qp_wr / depth;
271 if (!ep->rep_max_requests)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000272 return -EINVAL;
David Brazdil0f672f62019-12-10 10:32:29 +0000273 ep->rep_attr.cap.max_send_wr = ep->rep_max_requests * depth;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000274 }
275 ep->rep_attr.cap.max_send_wr += RPCRDMA_BACKWARD_WRS;
276 ep->rep_attr.cap.max_send_wr += 1; /* for ib_drain_sq */
David Brazdil0f672f62019-12-10 10:32:29 +0000277 ep->rep_attr.cap.max_recv_wr = ep->rep_max_requests;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000278 ep->rep_attr.cap.max_recv_wr += RPCRDMA_BACKWARD_WRS;
279 ep->rep_attr.cap.max_recv_wr += 1; /* for ib_drain_rq */
280
David Brazdil0f672f62019-12-10 10:32:29 +0000281 ia->ri_max_segs =
282 DIV_ROUND_UP(RPCRDMA_MAX_DATA_SEGS, ia->ri_max_frwr_depth);
283 /* Reply chunks require segments for head and tail buffers */
284 ia->ri_max_segs += 2;
285 if (ia->ri_max_segs > RPCRDMA_MAX_HDR_SEGS)
286 ia->ri_max_segs = RPCRDMA_MAX_HDR_SEGS;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000287 return 0;
288}
289
David Brazdil0f672f62019-12-10 10:32:29 +0000290/**
291 * frwr_maxpages - Compute size of largest payload
292 * @r_xprt: transport
293 *
294 * Returns maximum size of an RPC message, in pages.
295 *
296 * FRWR mode conveys a list of pages per chunk segment. The
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000297 * maximum length of that list is the FRWR page list depth.
298 */
David Brazdil0f672f62019-12-10 10:32:29 +0000299size_t frwr_maxpages(struct rpcrdma_xprt *r_xprt)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000300{
301 struct rpcrdma_ia *ia = &r_xprt->rx_ia;
302
303 return min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS,
David Brazdil0f672f62019-12-10 10:32:29 +0000304 (ia->ri_max_segs - 2) * ia->ri_max_frwr_depth);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000305}
306
307/**
David Brazdil0f672f62019-12-10 10:32:29 +0000308 * frwr_map - Register a memory region
309 * @r_xprt: controlling transport
310 * @seg: memory region co-ordinates
311 * @nsegs: number of segments remaining
312 * @writing: true when RDMA Write will be used
313 * @xid: XID of RPC using the registered memory
314 * @mr: MR to fill in
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000315 *
David Brazdil0f672f62019-12-10 10:32:29 +0000316 * Prepare a REG_MR Work Request to register a memory region
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000317 * for remote access via RDMA READ or RDMA WRITE.
David Brazdil0f672f62019-12-10 10:32:29 +0000318 *
319 * Returns the next segment or a negative errno pointer.
320 * On success, @mr is filled in.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000321 */
David Brazdil0f672f62019-12-10 10:32:29 +0000322struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt,
323 struct rpcrdma_mr_seg *seg,
324 int nsegs, bool writing, __be32 xid,
325 struct rpcrdma_mr *mr)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000326{
327 struct rpcrdma_ia *ia = &r_xprt->rx_ia;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000328 struct ib_reg_wr *reg_wr;
David Brazdil0f672f62019-12-10 10:32:29 +0000329 struct ib_mr *ibmr;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000330 int i, n;
331 u8 key;
332
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000333 if (nsegs > ia->ri_max_frwr_depth)
334 nsegs = ia->ri_max_frwr_depth;
335 for (i = 0; i < nsegs;) {
336 if (seg->mr_page)
337 sg_set_page(&mr->mr_sg[i],
338 seg->mr_page,
339 seg->mr_len,
340 offset_in_page(seg->mr_offset));
341 else
342 sg_set_buf(&mr->mr_sg[i], seg->mr_offset,
343 seg->mr_len);
344
345 ++seg;
346 ++i;
David Brazdil0f672f62019-12-10 10:32:29 +0000347 if (ia->ri_mrtype == IB_MR_TYPE_SG_GAPS)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000348 continue;
349 if ((i < nsegs && offset_in_page(seg->mr_offset)) ||
350 offset_in_page((seg-1)->mr_offset + (seg-1)->mr_len))
351 break;
352 }
353 mr->mr_dir = rpcrdma_data_dir(writing);
354
David Brazdil0f672f62019-12-10 10:32:29 +0000355 mr->mr_nents =
356 ib_dma_map_sg(ia->ri_id->device, mr->mr_sg, i, mr->mr_dir);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000357 if (!mr->mr_nents)
358 goto out_dmamap_err;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000359
David Brazdil0f672f62019-12-10 10:32:29 +0000360 ibmr = mr->frwr.fr_mr;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000361 n = ib_map_mr_sg(ibmr, mr->mr_sg, mr->mr_nents, NULL, PAGE_SIZE);
362 if (unlikely(n != mr->mr_nents))
363 goto out_mapmr_err;
364
David Brazdil0f672f62019-12-10 10:32:29 +0000365 ibmr->iova &= 0x00000000ffffffff;
366 ibmr->iova |= ((u64)be32_to_cpu(xid)) << 32;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000367 key = (u8)(ibmr->rkey & 0x000000FF);
368 ib_update_fast_reg_key(ibmr, ++key);
369
David Brazdil0f672f62019-12-10 10:32:29 +0000370 reg_wr = &mr->frwr.fr_regwr;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000371 reg_wr->mr = ibmr;
372 reg_wr->key = ibmr->rkey;
373 reg_wr->access = writing ?
374 IB_ACCESS_REMOTE_WRITE | IB_ACCESS_LOCAL_WRITE :
375 IB_ACCESS_REMOTE_READ;
376
377 mr->mr_handle = ibmr->rkey;
378 mr->mr_length = ibmr->length;
379 mr->mr_offset = ibmr->iova;
David Brazdil0f672f62019-12-10 10:32:29 +0000380 trace_xprtrdma_mr_map(mr);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000381
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000382 return seg;
383
384out_dmamap_err:
David Brazdil0f672f62019-12-10 10:32:29 +0000385 mr->mr_dir = DMA_NONE;
386 trace_xprtrdma_frwr_sgerr(mr, i);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000387 return ERR_PTR(-EIO);
388
389out_mapmr_err:
David Brazdil0f672f62019-12-10 10:32:29 +0000390 trace_xprtrdma_frwr_maperr(mr, n);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000391 return ERR_PTR(-EIO);
392}
393
David Brazdil0f672f62019-12-10 10:32:29 +0000394/**
395 * frwr_wc_fastreg - Invoked by RDMA provider for a flushed FastReg WC
396 * @cq: completion queue (ignored)
397 * @wc: completed WR
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000398 *
David Brazdil0f672f62019-12-10 10:32:29 +0000399 */
400static void frwr_wc_fastreg(struct ib_cq *cq, struct ib_wc *wc)
401{
402 struct ib_cqe *cqe = wc->wr_cqe;
403 struct rpcrdma_frwr *frwr =
404 container_of(cqe, struct rpcrdma_frwr, fr_cqe);
405
406 /* WARNING: Only wr_cqe and status are reliable at this point */
407 trace_xprtrdma_wc_fastreg(wc, frwr);
408 /* The MR will get recycled when the associated req is retransmitted */
409}
410
411/**
412 * frwr_send - post Send WR containing the RPC Call message
413 * @ia: interface adapter
414 * @req: Prepared RPC Call
415 *
416 * For FRWR, chain any FastReg WRs to the Send WR. Only a
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000417 * single ib_post_send call is needed to register memory
418 * and then post the Send WR.
David Brazdil0f672f62019-12-10 10:32:29 +0000419 *
420 * Returns the result of ib_post_send.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000421 */
David Brazdil0f672f62019-12-10 10:32:29 +0000422int frwr_send(struct rpcrdma_ia *ia, struct rpcrdma_req *req)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000423{
424 struct ib_send_wr *post_wr;
425 struct rpcrdma_mr *mr;
426
427 post_wr = &req->rl_sendctx->sc_wr;
428 list_for_each_entry(mr, &req->rl_registered, mr_list) {
429 struct rpcrdma_frwr *frwr;
430
431 frwr = &mr->frwr;
432
433 frwr->fr_cqe.done = frwr_wc_fastreg;
434 frwr->fr_regwr.wr.next = post_wr;
435 frwr->fr_regwr.wr.wr_cqe = &frwr->fr_cqe;
436 frwr->fr_regwr.wr.num_sge = 0;
437 frwr->fr_regwr.wr.opcode = IB_WR_REG_MR;
438 frwr->fr_regwr.wr.send_flags = 0;
439
440 post_wr = &frwr->fr_regwr.wr;
441 }
442
443 /* If ib_post_send fails, the next ->send_request for
David Brazdil0f672f62019-12-10 10:32:29 +0000444 * @req will queue these MRs for recovery.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000445 */
446 return ib_post_send(ia->ri_id->qp, post_wr, NULL);
447}
448
David Brazdil0f672f62019-12-10 10:32:29 +0000449/**
450 * frwr_reminv - handle a remotely invalidated mr on the @mrs list
451 * @rep: Received reply
452 * @mrs: list of MRs to check
453 *
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000454 */
David Brazdil0f672f62019-12-10 10:32:29 +0000455void frwr_reminv(struct rpcrdma_rep *rep, struct list_head *mrs)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000456{
457 struct rpcrdma_mr *mr;
458
459 list_for_each_entry(mr, mrs, mr_list)
460 if (mr->mr_handle == rep->rr_inv_rkey) {
461 list_del_init(&mr->mr_list);
David Brazdil0f672f62019-12-10 10:32:29 +0000462 trace_xprtrdma_mr_remoteinv(mr);
463 rpcrdma_mr_put(mr);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000464 break; /* only one invalidated MR per RPC */
465 }
466}
467
David Brazdil0f672f62019-12-10 10:32:29 +0000468static void __frwr_release_mr(struct ib_wc *wc, struct rpcrdma_mr *mr)
469{
470 if (wc->status != IB_WC_SUCCESS)
471 rpcrdma_mr_recycle(mr);
472 else
473 rpcrdma_mr_put(mr);
474}
475
476/**
477 * frwr_wc_localinv - Invoked by RDMA provider for a LOCAL_INV WC
478 * @cq: completion queue (ignored)
479 * @wc: completed WR
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000480 *
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000481 */
David Brazdil0f672f62019-12-10 10:32:29 +0000482static void frwr_wc_localinv(struct ib_cq *cq, struct ib_wc *wc)
483{
484 struct ib_cqe *cqe = wc->wr_cqe;
485 struct rpcrdma_frwr *frwr =
486 container_of(cqe, struct rpcrdma_frwr, fr_cqe);
487 struct rpcrdma_mr *mr = container_of(frwr, struct rpcrdma_mr, frwr);
488
489 /* WARNING: Only wr_cqe and status are reliable at this point */
490 trace_xprtrdma_wc_li(wc, frwr);
491 __frwr_release_mr(wc, mr);
492}
493
494/**
495 * frwr_wc_localinv_wake - Invoked by RDMA provider for a LOCAL_INV WC
496 * @cq: completion queue (ignored)
497 * @wc: completed WR
498 *
499 * Awaken anyone waiting for an MR to finish being fenced.
500 */
501static void frwr_wc_localinv_wake(struct ib_cq *cq, struct ib_wc *wc)
502{
503 struct ib_cqe *cqe = wc->wr_cqe;
504 struct rpcrdma_frwr *frwr =
505 container_of(cqe, struct rpcrdma_frwr, fr_cqe);
506 struct rpcrdma_mr *mr = container_of(frwr, struct rpcrdma_mr, frwr);
507
508 /* WARNING: Only wr_cqe and status are reliable at this point */
509 trace_xprtrdma_wc_li_wake(wc, frwr);
510 __frwr_release_mr(wc, mr);
511 complete(&frwr->fr_linv_done);
512}
513
514/**
515 * frwr_unmap_sync - invalidate memory regions that were registered for @req
516 * @r_xprt: controlling transport instance
517 * @req: rpcrdma_req with a non-empty list of MRs to process
518 *
519 * Sleeps until it is safe for the host CPU to access the previously mapped
520 * memory regions. This guarantees that registered MRs are properly fenced
521 * from the server before the RPC consumer accesses the data in them. It
522 * also ensures proper Send flow control: waking the next RPC waits until
523 * this RPC has relinquished all its Send Queue entries.
524 */
525void frwr_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000526{
527 struct ib_send_wr *first, **prev, *last;
528 const struct ib_send_wr *bad_wr;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000529 struct rpcrdma_frwr *frwr;
530 struct rpcrdma_mr *mr;
David Brazdil0f672f62019-12-10 10:32:29 +0000531 int rc;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000532
533 /* ORDER: Invalidate all of the MRs first
534 *
535 * Chain the LOCAL_INV Work Requests and post them with
536 * a single ib_post_send() call.
537 */
538 frwr = NULL;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000539 prev = &first;
David Brazdil0f672f62019-12-10 10:32:29 +0000540 while ((mr = rpcrdma_mr_pop(&req->rl_registered))) {
541
542 trace_xprtrdma_mr_localinv(mr);
543 r_xprt->rx_stats.local_inv_needed++;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000544
545 frwr = &mr->frwr;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000546 frwr->fr_cqe.done = frwr_wc_localinv;
547 last = &frwr->fr_invwr;
David Brazdil0f672f62019-12-10 10:32:29 +0000548 last->next = NULL;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000549 last->wr_cqe = &frwr->fr_cqe;
David Brazdil0f672f62019-12-10 10:32:29 +0000550 last->sg_list = NULL;
551 last->num_sge = 0;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000552 last->opcode = IB_WR_LOCAL_INV;
David Brazdil0f672f62019-12-10 10:32:29 +0000553 last->send_flags = IB_SEND_SIGNALED;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000554 last->ex.invalidate_rkey = mr->mr_handle;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000555
556 *prev = last;
557 prev = &last->next;
558 }
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000559
560 /* Strong send queue ordering guarantees that when the
561 * last WR in the chain completes, all WRs in the chain
562 * are complete.
563 */
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000564 frwr->fr_cqe.done = frwr_wc_localinv_wake;
565 reinit_completion(&frwr->fr_linv_done);
566
567 /* Transport disconnect drains the receive CQ before it
568 * replaces the QP. The RPC reply handler won't call us
569 * unless ri_id->qp is a valid pointer.
570 */
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000571 bad_wr = NULL;
David Brazdil0f672f62019-12-10 10:32:29 +0000572 rc = ib_post_send(r_xprt->rx_ia.ri_id->qp, first, &bad_wr);
573 trace_xprtrdma_post_send(req, rc);
574
575 /* The final LOCAL_INV WR in the chain is supposed to
576 * do the wake. If it was never posted, the wake will
577 * not happen, so don't wait in that case.
578 */
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000579 if (bad_wr != first)
580 wait_for_completion(&frwr->fr_linv_done);
David Brazdil0f672f62019-12-10 10:32:29 +0000581 if (!rc)
582 return;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000583
David Brazdil0f672f62019-12-10 10:32:29 +0000584 /* Recycle MRs in the LOCAL_INV chain that did not get posted.
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000585 */
586 while (bad_wr) {
587 frwr = container_of(bad_wr, struct rpcrdma_frwr,
588 fr_invwr);
589 mr = container_of(frwr, struct rpcrdma_mr, frwr);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000590 bad_wr = bad_wr->next;
David Brazdil0f672f62019-12-10 10:32:29 +0000591
592 list_del_init(&mr->mr_list);
593 rpcrdma_mr_recycle(mr);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000594 }
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000595}
596
David Brazdil0f672f62019-12-10 10:32:29 +0000597/**
598 * frwr_wc_localinv_done - Invoked by RDMA provider for a signaled LOCAL_INV WC
599 * @cq: completion queue (ignored)
600 * @wc: completed WR
601 *
602 */
603static void frwr_wc_localinv_done(struct ib_cq *cq, struct ib_wc *wc)
604{
605 struct ib_cqe *cqe = wc->wr_cqe;
606 struct rpcrdma_frwr *frwr =
607 container_of(cqe, struct rpcrdma_frwr, fr_cqe);
608 struct rpcrdma_mr *mr = container_of(frwr, struct rpcrdma_mr, frwr);
609 struct rpcrdma_rep *rep = mr->mr_req->rl_reply;
610
611 /* WARNING: Only wr_cqe and status are reliable at this point */
612 trace_xprtrdma_wc_li_done(wc, frwr);
613 __frwr_release_mr(wc, mr);
614
615 /* Ensure @rep is generated before __frwr_release_mr */
616 smp_rmb();
617 rpcrdma_complete_rqst(rep);
618}
619
620/**
621 * frwr_unmap_async - invalidate memory regions that were registered for @req
622 * @r_xprt: controlling transport instance
623 * @req: rpcrdma_req with a non-empty list of MRs to process
624 *
625 * This guarantees that registered MRs are properly fenced from the
626 * server before the RPC consumer accesses the data in them. It also
627 * ensures proper Send flow control: waking the next RPC waits until
628 * this RPC has relinquished all its Send Queue entries.
629 */
630void frwr_unmap_async(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
631{
632 struct ib_send_wr *first, *last, **prev;
633 const struct ib_send_wr *bad_wr;
634 struct rpcrdma_frwr *frwr;
635 struct rpcrdma_mr *mr;
636 int rc;
637
638 /* Chain the LOCAL_INV Work Requests and post them with
639 * a single ib_post_send() call.
640 */
641 frwr = NULL;
642 prev = &first;
643 while ((mr = rpcrdma_mr_pop(&req->rl_registered))) {
644
645 trace_xprtrdma_mr_localinv(mr);
646 r_xprt->rx_stats.local_inv_needed++;
647
648 frwr = &mr->frwr;
649 frwr->fr_cqe.done = frwr_wc_localinv;
650 last = &frwr->fr_invwr;
651 last->next = NULL;
652 last->wr_cqe = &frwr->fr_cqe;
653 last->sg_list = NULL;
654 last->num_sge = 0;
655 last->opcode = IB_WR_LOCAL_INV;
656 last->send_flags = IB_SEND_SIGNALED;
657 last->ex.invalidate_rkey = mr->mr_handle;
658
659 *prev = last;
660 prev = &last->next;
661 }
662
663 /* Strong send queue ordering guarantees that when the
664 * last WR in the chain completes, all WRs in the chain
665 * are complete. The last completion will wake up the
666 * RPC waiter.
667 */
668 frwr->fr_cqe.done = frwr_wc_localinv_done;
669
670 /* Transport disconnect drains the receive CQ before it
671 * replaces the QP. The RPC reply handler won't call us
672 * unless ri_id->qp is a valid pointer.
673 */
674 bad_wr = NULL;
675 rc = ib_post_send(r_xprt->rx_ia.ri_id->qp, first, &bad_wr);
676 trace_xprtrdma_post_send(req, rc);
677 if (!rc)
678 return;
679
680 /* Recycle MRs in the LOCAL_INV chain that did not get posted.
681 */
682 while (bad_wr) {
683 frwr = container_of(bad_wr, struct rpcrdma_frwr, fr_invwr);
684 mr = container_of(frwr, struct rpcrdma_mr, frwr);
685 bad_wr = bad_wr->next;
686
687 rpcrdma_mr_recycle(mr);
688 }
689
690 /* The final LOCAL_INV WR in the chain is supposed to
691 * do the wake. If it was never posted, the wake will
692 * not happen, so wake here in that case.
693 */
694 rpcrdma_complete_rqst(req->rl_reply);
695}