--- drivers/infiniband/hw/siw/siw_qp_rx.c | 1560 +++++++++++++++++++++++++++++++++ 1 files changed, 1560 insertions(+), 0 deletions(-) create mode 100644 drivers/infiniband/hw/siw/siw_qp_rx.c
diff --git a/drivers/infiniband/hw/siw/siw_qp_rx.c b/drivers/infiniband/hw/siw/siw_qp_rx.c new file mode 100644 index 0000000..b5792fe --- /dev/null +++ b/drivers/infiniband/hw/siw/siw_qp_rx.c @@ -0,0 +1,1560 @@ +/* + * Software iWARP device driver for Linux + * + * Authors: Bernard Metzler <[email protected]> + * Fredy Neeser <[email protected]> + * + * Copyright (c) 2008-2011, IBM Corporation + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above copyright notice, + * this list of conditions and the following disclaimer. + * + * - Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * - Neither the name of IBM nor the names of its contributors may be + * used to endorse or promote products derived from this software without + * specific prior written permission. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include <linux/errno.h> +#include <linux/types.h> +#include <linux/net.h> +#include <linux/scatterlist.h> +#include <linux/highmem.h> +#include <net/sock.h> +#include <net/tcp_states.h> +#include <net/tcp.h> + +#include <rdma/iw_cm.h> +#include <rdma/ib_verbs.h> +#include <rdma/ib_smi.h> +#include <rdma/ib_user_verbs.h> +#include <rdma/ib_umem.h> + +#include "siw.h" +#include "siw_obj.h" +#include "siw_cm.h" + + +/* + * ---------------------------- + * DDP reassembly for Softiwarp + * ---------------------------- + * For the ordering of transmitted DDP segments, the relevant iWARP ordering + * rules are as follows: + * + * - RDMAP (RFC 5040): Section 7.5, Rule 17: + * "RDMA Read Response Message processing at the Remote Peer (reading + * the specified Tagged Buffer) MUST be started only after the RDMA + * Read Request Message has been Delivered by the DDP layer (thus, + * all previous RDMA Messages have been properly submitted for + * ordered Placement)." + * + * - DDP (RFC 5041): Section 5.3: + * "At the Data Source, DDP: + * o MUST transmit DDP Messages in the order they were submitted to + * the DDP layer, + * o SHOULD transmit DDP Segments within a DDP Message in increasing + * MO order for Untagged DDP Messages, and in increasing TO order + * for Tagged DDP Messages." + * + * Combining these rules implies that, although RDMAP does not provide + * ordering between operations that are generated from the two ends of an + * RDMAP stream, DDP *must not* transmit an RDMA Read Response Message before + * it has finished transmitting SQ operations that were already submitted + * to the DDP layer. It follows that an iWARP transmitter must fully + * serialize RDMAP messages belonging to the same QP. + * + * Given that a TCP socket receives DDP segments in peer transmit order, + * we obtain the following ordering of received DDP segments: + * + * (i) the received DDP segments of RDMAP messages for the same QP + * cannot be interleaved + * (ii) the received DDP segments of a single RDMAP message *should* + * arrive in order. + * + * The Softiwarp transmitter obeys rule #2 in DDP Section 5.3. + * With this property, the "should" becomes a "must" in (ii) above, + * which simplifies DDP reassembly considerably. + * The Softiwarp receiver currently relies on this property + * and reports an error if DDP segments of the same RDMAP message + * do not arrive in sequence. + */ + +static inline int siw_crc_rxhdr(struct siw_iwarp_rx *ctx) +{ + crypto_hash_init(&ctx->mpa_crc_hd); + + return siw_crc_array(&ctx->mpa_crc_hd, (u8 *)&ctx->hdr, + ctx->fpdu_part_rcvd); +} + + +/* + * siw_rx_umem_init() + * + * Given memory region @mr and tagged offset @t_off within @mr, + * resolve corresponding ib_umem_chunk memory chunk pointer + * and update receive context variables to point at receive position. + * returns 0 on sucess and failure otherwise. + * + * NOTE: This function expects virtual addresses. + * TODO: Function needs generalization to support relative adressing + * aka "ZBVA". + * + * @rctx: Receive Context to be updated + * @mr: Memory Region + * @t_off: Offset within Memory Region + * + */ +static int siw_rx_umem_init(struct siw_iwarp_rx *rctx, struct siw_mr *mr, + u64 t_off) +{ + struct ib_umem_chunk *chunk; + u64 off_mr; /* offset into MR */ + int psge_idx; /* Index of PSGE */ + + off_mr = t_off - (mr->mem.va & PAGE_MASK); + /* + * Equivalent to + * off_mr = t_off - mr->mem.va; + * off_mr += mr->umem->offset; + */ + + /* Skip pages not referenced by t_off */ + psge_idx = off_mr >> PAGE_SHIFT; + + list_for_each_entry(chunk, &mr->umem->chunk_list, list) { + if (psge_idx < chunk->nents) + break; + psge_idx -= chunk->nents; + } + if (psge_idx >= chunk->nents) { + dprint(DBG_MM|DBG_ON, "(QP%d): Short chunk list\n", + RX_QPID(rctx)); + return -EINVAL; + } + rctx->pg_idx = psge_idx; + rctx->pg_off = off_mr & ~PAGE_MASK; + rctx->umem_chunk = chunk; + + dprint(DBG_MM, "(QP%d): New chunk, idx %d\n", RX_QPID(rctx), psge_idx); + return 0; +} + + +/* + * siw_rx_umem() + * + * Receive data of @len into target referenced by @rctx. + * This function does not check if umem is within bounds requested by + * @len and @t_off. @umem_ends indicates if routine should + * not update chunk position pointers after the point it is + * currently receiving + * + * @rctx: Receive Context + * @len: Number of bytes to place + * @umen_ends: 1, if rctx chunk pointer should not be updated after len. + */ +static int siw_rx_umem(struct siw_iwarp_rx *rctx, int len, int umem_ends) +{ + struct scatterlist *p_list; + void *dest; + struct ib_umem_chunk *chunk = rctx->umem_chunk; + int pg_off = rctx->pg_off, + copied = 0, + bytes, + rv; + + while (len) { + bytes = min(len, (int)PAGE_SIZE - pg_off); + p_list = &chunk->page_list[rctx->pg_idx]; + + dest = kmap_atomic(sg_page(p_list), KM_SOFTIRQ0); + + rv = skb_copy_bits(rctx->skb, rctx->skb_offset, dest + pg_off, + bytes); + + dprint(DBG_RX, "(QP%d): Page #%d, " + "bytes=%u, rv=%d returned by skb_copy_bits()\n", + RX_QPID(rctx), rctx->pg_idx, bytes, rv); + + if (likely(!rv)) { + if (rctx->crc_enabled) + rv = siw_crc_sg(&rctx->mpa_crc_hd, p_list, + pg_off, bytes); + + rctx->skb_offset += bytes; + copied += bytes; + len -= bytes; + pg_off += bytes; + } + + kunmap_atomic(dest, KM_SOFTIRQ0); + + if (unlikely(rv)) { + rctx->skb_copied += copied; + rctx->skb_new -= copied; + copied = -EFAULT; + + dprint(DBG_RX|DBG_ON, "(QP%d): failed with %d\n", + RX_QPID(rctx), rv); + + goto out; + } + if (pg_off == PAGE_SIZE) { + /* + * end of page + */ + pg_off = 0; + /* + * reference next page chunk if + * - all pages in chunk used AND + * - current loop fills more into this umem + * OR the next receive will go into this umem + * starting at the position where we are leaving + * the routine. + */ + if (++rctx->pg_idx == chunk->nents && + (len > 0 || !umem_ends)) { + + rctx->pg_idx = 0; + chunk = mem_chunk_next(chunk); + } + } + } + /* + * store chunk position for resume + */ + rctx->umem_chunk = chunk; + rctx->pg_off = pg_off; + + rctx->skb_copied += copied; + rctx->skb_new -= copied; +out: + return copied; +} + +static inline int siw_rx_kva(struct siw_iwarp_rx *rctx, int len, void *kva) +{ + int rv = skb_copy_bits(rctx->skb, rctx->skb_offset, kva, len); + + if (likely(!rv)) { + rctx->skb_offset += len; + rctx->skb_copied += len; + rctx->skb_new -= len; + if (rctx->crc_enabled) { + rv = siw_crc_array(&rctx->mpa_crc_hd, kva, len); + if (rv) + goto done; + } + rv = len; + } +done: + return rv; +} + +/* + * siw_rresp_check_ntoh() + * + * Check incoming RRESP fragment header against expected + * header values and update expected values for potential next + * fragment. + * + * NOTE: This function must be called only if a RRESP DDP segment + * starts but not for fragmented consecutive pieces of an + * already started DDP segement. + */ +static inline int siw_rresp_check_ntoh(struct siw_iwarp_rx *rctx) +{ + struct iwarp_rdma_rresp *rresp = &rctx->hdr.rresp; + struct siw_wqe *wqe = rctx->dest.wqe; + + u32 sink_stag = be32_to_cpu(rresp->sink_stag); + u64 sink_to = be64_to_cpu(rresp->sink_to); + + if (rctx->first_ddp_seg) { + rctx->ddp_stag = wqe->wr.rread.sge[0].lkey; + rctx->ddp_to = wqe->wr.rread.sge[0].addr; + } + if (rctx->ddp_stag != sink_stag) { + dprint(DBG_RX|DBG_ON, + " received STAG=%08x, expected STAG=%08x\n", + sink_stag, rctx->ddp_stag); + /* + * Verbs: RI_EVENT_QP_LLP_INTEGRITY_ERROR_BAD_FPDU + */ + return -EINVAL; + } + if (rctx->ddp_to != sink_to) { + dprint(DBG_RX|DBG_ON, + " received TO=%016llx, expected TO=%016llx\n", + (unsigned long long)sink_to, + (unsigned long long)rctx->ddp_to); + /* + * Verbs: RI_EVENT_QP_LLP_INTEGRITY_ERROR_BAD_FPDU + */ + return -EINVAL; + } + if (!rctx->more_ddp_segs && (wqe->processed + rctx->fpdu_part_rem + != wqe->bytes)) { + dprint(DBG_RX|DBG_ON, + " RRESP length does not match RREQ, " + "peer sent=%d, expected %d\n", + wqe->processed + rctx->fpdu_part_rem, wqe->bytes); + return -EINVAL; + } + return 0; +} + +/* + * siw_write_check_ntoh() + * + * Check incoming WRITE fragment header against expected + * header values and update expected values for potential next + * fragment + * + * NOTE: This function must be called only if a WRITE DDP segment + * starts but not for fragmented consecutive pieces of an + * already started DDP segement. + */ +static inline int siw_write_check_ntoh(struct siw_iwarp_rx *rctx) +{ + struct iwarp_rdma_write *write = &rctx->hdr.rwrite; + + u32 sink_stag = be32_to_cpu(write->sink_stag); + u64 sink_to = be64_to_cpu(write->sink_to); + + if (rctx->first_ddp_seg) { + rctx->ddp_stag = sink_stag; + rctx->ddp_to = sink_to; + } else { + if (rctx->ddp_stag != sink_stag) { + dprint(DBG_RX|DBG_ON, + " received STAG=%08x, expected STAG=%08x\n", + sink_stag, rctx->ddp_stag); + /* + * Verbs: RI_EVENT_QP_LLP_INTEGRITY_ERROR_BAD_FPDU + */ + return -EINVAL; + } + if (rctx->ddp_to != sink_to) { + dprint(DBG_RX|DBG_ON, + " received TO=%016llx, expected TO=%016llx\n", + (unsigned long long)sink_to, + (unsigned long long)rctx->ddp_to); + /* + * Verbs: RI_EVENT_QP_LLP_INTEGRITY_ERROR_BAD_FPDU + */ + return -EINVAL; + } + } + return 0; +} + +/* + * siw_send_check_ntoh() + * + * Check incoming SEND fragment header against expected + * header values and update expected MSN if no next + * fragment expected + * + * NOTE: This function must be called only if a SEND DDP segment + * starts but not for fragmented consecutive pieces of an + * already started DDP segement. + */ +static inline int siw_send_check_ntoh(struct siw_iwarp_rx *rctx) +{ + struct iwarp_send *send = &rctx->hdr.send; + struct siw_wqe *wqe = rctx->dest.wqe; + + u32 ddp_msn = be32_to_cpu(send->ddp_msn); + u32 ddp_mo = be32_to_cpu(send->ddp_mo); + u32 ddp_qn = be32_to_cpu(send->ddp_qn); + + if (ddp_qn != RDMAP_UNTAGGED_QN_SEND) { + dprint(DBG_RX|DBG_ON, " Invalid DDP QN %d for SEND\n", + ddp_qn); + return -EINVAL; + } + if (ddp_msn != rctx->ddp_msn[RDMAP_UNTAGGED_QN_SEND]) { + dprint(DBG_RX|DBG_ON, " received MSN=%d, expected MSN=%d\n", + rctx->ddp_msn[RDMAP_UNTAGGED_QN_SEND], ddp_msn); + /* + * TODO: Error handling + * async_event= RI_EVENT_QP_RQ_PROTECTION_ERROR_MSN_GAP; + * cmpl_status= RI_WC_STATUS_LOCAL_QP_CATASTROPHIC; + */ + return -EINVAL; + } + if (ddp_mo != wqe->processed) { + dprint(DBG_RX|DBG_ON, " Received MO=%u, expected MO=%u\n", + ddp_mo, wqe->processed); + /* + * Verbs: RI_EVENT_QP_LLP_INTEGRITY_ERROR_BAD_FPDU + */ + return -EINVAL; + } + if (rctx->first_ddp_seg) { + /* initialize user memory write position */ + rctx->sge_idx = 0; + rctx->sge_off = 0; + } + if (wqe->bytes < wqe->processed + rctx->fpdu_part_rem) { + dprint(DBG_RX|DBG_ON, " Receive space short: %d < %d\n", + wqe->bytes - wqe->processed, rctx->fpdu_part_rem); + wqe->wc_status = IB_WC_LOC_LEN_ERR; + return -EINVAL; + } + return 0; +} + + +/* + * siw_srq_fetch_wqe() + * + * Get one RQ wqe from SRQ and inform user + * if SRQ lower watermark reached + */ +static inline struct siw_wqe *siw_srq_fetch_wqe(struct siw_srq *srq) +{ + struct siw_wqe *wqe = NULL; + int qlen; + + lock_srq(srq); + if (!list_empty(&srq->rq)) { + wqe = list_first_wqe(&srq->rq); + list_del_init(&wqe->list); + /* + * The SRQ wqe is counted for SRQ space until completed. + */ + qlen = srq->max_wr - (atomic_read(&srq->space) + 1); + if (srq->armed && qlen < srq->limit) { + srq->armed = 0; + dprint(DBG_RX, " SRQ(%p): SRQ limit event\n", srq); + siw_srq_event(srq, IB_EVENT_SRQ_LIMIT_REACHED); + } + } + unlock_srq(srq); + + return wqe; +} + +static inline struct siw_wqe *siw_get_rqe(struct siw_qp *qp) +{ + struct siw_wqe *wqe = NULL; + + if (!qp->srq) { + lock_rq(qp); + if (!list_empty(&qp->rq)) { + wqe = list_first_wqe(&qp->rq); + list_del_init(&wqe->list); + unlock_rq(qp); + } else { + unlock_rq(qp); + dprint(DBG_RX, " QP(%d): RQ empty!\n", QP_ID(qp)); + } + } else { + wqe = siw_srq_fetch_wqe(qp->srq); + if (wqe) { + siw_qp_get(qp); + wqe->qp = qp; + } else + dprint(DBG_RX, " QP(%d): SRQ empty!\n", QP_ID(qp)); + } + return wqe; +} + + +/* + * siw_proc_send: + * + * Process one incoming SEND and place data into memory referenced by + * receive wqe. + * + * Function supports partially received sends (suspending/resuming + * current receive wqe processing) + * + * return value: + * 0: reached the end of a DDP segment + * -EAGAIN: to be called again to finish the DDP segment + */ +int siw_proc_send(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + struct siw_wqe *wqe; + struct siw_sge *sge; + struct siw_mr *mr; + u32 data_bytes, /* all data bytes available */ + rcvd_bytes; /* sum of data bytes rcvd */ + int rv = 0; + + if (rctx->first_ddp_seg) { + WARN_ON(rx_wqe(qp) != NULL); + + wqe = siw_get_rqe(qp); + if (!wqe) + return -ENOENT; + + rx_wqe(qp) = wqe; + wqe->wr_status = SR_WR_INPROGRESS; + } else { + wqe = rx_wqe(qp); + if (!wqe) { + /* + * this is a siw bug! + */ + dprint(DBG_ON, "QP(%d): RQ failure\n", QP_ID(qp)); + return -EPROTO; + } + } + if (rctx->state == SIW_GET_DATA_START) { + rv = siw_send_check_ntoh(rctx); + if (rv) { + siw_qp_event(qp, IB_EVENT_QP_FATAL); + return rv; + } + if (!rctx->fpdu_part_rem) /* zero length SEND */ + return 0; + } + data_bytes = min(rctx->fpdu_part_rem, rctx->skb_new); + rcvd_bytes = 0; + + while (data_bytes) { + struct siw_pd *pd; + u32 sge_bytes; /* data bytes avail for SGE */ + + sge = &wqe->wr.sgl.sge[rctx->sge_idx]; + + if (!sge->len) { + /* just skip empty sge's */ + rctx->sge_idx++; + rctx->sge_off = 0; + continue; + } + sge_bytes = min(data_bytes, sge->len - rctx->sge_off); + + /* + * check with QP's PD if no SRQ present, SRQ's PD otherwise + */ + pd = qp->srq == NULL ? qp->pd : qp->srq->pd; + + rv = siw_check_sge(pd, sge, SR_MEM_LWRITE, rctx->sge_off, + sge_bytes); + if (rv) { + siw_qp_event(qp, IB_EVENT_QP_ACCESS_ERR); + break; + } + mr = siw_mem2mr(sge->mem.obj); + + if (mr->umem) { + /* + * Are we going to finish placing + * - the last fragment of the current SGE or + * - the last DDP segment (L=1) of the current + * RDMAP message? + * + * siw_rx_umem() must advance umem page_chunk position + * after sucessful receive only, if receive into + * current umem does not end. + * umem ends, if: + * - current SGE gets completely filled, OR + * - current MPA FPDU is last AND gets consumed now + */ + int umem_ends = + ((sge_bytes + rctx->sge_off == sge->len) || + (!rctx->more_ddp_segs && + rctx->fpdu_part_rcvd + sge_bytes == + rctx->fpdu_part_rem)) ? 1 : 0; + + if (rctx->sge_off == 0) { + /* + * started a new sge: update receive pointers + */ + rv = siw_rx_umem_init(rctx, mr, sge->addr); + if (rv) + break; + } + rv = siw_rx_umem(rctx, sge_bytes, umem_ends); + } else + rv = siw_rx_kva(rctx, sge_bytes, + (void *)(sge->addr + rctx->sge_off)); + if (rv != sge_bytes) { + wqe->processed += rcvd_bytes; + return -EINVAL; + } + rctx->sge_off += rv; + + if (rctx->sge_off == sge->len) { + rctx->sge_idx++; + rctx->sge_off = 0; + } + data_bytes -= rv; + rcvd_bytes += rv; + + rctx->fpdu_part_rem -= rv; + rctx->fpdu_part_rcvd += rv; + } + wqe->processed += rcvd_bytes; + + if (!rctx->fpdu_part_rem) + return 0; + + return (rv < 0) ? rv : -EAGAIN; +} + +/* + * siw_proc_write: + * + * Place incoming WRITE after referencing and checking target buffer + + * Function supports partially received WRITEs (suspending/resuming + * current receive processing) + * + * return value: + * 0: reached the end of a DDP segment + * -EAGAIN: to be called again to finish the DDP segment + */ + +int siw_proc_write(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + struct siw_dev *dev = qp->hdr.sdev; + struct siw_mem *mem; + struct siw_mr *mr; + int bytes, + rv; + + if (rctx->state == SIW_GET_DATA_START) { + + if (!rctx->fpdu_part_rem) /* zero length WRITE */ + return 0; + + rv = siw_write_check_ntoh(rctx); + if (rv) { + siw_qp_event(qp, IB_EVENT_QP_FATAL); + return rv; + } + } + bytes = min(rctx->fpdu_part_rem, rctx->skb_new); + + /* + * NOTE: bytes > 0 is always true, since this routine + * gets only called if so. + */ + if (rctx->first_ddp_seg) { + /* DEBUG Code, to be removed */ + if (rx_mem(qp) != NULL) { + dprint(DBG_RX|DBG_ON, "(QP%d): Stale rctx state!\n", + QP_ID(qp)); + return -EFAULT; + } + rx_mem(qp) = siw_mem_id2obj(dev, rctx->ddp_stag >> 8); + } + if (rx_mem(qp) == NULL) { + dprint(DBG_RX|DBG_ON, "(QP%d): " + "Sink STag not found or invalid, STag=0x%08x\n", + QP_ID(qp), rctx->ddp_stag); + return -EINVAL; + } + mem = rx_mem(qp); + /* + * Rtag not checked against mem's tag again because + * hdr check guarantees same tag as before if fragmented + */ + rv = siw_check_mem(qp->pd, mem, rctx->ddp_to + rctx->fpdu_part_rcvd, + SR_MEM_RWRITE, bytes); + if (rv) { + siw_qp_event(qp, IB_EVENT_QP_ACCESS_ERR); + return rv; + } + mr = siw_mem2mr(mem); + + if (mr->umem) { + /* + * Are we going to place the last piece of the last + * DDP segment of the current RDMAP message? + * + * It is last if: + * - rctx->fpdu_part_rem <= rctx->skb_new AND + * - payload_rem (of current DDP segment) <= rctx->skb_new + */ + int last_write = ((rctx->fpdu_part_rem <= rctx->skb_new) && + !rctx->more_ddp_segs) ? 1 : 0; + + if (rctx->first_ddp_seg) { + rv = siw_rx_umem_init(rctx, mr, rctx->ddp_to); + if (rv) + return -EINVAL; + + } + rv = siw_rx_umem(rctx, bytes, last_write); + } else + rv = siw_rx_kva(rctx, bytes, + (void *)(rctx->ddp_to + + rctx->fpdu_part_rcvd)); + + if (rv != bytes) + return -EINVAL; + + rctx->fpdu_part_rem -= rv; + rctx->fpdu_part_rcvd += rv; + + if (!rctx->fpdu_part_rem) { + rctx->ddp_to += rctx->fpdu_part_rcvd; + return 0; + } + + return (rv < 0) ? rv : -EAGAIN; +} + +/* + * inbound RREQ's cannot carry user data. + */ +int siw_proc_rreq(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + if (!rctx->fpdu_part_rem) + return 0; + + dprint(DBG_ON|DBG_RX, "(QP%d): RREQ with MPA len %d\n", QP_ID(qp), + be16_to_cpu(rctx->hdr.ctrl.mpa_len)); + + return -EPROTO; +} + +static inline struct siw_wqe *siw_get_irqe(struct siw_qp *qp) +{ + struct siw_wqe *wqe = NULL; + + if (atomic_dec_return(&qp->irq_space) >= 0) { + wqe = siw_freeq_wqe_get(qp); + if (wqe) { + INIT_LIST_HEAD(&wqe->list); + wqe->processed = 0; + siw_qp_get(qp); + wqe->qp = qp; + wr_type(wqe) = SIW_WR_RDMA_READ_RESP; + } else + atomic_inc(&qp->irq_space); + } else + atomic_inc(&qp->irq_space); + + return wqe; +} + +/* + * siw_init_rresp: + * + * Process inbound RDMA READ REQ. Produce a pseudo READ RESPONSE WQE. + * Put it at the tail of the IRQ, if there is another WQE currently in + * transmit processing. If not, make it the current WQE to be processed + * and schedule transmit processing. + * + * Can be called from softirq context and from process + * context (RREAD socket loopback case!) + * + * return value: + * 0: success, + * failure code otherwise + */ + +int siw_init_rresp(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + struct siw_wqe *rsp; + + rsp = siw_get_irqe(qp); + if (rsp) { + rsp->wr.rresp.sge.len = be32_to_cpu(rctx->hdr.rreq.read_size); + rsp->bytes = rsp->wr.rresp.sge.len; /* redundant */ + + rsp->wr.rresp.sge.addr = be64_to_cpu(rctx->hdr.rreq.source_to); + rsp->wr.rresp.num_sge = rsp->bytes ? 1 : 0; + + rsp->wr.rresp.sge.mem.obj = NULL; /* defer lookup */ + rsp->wr.rresp.sge.lkey = + be32_to_cpu(rctx->hdr.rreq.source_stag); + + rsp->wr.rresp.raddr = be64_to_cpu(rctx->hdr.rreq.sink_to); + rsp->wr.rresp.rtag = be32_to_cpu(rctx->hdr.rreq.sink_stag); + + } else { + dprint(DBG_RX|DBG_ON, "(QP%d): IRD exceeded!\n", QP_ID(qp)); + return -EPROTO; + } + rsp->wr_status = SR_WR_QUEUED; + + /* + * Insert into IRQ + * + * TODO: Revisit ordering of genuine SQ WRs and Read Response + * pseudo-WRs. RDMAP specifies that there is no ordering among + * the two directions of transmission, so there is a degree of + * freedom. + * + * The current logic favours Read Responses over SQ work requests + * that are queued but not already in progress. + */ + lock_sq(qp); + if (!tx_wqe(qp)) { + tx_wqe(qp) = rsp; + unlock_sq(qp); + /* + * schedule TX work, even if SQ was supended due to + * ORD limit: it is always OK (and may even prevent peers + * from appl lock) to send RRESPONSE's + */ + siw_sq_queue_work(qp); + } else { + list_add_tail(&rsp->list, &qp->irq); + unlock_sq(qp); + } + return 0; +} + +/* + * siw_proc_rresp: + * + * Place incoming RRESP data into memory referenced by RREQ WQE. + * + * Function supports partially received RRESP's (suspending/resuming + * current receive processing) + */ +int siw_proc_rresp(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + struct siw_wqe *wqe; + struct siw_mr *mr; + struct siw_sge *sge; + int bytes, + rv; + + if (rctx->first_ddp_seg) { + WARN_ON(rx_wqe(qp) != NULL); + /* + * fetch pending RREQ from orq + */ + lock_orq(qp); + if (!list_empty(&qp->orq)) { + wqe = list_first_entry(&qp->orq, struct siw_wqe, list); + list_del_init(&wqe->list); + } else { + unlock_orq(qp); + dprint(DBG_RX|DBG_ON, "(QP%d): ORQ empty\n", + QP_ID(qp)); + /* + * TODO: Should generate an async error + */ + rv = -ENODATA; /* or -ENOENT ? */ + goto done; + } + unlock_orq(qp); + + rx_wqe(qp) = wqe; + + if (wr_type(wqe) != SIW_WR_RDMA_READ_REQ || wqe->processed) { + WARN_ON(wqe->processed); + WARN_ON(wr_type(wqe) != SIW_WR_RDMA_READ_REQ); + rv = -EINVAL; + goto done; + } + + wqe->wr_status = SR_WR_INPROGRESS; + + rv = siw_rresp_check_ntoh(rctx); + if (rv) { + siw_qp_event(qp, IB_EVENT_QP_FATAL); + goto done; + } + } else { + wqe = rx_wqe(qp); + if (!wqe) { + WARN_ON(1); + rv = -ENODATA; + goto done; + } + } + if (!rctx->fpdu_part_rem) /* zero length RRESPONSE */ + return 0; + + bytes = min(rctx->fpdu_part_rem, rctx->skb_new); + sge = wqe->wr.rread.sge; /* there is only one */ + + /* + * check target memory which resolves memory on first fragment + */ + rv = siw_check_sge(qp->pd, sge, SR_MEM_LWRITE, wqe->processed, bytes); + if (rv) { + dprint(DBG_RX|DBG_ON, "(QP%d): siw_check_sge failed: %d\n", + QP_ID(qp), rv); + wqe->wc_status = IB_WC_LOC_PROT_ERR; + siw_qp_event(qp, IB_EVENT_QP_ACCESS_ERR); + goto done; + } + mr = siw_mem2mr(sge->mem.obj); + + if (mr->umem) { + /* + * Are we going to finish placing the last DDP segment (L=1) + * of the current RDMAP message? + * + * NOTE: siw_rresp_check_ntoh() guarantees that the + * last inbound RDMAP Read Response message exactly matches + * with the RREQ WR. + */ + int is_last = (bytes + wqe->processed == wqe->bytes) ? 1 : 0; + + if (rctx->first_ddp_seg) { + rv = siw_rx_umem_init(rctx, mr, sge->addr); + if (rv) { + wqe->wc_status = IB_WC_LOC_PROT_ERR; + goto done; + } + } + rv = siw_rx_umem(rctx, bytes, is_last); + } else + rv = siw_rx_kva(rctx, bytes, + (void *)(sge->addr + wqe->processed)); + if (rv != bytes) { + wqe->wc_status = IB_WC_GENERAL_ERR; + rv = -EINVAL; + goto done; + } + rctx->fpdu_part_rem -= rv; + rctx->fpdu_part_rcvd += rv; + + wqe->processed += rv; + + if (!rctx->fpdu_part_rem) { + rctx->ddp_to += rctx->fpdu_part_rcvd; + return 0; + } +done: + return (rv < 0) ? rv : -EAGAIN; +} + +static void siw_drain_pkt(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + char buf[128]; + int len; + + dprint(DBG_ON|DBG_RX, " (QP%d): drain %d bytes\n", + QP_ID(qp), rctx->fpdu_part_rem); + + while (rctx->fpdu_part_rem) { + len = min(rctx->fpdu_part_rem, 128); + + skb_copy_bits(rctx->skb, rctx->skb_offset, + buf, rctx->fpdu_part_rem); + + rctx->skb_copied += len; + rctx->skb_offset += len; + rctx->skb_new -= len; + rctx->fpdu_part_rem -= len; + rctx->fpdu_part_rcvd += len; + } +} + +int siw_proc_unsupp(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + WARN_ON(1); + siw_drain_pkt(qp, rctx); + return 0; +} + + +int siw_proc_terminate(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + dprint(DBG_ON, " (QP%d): RX Terminate: type=%d, layer=%d, code=%d\n", + QP_ID(qp), + __rdmap_term_etype(&rctx->hdr.terminate), + __rdmap_term_layer(&rctx->hdr.terminate), + __rdmap_term_ecode(&rctx->hdr.terminate)); + + siw_drain_pkt(qp, rctx); + return 0; +} + + +static int siw_get_trailer(struct siw_qp *qp, struct siw_iwarp_rx *rctx) +{ + struct sk_buff *skb = rctx->skb; + u8 *tbuf = (u8 *)&rctx->trailer.crc - rctx->pad; + int avail; + + avail = min(rctx->skb_new, rctx->fpdu_part_rem); + + skb_copy_bits(skb, rctx->skb_offset, + tbuf + rctx->fpdu_part_rcvd, avail); + + rctx->fpdu_part_rcvd += avail; + rctx->fpdu_part_rem -= avail; + + rctx->skb_new -= avail; + rctx->skb_offset += avail; + rctx->skb_copied += avail; + + dprint(DBG_RX, " (QP%d): %d remaining (%d)\n", QP_ID(qp), + rctx->fpdu_part_rem, avail); + + if (!rctx->fpdu_part_rem) { + __be32 crc_in, crc_own = 0; + /* + * check crc if required + */ + if (!rctx->crc_enabled) + return 0; + + if (rctx->pad && siw_crc_array(&rctx->mpa_crc_hd, + tbuf, rctx->pad) != 0) + return -EINVAL; + + crypto_hash_final(&rctx->mpa_crc_hd, (u8 *)&crc_own); + + /* + * CRC32 is computed, transmitted and received directly in NBO, + * so there's never a reason to convert byte order. + */ + crc_in = rctx->trailer.crc; + + if (crc_in != crc_own) { + dprint(DBG_RX|DBG_ON, + " (QP%d): CRC ERROR in:=%08x, own=%08x\n", + QP_ID(qp), crc_in, crc_own); + return -EINVAL; + } + return 0; + } + return -EAGAIN; +} + + +static int siw_get_hdr(struct siw_iwarp_rx *rctx) +{ + struct sk_buff *skb = rctx->skb; + struct iwarp_ctrl *c_hdr = &rctx->hdr.ctrl; + u8 opcode; + + int bytes; + + if (rctx->fpdu_part_rcvd < sizeof(struct iwarp_ctrl)) { + /* + * copy first fix part of iwarp hdr + */ + bytes = min_t(int, rctx->skb_new, sizeof(struct iwarp_ctrl) + - rctx->fpdu_part_rcvd); + + skb_copy_bits(skb, rctx->skb_offset, + (char *)c_hdr + rctx->fpdu_part_rcvd, bytes); + + rctx->fpdu_part_rcvd += bytes; + + rctx->skb_new -= bytes; + rctx->skb_offset += bytes; + rctx->skb_copied += bytes; + + if (!rctx->skb_new || + rctx->fpdu_part_rcvd < sizeof(struct iwarp_ctrl)) + return -EAGAIN; + + if (__ddp_version(c_hdr) != DDP_VERSION) { + dprint(DBG_RX|DBG_ON, " dversion %d\n", + __ddp_version(c_hdr)); + return -EINVAL; + } + if (__rdmap_version(c_hdr) != RDMAP_VERSION) { + dprint(DBG_RX|DBG_ON, " rversion %d\n", + __rdmap_version(c_hdr)); + return -EINVAL; + } + opcode = __rdmap_opcode(c_hdr); + + if (opcode > RDMAP_TERMINATE) { + dprint(DBG_RX|DBG_ON, " opcode %d\n", opcode); + return -EINVAL; + } + dprint(DBG_RX, "(QP%d): New Header, opcode:%d\n", + RX_QPID(rctx), opcode); + } else + opcode = __rdmap_opcode(c_hdr); + /* + * figure out len of current hdr: variable length of + * iwarp hdr forces us to copy hdr information + */ + bytes = min(rctx->skb_new, + iwarp_pktinfo[opcode].hdr_len - rctx->fpdu_part_rcvd); + + skb_copy_bits(skb, rctx->skb_offset, + (char *)c_hdr + rctx->fpdu_part_rcvd, bytes); + + rctx->fpdu_part_rcvd += bytes; + + rctx->skb_new -= bytes; + rctx->skb_offset += bytes; + rctx->skb_copied += bytes; + + if (rctx->fpdu_part_rcvd == iwarp_pktinfo[opcode].hdr_len) { + /* + * HDR receive completed. Check if the current DDP segment + * starts a new RDMAP message or continues a previously + * started RDMAP message. + * + * Note well from the comments on DDP reassembly: + * - Support for unordered reception of DDP segments + * (or FPDUs) from different RDMAP messages is not needed. + * - Unordered reception of DDP segments of the same + * RDMAP message is not supported. It is probably not + * needed with most peers. + */ + siw_dprint_hdr(&rctx->hdr, RX_QPID(rctx), "HDR received"); + + if (rctx->more_ddp_segs != 0) { + rctx->first_ddp_seg = 0; + if (rctx->prev_rdmap_opcode != opcode) { + dprint(DBG_ON, + "packet intersection: %d <> %d\n", + rctx->prev_rdmap_opcode, opcode); + return -EPROTO; + } + } else { + rctx->prev_rdmap_opcode = opcode; + rctx->first_ddp_seg = 1; + } + rctx->more_ddp_segs = + c_hdr->ddp_rdmap_ctrl & DDP_FLAG_LAST ? 0 : 1; + + return 0; + } + return -EAGAIN; +} + +static inline int siw_fpdu_payload_len(struct siw_iwarp_rx *rctx) +{ + return be16_to_cpu(rctx->hdr.ctrl.mpa_len) - rctx->fpdu_part_rcvd + + MPA_HDR_SIZE; +} + +static inline int siw_fpdu_trailer_len(struct siw_iwarp_rx *rctx) +{ + int mpa_len = be16_to_cpu(rctx->hdr.ctrl.mpa_len) + MPA_HDR_SIZE; + + return MPA_CRC_SIZE + (-mpa_len & 0x3); +} + +/* + * siw_rreq_complete() + * + * Complete the current READ REQUEST after READ RESPONSE processing. + * It may complete consecutive WQE's which were already SQ + * processed before but are awaiting completion due to completion + * ordering (see verbs 8.2.2.2). + * The READ RESPONSE may also resume SQ processing if it was stalled + * due to ORD exhaustion (see verbs 8.2.2.18) + * Function stops completion when next READ REQUEST found or ORQ empty. + */ +static void siw_rreq_complete(struct siw_wqe *wqe, int error) +{ + struct siw_qp *qp = wqe->qp; + int num_wc = 1; + enum ib_send_flags flags; + LIST_HEAD(c_list); + + flags = wr_flags(wqe); + + if (flags & IB_SEND_SIGNALED) + list_add(&wqe->list, &c_list); + else { + atomic_inc(&qp->sq_space); + siw_wqe_put(wqe); + num_wc = 0; + } + + lock_orq(qp); + + /* More WQE's to complete following this RREQ? */ + if (!list_empty(&qp->orq)) { + struct list_head *pos, *n; + list_for_each_safe(pos, n, &qp->orq) { + wqe = list_entry_wqe(pos); + if (wr_type(wqe) == SIW_WR_RDMA_READ_REQ) + break; + flags |= wr_flags(wqe); + num_wc++; + dprint(DBG_WR|DBG_ON, + "(QP%d): Resume completion, wr_type %d\n", + QP_ID(qp), wr_type(wqe)); + list_move_tail(pos, &c_list); + } + } + unlock_orq(qp); + + if (num_wc) + siw_sq_complete(&c_list, qp, num_wc, flags); + + /* + * Check if SQ processing was stalled due to ORD limit + */ + if (ORD_SUSPEND_SQ(qp)) { + lock_sq(qp); + + wqe = siw_next_tx_wqe(qp); + + if (wqe && !tx_wqe(qp)) { + WARN_ON(wr_type(wqe) != SIW_WR_RDMA_READ_REQ); + list_del_init(&wqe->list); + tx_wqe(qp) = wqe; + + list_add_tail(&wqe->list, &qp->orq); + + unlock_sq(qp); + + dprint(DBG_RX, "(QP%d): SQ resume (%d)\n", + QP_ID(qp), atomic_read(&qp->sq_space)); + + siw_sq_queue_work(qp); + } else { + /* only new ORQ space if not next RREQ queued */ + atomic_inc(&qp->orq_space); + unlock_sq(qp); + } + } else + atomic_inc(&qp->orq_space); +} + +/* + * siw_rdmap_complete() + * + * complete processing of an RDMA message after receiving all + * DDP segmens + * + * o SENDs + RRESPs will need for completion, + * o RREQs need for READ RESPONSE initialization + * o WRITEs need memory dereferencing + * + * TODO: Could siw_[s,r]_complete() fail? (CQ full) + */ +static inline int siw_rdmap_complete(struct siw_qp *qp, + struct siw_iwarp_rx *rctx) +{ + struct siw_wqe *wqe; + u8 opcode = __rdmap_opcode(&rctx->hdr.ctrl); + int rv = 0; + + switch (opcode) { + + case RDMAP_SEND_SE: + wr_flags(rx_wqe(qp)) |= IB_SEND_SOLICITED; + case RDMAP_SEND: + rctx->ddp_msn[RDMAP_UNTAGGED_QN_SEND]++; + + wqe = rx_wqe(qp); + + wqe->wc_status = IB_WC_SUCCESS; + wqe->wr_status = SR_WR_DONE; + + siw_rq_complete(wqe, qp); + + break; + + case RDMAP_RDMA_READ_RESP: + rctx->ddp_msn[RDMAP_UNTAGGED_QN_RDMA_READ]++; + + wqe = rx_wqe(qp); + + wqe->wc_status = IB_WC_SUCCESS; + wqe->wr_status = SR_WR_DONE; + + siw_rreq_complete(wqe, 0); + + break; + + case RDMAP_RDMA_READ_REQ: + rv = siw_init_rresp(qp, rctx); + + break; + + case RDMAP_RDMA_WRITE: + /* + * Free References from memory object if + * attached to receive context (inbound WRITE) + * While a zero-length WRITE is allowed, the + * current implementation does not create + * a memory reference (it is unclear if memory + * rights should be checked in that case!). + * + * TODO: check zero length WRITE semantics + */ + if (rx_mem(qp)) + siw_mem_put(rx_mem(qp)); + break; + + default: + break; + + } + rctx->umem_chunk = NULL; /* DEBUG aid, tentatively */ + rx_wqe(qp) = NULL; /* also clears MEM object for WRITE */ + + return rv; +} + +/* + * siw_rdmap_error() + * + * Abort processing of RDMAP message after failure. + * SENDs + RRESPs will need for receive completion, if + * already started. + * + * TODO: WRITE need local error to be surfaced. + * + */ +static inline void +siw_rdmap_error(struct siw_qp *qp, struct siw_iwarp_rx *rctx, int status) +{ + struct siw_wqe *wqe; + u8 opcode = __rdmap_opcode(&rctx->hdr.ctrl); + + switch (opcode) { + + case RDMAP_SEND_SE: + case RDMAP_SEND: + rctx->ddp_msn[RDMAP_UNTAGGED_QN_SEND]++; + + wqe = rx_wqe(qp); + if (!wqe) + return; + + if (opcode == RDMAP_SEND_SE) + wr_flags(wqe) |= IB_SEND_SOLICITED; + + if (!wqe->wc_status) + wqe->wc_status = IB_WC_GENERAL_ERR; + + wqe->wr_status = SR_WR_DONE; + siw_rq_complete(wqe, qp); + + break; + + case RDMAP_RDMA_READ_RESP: + /* + * A READ RESPONSE may flush consecutive WQE's + * which were SQ processed before + */ + rctx->ddp_msn[RDMAP_UNTAGGED_QN_RDMA_READ]++; + + if (rctx->state == SIW_GET_HDR || status == -ENODATA) + /* eventual RREQ left untouched */ + break; + + wqe = rx_wqe(qp); + if (wqe) { + if (status) + wqe->wc_status = status; + else + wqe->wc_status = IB_WC_GENERAL_ERR; + + wqe->wr_status = SR_WR_DONE; + /* + * All errors turn the wqe into signalled. + */ + wr_flags(wqe) |= IB_SEND_SIGNALED; + siw_rreq_complete(wqe, status); + } + break; + + case RDMAP_RDMA_WRITE: + /* + * Free References from memory object if + * attached to receive context (inbound WRITE) + * While a zero-length WRITE is allowed, the + * current implementation does not create + * a memory reference (it is unclear if memory + * rights should be checked in that case!). + * + * TODO: check zero length WRITE semantics + */ + if (rx_mem(qp)) + siw_mem_put(rx_mem(qp)); + break; + + default: + break; + } + rctx->umem_chunk = NULL; /* DEBUG aid, tentatively */ + rx_wqe(qp) = NULL; /* also clears MEM object for WRITE */ +} + +/* + * siw_tcp_rx_data() + * + * Main routine to consume inbound TCP payload + * + * @rd_desc: read descriptor + * @skb: socket buffer + * @off: offset in skb + * @len: skb->len - offset : payload in skb + */ +int siw_tcp_rx_data(read_descriptor_t *rd_desc, struct sk_buff *skb, + unsigned int off, size_t len) +{ + struct siw_qp *qp = rd_desc->arg.data; + struct siw_iwarp_rx *rctx = &qp->rx_ctx; + int rv; + + rctx->skb = skb; + rctx->skb_new = skb->len - off; + rctx->skb_offset = off; + rctx->skb_copied = 0; + + dprint(DBG_RX, "(QP%d): new data %d, rx-state %d\n", QP_ID(qp), + rctx->skb_new, rctx->state); + + if (unlikely(rctx->rx_suspend == 1 || + qp->attrs.state != SIW_QP_STATE_RTS)) { + dprint(DBG_RX|DBG_ON, "(QP%d): failed. state rx:%d, qp:%d\n", + QP_ID(qp), qp->rx_ctx.state, qp->attrs.state); + return 0; + } + while (rctx->skb_new) { + + switch (rctx->state) { + + case SIW_GET_HDR: + rv = siw_get_hdr(rctx); + if (!rv) { + if (rctx->crc_enabled && + siw_crc_rxhdr(rctx) != 0) { + rv = -EINVAL; + break; + } + rctx->fpdu_part_rem = + siw_fpdu_payload_len(rctx); + + if (rctx->fpdu_part_rem) + rctx->pad = -rctx->fpdu_part_rem & 0x3; + else + rctx->pad = 0; + + rctx->state = SIW_GET_DATA_START; + rctx->fpdu_part_rcvd = 0; + } + break; + + case SIW_GET_DATA_MORE: + /* + * Another data fragment of the same DDP segment. + * Headers will not be checked again by the + * opcode-specific data receive function below. + * Setting first_ddp_seg = 0 avoids repeating + * initializations that may occur only once per + * DDP segment. + */ + rctx->first_ddp_seg = 0; + + case SIW_GET_DATA_START: + /* + * Headers will be checked by the opcode-specific + * data receive function below. + */ + rv = siw_rx_data(qp, rctx); + if (!rv) { + rctx->fpdu_part_rem = + siw_fpdu_trailer_len(rctx); + rctx->fpdu_part_rcvd = 0; + rctx->state = SIW_GET_TRAILER; + } else + rctx->state = SIW_GET_DATA_MORE; + + break; + + case SIW_GET_TRAILER: + /* + * read CRC + any padding + */ + rv = siw_get_trailer(qp, rctx); + if (!rv) { + /* + * FPDU completed. + * complete RDMAP message if last fragment + */ + rctx->state = SIW_GET_HDR; + rctx->fpdu_part_rcvd = 0; + + if (!(rctx->hdr.ctrl.ddp_rdmap_ctrl + & DDP_FLAG_LAST)) + /* more frags */ + break; + + rv = siw_rdmap_complete(qp, rctx); + if (rv) + break; + } + break; + + default: + WARN_ON(1); + rv = -EAGAIN; + } + + if (unlikely(rv != 0 && rv != -EAGAIN)) { + /* + * TODO: implement graceful error handling including + * generation (and processing) of TERMINATE + * messages. + * + * for now we are left with a bogus rx status + * unable to receive any further byte. + * BUT: code must handle difference between + * + * o protocol syntax (FATAL, framing lost) + * o crc (FATAL, framing lost since we do not + * trust packet header (??)) + * o local resource (maybe non fatal, framing + * not lost) + * + * errors. + */ + siw_rdmap_error(qp, rctx, rv); + + dprint(DBG_RX|DBG_ON, + "(QP%d): RX ERROR %d at RX state %d\n", + QP_ID(qp), rv, rctx->state); + + siw_dprint_rctx(rctx); + /* + * Calling siw_cm_queue_work() is safe without + * releasing qp->state_lock because the QP state + * will be transitioned to SIW_QP_STATE_ERROR + * by the siw_work_handler() workqueue handler + * after we return from siw_qp_llp_data_ready(). + */ + siw_qp_cm_drop(qp, 1); + + break; + } + if (rv) { + dprint(DBG_RX, "(QP%d): " + "Misaligned FPDU: State: %d, missing: %d\n", + QP_ID(qp), rctx->state, rctx->fpdu_part_rem); + break; + } + } + return rctx->skb_copied; +} -- 1.5.4.3 -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
