diff options
author | Dean Luick <dean.luick@intel.com> | 2016-02-03 14:35:49 -0800 |
---|---|---|
committer | Doug Ledford <dledford@redhat.com> | 2016-03-10 20:37:56 -0500 |
commit | 7b0b01aa8f48cd237322cbffa05662a9c6b156f8 (patch) | |
tree | 02482f328f5dc0b7f509af19628c8f88c48a3e52 /drivers/staging/rdma/hfi1/rc.c | |
parent | staging/rdma/hfi1: Fix fabric serdes reset by re-downloading firmware (diff) | |
download | linux-dev-7b0b01aa8f48cd237322cbffa05662a9c6b156f8.tar.xz linux-dev-7b0b01aa8f48cd237322cbffa05662a9c6b156f8.zip |
staging/rdma/hfi1: Split last 8 bytes of copy to user buffer
Copy the last 8 bytes of user mode RC WRITE_ONLY and WRITE_LAST
opcodes separately from the rest of the data.
It is a de-facto standard for some MPI implementations to use a
poll on the last few bytes of a verbs message to indicate that
the message has been received rather than follow the required
function method. The driver uses the kernel memcpy routine, which
becomes "rep movsb" on modern machines. This copy, while very
fast, does not guarantee in-order copy completion and the result
is an occasional perceived corrupted packet. Avoid the issue by
splitting the last 8 bytes to copy from the verbs opcodes where it
matters and performing an in-order byte copy.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Jubin John <jubin.john@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Diffstat (limited to 'drivers/staging/rdma/hfi1/rc.c')
-rw-r--r-- | drivers/staging/rdma/hfi1/rc.c | 17 |
1 files changed, 11 insertions, 6 deletions
diff --git a/drivers/staging/rdma/hfi1/rc.c b/drivers/staging/rdma/hfi1/rc.c index 50559fd14a70..371edc3dd4f6 100644 --- a/drivers/staging/rdma/hfi1/rc.c +++ b/drivers/staging/rdma/hfi1/rc.c @@ -1539,7 +1539,7 @@ read_middle: qp->s_rdma_read_len -= pmtu; update_last_psn(qp, psn); spin_unlock_irqrestore(&qp->s_lock, flags); - hfi1_copy_sge(&qp->s_rdma_read_sge, data, pmtu, 0); + hfi1_copy_sge(&qp->s_rdma_read_sge, data, pmtu, 0, 0); goto bail; case OP(RDMA_READ_RESPONSE_ONLY): @@ -1583,7 +1583,7 @@ read_last: if (unlikely(tlen != qp->s_rdma_read_len)) goto ack_len_err; aeth = be32_to_cpu(ohdr->u.aeth); - hfi1_copy_sge(&qp->s_rdma_read_sge, data, tlen, 0); + hfi1_copy_sge(&qp->s_rdma_read_sge, data, tlen, 0, 0); WARN_ON(qp->s_rdma_read_sge.num_sge); (void) do_rc_ack(qp, aeth, psn, OP(RDMA_READ_RESPONSE_LAST), 0, rcd); @@ -1977,6 +1977,7 @@ void hfi1_rc_rcv(struct hfi1_packet *packet) unsigned long flags; u32 bth1; int ret, is_fecn = 0; + int copy_last = 0; bth0 = be32_to_cpu(ohdr->bth[0]); if (hfi1_ruc_check_hdr(ibp, hdr, rcv_flags & HFI1_HAS_GRH, qp, bth0)) @@ -2081,7 +2082,7 @@ send_middle: qp->r_rcv_len += pmtu; if (unlikely(qp->r_rcv_len > qp->r_len)) goto nack_inv; - hfi1_copy_sge(&qp->r_sge, data, pmtu, 1); + hfi1_copy_sge(&qp->r_sge, data, pmtu, 1, 0); break; case OP(RDMA_WRITE_LAST_WITH_IMMEDIATE): @@ -2109,8 +2110,10 @@ send_last_imm: wc.ex.imm_data = ohdr->u.imm_data; wc.wc_flags = IB_WC_WITH_IMM; goto send_last; - case OP(SEND_LAST): case OP(RDMA_WRITE_LAST): + copy_last = ibpd_to_rvtpd(qp->ibqp.pd)->user; + /* fall through */ + case OP(SEND_LAST): no_immediate_data: wc.wc_flags = 0; wc.ex.imm_data = 0; @@ -2126,7 +2129,7 @@ send_last: wc.byte_len = tlen + qp->r_rcv_len; if (unlikely(wc.byte_len > qp->r_len)) goto nack_inv; - hfi1_copy_sge(&qp->r_sge, data, tlen, 1); + hfi1_copy_sge(&qp->r_sge, data, tlen, 1, copy_last); rvt_put_ss(&qp->r_sge); qp->r_msn++; if (!test_and_clear_bit(RVT_R_WRID_VALID, &qp->r_aflags)) @@ -2163,8 +2166,10 @@ send_last: (bth0 & IB_BTH_SOLICITED) != 0); break; - case OP(RDMA_WRITE_FIRST): case OP(RDMA_WRITE_ONLY): + copy_last = 1; + /* fall through */ + case OP(RDMA_WRITE_FIRST): case OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE): if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE))) goto nack_inv; |