[ofa-general] eHCA issues

Anton Blanchard anton at samba.org
Sat Sep 29 22:37:26 PDT 2007


Hi,

Ive been trying to get the DAPL tests running on eHCA. One issue was
traced back to a negative value in ->max_cqe. Looking at all the device
attributes max_pd and max_ah are negative too.

        props->max_cqe         = min_t(int, rblock->max_cqe, INT_MAX);

Can this really be negative or should we be doing:

        props->max_cqe         = min_t(unsigned int, rblock->max_cqe, INT_MAX);

Anton

hca_id: ehca0
        vendor_id:                      0x5076
        vendor_part_id:                 0
        hw_ver:                         0x1000003
        phys_port_cnt:                  1
        max_mr_size:                    0xb70000000
        page_size_cap:                  0x0
        max_qp:                         16239
        max_qp_wr:                      32768
        device_cap_flags:               0x00005800
        max_sge:                        252
        max_sge_rd:                     0
        max_cq:                         16380
        max_cqe:                        -64
        max_mr:                         61382
        max_pd:                         -1
        max_qp_rd_atom:                 0
        max_ee_rd_atom:                 0
        max_res_rd_atom:                0
        max_qp_init_rd_atom:            0
        max_ee_init_rd_atom:            0
        atomic_cap:                     ATOMIC_NONE (0)
        max_ee:                         0
        max_rdd:                        0
        max_mw:                         61382
        max_raw_ipv6_qp:                0
        max_raw_ethy_qp:                0
        max_mcast_grp:                  32
        max_mcast_qp_attach:            8
        max_total_mcast_qp_attach:      256
        max_ah:                         -1
        max_fmr:                        61382
        max_map_per_fmr:                0
        max_srq:                        0
        max_pkeys:                      16
        local_ca_ack_delay:             0



More information about the general mailing list