[ofa-general] vendstat issue with shark switch

[Hal Rosenstock]

Hi again Vincent,

On Wed, Dec 17, 2008 at 2:56 AM, Vincent Ficet
<jean-vincent.ficet at bull.net> wrote:
> Hello Hal,
>
>> ClassPortInfo is a required attribute for any supported class but I'm
>> not 100% sure what is going on. Those vendstat options are for IS-3
>> and this is IS-4 although there is some device validation supported in
>> the vendstat command.
>>
>> Can you show the output with -d option ?
>>
>>
>
> This is what I have:
>
> [root at inti0 sbin]# vendstat -dd -N 2
> ibwarn: [8822] umad_init: umad_init
> ibwarn: [8822] umad_open_port: ca (null) port 0
> ibwarn: [8822] umad_get_cas_names: max 20
> ibwarn: [8822] umad_get_cas_names: return 1 cas
> ibwarn: [8822] resolve_ca_name: checking ca 'mlx4_0'
> ibwarn: [8822] resolve_ca_port: checking ca 'mlx4_0'
> ibwarn: [8822] umad_get_ca: ca_name mlx4_0
> ibwarn: [8822] umad_get_ca: opened mlx4_0
> ibwarn: [8822] resolve_ca_port: checking port 0
> ibwarn: [8822] resolve_ca_port: checking port 1
> ibwarn: [8822] resolve_ca_port: found active port 1
> ibwarn: [8822] resolve_ca_name: found ca mlx4_0 with port 1 type 1
> ibwarn: [8822] resolve_ca_name: found ca mlx4_0 with active port 1
> ibwarn: [8822] umad_open_port: opening mlx4_0 port 1
> ibwarn: [8822] dev_to_umad_id: mapped mlx4_0 1 to 0
> ibwarn: [8822] umad_open_port: opened /dev/infiniband/umad0 fd 3 portid 0
> ibwarn: [8822] umad_register: fd 3 mgmt_class 1 mgmt_version 1 rmpp_version
> 0 method_mask (nil)
> ibwarn: [8822] umad_register: fd 3 registered to use agent 0 qp 0
> ibwarn: [8822] umad_register: fd 3 mgmt_class 129 mgmt_version 1
> rmpp_version 0 method_mask (nil)
> ibwarn: [8822] umad_register: fd 3 registered to use agent 1 qp 0
> ibwarn: [8822] umad_register: fd 3 mgmt_class 3 mgmt_version 2 rmpp_version
> 1 method_mask (nil)
> ibwarn: [8822] umad_register: fd 3 registered to use agent 2 qp 1
> ibwarn: [8822] umad_register: fd 3 mgmt_class 10 mgmt_version 1 rmpp_version
> 0 method_mask (nil)
> ibwarn: [8822] umad_register: fd 3 registered to use agent 3 qp 1
> ibwarn: [8822] ib_vendor_call: route Lid 2 data 0x7fff19b91c60
> ibwarn: [8822] ib_vendor_call: class 0xa method 0x1 attr 0x1 mod 0x0 datasz
> 232 off 24 res_ex 1
> ibwarn: [8822] mad_rpc_rmpp: rmpp (nil) data 0x7fff19b91c60
> ibwarn: [8822] umad_set_addr: umad 0x7fff19b912b0 dlid 2 dqp 1 sl 0, qkey
> 80010000
> ibwarn: [8822] _do_madrpc: >>> sending: len 256 pktsz 320
> send buf
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0001 8001 0000 0002 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 010a 0101 0000 0000 0960 1a1a 25a8 c4c0
> 0001 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> ibwarn: [8822] umad_send: fd 3 agentid 3 umad 0x7fff19b912b0 timeout 1000
> ibwarn: [8822] umad_recv: fd 3 umad 0x7fff19b90eb0 timeout 1000
> ibwarn: [8822] umad_recv: mad received by agent 3 length 320
> ibwarn: [8822] _do_madrpc: rcv buf:
> rcv buf
> 010a 0181 000c 0000 0000 009f 25a8 c4c0
> 0001 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> 0000 0000 0000 0000 0000 0000 0000 0000
> ibwarn: [8822] mad_rpc_rmpp: MAD completed with error status 0xc; dport (Lid
> 2)
> vendstat: iberror: [pid 8822] main: failed: classportinfo query

This looks like a timeout error on the ClassPortInfo attribute in the
vendor class Mellanox uses. So either that class is not implemented in
the firmware or this (required) attribute is not implemented. Can you
contact Mellanox about this ?

Also, can you try the following experiment:

rebuild vendstat with #if 0'ing out the following lines:

        memset(&buf, 0, sizeof(buf));
        /* vendor ClassPortInfo is required attribute if class supported */
        call.attrid = CLASS_PORT_INFO;
        if (!ib_vendor_call(&buf, &portid, &call))
                IBERROR("classportinfo query");

I'm not sure whether this tool should support IS-IV or not. Currently,
as the man page says, it's intended for IS-III only. It may be that
the current firmware does not yet support this but will sometime in
the future.

-- Hal

> Cheers,
>
> Vincent
>