[openib-general] Re: Send SA request over umad problem.

[Hal Rosenstock]

On Tue, 2005-08-02 at 10:33, Liran Sorani wrote:
> Hi ,
> I'm working on the SM group at Mellanox.
> While testing SM-gen2 on a loopback , I've encountered a basic problem
> trying to send an SA query (single mad) over osm_vendor (gen2).
> Trying to send the request using osm_vendor_send , passed succesfully
> , BUT got from the receiver (umad_recv) an error (in  an endless loop
> ): "No space left on device". The MAD request was simple GSI - SA
> request of ClassPortInfo , here are the details , I've truned on debug
> mode of vendor_lib and umad (marked in red the important lines in the
> log ):
> ...
> Aug 02 03:35:49 [401776C0] -> osm_vendor_send: [
> warn: [19219] umad_set_addr_net: umad 0x80810d0 dlid 1 dqp 1 sl, qkey
> 0
> warn: [19219] umad_dump: agent id 0 status 0 timeout 0
> warn: [19219] umad_addr_dump: qpn 1 qkey 0x80010000 lid 0x1 sl 0
> grh_present 0 gid_index 0 hop_limit 0 traffic_class 0 flow_label 0x0
> Gid 0x00000000000000000000000000000000
> Aug 02 03:35:49 [401776C0] -> osm_vendor_send: RMPP 0 length 256
> warn: [19219] umad_send: portid 0 agentid 0 umad 0x80810d0 timeout
> 1000
> Aug 02 03:35:49 [401776C0] -> osm_vendor_send: Completed Sending
> Request p_madw = 0x80807dc.
> Aug 02 03:35:49 [401776C0] -> osm_vendor_send: ]
> Aug 02 03:35:49 [401776C0] -> __osmv_send_sa_req: Waiting for async
> event.
> warn: [19219] umad_recv: read returned 356 > sizeof umad 56 + length
> 256 (No space left on device)
> Aug 02 03:35:49 [40D7EBB0] -> umad_receiver: recv error No space left
> on device
> warn: [19219] umad_recv: portid 0 umad 0x8080e28 timeout 4294967295
> warn: [19219] umad_recv: read returned 356 > sizeof umad 56 + length
> 256 (No space left on device)
> Aug 02 03:35:49 [40D7EBB0] -> umad_receiver: recv error No space left
> on device
> warn: [19219] umad_recv: portid 0 umad 0x8080e28 timeout 4294967295
> warn: [19219] umad_recv: read returned 356 > sizeof umad 56 + length
> 256 (No space left on device)
> Aug 02 03:35:49 [40D7EBB0] -> umad_receiver: recv error No space left
> on device
> warn: [19219] umad_recv: portid 0 umad 0x8080e28 timeout 4294967295
> warn: [19219] umad_recv: read returned 356 > sizeof umad 56 + length
> 256 (No space left on device)
> Aug 02 03:35:49 [40D7EBB0] -> umad_receiver: recv error No space left
> on device
> ...

What SM is this talking to ? What does the SA response look like ? I
have a theory as to what is going on. Just want to see if it is accurate
before I spend more time on it.

-- Hal