Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 911928

Summary: infiniband fails to get UDI
Product: [Retired] Red Hat Hardware Certification Program Reporter: Brian Brock <bbrock>
Component: Test Suite (tests)Assignee: Greg Nichols <gnichols>
Status: CLOSED CURRENTRELEASE QA Contact: Red Hat Kernel QE team <kernel-qe>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 6.1CC: czhang, dledford, rlandry
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: v7-1.7.0-R9 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
: 915482 (view as bug list) Environment:
Last Closed: 2013-03-28 06:20:51 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 904887    
Attachments:
Description Flags
v7 results.xml with infiniband test results
none
infiniband patch to plan the test based on loaded modules, and log without UDI set none

Description Brian Brock 2013-02-16 20:04:06 UTC
v7  run -t infiniband fails, looks like udev doesn't give the device id like v7 expects.

v7-1.7.0-7.el7

can't seem to work around it with any set of options.

$ v7 plan -t infiniband
$ v7 run -t infiniband

running with -v <GID> and -i <GID> as returned by `ibstatus` didn't change the results.

expected PASS on infiniband tests

actual test output:

<output>
Test Parameters: DEBUG=off RUNMODE=normal UDI= DEVICE= OUTPUTFILE=/var/log/v7/runs/2/infiniband/output.log 
<output name="Log Hardware Info" description="log hardware info from udev">
Warning: fail to get UDI
<summary>FAIL</summary>
</output>
<output name="Log Status" description="log infiniband status">
Infiniband device 'mlx4_0' port 1 status:
        default gid:     fe80:0000:0000:0002:0002:c903:0031:7b11
        base lid:        0xa
        sm lid:          0x6
        state:           4: ACTIVE
        phys state:      5: LinkUp
        rate:            40 Gb/sec (4X QDR)
        link_layer:      InfiniBand

Infiniband device 'mlx4_0' port 2 status:
        default gid:     fe80:0000:0000:0000:0202:c9ff:fe31:7b11
        base lid:        0x0
        sm lid:          0x0
        state:           4: ACTIVE
        phys state:      5: LinkUp
        rate:            10 Gb/sec (1X QDR)
        link_layer:      Ethernet

CA 'mlx4_0'
        CA type: MT4099
        Number of ports: 2
        Firmware version: 2.11.500
        Hardware version: 0
        Node GUID: 0x0002c90300317b10
        System image GUID: 0x0002c90300317b13
        Port 1:
                State: Active
                Physical state: LinkUp
                Rate: 40
                Base lid: 10
                LMC: 0
                SM lid: 6
                Capability mask: 0x02514868
                Port GUID: 0x0002c90300317b11
                Link layer: InfiniBand
        Port 2:
                State: Active
                Physical state: LinkUp
                Rate: 10
                Base lid: 0
                LMC: 0
                SM lid: 0
                Capability mask: 0x00010000
                Port GUID: 0x0202c9fffe317b11
                Link layer: Ethernet
<summary>PASS</summary>
</output>
</output>

Comment 1 Greg Nichols 2013-02-18 14:17:21 UTC
Could you attach results.xml?

Comment 2 Brian Brock 2013-02-18 17:34:01 UTC
Created attachment 699009 [details]
v7 results.xml with infiniband test results

v7 results from runs including infiniband tests.

Comment 3 Greg Nichols 2013-02-18 19:57:29 UTC
Created attachment 699104 [details]
infiniband patch to plan the test based on loaded modules, and log without UDI set

Comment 7 Doug Ledford 2013-02-19 15:11:43 UTC
There are a number of cards that can be listed as type network and still be capable of RDMA (not necessarily InfiniBand, but that too).  All iWARP cards will be class network, any cards that are RoCE/IBoE capable but not InfiniBand capable will show as class network, and as you see here, the dual mode Ethernet/InfiniBand cards from Mellanox are class network.  Given that the cert should probably cover all RDMA hardware, of which InfiniBand is a subset, I would skip the UDI item and focus on RDMA capability.

One thing that is true, is that for any RDMA capable hardware to be up and running, ib_core must be loaded.  The next thing you know for sure is that if the output of ibstatus shows a device with type Ethernet, then it is either an iWARP or RoCE/IBoE interface (ibstat can show a normal Ethernet interface if there happens to be one on a card that also has an InfiniBand interface and we are not running the HPN add on channel on rhel6, ibstatus can not be fooled in this way).  I would build off that information and go from there.

Comment 11 Brian Brock 2013-02-25 17:28:14 UTC
fix verified in R11