Bug 1072001 - SRIOV, VF failed to be initializaed
SRIOV, VF failed to be initializaed
Status: RESOLVED FIXED
Classification: openSUSE
Product: openSUSE Tumbleweed
Classification: openSUSE
Component: KVM
Current
x86-64 Other
: P5 - None : Minor (vote)
: ---
Assigned To: Liang Yan
E-mail List
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2017-12-08 19:08 UTC by Liang Yan
Modified: 2017-12-08 19:20 UTC (History)
0 users

See Also:
Found By: ---
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Liang Yan 2017-12-08 19:08:40 UTC
2017-12-07T18:47:30.255226-05:00 linux-dv96 kernel: [    3.980332] mlx4_core: Mellanox ConnectX core driver v4.0-0
2017-12-07T18:47:30.255228-05:00 linux-dv96 kernel: [    3.980347] mlx4_core: Initializing 0000:00:07.0
2017-12-07T18:47:30.255229-05:00 linux-dv96 kernel: [    3.993346] mlx4_core 0000:00:07.0: Detected virtual function - running in slave mode
2017-12-07T18:47:30.255229-05:00 linux-dv96 kernel: [    4.008190] XFS (vda3): Mounting V5 Filesystem
2017-12-07T18:47:30.255230-05:00 linux-dv96 kernel: [    4.018357] XFS (vda3): Ending clean mount
2017-12-07T18:47:30.285960-05:00 linux-dv96 kernel: [    4.299731] No iBFT detected.
2017-12-07T18:48:00.010059-05:00 linux-dv96 kernel: [   34.020111] mlx4_core 0000:00:07.0: recovering from previously mis-behaved VM
2017-12-07T18:48:06.839133-05:00 linux-dv96 kernel: [   40.853044] random: crng init done
2017-12-07T18:48:30.042710-05:00 linux-dv96 kernel: [   64.052119] mlx4_core 0000:00:07.0: Communication channel is offline.
2017-12-07T18:48:30.042739-05:00 linux-dv96 kernel: [   64.052122] mlx4_core 0000:00:07.0: PF is not responsive, skipping initialization
2017-12-07T18:48:30.042743-05:00 linux-dv96 kernel: [   64.052125] mlx4_core 0000:00:07.0: Failed to initialize slave
2017-12-07T18:48:30.042745-05:00 linux-dv96 kernel: [   64.055676] mlx4_core: probe of 0000:00:07.0 failed with error -5
Comment 1 Liang Yan 2017-12-08 19:13:59 UTC
when trying to use vf inside a vm, could not initialize it because no response from PF.
Comment 2 Liang Yan 2017-12-08 19:17:26 UTC
Turns out it is related to a kernel parameter pci=nocrs


pci=nocrs

Discard pci ACPI information. May fix boot problems.


VF needs PF acpi information during initialization
Comment 3 Liang Yan 2017-12-08 19:20:23 UTC
After remove pci=nocrs from command line, make it work

linux-dv96:~ # dmesg | grep mlx
[    8.315629] mlx4_core: Mellanox ConnectX core driver v4.0-0
[    8.315651] mlx4_core: Initializing 0000:00:07.0
[    8.318090] mlx4_core 0000:00:07.0: Detected virtual function - running in slave mode
[    8.318160] mlx4_core 0000:00:07.0: Sending reset
[    8.318350] mlx4_core 0000:00:07.0: Sending vhcr0
[    8.323503] mlx4_core 0000:00:07.0: HCA minimum page size:512
[    8.325388] mlx4_core 0000:00:07.0: Timestamping is not supported in slave mode
[    8.435709] mlx4_en: Mellanox ConnectX HCA Ethernet driver v4.0-0
[    8.527556] <mlx4_ib> mlx4_ib_add: mlx4_ib: Mellanox ConnectX InfiniBand driver v4.0-0
[    8.544666] <mlx4_ib> mlx4_ib_add: counter index 2 for port 1 allocated 0
[    8.544668] <mlx4_ib> mlx4_ib_add: counter index 3 for port 2 allocated 0
[    8.760133] mlx4_core 0000:00:07.0: mlx4_ib: multi-function enabled
[    8.760136] mlx4_core 0000:00:07.0: mlx4_ib: operating in qp1 tunnel mode