NVMe Virtual Controller with SPDK

[subbu_work]

Hi All,
As part of our product unit testing, I thought of using the VirtualBox NVMe Controller and the NVMe NS as a target device. The VM guest OS as Linux Fedora.
Our product uses the SPDK uio_pci_generic driver to expose the PCIe config to user space and using SPDK user space driver we run the IO to the device.
The issue is controller hangs and throws timed out error, when SPDK probe for the NVMe Controller (as part of this SPDK disable and enable the controller using the UIO driver).
below are the erro log from SPDK. Please let me know if there is anyone facing this issue or is there any fix available for this?
nvme_ctrlr.c: 979:nvme_ctrlr_process_init: ***ERROR*** Initialization timed out in state 1
nvme_ctrlr.c: 342:nvme_ctrlr_shutdown: ***ERROR*** did not shutdown within 5 seconds

Please find the below detailed log

==========FULL-LOG-BELOW=========

Starting Intel(R) DPDK initialization ...
[ DPDK EAL parameters: test_fw -c 1 -n 4 -m 1024 --master-lcore=0 --file-prefix=rte0 --proc-type=auto ]

EAL: Detected 2 lcore(s)

EAL: Auto-detected process type: PRIMARY

EAL: Probing VFIO support...
done.
Occupied cpu core mask is 0x1
Occupied cpu socket mask is 0x1

level:5781:399491539:6:/vagrant/src/infra/infra.c:infra_init:31:entering
level:5781:399603212:6:/vagrant/src/infra/mem.c:mem_init:62:Initializing Memory for Controllers
level:5781:400158114:6:/vagrant/src/infra/infra.c:infra_init:35:leaving
level:5781:400563570:7:/vagrant/src/boss/sbd/sbd.c:sbd_init:44:entering
level:5781:401294494:6:/vagrant/src/boss/sbd/pbd_nvme.c:pbd_nvme_init:550:Initializing NVMe Controllers ...
EAL: PCI device 0000:00:0e.0 on NUMA socket 0
EAL: probe driver: 80ee:4e56 SPDK NVMe
level:5781:404243951:7:/vagrant/src/boss/sbd/pbd_nvme.c:pbd_nvme_probe_cb:257:entering
level:5781:404264679:3:/vagrant/src/boss/sbd/pbd_nvme.c:pbd_nvme_probe_cb:264:Attaching to 0000:00:0e.00
level:5781:404270008:7:/vagrant/src/boss/sbd/pbd_nvme.c:pbd_nvme_probe_cb:265:leaving
nvme_ctrlr.c: 979:nvme_ctrlr_process_init: ***ERROR*** Initialization timed out in state 1
nvme_ctrlr.c: 342:nvme_ctrlr_shutdown: ***ERROR*** did not shutdown within 5 seconds


Thanks & Regards,
Subbu

[michaln]

Is anyone facing this issue -- that's a "no" with 99.99% confidence. What you're running there sounds just a bit too exotic.

If you can provide a reproduction scenario, someone can look into this next year. If you can't, then the problem may or may not get fixed as part of some other fix.

[aeichner]

I was able to reproduce the issues using the SPDK examples, should be fixed with the next maintenance release. Thanks for the report but please open a ticket in our public bug tracker the next time because it is more likely to get attention there. Happy new year!

[subbu_work]

Thanks a lot folks!!! and Happy new year.

Regards,
Subbu