[vfio-users] Problem with vfio-pci device type and memory > 10G (or 8G)

Anjali Kulkarni anjali at juniper.net
Tue Dec 13 23:01:09 UTC 2016


To add, this works perfectly fine (only change from non-working command is memory is changed from 12288 to 4096):


/usr/bin/qemu-system-x86_64 -drive if=ide,file=/home/me/myimage.img --enable-kvm -m 4096 -smp 2,sockets=1,cores=1,threads=2 -device vfio-pci,host=03:10.0,id=hostdev0,bus=pci.0,addr=0x6 -nographic

From: Anjali Kulkarni <anjali at juniper.net<mailto:anjali at juniper.net>>
Date: Tuesday, December 13, 2016 at 2:58 PM
To: "vfio-users at redhat.com<mailto:vfio-users at redhat.com>" <vfio-users at redhat.com<mailto:vfio-users at redhat.com>>
Subject: Problem with vfio-pci device type and memory > 10G (or 8G)

Hi,

I am observing that with memory of 10G or 12G, with a vfio-pci device type, qemu hangs and system becomes slow and sluggish, and eventually you need to kill the hung qemu process which never returns.
For eg.:

/usr/bin/qemu-system-x86_64 -drive if=ide,file=/home/me/myimage.img --enable-kvm -m 12288 -smp 2,sockets=1,cores=1,threads=2 -device vfio-pci,host=03:10.0,id=hostdev0,bus=pci.0,addr=0x6 -nographic


Whereas with 8G or less of memory, it works fine.


If I remove -device vfio-pci..., everything works fine.

Any idea if someone has seen this issue or how to debug or what the problem could be? Looks like there is a memory limitation somewhere.


Anjali


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/vfio-users/attachments/20161213/6c06b96f/attachment.htm>


More information about the vfio-users mailing list