Hi, as I understand that udev workaround does not help with forwarded host usb devices? Is it possible manually reduce poll interval for usb device to reduce cpu usage?
2013/2/19 Michael Tokarev <m...@tls.msk.ru>: > Control: retitle -1 kvm uses cpu when guest system with USB devices is idle > Control: severity -1 minor > Control: tag -1 + confirmed wontfix > > 19.02.2013 11:39, Alexander Kudrevatykh wrote: >> Package: qemu-kvm >> Version: 1.1.2+dfsg-5 >> Severity: normal >> >> kvm with debian stable guest uses about 10% cpu on idle >> >> there is kvm command line from ps output (started from libvirt) >> /usr/bin/kvm -S -M pc-1.1 -enable-kvm -m 1024 -smp >> 1,sockets=1,cores=1,threads=1 -name cups -uuid >> ecfcb338-bd8d-cb88-3f02-7a1e52b13609 -nodefconfig -nodefaults -chardev >> socket,id=charmonitor,path=/var/lib/libvirt/qemu/cups.monitor,server,nowait >> -mon chardev=charmonitor,id=monitor,mode=control -rtc base=utc -no-shutdown >> -kernel /var/lib/libvirt/qemu/dkb/vmlinuz-2.6.32-5-amd64 -initrd >> /var/lib/libvirt/qemu/dkb/initrd.img-2.6.32-5-amd64 -append >> root=UUID=398df502-1157-440c-ba88-6ec93d68b490 -device >> ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x7 -device >> ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,addr=0x8 -device >> ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x9 -device >> ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0xa -drive >> file=/dev/second_volume_group/cups,if=none,id=drive-virtio-disk0,format=raw >> -device >> virtio-blk-pci,scsi=off,bus=pci.0,addr=0x5,drive=drive-virtio-disk0,id=virtio-disk0,bootindex=1 >> -netdev tap,fd=29,id=hostnet0 -device virtio-net-pci,netd > e > v >> =hostnet0,id=net0,mac=52:54:00:f0:b4:30,bus=pci.0,addr=0x3 -chardev >> pty,id=charserial0 -device isa-serial,chardev=charserial0,id=serial0 -vnc >> 127.0.0.1:0 -vga cirrus -device usb-host,hostbus=1,hostaddr=7,id=hostdev0 >> -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x6 > > This is a well-known effect of using USB devices. USB spec requires > for the machine to poll devices for (in)activity. So the guest pokes > USB devices frequently -- for example for (virtual) mouse/tablet it > does that 250 times per second. If we lower this value, guest mouse > will be sluggish. > > There's a more or less workaround in later linux guests, an udev rule > that allows qemu-emulated mouse/tablet/keyboard to enter (vurtual) > sleep mode when not doing anything (the same as for real hardware), -- > this takes care of CPU usage, but it makes mouse react slowly to the > first event after some idle time, ie, when it got a chance to enter > that sleep mode. > > Another workaround is to remove USB devices entirely (start the guest > without any USB device but with USB bus), and (again, virtually) hot- > plug any needed devices when the guest is actually in use (and hot- > remove again when it isn't in use). > > This is how USB works, there's almost nothing else we can do about > it. > > Tagging this bugreport as "wontfix" (and minor, since the bad effect > is really minimal) instead of closing it right away - just to prevent > other bugreports to be filed about this very issue. > > Thanks, > > /mjt -- С уважением, Кудреватых Александр -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
