Dimitry Golubovsky wrote: > Anthony, > > On 6/13/07, Anthony Liguori <[EMAIL PROTECTED]> wrote: > >> One thing to consider is that if a userspace process can create KVM >> guests, they are capable of pinning large quantities of physical >> memory. This could be used as a DoS attack so consider VM creation a >> privileged operation. > > No, that's not what is intended. I was asking about possibility to run > KVM at users privileges after some necessary actions have been > completed, and tried to compile a list of such actions. > > That is, > > - adjust RTC (I just added this to the system startup script)
You can also just run guests with -no-rtc. > - create a tap > - add tap to the bridge (if bridging is used)/adjust iptables if no > bridging (another example in qemu wiki) You may want to look at VDE (http://vde.sf.net). One of it's purposes was to allow non-privileged users to access bridges. > - open /dev/kvm (as it has been found, group membership is sufficient > if group can write to /dev/kvm) > > After that, process privileges might be dropped to those of the user > who logged (ssh'd) in. Images of disk volumes and CDs may then be > assigned proper permissions, so users may be more flexible on what to > run, and regular Unix filesystem mechanisms will control access. Yup. > BTW if qemu_system_x86-64 runs at user privileges, can the memory > consumed be subject to whatever per-user limits that may be set > systemwide? I don't think so although as Avi mentioned, this will be addressed when the overcommit patches are merged. Regards, Anthony Liguori ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ kvm-devel mailing list kvm-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/kvm-devel
