On Tue, 2019-03-26 at 16:20 +0000, Daniel P. Berrangé wrote: > On Tue, Mar 26, 2019 at 05:06:15PM +0100, Andrea Bolognani wrote: > > + # Try to integrate gracefully with downstream packages by running QEMU > > + # processes under the same user and group they would > > + case $(grep ^ID= /etc/os-release 2>/dev/null) in > > + *arch*) > > + default_qemu_user=nobody > > + default_qemu_group=nobody > > + ;; > > + *centos*|*fedora*|*gentoo*|*rhel*|*suse*) > > + default_qemu_user=qemu > > + default_qemu_group=qemu > > + ;; > > + *debian*) > > + default_qemu_user=libvirt-qemu > > + default_qemu_group=libvirt-qemu > > + ;; > > + *ubuntu*) > > + default_qemu_user=libvirt-qemu > > + default_qemu_group=kvm > > + ;; > > Or this needs to gain... > > *) > default_qemu_user=root > default_qemu_group=root > > > + if ! getent passwd "$default_qemu_user" >/dev/null 2>&1 || \ > > + ! getent group "$default_qemu_group" >/dev/null 2>&1; then > > otherwise these are passing an empty string to getent which > feels dubious. Even if empty string works I wouldn't want to > rely on that. A (quoted) empty string work exactly the way you'd expect it to, in that it will make the command fail every single time. But sure, I can do the more obviously correct thing and add the default case above if you prefer that :) > > + default_qemu_user=root > > + default_qemu_group=root > > We ought to issue a warning if the expected user/group doesn't > actually exist. You got it. v3 coming right up. -- Andrea Bolognani / Red Hat / Virtualization -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list
