On Fri, May 10, 2019 at 02:37:41PM +0200, Laurent Vivier wrote: > On 10/05/2019 14:27, Markus Armbruster wrote: > > Laurent Vivier <lviv...@redhat.com> writes: > > The new rng-builtin is considerably simpler than both rng-random and > > rng-egd. Moreover, it just works, whereas rng-random is limited to > > CONFIG_POSIX, and rng-egd needs egd running (which I suspect basically > > nobody does). Have we considered deprecating these two backends in > > favor of rng-builtin? > > I have several bugzilla involving these backends: as there are blocking, the > virtio-rng device in the guest can hang, or crash during hot-unplug. From my > point of view, life would be easier without them...
Are you sure about that ? The EGD impl looks like it is requesting entropy in an async manner. Any problem with rng-random would also affect rng-builtin, as depending on platform / build options, rng-builtin may just use /dev/urandom directly. It should only block with /dev/random really and that's only with Linux's impl of /dev/random - some OS effectively have /dev/random behave identically to /dev/urandom. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
