On Thu, 2012-06-28 at 21:48 +0200, intrigeri wrote: > Hi, > > firstname.lastname@example.org wrote (27 Jun 2012 11:00:22 GMT) : > > On Wed, Jun 27, 2012 at 04:32:31AM +0100, Ben Hutchings wrote: > > >> Yes, but I think it would make more sense to emulate a USB storage > >> device in qemu rather than the host kernel. > > I do agree. > > bertagaz and I have spent a bit more time testing and comparing the > available options. Our results are summed up there: > https://tails.boum.org/todo/automated_builds_and_tests/USB/ > > tl;dr --> as far as Wheezy is concerned: > * qemu-kvm emulates just fine a USB 2.0 mass storage device, and > knows how to boot from it; personally, I'd rather use that than > a dedicated kernel module. > * with qemu-kvm on the command-line: no need for an additional > kernel module > * with a libvirt stack: a missing interface in some abstraction > layer makes it a pain to use the qemu-kvm USB emulation of > removable mass storage devices. > > We are going to request the missing interface to upstream libvirt, > but it's unlikely the result thereof is ready in time for Wheezy. > > So, with my Tails developer hat, we would be happy to use > g_mass_storage at least from now to the Wheezy+1 release. [...] So you're asking for extra kernel modules to work around a libvirt deficiency, not even a missing feature in qemu? I think this is ridiculous. Ben. -- Ben Hutchings Lowery's Law: If it jams, force it. If it breaks, it needed replacing anyway.
Description: This is a digitally signed message part