On Sat, 2015-04-04 at 17:26 -0400, Daniel Kahn Gillmor wrote:
> Control: affects 780818 lvm2 partman-base
>
> On Sat 2015-03-21 07:26:46 -0400, Bastian Blank wrote:
> > I don't think it is a bug in lvm to try reading a device.
> >
> >> lvm is a prime culprit of trying to scan this device, which is just not
> >> going to be helpful to users, since it won't contain a PV.
> >
> > How does it know that as a fact?
> >
> > Re-assigning to linux, as it allowes read request on devices not
> > supporting it.
>
> I understand that the issue might be caused by the kernel itself. I'm
> noting here that this bug affects both lvm2 (because it causes vgscan to
> hang for quite a while, at least) as well as causing the partition
> manager during d-i to take ages on any system that has this kind of
> device.
Ideally the driver would avoid doing whatever it is that results in a
hang, or would expose the RPMB only if it's really accessible. I
suspect that where access to the RPMB hangs this is because the system
firmware (BIOS/EFI) has intentionally disabled access before handing
over to the OS.
Hiding RPMBs seems to be a popular workaround, and I doubt that it will
be useful to access them under a Debian system or other free operating
system, so that's what I intend to do. If we need to change that later
we can add a module parameter to override this.
Ben.
--
Ben Hutchings
Power corrupts. Absolute power is kind of neat.
- John Lehman, Secretary of the US Navy 1981-1987
Attachment:
signature.asc
Description: This is a digitally signed message part