Re: [PATCH 26/30] Lock down ftrace

[Jiri Kosina]

On Fri, 10 Nov 2017, David Howells wrote:

> > I fail to see how this fits into the secure boot security model, could you 
> > please explain?
> 
> The idea is to prevent cryptographic data for filesystems and other things
> from being read out of the kernel memory as well as to prevent unauthorised
> modification of kernel memory.

Then it would make sense to actually lock down dumping of registers / 
function arguments (kprobes can currently do that, ftrace eventually could 
as well I guess), but disabling the whole ftrace altogether seems like a 
totally unnecessary overkill.

> > Secure boot is about having a constant proof / verification that the code 
> > you're running in ring0 can be trusted (IOW is the one that has been 
> > signed and verified by the whole boot chain).
> > 
> > Checking execution patterns doesn't seem to fit at all.
> 
> I'll defer this question to Alexei since he suggested I needed to deal 
> with this too.

Thanks.

-- 
Jiri Kosina
SUSE Labs