[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Re: [RFC PATCH 15/33] move segment checks to subarch

On Tue, 2006-07-18 at 12:25 -0700, Chris Wright wrote:
> * Rusty Russell (rusty@xxxxxxxxxxxxxxx) wrote:
> > On Tue, 2006-07-18 at 00:00 -0700, Chris Wright wrote:
> > > plain text document attachment (i386-segments)
> > > We allow for the fact that the guest kernel may not run in ring 0.
> > > This requires some abstraction in a few places when setting %cs or
> > > checking privilege level (user vs kernel).
> > 
> > Zach had an alternate patch for this, which didn't assume the kernel ran
> > in a compile-time known ring, but is otherwise very similar.  I've put
> > it below for discussion (but Zach now tells me the asm parts are not
> > required: Zach, can you mod this patch and comment?).
> This patch also doesn't have a compile time known ring, it's using
> get_kernel_cs() because the Xen method for booting native is dynamic and
> would resolve to ring 0 in that case (XENFEAT_supervisor_mode_kernel).

I was referring to the different ways the two patches figure out whether
we're in user mode:

 static inline int user_mode(struct pt_regs *regs)
       return (regs->xcs & USER_MODE_MASK) != 0;

Where you have for native:
        #define USER_MODE_MASK 3
vs Xen:
        #define USER_MODE_MASK 2

Zach's patch does this:

 static inline int user_mode(struct pt_regs *regs)
        return (regs->xcs & SEGMENT_RPL_MASK) == 3;

I'm no x86pert, but the latter seems more generic to me (user mode is
ring 3, vs. usermode is anything >= 2).  Perhaps they are in fact

Help! Save Australia from the worst of the DMCA: http://linux.org.au/law

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.