Re: [PATCH 2/3] x86/traps: Print non-canonical address on #GP

[Sean Christopherson <sean.j.christopherson@intel.com>]

On Thu, Nov 14, 2019 at 10:41:06AM -0800, Andy Lutomirski wrote:
> On Thu, Nov 14, 2019 at 10:20 AM Sean Christopherson
> <sean.j.christopherson@intel.com> wrote:
> >
> > On Thu, Nov 14, 2019 at 10:00:35AM -0800, Andy Lutomirski wrote:
> > > On Thu, Nov 14, 2019 at 9:46 AM Sean Christopherson
> > > <sean.j.christopherson@intel.com> wrote:
> > > > > +     /*
> > > > > +      * For the user half, check against TASK_SIZE_MAX; this way, if the
> > > > > +      * access crosses the canonical address boundary, we don't miss it.
> > > > > +      */
> > > > > +     if (addr_ref <= TASK_SIZE_MAX)
> > > >
> > > > Any objection to open coding the upper bound instead of using
> > > > TASK_SIZE_MASK to make the threshold more obvious?
> > > >
> > > > > +             return;
> > > > > +
> > > > > +     pr_alert("dereferencing non-canonical address 0x%016lx\n", addr_ref);
> > > >
> > > > Printing the raw address will confuse users in the case where the access
> > > > straddles the lower canonical boundary.  Maybe combine this with open
> > > > coding the straddle case?  With a rough heuristic to hedge a bit for
> > > > instructions whose operand size isn't accurately reflected in opnd_bytes.
> > > >
> > > >         if (addr_ref > __VIRTUAL_MASK)
> > > >                 pr_alert("dereferencing non-canonical address 0x%016lx\n", addr_ref);
> > > >         else if ((addr_ref + insn->opnd_bytes - 1) > __VIRTUAL_MASK)
> > > >                 pr_alert("straddling non-canonical boundary 0x%016lx - 0x%016lx\n",
> > > >                          addr_ref, addr_ref + insn->opnd_bytes - 1);
> > > >         else if ((addr_ref + PAGE_SIZE - 1) > __VIRTUAL_MASK)
> > > >                 pr_alert("potentially straddling non-canonical boundary 0x%016lx - 0x%016lx\n",
> > > >                          addr_ref, addr_ref + PAGE_SIZE - 1);
> > >
> > > This is unnecessarily complicated, and I suspect that Jann had the
> > > right idea but just didn't quite explain it enough.  The secret here
> > > is that TASK_SIZE_MAX is a full page below the canonical boundary
> > > (thanks, Intel, for screwing up SYSRET), so, if we get #GP for an
> > > address above TASK_SIZE_MAX,
> >
> > Ya, I followed all that.  My point is that if "addr_ref + insn->opnd_bytes"
> > straddles the boundary then it's extremely likely the #GP is due to a
> > non-canonical access, i.e. the pr_alert() doesn't have to hedge (as much).
> 
> I suppose.  But I don't think we have a real epidemic of failed
> accesses to user memory between TASK_SIZE_MAX and the actual boundary
> that get #GP instead of #PF but fail for a reason other than
> non-canonicality :)

No argument there.

> I think we should just go back in time and fix x86_64 to either give
> #PF or at least give some useful page fault for a non-canonical
> address. The only difficulties I'm aware of is that Intel CPUs would
> either need to be redesigned better or would have slightly odd
> semantics for jumps to non-canonical addresses -- #PF in Intel's model
> of "RIP literally *can't* have a non-canonical value" would be a bit
> strange.  Also, my time machine is out of commission.

If you happen to fix your time machine, just go back a bit further and
change protected mode to push the faulting address onto the stack.