Re: [PATCH v2 6/7] flask: drop dead compat translation code

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Wed, Jul 15, 2020 at 08:42:44AM +0200, Jan Beulich wrote:
> On 14.07.2020 16:58, Roger Pau Monné wrote:
> > On Wed, Jul 01, 2020 at 12:28:07PM +0200, Jan Beulich wrote:
> >> Translation macros aren't needed at all (or else a devicetree_label
> >> entry would have been missing), and userlist has been removed quite some
> >> time ago.
> >>
> >> No functional change.
> >>
> >> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> >>
> >> --- a/xen/include/xlat.lst
> >> +++ b/xen/include/xlat.lst
> >> @@ -148,14 +148,11 @@
> >>  ? xenoprof_init                   xenoprof.h
> >>  ? xenoprof_passive                xenoprof.h
> >>  ? flask_access                    xsm/flask_op.h
> >> -! flask_boolean                   xsm/flask_op.h
> >>  ? flask_cache_stats               xsm/flask_op.h
> >>  ? flask_hash_stats                xsm/flask_op.h
> >> -! flask_load                      xsm/flask_op.h
> >>  ? flask_ocontext                  xsm/flask_op.h
> >>  ? flask_peersid                   xsm/flask_op.h
> >>  ? flask_relabel                   xsm/flask_op.h
> >>  ? flask_setavc_threshold          xsm/flask_op.h
> >>  ? flask_setenforce                xsm/flask_op.h
> >> -! flask_sid_context               xsm/flask_op.h
> >>  ? flask_transition                xsm/flask_op.h
> > 
> > Shouldn't those become checks then?
> 
> No, checking will never succeed for structures containing
> XEN_GUEST_HANDLE(). But there's no point in generating xlat macros
> when they're never used. There are two fundamentally different
> strategies for handling the compat hypercalls: One is to wrap a
> translation layer around the native hypercall. That's where the
> xlat macros come into play. The other, used here, is to compile
> the entire hypercall function a second time, arranging for the
> compat structures to get used in place of the native ones. There
> are no xlat macros involved here, all that's needed are correctly
> translated structures. (For completeness, x86's MCA hypercall
> uses yet another, quite adhoc strategy for handling, but also not
> involving any xlat macro use. Hence the consideration there to
> possibly drop the respective lines from the file here.)

Thanks, I think this explanation is helpful and I wonder whether it
would be possible to have something along this lines in a file or as a
comment somewhere, maybe at the top of xlat.lst?

Also could you add a line to the commit message noting that flask code
doesn't use any of the translation macros because it follows a
different approach to compat handling?

IMO the compat code is complicated to understand, and it also seems to
be mostly undocumented.

For the patch:

Reviewed-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>

Thanks.