Re: [Qemu-devel] [Qemu-ppc] [PATCH v2] spapr: quantify error messages regarding capability settings

[David Gibson <david@gibson.dropbear.id.au>]

[-- Attachment #1: Type: text/plain, Size: 3895 bytes --]

On Fri, Aug 02, 2019 at 11:03:48AM +1000, Daniel Black wrote:
> On Thu, 1 Aug 2019 12:41:59 +0200
> Greg Kurz <groug@kaod.org> wrote:
> 
> > On Thu,  1 Aug 2019 13:38:19 +1000
> > Daniel Black <daniel@linux.ibm.com> wrote:
> > 
> > > Its not immediately obvious how cap-X=Y setting need to be applied
> > > to the command line so, for spapr capability error messages, this
> > > has been clarified to:
> > > 
> ...
> > > index bbb001f84a..1c0222a081 100644
> > > --- a/hw/ppc/spapr_caps.c
> > > +++ b/hw/ppc/spapr_caps.c
> > > @@ -37,6 +37,8 @@
> > >  
> > >  #include "hw/ppc/spapr.h"
> > >  
> > > +#define CAPABILITY_ERROR(X) "appending -machine " X  
> > 
> > I would make that:
> > 
> > #define CAPABILITY_HINT() "try appending -machine " X
> > 
> > because it is really an hint for the user, not an
> > error,
> 
> Works for me. At the lowest layer it is a hint.

Oh.. of course it is.  Which means we should be using the
error_append_hint() system that's for exactly this sort of
information.

Sorry I didn't think of that earlier.

> 
> > and all original strings have "try",
> 
> True.
> 
> > except...
> 
> 
> > > @@ -249,11 +255,13 @@ static void
> > > cap_safe_cache_apply(SpaprMachineState *spapr, uint8_t val, if
> > > (tcg_enabled() && val) { /* TCG only supports broken, allow other
> > > values and print a warning */ error_setg(&local_err,
> > > -                   "TCG doesn't support requested feature,
> > > cap-cfpc=%s",
> > > +                   "TCG doesn't support requested feature, "
> > > +                   CAPABILITY_ERROR("cap-cfpc=%s"),  
> > 
> > ... this one, but it doesn't look like a hint to me. It just tells
> > which is the unsupported cap.
> 
> This is one of 3 that local_error (commit
> 006e9d3618698eeef2f3e07628d22cb6f5c2a039) - intentionally just a
> warning and to TLDR the commit/Suraj conversation; defaults apply
> to all machine types; hardware security measures don't make sense in
> TCG; hence warning.
> 
> For every function with CAPABILITY_[ERROR|HINT] its called by
> spapr_caps_apply, has its errp as &error_fatal (intentionally - spoke
> to Suraj - migrations to machines without capabilities need to fail and
> defaults (kvm) should be secure unless explicitly disabled).
> 
> > >                     cap_cfpc_possible.vals[val]);
> > >      } else if (kvm_enabled() && (val > kvm_val)) {
> > >          error_setg(errp,
> > > -"Requested safe cache capability level not supported by kvm, try
> > > cap-cfpc=%s", +"Requested safe cache capability level not supported
> > > by kvm, try "
> > > +                   CAPABILITY_ERROR("cap-cfpc=%s"),
> > >                     cap_cfpc_possible.vals[kvm_val]);  
> > 
> > Also, we have a dedicated API for hints, which are only printed under
> > the monitor but ignored under QMP.
> 
> Ok.
>  
> > Not sure why it isn't used here but it should be something like:
> 
> If error_append_hint should be used for fatal errors (all that use
> errp), then this patten should be applied further to
> CAPABILITY_[HINT|ERROR] functions.
> 
> If error_append_hint needs to apply to warnings
> cap_[cfpc/sbbc/ibs]_apply functions need to use it.
> 
> Would I be right in I'm assuming that the below pattern needs to apply
> to both of these cases?
> 
> >         error_setg(errp, 
> >                    "Requested safe cache capability level not
> > supported by kvm");
> >         error_append_hint(errp,
> > CAPABILITY_HINT("cap-cfpc=%s") "\n", cap_cfpc_possible.vals[kvm_val]);
> 
> This is going a little beyond the scope of fixing a message, ok, but
> lets not extend the scope too much more.
> 

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]