Re: [PATCH v4 3/4] tools: add tools support for Intel CDP

[Ian Campbell <ian.campbell@citrix.com>]

On Thu, 2015-09-24 at 05:50 -0600, Jan Beulich wrote:
> > > > On 24.09.15 at 13:00, <ian.campbell@citrix.com> wrote:
> > On Thu, 2015-09-17 at 17:35 +0800, He Chen wrote:
> > > diff --git a/tools/libxl/libxl_psr.c b/tools/libxl/libxl_psr.c
> > > index 3378239..62963cf 100644
> > > --- a/tools/libxl/libxl_psr.c
> > > +++ b/tools/libxl/libxl_psr.c
> > > @@ -87,6 +87,9 @@ static void libxl__psr_cat_log_err_msg(libxl__gc
> > > *gc,
> > > int err)
> > >      case EEXIST:
> > >          msg = "The same CBM is already set to this domain";
> > >          break;
> > > +    case EINVAL:
> > > +        msg = "Unable to set code or data CBM when CDP is disabled";
> > > +        break;
> > 
> > These overloading of the errno values are getting a bit thinly
> > stretched.
> > The more so that EINVAL has a widely used more generic meaning.
> > 
> > Hypervisor maintainers, what is your opinion of this?
> > 
> > Since this is a sysctl I suppose we could consider adding a new PSR
> > specific error type with appropriate codes?
> 
> I'd prefer using recognizable -E... values;

_If_ the -E values somehow map sensibly onto the PSR errors, otherwise they
aren't really recognisable any more.

>  a specific error type
> to me seems to go too far. Surely out of the several dozen
> possibilities a handful of not-so-common ones can be picked?

I was thinking in particular EINVAL was not in the not-so-common bracket.

The current code already uses 9 values FWIW.

Ian.