Re: [Xen-devel] [PATCH v2] xen: Implement hypercall for tracing of program counters

[Wei Liu <wei.liu2@xxxxxxxxxx>]

On Mon, Jul 31, 2017 at 11:13:33AM +0100, Julien Grall wrote:
> 
> 
> On 31/07/17 10:56, Wei Liu wrote:
> > On Mon, Jul 31, 2017 at 09:22:35AM +0100, Julien Grall wrote:
> > > Hi,
> > > 
> > > On 07/28/2017 04:37 PM, Wei Liu wrote:
> > > > On Wed, Jul 26, 2017 at 12:43:45PM +0200, Felix Schmoll wrote:
> > > > > +        d->tracing_buffer = NULL;
> > > > > +
> > > > > +        if ( copy_to_guest(buf, temp, d->tracing_buffer_pos) )
> > > > > +            ret = -EFAULT;
> > > > > +
> > > > > +        xfree(temp);
> > > > > +
> > > > > +        ret = d->tracing_buffer_pos;
> > > > > +        break;
> > > > > +    }
> > > > > +
> > > > > +    default:
> > > > > +        ret = -ENOSYS;
> > > > 
> > > > EINVAL
> > > 
> > > Should not it be -EOPNOTSUPP to match return error when CONFIG_TRACE_PC is
> > > not?
> > 
> > AIUI EOPNOTSUPP means "This is a valid operation but I am not configured
> > to support it" while EINVAL means "This is an invalid value
> > (operation)".
> 
> Fair enough. However, you impose the caller to check -EINVAL and -EOPNOTSUPP
> in order to know if an operation can be done.
> 
> I first thought all the other hypercalls use -EOPNOTSUPP, but in face they
> use -ENOSYS. It would be to be consistent with the rest rather than
> reinventing our own.
> 

Oh, if that is already an established convention, I think using ENOSYS
is fine. Felix, please keep it as ENOSYS.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel