Re: [kvm-unit-tests PATCH v9 12/12] s390x: css: ssch/tsch with sense and interrupt

[Cornelia Huck <cohuck@redhat.com>]

On Wed, 17 Jun 2020 13:55:52 +0200
Pierre Morel <pmorel@linux.ibm.com> wrote:

> On 2020-06-17 11:54, Cornelia Huck wrote:
> > On Mon, 15 Jun 2020 11:32:01 +0200
> > Pierre Morel <pmorel@linux.ibm.com> wrote:

(...)

> >> +int start_subchannel(unsigned int sid, int code, void *data, int count,
> >> +		     unsigned char flags)
> >> +{
> >> +	int cc;
> >> +	struct ccw1 *ccw = &unique_ccw;  
> > 
> > Hm... it might better to call this function "start_single_ccw" or
> > something like that.  
> 
> You are right.
> I will rework this.
> What about differentiating this badly named "start_subchannel()" into:
> 
> ccw_setup_ccw(ccw, data, cnt, flgs);
> ccw_setup_orb(orb, ccw, flgs)
> ccw_start_request(schid, orb);
> 
> would be much clearer I think.

Not sure about ccw_setup_ccw; might get a bit non-obvious if you're
trying to build a chain.

Let's see how this turns out.

(...)

> I will rework this.
> 
> - rework the start_subchannel()
> - rework the read_len() if we ever need this

I think checking the count after the request concluded is actually a
good idea. In the future, we could also add a check that it matches the
requested length for a request where SLI was not specified.

> 
> Also thinking to put the irq_io routine inside the library, it will be 
> reused by other tests.

Yes, that probably makes sense as well.