Re: [PATCH 1/1] scsi: storvsc: Fix a build issue reported by kbuild test robot

[James Bottomley]

On Thu, 2016-01-28 at 19:07 +0000, KY Srinivasan wrote:
> 
> > -----Original Message-----
> > From: Olaf Hering [mailto:olaf@xxxxxxxxx]
> > Sent: Thursday, January 28, 2016 7:56 AM
> > To: KY Srinivasan <kys@xxxxxxxxxxxxx>
> > Cc: gregkh@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx;
> > devel@xxxxxxxxxxxxxxxxxxxxxx; ohering@xxxxxxxx;
> > jbottomley@xxxxxxxxxxxxx; hch@xxxxxxxxxxxxx; 
> > linux-scsi@xxxxxxxxxxxxxxx;
> > apw@xxxxxxxxxxxxx; vkuznets@xxxxxxxxxx; jasowang@xxxxxxxxxx;
> > martin.petersen@xxxxxxxxxx; hare@xxxxxxx
> > Subject: Re: [PATCH 1/1] scsi: storvsc: Fix a build issue reported
> > by kbuild test
> > robot
> > 
> > On Wed, Jan 27, K. Y. Srinivasan wrote:
> > 
> > > +	depends on SCSI_FC_ATTRS
> > 
> > I think 'depends' instead of 'select' will cause HYPERV_STORAGE to
> > disapepar during make oldconfig if SCSI_FC_ATTRS was not set
> > before.
> > Not sure what the policy of 'depends' vs.  'select' actually is. 
> >  If
> > SCSI_FC_ATTRS is supposed to be a library kind of thing then
> > 'select'
> > might be the correct way.
> 
> The current build issue is because the Hyper-V storage is configured
> to be built with
> the kernel while the SCSI_FC_AATRS is configured as a module. This
> patch fixes that issue.
> I too am not sure what the policy of using "depends" vs " select" is.
> James, what would be your recommendation here.

Oh, you don't care if FC_ATTRS are enabled, but if they are you need to
exclude the case where storvsc is built in but FC_ATTRS is a module? 
 This is the accepted way of doing it:

depends on m || SCSI_FC_ATTRS != m

James