Re: linux-next: Tree for April 1 (virtio_blk warning)

[Randy Dunlap <randy.dunlap@oracle.com>]

On Tue, 06 Apr 2010 14:15:32 -0400 john cooper wrote:

> Rusty Russell wrote:
> > On Fri, 2 Apr 2010 02:43:12 am Randy Dunlap wrote:
> >> On Thu, 1 Apr 2010 17:23:17 +1100 Stephen Rothwell wrote:
> >>
> >>> Hi all,
> >>>
> >>> Changes since 20100331:
> >>
> >> drivers/block/virtio_blk.c:228:13: warning: multi-character character constant
> >>
> >> due to:
> >>
> >> 	if (cmd == 'VBID') {
> > 
> > John?  That looks suspiciously like untested code.
> 
> It does work as advertised although gcc is obliged to squawk
> at multi-byte character constants due to portability concerns.

Thanks.  I wasn't aware of that gcc extension.

> Note I'd intended that only as an example for the benefit of
> completeness in the associated patch set, and to stick out
> like a sore thumb of sorts.  Marc Haber had suggested exposing
> the id string to the guest userland via /sys which is what I
> had expected to displace the loose ioctl example above.  In
> fact that discussion is what prodded revisiting this issue in
> the first place.
> 
> Unsure whether Marc (cc'ed here) is still planning to pursue
> that interface so unless I hear something to the contrary in
> the next day or so I'll package up a suitable ioctl interface
> and forward a patch.  It doesn't hurt to provide ioctl access
> to the info in addition to exposing it via /sys, but I'd
> hoped the latter would overshadow the need to do so.

---
~Randy