Re: [Qemu-devel] [Nbd] [PATCH 2/2] NBD proto: add GET_LBA_STATUS extension

From: Wouter Verhelst <w@uter.be>
To: Kevin Wolf <kwolf@redhat.com>
Cc: nbd-general@lists.sourceforge.net, qemu-devel@nongnu.org,
	Stefan Hajnoczi <stefanha@redhat.com>,
	Paolo Bonzini <pbonzini@redhat.com>,
	"Denis V. Lunev" <den@openvz.org>
Subject: Re: [Qemu-devel] [Nbd] [PATCH 2/2] NBD proto: add GET_LBA_STATUS extension
Date: Tue, 29 Mar 2016 11:53:54 +0200	[thread overview]
Message-ID: <20160329095354.GB6414@grep.be> (raw)
In-Reply-To: <20160329093835.GB4600@noname.redhat.com>

On Tue, Mar 29, 2016 at 11:38:35AM +0200, Kevin Wolf wrote:
> Am 24.03.2016 um 17:47 hat Wouter Verhelst geschrieben:
> > On Thu, Mar 24, 2016 at 05:07:47PM +0100, Kevin Wolf wrote:
> > > Am 24.03.2016 um 17:04 hat Eric Blake geschrieben:
> > > > On 03/24/2016 09:53 AM, Wouter Verhelst wrote:
> > > > > On Thu, Mar 24, 2016 at 04:33:42PM +0100, Paolo Bonzini wrote:
> > > > >> On 24/03/2016 16:25, Eric Blake wrote:
> > > > >>>> However, let's make these bits, so that
> > > > >>>>
> > > > >>>> NBD_STATE_ALLOCATED (0x1), LBA extent is present on the block device
> > > > >>>> NBD_STATE_ZERO (0x2), LBA extent will read as zeroes
> > > > >>>
> > > > >>> Should we flip the sense and call this NBD_STATE_UNALLOCATED (0 means
> > > > >>> allocated, 1 means not present), so that an overall status of 0 is a
> > > > >>> safe default?
> > > > >>
> > > > >> Double negations are evil (and don't work the same in all languages), so
> > > > >> I think it's a worse option.
> > > > > 
> > > > > I agree that a bit which says "unallocated" is confusing in that manner,
> > > > > but that just means we need a better name (one that doesn't contain
> > > > > "un-" or "not")
> > > > > 
> > > > > I like the idea of having zero be the "sensible" default, although I'm
> > > > > quite unable to come up with a better name myself.
> > > > 
> > > > NBD_STATE_TRIM, perhaps? (0 for present, 1 for trimmed or unallocated);
> > > > matches well that we have NBD_CMD_TRIM for requesting the creation of
> > > > such a state.
> > > 
> > > How about NBD_STATE_HOLE?
> > 
> > Both will work, although I like NBD_STATE_TRIM slightly better because
> > it indeed nicely references NBD_CMD_TRIM.
> 
> I just thought that "trim" sounds more like an action than a status, and
> while the reason for a hole to exist can be a previous TRIM command,
> another option is that it's an area in an image that just has never been
> written to. In that case "trim" would be a misnomer.

Point. It could be "TRIMMED" instead, I suppose.

> > However, I also think it should then be made clear that issuing
> > NBD_CMD_TRIM doesn't *require* that GET_BLOCK returns NBD_STATE_TRIM for
> > that region if the backend storage format dosn't support that, to avoid
> > confusion later on.
> 
> Good point. That might be another reason for not calling the status
> "trim".

Also a good point...

-- 
< ron> I mean, the main *practical* problem with C++, is there's like a dozen
       people in the world who think they really understand all of its rules,
       and pretty much all of them are just lying to themselves too.
 -- #debian-devel, OFTC, 2016-02-12