Linux Virtualization - Aug 2013 - [PATCH] percpu ida: Switch to cpumask_t, add some comments

On Wed, 28 Aug 2013 14:00:10 -0700 Kent Overstreet <kmo at daterainc.com>
wrote:
> On Wed, Aug 28, 2013 at 01:25:50PM -0700, Andrew Morton wrote:
> > On Wed, 28 Aug 2013 12:55:17 -0700 Kent Overstreet <kmo at
daterainc.com> wrote:
> > 
> > > Fixup patch, addressing Andrew's review feedback:
> > 
> > Looks reasonable.
> > 
> > >  lib/idr.c           | 38 +++++++++++++++++++++-----------------
> > 
> > I still don't think it should be in this file.
> > 
> > You say that some as-yet-unmerged patches will tie the new code into
> > the old ida code.  But will it do it in a manner which requires that
> > the two reside in the same file?
> 
> Not require, no - but it's just intimate enough with my ida rewrite
that
> I think it makes sense; it makes some use of stuff that should be
> internal to the ida code.
> 
> Mostly just sharing the lock though, since I got rid of the ida
> interfaces that don't do locking, but percpu ida needs a lock that also
> covers what ida needs.
> 
> It also makes use of a ganged allocation interface, but there's no real
> reason ida can't expose that, it's just unlikely to be useful to
> anything but percpu ida.
> 
> The other reason I think it makes sense to live in idr.c is more for
> users of the code; as you pointed out as far as the user's perspective
> percpu ida isn't doing anything fundamentally different from ida, so I
> think it makes sense for the code to live in the same place as a
> kindness to future kernel developers who are trying to find their way
> around the various library code.
I found things to be quite the opposite - it took 5 minutes of staring,
head-scratching, double-checking and penny-dropping before I was
confident that the newly-added code actually has nothing at all to do
with the current code.  Putting it in the same file was misleading, and
I got misled.

On Wed, Aug 28, 2013 at 02:10:19PM -0700, Andrew Morton
wrote:
> On Wed, 28 Aug 2013 14:00:10 -0700 Kent Overstreet <kmo at
daterainc.com> wrote:
> 
> > On Wed, Aug 28, 2013 at 01:25:50PM -0700, Andrew Morton wrote:
> > > On Wed, 28 Aug 2013 12:55:17 -0700 Kent Overstreet <kmo at
daterainc.com> wrote:
> > > 
> > > > Fixup patch, addressing Andrew's review feedback:
> > > 
> > > Looks reasonable.
> > > 
> > > >  lib/idr.c           | 38
+++++++++++++++++++++-----------------
> > > 
> > > I still don't think it should be in this file.
> > > 
> > > You say that some as-yet-unmerged patches will tie the new code
into
> > > the old ida code.  But will it do it in a manner which requires
that
> > > the two reside in the same file?
> > 
> > Not require, no - but it's just intimate enough with my ida
rewrite that
> > I think it makes sense; it makes some use of stuff that should be
> > internal to the ida code.
> > 
> > Mostly just sharing the lock though, since I got rid of the ida
> > interfaces that don't do locking, but percpu ida needs a lock that
also
> > covers what ida needs.
> > 
> > It also makes use of a ganged allocation interface, but there's no
real
> > reason ida can't expose that, it's just unlikely to be useful
to
> > anything but percpu ida.
> > 
> > The other reason I think it makes sense to live in idr.c is more for
> > users of the code; as you pointed out as far as the user's
perspective
> > percpu ida isn't doing anything fundamentally different from ida,
so I
> > think it makes sense for the code to live in the same place as a
> > kindness to future kernel developers who are trying to find their way
> > around the various library code.
> 
> I found things to be quite the opposite - it took 5 minutes of staring,
> head-scratching, double-checking and penny-dropping before I was
> confident that the newly-added code actually has nothing at all to do
> with the current code.  Putting it in the same file was misleading, and
> I got misled.
Ok... and I could see how the fact that it currently _doesn't_ have
anything to do with the existing code would be confusing...

Do you think that if/when it's making use of the ida rewrite it'll be
ok? Or would you still prefer to have it in a new file (and if so, any
preference on the naming?)

On Wed, 28 Aug 2013 14:23:58 -0700 Kent Overstreet <kmo at daterainc.com>
wrote:
> > I found things to be quite the opposite - it took 5 minutes of
staring,
> > head-scratching, double-checking and penny-dropping before I was
> > confident that the newly-added code actually has nothing at all to do
> > with the current code.  Putting it in the same file was misleading,
and
> > I got misled.
> 
> Ok... and I could see how the fact that it currently _doesn't_ have
> anything to do with the existing code would be confusing...
> 
> Do you think that if/when it's making use of the ida rewrite it'll
be
> ok? Or would you still prefer to have it in a new file
I'm constitutionally reluctant to ever assume that any out-of-tree code
will be merged.  Maybe you'll get hit by a bus, and maybe the code
sucks ;)

Are you sure that the two things are so tangled together that they must
live in the same file?  If there's some nice layering between ida and
percpu_ida then perhaps such a physical separation would remain
appropriate?
> (and if so, any preference on the naming?)
percpu_ida.c?