Re: [PATCH v3] lib/idr.c rewrite, percpu ida/tag allocator

[Michael S. Tsirkin]

On Sun, Jul 07, 2013 at 07:09:35PM -0700, Nicholas A. Bellinger wrote:
> On Sun, 2013-07-07 at 14:48 +0300, Michael S. Tsirkin wrote:
> > On Sat, Jul 06, 2013 at 03:59:06PM -0700, Nicholas A. Bellinger wrote:
> > > On Fri, 2013-07-05 at 22:08 -0700, Kent Overstreet wrote:
> > > > Previous posting: http://thread.gmane.org/gmane.linux.kernel/1511216
> > > > 
> > > > The only real change since the last version is that I've reworked the
> > > > new ida implementation to not use one giant allocation - it's still
> > > > logically one big arary, but it's implemented as an array of arrays.
> > > > 
> > > > With that, it scales up to INT_MAX allocated ids just fine. Benchmarks
> > > > are included in that patch.
> > > > 
> > > > Patch series is available in my git repo:
> > > > git://evilpiepirate.org/~kent/linux-bcache.git idr
> > > > 
> > > > Andrew, want to pick this up for 3.12?
> > > 
> > > Just FYI that the target-pending/target-per-cpu-ida branch utilizing
> > > per-cpu ida logic for vhost-scsi has been rebased to include this latest
> > > series.  Thanks Kent!
> > > 
> > > Andrew, what are your plans for this..?
> > > 
> > > Would you consider allowing this series be merged through target-pending
> > > in order to allow us to start taking advantage of per-cpu ida tag
> > > pre-allocation optimizations for v3.12 target code..? 
> > > 
> > > Another option would be to merge the per-cpu ida specific piece now for
> > > v3.11 for which the risk is low given it doesn't touch any existing
> > > code, and then include Kent's full idr rewrite separately as v3.12 item.
> > > 
> > > WDYT..?
> > 
> > Takes as is, this conflicts with cleanup patches by Asias in vhost-next
> > that I was going to send to Linus for 3.11.
> 
> So the proposed vhost-scsi per-cpu ida/tag pre-allocation changes should
> be considered v3.12 items.
> 
> > Merging new target code for 3.11 and vhost code for 3.12
> > helps keep the dependencies simple, so I'm all for it.
> > 
> 
> My point above is that having (at least) Kent's per-cpu ida code merged
> for v3.11 would make it easier to move forward during the v3.12 cycle on
> per-cpu ida/tag conversions for vhost-scsi + other target fabric
> drivers.
> 
> --nab

Exactly. I think it's a good idea, too.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/