[PATCH v4 0/3] arm: mvebu: fix resource leak

[Jason Cooper]

On Tue, Aug 27, 2013 at 11:01:20AM +0800, Jisheng Zhang wrote:
> Dear Jason,
> 
> On Mon, 26 Aug 2013 13:45:23 -0700
> Jason Cooper <jason at lakedaemon.net> wrote:
> 
> > On Mon, Aug 26, 2013 at 07:36:55PM +0800, Jisheng Zhang wrote:
> > > These patches try to fix resource leak by adding missing of_node_put(),
> > > iounmap or using devm_ioremap_resource() if available.
> > > 
> > > v4:
> > >   - re-generate since Ezequiel's patches add DT support to the
> > > mvebu-mbus driver
> > 
> > grrr.  I hate to ask this, but can you please rebase patches 1 and 2
> > against mvebu/fixes-non-critical?  No need to add back in the mvebu-mbus
> > hunk.
> 
> patch 1 also fixes one similar trivial issue introduced since 994c8c94b419e
> "ARM: mvebu: Remove the harcoded BootROM window allocation" in linux-next tree
> 
> patch 2 is updated to fix similar trivial issue introduce since 6839cfa82f99
> "bus: mvebu-mbus: Introduce device tree binding"
> 
> These two commits aren't included in mvebu/fixes-non-criticial yet. Could you
> please give suggestion?

Yes, that's correct.  We prefer to have patch submitters base off of a
mainline tag (eg v3.11-rc7).  conflicts between patchsets are then
caught and resolved when branches are merged.  If done correctly, the
merge resolution should be obvious in most cases.

The upstream maintainers _prefer_ to see those conflicts because it
gives them a better sense of who is tinkering in the same code-paths.

Trying to base patches off of disparate branches in order to
'pre-resolve' those conflicts creates unnecessary dependencies and
non-obvious merge-resolutions.

In this case I asked you base off of mvebu/fixes-non-critical because
that is where I will be applying them for queueing to arm-soc.  You
could also base off of v3.11-rc7, there's nothing in
mvebu/fixes-non-critical that should conflict with your changes.

thx,

Jason.