[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20170104001145.GA15099@roeck-us.net>
Date: Tue, 3 Jan 2017 16:11:45 -0800
From: Guenter Roeck <linux@...ck-us.net>
To: Andy Shevchenko <andy.shevchenko@...il.com>
Cc: "wim@...ana.be" <wim@...ana.be>,
"linux-watchdog@...r.kernel.org" <linux-watchdog@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: Auto-conversion of watchdog drivers to use devm functions
On Wed, Jan 04, 2017 at 01:55:39AM +0200, Andy Shevchenko wrote:
> On Mon, Jan 2, 2017 at 1:53 AM, Guenter Roeck <linux@...ck-us.net> wrote:
> > Hi Wim,
> >
> > With heavy support by Julia Lawall, I created a number of coccinelle scripts
> > to auto-convert watchdog platform drivers to use devm functions. The result
> > is quite impressive - many drivers won't even need remove functions anymore
> > after the conversion.
> >
> > Question is now how to submit the patches. There are 12 rule files affecting
> > a total of 62 drivers. Should I submit one patch per driver (62 patches),
> > one patch per rule (12), or one patch per rule file per driver (183) ?
>
> One per driver.
>
> > Also, how should I refer to the rules ? The total number of rule lines
> > is more than 1,000, which seems to be a bit much for the commit log (even
> > though not all rules apply to all files).
>
> No need, perhaps some URL?
>
Yes, that is what I ended up doing - I created a repository on github and
pushed the semantic patches as well as the scripts used to generate the
actual patches into it. I am going to reference that repository in the
patches, together with a summary of the changes made.
An interesting exercise is to let the semantic patches run on the entire
kernel. It generates almost 1,500 patches.
Thanks,
Guenter
Powered by blists - more mailing lists