Date:   Mon, 12 Sep 2016 10:40:48 -0400
From:   Paul Gortmaker <paul.gortmaker@...driver.com>
To:     <linux-kernel@...r.kernel.org>
CC:     Paul Gortmaker <paul.gortmaker@...driver.com>,
        Boris BREZILLON <boris.brezillon@...e-electrons.com>,
        Florian Lobmaier <florian.lobmaier@....com>,
        Laxman Dewangan <ldewangan@...dia.com>,
        Lee Jones <lee.jones@...aro.org>,
        Maxime Ripard <maxime.ripard@...e-electrons.com>,
        Mika Westerberg <mika.westerberg@...ux.intel.com>,
        Samuel Ortiz <sameo@...ux.intel.com>,
        Sourav Poddar <sourav.poddar@...com>,
        Thor Thayer <tthayer@...nsource.altera.com>,
        Tony Lindgren <tony@...mide.com>, <linux-omap@...r.kernel.org>
Subject: [PATCH v2 0/6] mfd: trivial demodularization of non-modular drivers.

I grouped this little series of mfd commits together because of the
common theme of their relatively trivial use of modular infrastructure
in these non-modular drivers.

What that means is there are no orphaned __exit functions to remove;
no unused ".remove" functions linked into the driver struct to be
deleted -- so in the end we have binary equivalence between the
pre-patched code and the post-patched code. Easy to review and low risk.

More specifically - the only thing we are doing here is mapping the
initcall that binds in the driver from the module variant to the
built-in variant ; something that CPP would do for us anyway -- and
then throwing away any MODULE_ tags which are no-ops in non-modular
builds (while preserving any author/desc. info they had in comments).
And in doing that, we can toss the include of module.h in favour of
init.h (asssuming it wasn't included already).

One exception is patch #2; it converts a driver from bool to tristate,
as was asked per the v1 review[1].  This has been build tested, but
I don't have the specific hardware for run time testing.

For anyone new to the underlying goal of this cleanup, we are trying to
make driver code consistent with the Makefiles/Kconfigs that control them.

This means not using modular functions/macros for drivers that can never
be built as a module.  Some of the other downfalls this leads to are:

 (1) it is easy to accidentally write unused module_exit and remove code
 (2) it can be misleading when reading the source, thinking it can be
     modular when the Makefile and/or Kconfig prohibit it
 (3) it requires the include of the module.h header file which in turn
     includes nearly everything else, thus adding to CPP overhead.
 (4) it gets copied/replicated into other drivers and spreads like weeds.

Build tested on linux-next for arm, arm64 and x86-64 to ensure no typos
or similar issues crept in.

[1] https://lkml.kernel.org/r/20160705012544.14143-1-paul.gortmaker@windriver.com

Paul.
---

[v1 --> v2: make one driver tristate, add some Ackd-by tags.]

Cc: Boris BREZILLON <boris.brezillon@...e-electrons.com>
Cc: Florian Lobmaier <florian.lobmaier@....com>
Cc: Laxman Dewangan <ldewangan@...dia.com>
Cc: Lee Jones <lee.jones@...aro.org>
Cc: Maxime Ripard <maxime.ripard@...e-electrons.com>
Cc: Mika Westerberg <mika.westerberg@...ux.intel.com>
Cc: Samuel Ortiz <sameo@...ux.intel.com>
Cc: Sourav Poddar <sourav.poddar@...com>
Cc: Thor Thayer <tthayer@...nsource.altera.com>
Cc: Tony Lindgren <tony@...mide.com>
Cc: linux-omap@...r.kernel.org

Paul Gortmaker (6):
  mfd: altera-a10sr: make it explicitly non-modular
  mfd: as3722: convert MFD_AS3722 from bool to tristate
  mfd: intel_msic: Make it explicitly non-modular
  mfd: smsc-ece1099: Make it explicitly non-modular
  mfd: sun6i-prcm: Make it explicitly non-modular
  mfd: twl-core: Make it explicitly non-modular

 drivers/mfd/Kconfig        |  2 +-
 drivers/mfd/altera-a10sr.c | 14 ++++++--------
 drivers/mfd/intel_msic.c   |  9 ++-------
 drivers/mfd/smsc-ece1099.c | 11 ++---------
 drivers/mfd/sun6i-prcm.c   |  8 ++------
 drivers/mfd/twl-core.c     |  9 +--------
 6 files changed, 14 insertions(+), 39 deletions(-)

-- 
2.8.4

Powered by blists - more mailing lists