From mboxrd@z Thu Jan  1 00:00:00 1970
From: marek.vasut@gmail.com (Marek Vasut)
Date: Tue, 20 Dec 2011 14:43:02 +0100
Subject: [PATCH 0/12] Add clk_prepare/clk_unprepare for mxs clock api
In-Reply-To: <1324382443-4439-1-git-send-email-shawn.guo@linaro.org>
References: <1324382443-4439-1-git-send-email-shawn.guo@linaro.org>
Message-ID: <201112201443.03202.marek.vasut@gmail.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

> The mutex locking problem on mxs platform has been reported on mailing
> list for a few time.  The cause is that a mutex is being held in mxs
> clock api clk_enable/clk_disable which are being called by some mxs
> clock client drivers like amba-pl011 and mxs-saif in atomic context.
> 
> This series takes Russell's suggestion to:
> 
>  * Convert all mxs clock client drivers to clk_prepare/clk_unprepare.
>    (I hope the series has covered all the drivers, but appreciate any
>    reminding on any missing ones.)

serial/amba-pl011.c isn't converted, that's where I originally observed the 
issue.
> 
>  * Add clk_prepare/clk_unprepare for mxs clock api to contain the
>    non-atomic portion of existing clk_enable/clk_disable implemention.
>    (As the whole existing clk_enable/clk_disable are non-atomic,
>    clk_prepare/clk_unprepare become a renaming of the existing
>    clk_enable/clk_disable, and the new clk_enable/clk_disable become
>    a pair of dummy functions.)
> 
> Thus, we can ensure that no mutex will be held for atomic context, and
> the problem people keep reporting gets fixed with one step move toward
> common clk framework while not ncessarily waiting for it.
> 
> If driver maintainers have no objection, after the review process, I
> would like to send the series through arm-soc tree to ease the merge
> process.
> 
> Those two patches from Richard Zhao are included here for completeness,
> and they have been picked up by Sascha on imx tree.  But since imx and
> mxs will both go through arm-soc, it should easy to sort out how we
> should merge them.
> 
> Regards,
> Shawn
> 
> Richard Zhao (2):
>       clk: add helper functions clk_prepare_enable and
> clk_disable_unprepare net: fec: add clk_prepare/clk_unprepare
> 
> Shawn Guo (10):
>       arm/mxs: convert platform code to clk_prepare/clk_unprepare
>       dma: mxs-dma: convert to clk_prepare/clk_unprepare
>       mmc: mxs-mmc: convert to clk_prepare/clk_unprepare
>       mtd: gpmi-lib: convert to clk_prepare/clk_unprepare
>       net: flexcan: convert to clk_prepare/clk_unprepare
>       serial: mxs-auart: convert to clk_prepare/clk_unprepare
>       video: mxsfb: convert to clk_prepare/clk_unprepare
>       ASoC: mxs-saif: convert to clk_prepare/clk_unprepare
>       clk: add config option HAVE_CLK_PREPARE into Kconfig
>       arm/mxs: select HAVE_CLK_PREPARE for clock
> 
>  arch/arm/Kconfig                      |    1 +
>  arch/arm/mach-mxs/clock-mx23.c        |   10 +++++-----
>  arch/arm/mach-mxs/clock-mx28.c        |   10 +++++-----
>  arch/arm/mach-mxs/clock.c             |   33
> +++++++++++++++++++++++---------- arch/arm/mach-mxs/mach-mx28evk.c      | 
>   2 +-
>  arch/arm/mach-mxs/system.c            |    2 +-
>  arch/arm/mach-mxs/timer.c             |    2 +-
>  drivers/clk/Kconfig                   |    3 +++
>  drivers/dma/mxs-dma.c                 |    8 ++++----
>  drivers/mmc/host/mxs-mmc.c            |   10 +++++-----
>  drivers/mtd/nand/gpmi-nand/gpmi-lib.c |   12 ++++++------
>  drivers/net/can/flexcan.c             |   10 +++++-----
>  drivers/net/ethernet/freescale/fec.c  |   10 +++++-----
>  drivers/tty/serial/mxs-auart.c        |    8 ++++----
>  drivers/video/mxsfb.c                 |    8 ++++----
>  include/linux/clk.h                   |   22 ++++++++++++++++++++++
>  sound/soc/mxs/mxs-saif.c              |    4 ++--
>  17 files changed, 97 insertions(+), 58 deletions(-)