firefly-linux-kernel-4.4.55.git
9 years agommc: block: fix format string warning
Asaf Vertz [Tue, 20 Jan 2015 16:33:35 +0000 (18:33 +0200)]
mmc: block: fix format string warning

Fixed the following warning (reported by cppcheck):
[drivers/mmc/card/block.c:2149]: (warning) %d in format string (no. 1)
requires 'int' but the argument type is 'unsigned int'.

Signed-off-by: Asaf Vertz <asaf.vertz@tandemg.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agoMerge branch 'tmio' into next
Ulf Hansson [Tue, 20 Jan 2015 09:44:08 +0000 (10:44 +0100)]
Merge branch 'tmio' into next

9 years agommc: tmio: remove TMIO_MMC_HAVE_CTL_DMA_REG flag
Kuninori Morimoto [Tue, 13 Jan 2015 05:00:39 +0000 (05:00 +0000)]
mmc: tmio: remove TMIO_MMC_HAVE_CTL_DMA_REG flag

tmio_mmc_host has .enable_dma callback now.
We don't need TMIO_MMC_HAVE_CTL_DMA_REG anymore.
Let's remove it

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sh_mobile_sdhi: enable 32bit DMA access
Kuninori Morimoto [Tue, 13 Jan 2015 05:00:07 +0000 (05:00 +0000)]
mmc: sh_mobile_sdhi: enable 32bit DMA access

Now, sh_mobile_sdhi can use 32bit DMA access in R-Cer Gen2.
Let's use it.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sh_mobile_sdhi: use .enable_dma
Kuninori Morimoto [Tue, 13 Jan 2015 04:59:56 +0000 (04:59 +0000)]
mmc: sh_mobile_sdhi: use .enable_dma

tmio_mmc_dma has .enable_dma callback now.
let's use it.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sh_mobile_sdhi: add new macro for mmc_host to sh_mobile_sdhi
Kuninori Morimoto [Tue, 13 Jan 2015 04:59:43 +0000 (04:59 +0000)]
mmc: sh_mobile_sdhi: add new macro for mmc_host to sh_mobile_sdhi

getting struct sh_mobile_sdhi from struct mmc_host needs to use
complex container_of(). This patch adds new host_to_priv macro to
get it easily.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sh_mobile_sdhi: tidyup mmc_data->bus_shift for latest SoC
Kuninori Morimoto [Tue, 13 Jan 2015 04:59:33 +0000 (04:59 +0000)]
mmc: sh_mobile_sdhi: tidyup mmc_data->bus_shift for latest SoC

Latest SDHI on Renesas has expand register mapping.
update mmc_data->bus_shift for it.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sh_mobile_sdhi: remove .init/.cleanup
Kuninori Morimoto [Tue, 13 Jan 2015 04:59:24 +0000 (04:59 +0000)]
mmc: sh_mobile_sdhi: remove .init/.cleanup

No one is using .init/.cleanup callback function.
Let's remove these.
sdhi_ops and .cd_wakeup are also removed

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: enable SoC specific DMA buswidth settings
Kuninori Morimoto [Tue, 13 Jan 2015 04:59:14 +0000 (04:59 +0000)]
mmc: tmio: enable SoC specific DMA buswidth settings

Renesas SDHI which is based on TMIO driver has 2 type SoC. 1st one is
used as SH-Mobile series, and 2nd is R-Car series. R-Car series SoC has
DMA buswidth setting register which enables 32bit access.
This patch adds .dma_buswidth and enables it.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: add .enable_dma on tmio_mmc_dma
Kuninori Morimoto [Tue, 13 Jan 2015 04:59:05 +0000 (04:59 +0000)]
mmc: tmio: add .enable_dma on tmio_mmc_dma

Some controllers need DMA special register/setting.
This patch adds new .enable_dma callback for it.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_data has .dma_rx_offset
Kuninori Morimoto [Tue, 13 Jan 2015 04:58:56 +0000 (04:58 +0000)]
mmc: tmio: tmio_mmc_data has .dma_rx_offset

Current .dma_rx_offset is implemented under tmio_mmc_dma.
It goes to tmio_mmc_data by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_data has .alignment_shift
Kuninori Morimoto [Tue, 13 Jan 2015 04:58:46 +0000 (04:58 +0000)]
mmc: tmio: tmio_mmc_data has .alignment_shift

Current .alignment_shift is implemented under tmio_mmc_dma.
It goes to tmio_mmc_data by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_host has .bus_shift
Kuninori Morimoto [Tue, 13 Jan 2015 04:58:20 +0000 (04:58 +0000)]
mmc: tmio: tmio_mmc_host has .bus_shift

Current .bus_shift is implemented under tmio_mmc_data.
It goes to tmio_mmc_host by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_host has .multi_io_quirk
Kuninori Morimoto [Tue, 13 Jan 2015 04:58:10 +0000 (04:58 +0000)]
mmc: tmio: tmio_mmc_host has .multi_io_quirk

Current .multi_io_quirk is implemented under tmio_mmc_data.
It goes to tmio_mmc_host by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_host has .clk_disable
Kuninori Morimoto [Tue, 13 Jan 2015 04:58:01 +0000 (04:58 +0000)]
mmc: tmio: tmio_mmc_host has .clk_disable

Current .clk_disable is implemented under tmio_mmc_data.
It goes to tmio_mmc_host by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_host has .clk_enable
Kuninori Morimoto [Tue, 13 Jan 2015 04:57:52 +0000 (04:57 +0000)]
mmc: tmio: tmio_mmc_host has .clk_enable

Current .clk_enable is implemented under tmio_mmc_data.
It goes to tmio_mmc_host by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_host has .write16_hook
Kuninori Morimoto [Tue, 13 Jan 2015 04:57:42 +0000 (04:57 +0000)]
mmc: tmio: tmio_mmc_host has .write16_hook

Current .write16_hook is implemented under tmio_mmc_data.
It goes to tmio_mmc_host by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: tmio_mmc_host has .dma
Kuninori Morimoto [Tue, 13 Jan 2015 04:57:33 +0000 (04:57 +0000)]
mmc: tmio: tmio_mmc_host has .dma

Current .dma is implemented under tmio_mmc_data.
It goes to tmio_mmc_host by this patch.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: tmio: add tmio_mmc_host_alloc/free()
Kuninori Morimoto [Tue, 13 Jan 2015 04:57:22 +0000 (04:57 +0000)]
mmc: tmio: add tmio_mmc_host_alloc/free()

Current tmio_mmc driver is using tmio_mmc_data for driver/platform
specific data/callback, and it is needed for tmio_mmc_host_probe()
function. Because of this style, include/linux/mfd/tmio.h header has
tmio driver/framework specific data which is not needed from platform.

This patch adds new tmio_mmc_host_alloc/free() as cleanup preparation.
tmio driver specific data/callback will be implemented in tmio_mmc_host,
and platform specific data/callback will be implemented in tmio_mmc_data
in this cleanup.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: host: add new f_sdh30
Vincent Yang [Tue, 20 Jan 2015 08:05:18 +0000 (16:05 +0800)]
mmc: sdhci: host: add new f_sdh30

This patch adds new host controller driver for
Fujitsu SDHCI controller f_sdh30.

Signed-off-by: Vincent Yang <Vincent.Yang@tw.fujitsu.com>
Signed-off-by: Andy Green <andy.green@linaro.org>
Signed-off-by: Tetsuya Takinishi <t.takinishi@jp.fujitsu.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: add a quirk for single block transactions
Vincent Yang [Tue, 20 Jan 2015 08:05:17 +0000 (16:05 +0800)]
mmc: sdhci: add a quirk for single block transactions

This patch defines a quirk to disable the block count
for single block transactions.
It is a preparation and will be used by Fujitsu
SDHCI controller f_sdh30 driver.

Signed-off-by: Vincent Yang <Vincent.Yang@tw.fujitsu.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: add a quirk for tuning work around
Vincent Yang [Tue, 20 Jan 2015 08:05:16 +0000 (16:05 +0800)]
mmc: sdhci: add a quirk for tuning work around

This patch defines a quirk for tuning work
around for some sdhci host controller. It sets
both SDHCI_CTRL_EXEC_TUNING and SDHCI_CTRL_TUNED_CLK
for tuning.
It is a preparation and will be used by Fujitsu
SDHCI controller f_sdh30 driver.

Signed-off-by: Vincent Yang <Vincent.Yang@tw.fujitsu.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: add a voltage switch callback function
Vincent Yang [Tue, 20 Jan 2015 08:05:15 +0000 (16:05 +0800)]
mmc: sdhci: add a voltage switch callback function

This patch adds a callback function to do
controller-specific actions when switching voltages.
It is a preparation and will be used by Fujitsu
SDHCI controller f_sdh30 driver.

Signed-off-by: Vincent Yang <Vincent.Yang@tw.fujitsu.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: dw_mmc: Power on vqmmc in MMC_POWER_ON, not MMC_POWER_UP
Doug Anderson [Tue, 13 Jan 2015 23:58:44 +0000 (15:58 -0800)]
mmc: dw_mmc: Power on vqmmc in MMC_POWER_ON, not MMC_POWER_UP

If we power up vqmmc in MMC_POWER_ON then we end up turning it on
before mmc_power_up() sets the signal voltage.  That's not so great
since we might be powering it up at the wrong voltage.

Note that this is how Yuvaraj originally coded things up in
<https://patchwork.kernel.org/patch/4401231/> but he changed it on my
suggestion.  Apparently I was wrong.

Reported-by: Alexandru M Stan <amstan@chromium.org>
Signed-off-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: host: sdhci: Added a space before (
Mohammad Jamal [Tue, 13 Jan 2015 15:17:24 +0000 (20:47 +0530)]
mmc: host: sdhci: Added a space before (

This patch solves the coding style issue by adding a space
before (

Signed-off-by: Mohammad Jamal <md.jamalmohiuddin@gmail.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-pxav3: fix pm unbalanced issue in -> remove()
Jisheng Zhang [Sun, 4 Jan 2015 15:15:48 +0000 (23:15 +0800)]
mmc: sdhci-pxav3: fix pm unbalanced issue in -> remove()

This patch calls pm_runtime_put_noidle() to restore the device's usage
counter in the ->remove() implementation.

Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-pxav3: fix unbalanced clock issues during probe
Jisheng Zhang [Sun, 4 Jan 2015 15:15:47 +0000 (23:15 +0800)]
mmc: sdhci-pxav3: fix unbalanced clock issues during probe

Commit 0dcaa2499b7d ("sdhci-pxav3: Fix runtime PM initialization") tries
to fix one hang issue caused by calling sdhci_add_host() on a suspended
device. The fix enables the clock twice, once by clk_prepare_enable() and
another by pm_runtime_get_sync(), meaning that the clock will never be
gated at runtime PM suspend. I observed the power consumption regression on
Marvell BG2Q SoCs.

In fact, the fix is not correct. There still be a very small window
during which a runtime suspend might somehow occur after pm_runtime_enable()
but before pm_runtime_get_sync().

This patch fixes all of the two problems by just incrementing the usage
counter before pm_runtime_enable(). It also adjust the order of disabling
runtime pm and storing the usage count in the error path to handle clock
gating properly.

Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Cc: <stable@vger.kernel.org> # v3.11+
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: Always init buf_ready_int
Adrian Hunter [Fri, 5 Dec 2014 17:41:10 +0000 (19:41 +0200)]
mmc: sdhci: Always init buf_ready_int

There is no point making the initialization
of buf_ready_int conditional on host version.
Simplify by just doing it always. Note that
the other conditional initializations will be
removed when the new way of doing re-tuning
is taken into use.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: Move mmc_card_removed() into mmc_start_request()
Adrian Hunter [Fri, 5 Dec 2014 17:41:02 +0000 (19:41 +0200)]
mmc: core: Move mmc_card_removed() into mmc_start_request()

Both callers of mmc_start_request() call mmc_card_removed()
so move that call into mmc_start_request().

This patch is preparation for adding re-tuning support.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: Simplify by adding mmc_execute_tuning()
Adrian Hunter [Fri, 5 Dec 2014 17:40:59 +0000 (19:40 +0200)]
mmc: core: Simplify by adding mmc_execute_tuning()

For each MMC, SD and SDIO there is code that
holds the clock, calls ops->execute_tuning, and
releases the clock. Simplify the code a bit by
providing a separate function to do that.

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: omap_hsmmc: use mmc_of_parse to parse common mmc configuration.
NeilBrown [Mon, 12 Jan 2015 19:23:18 +0000 (08:23 +1300)]
mmc: omap_hsmmc: use mmc_of_parse to parse common mmc configuration.

This ensures that all standard options are available to hsmmc,
In particular, I need cap-power-off-card.

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: omap_hsmmc: use slot-gpio library for gpio support.
NeilBrown [Mon, 12 Jan 2015 19:23:18 +0000 (08:23 +1300)]
mmc: omap_hsmmc: use slot-gpio library for gpio support.

Using the common code removes some code duplication, and
makes it easier to switch to using mmc_of_parse() which
will remove more duplication.

This uses the new mmc_gpio_request_cd_isr to provide a non-standard
interrupt service routine for card-detect interrupts.

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: slot-gpio: Allow host driver to provide isr for card-detect interrupts
NeilBrown [Mon, 12 Jan 2015 19:23:18 +0000 (08:23 +1300)]
mmc: slot-gpio: Allow host driver to provide isr for card-detect interrupts

One of the reasons omap_hsmmc doesn't use the slot-gpio library
is that it has some non-standard functionality in the card-detect
interrupt service routine.

To make it possible for omap_hsmmc (and maybe others) to be converted
to use slot-gpio, add 'mmc_gpio_request_cd_isr' which provide an
alternate isr to be register by the slot-gpio code.

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-acpi: fix copy'n'paste typos in the comments
Andy Shevchenko [Mon, 12 Jan 2015 17:37:25 +0000 (19:37 +0200)]
mmc: sdhci-acpi: fix copy'n'paste typos in the comments

Just fix the comments, no functional changes.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sd: add reset bus_ops callback
Johan Rudholm [Mon, 12 Jan 2015 14:38:06 +0000 (15:38 +0100)]
mmc: sd: add reset bus_ops callback

Enable power cycle and re-initialization of SD cards via the reset
bus_ops. Power cycling a buggy SD card sometimes helps it get back on
track.

Signed-off-by: Johan Rudholm <johanru@axis.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: refactor the hw_reset routines
Johan Rudholm [Mon, 12 Jan 2015 14:38:05 +0000 (15:38 +0100)]
mmc: core: refactor the hw_reset routines

Move the (e)MMC specific hw_reset code from core.c into mmc.c. Call the
code from the new bus_ops member "reset". This also allows for adding
a SD card specific reset procedure.

Signed-off-by: Johan Rudholm <johanru@axis.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: always check status after reset
Johan Rudholm [Mon, 12 Jan 2015 14:38:04 +0000 (15:38 +0100)]
mmc: core: always check status after reset

Always check if the card is alive after a successful reset. This allows
us to remove mmc_hw_reset_check(), leaving mmc_hw_reset() as the only
card reset interface.

Signed-off-by: Johan Rudholm <johanru@axis.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: Add SDIO function devicetree subnode parsing
Sascha Hauer [Mon, 30 Jun 2014 09:07:25 +0000 (11:07 +0200)]
mmc: Add SDIO function devicetree subnode parsing

This adds SDIO devicetree subnode parsing to the mmc core. While
SDIO devices are runtime probable they sometimes need nonprobable
additional information on embedded systems, like an additional gpio
interrupt or a clock. This patch makes it possible to supply this
information from the devicetree. SDIO drivers will find a pointer
to the devicenode in their devices of_node pointer.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
[hdegoede@redhat.com: Misc. cleanups]
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: Add SDIO function subnode DT documentation
Sascha Hauer [Mon, 30 Jun 2014 09:07:24 +0000 (11:07 +0200)]
mmc: Add SDIO function subnode DT documentation

While SDIO devices are runtime probable they sometimes need nonprobable
additional information on embedded systems, like an additional gpio
interrupt or a clock. This binding describes how to add child nodes to the
devicetree to supply this information.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
[hdegoede@redhat.com: Documented the need for #address- and #size-cells]
[hdegoede@redhat.com: Added a real world example]
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-acpi: Remove redundant runtime PM idle callback
Ulf Hansson [Fri, 2 Jan 2015 13:47:00 +0000 (14:47 +0100)]
mmc: sdhci-acpi: Remove redundant runtime PM idle callback

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: use pipeline mmc requests to improve performance
Haibo Chen [Tue, 9 Dec 2014 09:04:05 +0000 (17:04 +0800)]
mmc: sdhci: use pipeline mmc requests to improve performance

This patch is based on the patches by Per Forlin, Tony Lin and Ryan QIAN.

This patch complete the API 'post_req' and 'pre_req' in sdhci host side,

Test Env:
1. i.MX6Q-SABREAUTO board, CPU @ 996MHz, use ADMA in uSDHC controller.
2. Test command:
$ echo 1 > /proc/sys/vm/drop_caches
write to sd card:
$ dd if=/dev/zero of=/dev/mmcblk0 bs=1M count=2000 conv=fsync
read the sd card:
$ dd if=/dev/mmcblk0 of=/dev/null bs=1M count=2000

3. TOSHIBA 16GB SD3.0 card, running at 4 bit, SDR104 @ 198MHZ
Performance with and without this patch:
      -------------------------------------------------
  |                    | read speed | write speed |
  |------------------------------------------------
  | with this patch    | ~76.7 MB/s |  ~23.3 MB/s |
  |------------------------------------------------
  |without this patch  | ~60.5 MB/s |  ~22.5 MB/s |
  -------------------------------------------------

4. SanDisk 8GB SD3.0 card, running at 4 bit, DDR50 @ 50MHZ
Performance with and without this patch:
      -------------------------------------------------
  |                    | read speed | write speed |
  |------------------------------------------------
  | with this patch    | ~40.5 MB/s |  ~15.6 MB/s |
  |------------------------------------------------
  |without this patch  | ~36.1 MB/s |  ~14.1 MB/s |
  -------------------------------------------------

5. Kingston 8GB SD2.0 card, running at 4 bit, High-speed @ 50MHZ
Performance with and without this patch:
      -------------------------------------------------
  |                    | read speed | write speed |
  |------------------------------------------------
  | with this patch    | ~22.7 MB/s |  ~8.2 MB/s  |
  |------------------------------------------------
  |without this patch  | ~21.3 MB/s |  ~8.0 MB/s  |
  -------------------------------------------------

6. About eMMC, Sandisk 8GB eMMC on i.MX6DL-sabresd board, CPU @ 792MHZ,
   eMMC running at 8 bit, DDR52 @ 52MHZ.
Performance with and without this patch:
      -------------------------------------------------
  |                    | read speed | write speed |
  |------------------------------------------------
  | with this patch    | ~37.3 MB/s |  ~10.5 MB/s |
  |------------------------------------------------
  |without this patch  | ~33.4 MB/s |  ~10.5 MB/s |
  -------------------------------------------------

Signed-off-by: Haibo Chen <haibo.chen@freescale.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: Remove redundant ADMA page boundary warnings
Andrew Gabbasov [Mon, 22 Dec 2014 12:23:21 +0000 (06:23 -0600)]
mmc: sdhci: Remove redundant ADMA page boundary warnings

The bounce buffer, used for misaligned bytes for ADMA access,
resides wholly within the (align_sz)-aligned word, just by construction.
The page addresses are aligned to (align_sz), either for 4 or 8 bytes
alignment, so that the aligned word resides wholly within a single page
and can't cross the page boundary. So, the bounce buffer can't cross
the page boundary too. That's why the warnings are never hit, and can
be safely removed.

Signed-off-by: Andrew Gabbasov <andrew_gabbasov@mentor.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: Make tuning block patterns static
Ulf Hansson [Mon, 1 Dec 2014 15:53:34 +0000 (16:53 +0100)]
mmc: core: Make tuning block patterns static

Since previous patches removed the need for the tuning block patterns
to be exported, let's move them close to the mmc_send_tuning() API.

Those are now intended to be used only by the mmc core.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
9 years agommc: dw_mmc: Convert to mmc_send_tuning()
Ulf Hansson [Mon, 1 Dec 2014 15:13:39 +0000 (16:13 +0100)]
mmc: dw_mmc: Convert to mmc_send_tuning()

Instead of having a local hack taking care of sending the tuning
command and as well to verify the response pattern, let's convert to
the common mmc_send_tuning() API.

This change affects the Exynos variant, since it's the only one which
support the dw_mmc's ->execute_tuning() callback.

It's seems like dw_mmc internal logic expects failed data transfers to
be ended using a stop command. Let the tuning requests also fall into
this category, since there are data transfer involved.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: Alim Akhtar <alim.akhtar@samsung.com>
Tested-by: Alim Akhtar <alim.akhtar@samsung.com>
9 years agommc: mmci: Do pm_runtime_put() after the host has been added
Ulf Hansson [Thu, 11 Dec 2014 13:35:55 +0000 (14:35 +0100)]
mmc: mmci: Do pm_runtime_put() after the host has been added

Previously the pm_runtime_put() caused the device to be runtime PM
suspended, but then immediately being resumed when we add the host.

Prevent this unnecessary runtime PM suspend/resume cycle during
->probe() by moving the call to pm_runtime_put() after mmc_add_host().

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-pci: Remove redundant runtime PM idle callback
Ulf Hansson [Wed, 10 Dec 2014 15:49:23 +0000 (16:49 +0100)]
mmc: sdhci-pci: Remove redundant runtime PM idle callback

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-esdhc-imx: Fixup runtime PM conditions during ->probe()
Ulf Hansson [Thu, 11 Dec 2014 14:12:25 +0000 (15:12 +0100)]
mmc: sdhci-esdhc-imx: Fixup runtime PM conditions during ->probe()

By configure runtime PM prior we enable the use of it, we close the gap
for strange and unhandled conditions. Moreover it makes us rely on the
driver core, after finalized ->probe(), to request an inactive device
to become runtime PM idle/suspended, which earlier potentially could
happen already at pm_runtime_set_autosuspend_delay().

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-esdhc-imx: Fixup ->remove()
Ulf Hansson [Thu, 11 Dec 2014 13:56:15 +0000 (14:56 +0100)]
mmc: sdhci-esdhc-imx: Fixup ->remove()

Since sdhci_remove_host() access the device, we need to make sure it's
runtime PM resumed.

To handle clock gating properly in remove, we first have to make sure
the clocks are ungated.

To fix both these issues, by invoking pm_runtime_get_sync() early in
->remove() callback. Later once we disabled runtime PM, we can restore
the usage count.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: mvsdio: don't use devm_pinctrl_get_select_default() in probe
Wolfram Sang [Mon, 22 Dec 2014 22:16:54 +0000 (23:16 +0100)]
mmc: mvsdio: don't use devm_pinctrl_get_select_default() in probe

Since commit ab78029ecc34 (drivers/pinctrl: grab default handles from device
core), we can rely on device core for setting the default pins.

Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: dw_mmc: remove unnecessary debug message
Beomho Seo [Tue, 23 Dec 2014 12:07:33 +0000 (21:07 +0900)]
mmc: dw_mmc: remove unnecessary debug message

This patch remove unnecessary 'out of memory' message on dw mmc driver.

Signed-off-by: Beomho Seo <beomho.seo@samsung.com>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: rtsx: swap function position to avoid pre declaration
Micky Ching [Tue, 23 Dec 2014 01:19:46 +0000 (09:19 +0800)]
mmc: rtsx: swap function position to avoid pre declaration

move function sd_read_data()/sd_write_data() behind
sd_send_cmd_get_rsp() to avoid pre-declaration.

Signed-off-by: Micky Ching <micky_ching@realsil.com.cn>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: rtsx: add support for sdio card
Micky Ching [Tue, 23 Dec 2014 01:19:45 +0000 (09:19 +0800)]
mmc: rtsx: add support for sdio card

Modify transfer mode for support sdio card,
send cmd and data at the same time for read data transfer,
but send data after cmd for write data transfer.

Signed-off-by: Micky Ching <micky_ching@realsil.com.cn>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: rtsx: add helper function to simplify code
Micky Ching [Tue, 23 Dec 2014 01:19:44 +0000 (09:19 +0800)]
mmc: rtsx: add helper function to simplify code

To support sdio card, transfer mode need make a change,
this need to split code and use it in different place,
Add new function to simplify repeat operation.

Signed-off-by: Micky Ching <micky_ching@realsil.com.cn>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: rtsx: init cookie at probe/card_event
Micky Ching [Tue, 23 Dec 2014 01:19:43 +0000 (09:19 +0800)]
mmc: rtsx: init cookie at probe/card_event

host->cookie is used for handle async request,
we should init it to negative value when new card inserted,
make cookie value invalid.

Signed-off-by: Micky Ching <micky_ching@realsil.com.cn>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: rtsx: add dump_reg_range to simplify dump register
Micky Ching [Tue, 23 Dec 2014 01:19:42 +0000 (09:19 +0800)]
mmc: rtsx: add dump_reg_range to simplify dump register

Add a new function to dump register within a range.
We print 1 register a line before this patch,
this may make debug info too long when we add more register to dump.

The new dump_reg_range() dump to 8 register a line,
and it is easy to use.

Signed-off-by: Micky Ching <micky_ching@realsil.com.cn>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: slot-gpio: Rework how to handle allocation of slot-gpio data
Ulf Hansson [Thu, 18 Dec 2014 14:44:36 +0000 (15:44 +0100)]
mmc: slot-gpio: Rework how to handle allocation of slot-gpio data

By moving the allocation of the slot-gpio data into mmc_alloc_host(),
we can remove the slot-gpio internal calls to mmc_gpio_alloc().

This means mmc_gpio_alloc() has now only one caller left, which
consequence allow us to simplify and remove some of the slot-gpio code.

Additionally, this makes the slot-gpio mutex redundant, so let's remove
it.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: Free all resources for the class device at ->dev_release()
Ulf Hansson [Thu, 18 Dec 2014 14:44:35 +0000 (15:44 +0100)]
mmc: core: Free all resources for the class device at ->dev_release()

To be consistent when freeing data, let's move the idr_remove() call
from mmc_free_host() into the ->dev_release() callback for the class
device.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: slot-gpio: Make mmc_gpio_alloc() available for MMC core
Ulf Hansson [Thu, 18 Dec 2014 14:44:34 +0000 (15:44 +0100)]
mmc: slot-gpio: Make mmc_gpio_alloc() available for MMC core

As a step in moving slot-gpio functions/structs closer to the MMC core,
let's add a local header file for slot-gpio.

In this initial step we move mmc_gpio_alloc() into the header file,
to make it available for the MMC core. Following patches will show the
utilization of it.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: slot-gpio: Use the parent device while allocating data
Ulf Hansson [Thu, 18 Dec 2014 14:44:33 +0000 (15:44 +0100)]
mmc: slot-gpio: Use the parent device while allocating data

We had a mix of using the class device and the parent device while
allocating data through the devm_* managed functions.

Let's be more consistent and always use the parent device.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: slot-gpio: Remove option to explicitly free requested CD/WP GPIOs
Ulf Hansson [Thu, 18 Dec 2014 14:44:32 +0000 (15:44 +0100)]
mmc: slot-gpio: Remove option to explicitly free requested CD/WP GPIOs

The slot-gpio uses the devm*_ managed functions. Still it provide APIs
to explicitly free requested CD/WP GPIOs, but these API isn't being
used.

Therefore let's simplify slot-gpio by removing these unused APIs. If it
later turns out we need some of them, we can always consider to restore
the code.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-sirf: Remove redundant use of mmc_gpio_free_cd()
Ulf Hansson [Thu, 18 Dec 2014 14:44:31 +0000 (15:44 +0100)]
mmc: sdhci-sirf: Remove redundant use of mmc_gpio_free_cd()

The MMC core makes use of the devm_* functions while requesting the
CD/WP GPIOs, let's rely on that.

Cc: Barry Song <baohua@kernel.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: mvsdio: Remove redundant use of mmc_gpio_free_cd|ro()
Ulf Hansson [Thu, 18 Dec 2014 14:44:30 +0000 (15:44 +0100)]
mmc: mvsdio: Remove redundant use of mmc_gpio_free_cd|ro()

The MMC core makes use of the devm_* functions while requesting the
CD/WP GPIOs, let's rely on that.

Cc: Nicolas Pitre <nico@fluxnic.net>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: Return error at failure of request CD/WP in mmc_of_parse()
Ulf Hansson [Thu, 18 Dec 2014 09:41:44 +0000 (10:41 +0100)]
mmc: core: Return error at failure of request CD/WP in mmc_of_parse()

Instead of just printing an error when mmc_of_parse() fails to request
CD/WP GPIO pins, let's propagate all errors, except for -ENOENT.
Earlier only -EPROBE_DEFER was handled correctly.

As a side effect of this change and by relying on host drivers to
handle the errors during ->probe(), we don't need to free any data in
the error path.

This also means we are actually fixing a bug, since we remove the call
to mmc_gpio_free_cd() which wasn't the correct function to invoke to
handle cleanup. Instead that should have been mmc_gpiod_free_cd().

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-st: Do cleanup while mmc_of_parse() return an error
Ulf Hansson [Thu, 18 Dec 2014 09:41:43 +0000 (10:41 +0100)]
mmc: sdhci-st: Do cleanup while mmc_of_parse() return an error

At errors, balance sdhci_pltfm_init() with sdhci_pltfm_free(), to make
sure all data is freed properly.

Cc: Srinivas Kandagatla <srinivas.kandagatla@gmail.com>
Cc: Maxime Coquelin <maxime.coquelin@st.com>
Cc: Patrice Chotard <patrice.chotard@st.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Maxime Coquelin <maxime.coquelin@st.com>
9 years agommc: sdhci-s3c: Handle error from mmc_of_parse()
Ulf Hansson [Thu, 18 Dec 2014 09:41:42 +0000 (10:41 +0100)]
mmc: sdhci-s3c: Handle error from mmc_of_parse()

Since mmc_of_parse() may fail, let's deal with it and thus do proper
error handling.

Cc: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
9 years agommc: sdhci-of-esdhc: Handle error from mmc_of_parse()
Ulf Hansson [Thu, 18 Dec 2014 09:41:41 +0000 (10:41 +0100)]
mmc: sdhci-of-esdhc: Handle error from mmc_of_parse()

Since mmc_of_parse() may fail, let's deal with it and thus do proper
error handling.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-bcm-kona: Handle error from mmc_of_parse()
Ulf Hansson [Thu, 18 Dec 2014 09:41:40 +0000 (10:41 +0100)]
mmc: sdhci-bcm-kona: Handle error from mmc_of_parse()

Since mmc_of_parse() may fail, let's deal with it and thus do proper
error handling.

Cc: Christian Daudt <bcm@fixthebug.org>
Cc: Matt Porter <mporter@linaro.org>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
9 years agommc: moxart: Handle error from mmc_of_parse()
Ulf Hansson [Thu, 18 Dec 2014 09:41:39 +0000 (10:41 +0100)]
mmc: moxart: Handle error from mmc_of_parse()

Since mmc_of_parse() may fail, let's deal with it and thus do proper
error handling.

Cc: Jonas Jensen <jonas.jensen@gmail.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: dw_mmc: exynos: move definitions to header file
Seungwon Jeon [Mon, 22 Dec 2014 12:12:04 +0000 (17:42 +0530)]
mmc: dw_mmc: exynos: move definitions to header file

Move exynos related definition to header file. And this
also changes some of the registers name to match the standard
naming convention.

Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
[Alim: updated the commit message]
Signed-off-by: Alim Akhtar <alim.akhtar@samsung.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: dw_mmc: exynos: don't use if clock isn't available
Seungwon Jeon [Mon, 22 Dec 2014 12:12:03 +0000 (17:42 +0530)]
mmc: dw_mmc: exynos: don't use if clock isn't available

Add checking whether the clock is valid.

Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Alim Akhtar <alim.akhtar@samsung.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: dw_mmc: fix the max_blk_count in IDMAC
Seungwon Jeon [Mon, 22 Dec 2014 12:12:02 +0000 (17:42 +0530)]
mmc: dw_mmc: fix the max_blk_count in IDMAC

Even though 1MB is reserved for descriptor table in IDMAC,
the dw_mmc host driver is allowed to receive only maximum
128KB block length in one request. This is caused by setting
improper max_blk_count. It needs to be e adjusted so that
descriptor table is used fully. It is found that the performance
is improved with the increased the max_blk_count.

Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Acked-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Alim Akhtar <alim.akhtar@samsung.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: mxs-mmc: remove unneeded NULL check
Dan Carpenter [Tue, 16 Dec 2014 23:57:50 +0000 (02:57 +0300)]
mmc: mxs-mmc: remove unneeded NULL check

Static checkers complain about the inconsistent NULL checking here:

drivers/mmc/host/mxs-mmc.c:680 mxs_mmc_probe()
warn: variable dereferenced before check 'ssp->dmach' (see line 660)

The variable can't actually be NULL so we can remove the check.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sunxi: Removing unused code
David Lanzendörfer [Tue, 16 Dec 2014 14:11:15 +0000 (15:11 +0100)]
mmc: sunxi: Removing unused code

Removing a relict from reverse engineering of the Android driver code
in sunxi_mmc_clk_set_rate.

Signed-off-by: David Lanzendörfer <david.lanzendoerfer@o2s.ch>
Reported-by: <lixiang@allwinnertech.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sunxi: Correcting SDXC_HARDWARE_RESET bit
David Lanzendörfer [Tue, 16 Dec 2014 14:11:10 +0000 (15:11 +0100)]
mmc: sunxi: Correcting SDXC_HARDWARE_RESET bit

Fixing the register name in sunxi_mmc_reset_host since the
SDXC_HARDWARE_RESET bit is actually located within REG_GCTRL and not
REG_CMDR as it was pointed out by Allwinner.

Signed-off-by: David Lanzendörfer <david.lanzendoerfer@o2s.ch>
Reported-by: <lixiang@allwinnertech.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sunxi: Lock fix
David Lanzendörfer [Tue, 16 Dec 2014 14:11:04 +0000 (15:11 +0100)]
mmc: sunxi: Lock fix

1) Adding a comment in order to clarify the choice of the locks within
sunxi_mmc_handle_manual_stop

2) As <lixiang@allwinnertech.com> has pointed out the wait_dma variable
was not accessed within the spin lock block in sunxi_mmc_request and so
(even if it should never happend) it would have theoretically been
possible that some other function would access the variable at the same
time as the function. This has been changed now and the function is
using local variables outside the lock and copys the value over during
the lock phase.

Signed-off-by: David Lanzendörfer <david.lanzendoerfer@o2s.ch>
Reported-by: <lixiang@allwinnertech.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sunxi: Fix setup of last descriptor of dma transfer
Hans de Goede [Tue, 16 Dec 2014 14:10:59 +0000 (15:10 +0100)]
mmc: sunxi: Fix setup of last descriptor of dma transfer

The last descriptor might be the first descriptor as well, so use masking to
add the LD (last descriptor) bit and drop the DIC (disable interrupt on
completion) bit rather then hard assignment as hard assignment will override
the FD (first descriptor) bit if there is only 1 descriptor.

Also set the ER (end of ring) bit and clear buf_addr_ptr2 on the last
descriptor, like the android kernel code does.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: David Lanzendörfer <david.lanzendoerfer@o2s.ch>
Reported-by: <lixiang@allwinnertech.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: omap_hsmmc: remove prepare/complete system suspend support.
NeilBrown [Thu, 11 Dec 2014 21:43:08 +0000 (08:43 +1100)]
mmc: omap_hsmmc: remove prepare/complete system suspend support.

The only function of these 'prepare' and 'complete' is to
disable the 'card detect' irq during suspend.

The commit which added this,
commit a48ce884d5819d5df2cf1139ab3c43f8e9e419b3
    mmc: omap_hsmmc: Introduce omap_hsmmc_prepare/complete

justified it by the need to avoid the registration of new devices
during suspend.
However mmc_pm_notify will set ->rescan_disable in the 'prepare'
stage and clear it in the 'complete' stage, so no card detection
will actually happen.
Also the interrupt will be disabled before final suspend as part
of common suspend processing.

So this disabling of the interrupt is unnecessary, and interferes
with a transition to using common code for card-detect management.

Cc: Felipe Balbi <balbi@ti.com>
Cc: Venkatraman S <svenkatr@ti.com>
Cc: Chris Ball <cjb@laptop.org>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: toshsd: Fix unbalanced locking
Axel Lin [Thu, 11 Dec 2014 03:10:44 +0000 (11:10 +0800)]
mmc: toshsd: Fix unbalanced locking

Fix returning IRQ_HANDLED with spin_lock held.

Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: toshsd: use module_pci_driver to simplify the code
Wei Yongjun [Tue, 9 Dec 2014 01:08:05 +0000 (09:08 +0800)]
mmc: toshsd: use module_pci_driver to simplify the code

Use the module_pci_driver() macro to make the code simpler
by eliminating module_init and module_exit calls.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: Fix FSL ESDHC reset handling quirk
Alessio Igor Bogani [Tue, 9 Dec 2014 08:40:38 +0000 (09:40 +0100)]
mmc: sdhci: Fix FSL ESDHC reset handling quirk

The commit 0718e59ae259 ("mmc: sdhci: move FSL ESDHC reset handling quirk into
esdhc code") states that Freescale esdhc is the only controller which needs
the interrupt registers restored after a reset. So it moves
SDHCI_QUIRK_RESTORE_IRQS_AFTER_RESET quirk handling code into the
esdhc-imx driver only. Unfortunately the same controller is used in
other boards which use the of-esdhc driver instead (like powerpc P2020).

Restore interrupts after reset in the sdhci-of-esdhc driver also.

Signed-off-by: Alessio Igor Bogani <alessio.bogani@elettra.eu>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-sirf: add sirf tuning function (cmd 19)
Minda Chen [Thu, 4 Dec 2014 12:09:20 +0000 (20:09 +0800)]
mmc: sdhci-sirf: add sirf tuning function (cmd 19)

Add manual tuning function in CSR atlas7 SoC. It is mainly used
for the UHS-I SD card working SDR50 SDR104 mode.

The tuning principle can be seen in SD spec part1 v3.01 4.2.4.5
(tuning command).

SD host send the cmd19 and set the delay value(0-127).
and the sdcard return 64 bytes data. If the data is same with
the tuning data. The delay value is valid. Execute this commmand
128 times. And calculate the longest window of the valid values.
The value in the middle of this window is the best value.

Signed-off-by: Minda Chen <Minda.Chen@csr.com>
Signed-off-by: Barry Song <Baohua.Song@csr.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: dw_mmc: Protect read-modify-write of INTMASK with a lock
Doug Anderson [Tue, 2 Dec 2014 23:42:47 +0000 (15:42 -0800)]
mmc: dw_mmc: Protect read-modify-write of INTMASK with a lock

We're running into cases where our enabling of the SDIO interrupt in
dw_mmc doesn't actually take effect.  Specifically, adding patch like
this:

 +++ b/drivers/mmc/host/dw_mmc.c
 @@ -1076,6 +1076,9 @@ static void dw_mci_enable_sdio_irq(struct mmc_host *mmc, int enb)

      mci_writel(host, INTMASK,
           (int_mask | SDMMC_INT_SDIO(slot->id)));
 +    int_mask = mci_readl(host, INTMASK);
 +    if (!(int_mask & SDMMC_INT_SDIO(slot->id)))
 +      dev_err(&mmc->class_dev, "failed to enable sdio irq\n");
    } else {

...actually triggers the error message.  That's because the
dw_mci_enable_sdio_irq() unsafely does a read-modify-write of the
INTMASK register.

We can't just use the standard host->lock since that lock is not irq
safe and mmc_signal_sdio_irq() (called from interrupt context) calls
dw_mci_enable_sdio_irq().  Add a new irq-safe lock to protect INTMASK.

An alternate solution to this is to punt mmc_signal_sdio_irq() to the
tasklet and then protect INTMASK modifications by the standard host
lock.  This seemed like a bit more of a high-latency change.

Reported-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-by: James Hogan <james.hogan@imgtec.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: dw_mmc: Cleanup disable of low power mode w/ SDIO interrupts
Doug Anderson [Tue, 2 Dec 2014 23:42:46 +0000 (15:42 -0800)]
mmc: dw_mmc: Cleanup disable of low power mode w/ SDIO interrupts

In the patch (9623b5b mmc: dw_mmc: Disable low power mode if SDIO
interrupts are used) I added code that disabled the low power mode of
dw_mmc when SDIO interrupts are used.  That code worked but always
felt a little hacky because we ended up disabling low power as a side
effect of the first enable_sdio_irq() call.  That wouldn't be so bad
except that disabling low power involves a complicated process of
writing to the CMD/CMDARG registers and that extra process makes it
difficult to cleanly the read-modify-write race in
dw_mci_enable_sdio_irq() (see future patch in the series).

Change the code to take advantage of the init_card() callback of the
mmc core to do this right at bootup.

Signed-off-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: core: Support the optional init_card() callback for MMC and SD
Doug Anderson [Tue, 2 Dec 2014 23:42:45 +0000 (15:42 -0800)]
mmc: core: Support the optional init_card() callback for MMC and SD

In (3fcb027 ARM: MXC: mxcmmc: work around a bug in the SDHC busy line
handling) the optional init_card() callback was added.  According to
the original change it was "for now only called from
mmc_sdio_init_card()".

This callback really ought to be called from the SD and MMC init
functions as well.  One current user of this callback
(mxcmci_init_card) will not work as expected if you insert an SDIO
card, then eject it and put a normal SD card in.  Specifically the
normal SD card will not get to run with 4-bit data.

I'd like to use the init_card() callback to handle a similar quirk on
dw_mmc when using SDIO Interrupts (the "low power" feature of the card
needs to be disabled), so that will add a second user of the function.

Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-by: Grant Grundler <grundler@chromium.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agoARM: OMAP2+: Make sure pandora_wl1251_init_card() applies to SDIO only
Doug Anderson [Tue, 2 Dec 2014 23:42:44 +0000 (15:42 -0800)]
ARM: OMAP2+: Make sure pandora_wl1251_init_card() applies to SDIO only

In preparation for having init_card() called for all card types (not
just SDIO), change pandora_wl1251_init_card() so it checks whether the
card type is SDIO.

Signed-off-by: Doug Anderson <dianders@chromium.org>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agoLinux 3.19-rc5
Linus Torvalds [Sun, 18 Jan 2015 06:02:20 +0000 (18:02 +1200)]
Linux 3.19-rc5

9 years agoMerge tag 'armsoc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm...
Linus Torvalds [Sun, 18 Jan 2015 06:00:40 +0000 (18:00 +1200)]
Merge tag 'armsoc-for-linus' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC fixes from Olof Johansson:
 "We've been sitting on our fixes branch for a while, so this batch is
  unfortunately on the large side.

  A lot of these are tweaks and fixes to device trees, fixing various
  bugs around clocks, reg ranges, etc.  There's also a few defconfig
  updates (which are on the late side, no more of those).

  All in all the diffstat is bigger than ideal at this time, but nothing
  in here seems particularly risky"

* tag 'armsoc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (31 commits)
  reset: sunxi: fix spinlock initialization
  ARM: dts: disable CCI on exynos5420 based arndale-octa
  drivers: bus: check cci device tree node status
  ARM: rockchip: disable jtag/sdmmc autoswitching on rk3288
  ARM: nomadik: fix up leftover device tree pins
  ARM: at91: board-dt-sama5: add phy_fixup to override NAND_Tree
  ARM: at91/dt: sam9263: Add missing clocks to lcdc node
  ARM: at91: sama5d3: dt: correct the sound route
  ARM: at91/dt: sama5d4: fix the timer reg length
  ARM: exynos_defconfig: Enable LM90 driver
  ARM: exynos_defconfig: Enable options for display panel support
  arm: dts: Use pmu_system_controller phandle for dp phy
  ARM: shmobile: sh73a0 legacy: Set .control_parent for all irqpin instances
  ARM: dts: berlin: correct BG2Q's SM GPIO location.
  ARM: dts: berlin: add broken-cd and set bus width for eMMC in Marvell DMP DT
  ARM: dts: berlin: fix io clk and add missing core clk for BG2Q sdhci2 host
  ARM: dts: Revert disabling of smc91x for n900
  ARM: dts: imx51-babbage: Fix ULPI PHY reset modelling
  ARM: dts: dra7-evm: fix qspi device tree partition size
  ARM: omap2plus_defconfig: use CONFIG_CPUFREQ_DT
  ...

9 years agoMerge tag 'clk-fixes-for-linus' of git://git.linaro.org/people/mike.turquette/linux
Linus Torvalds [Sun, 18 Jan 2015 03:29:11 +0000 (15:29 +1200)]
Merge tag 'clk-fixes-for-linus' of git://git.linaro.org/people/mike.turquette/linux

Pull clock driver fixes from Mike Turquette:
 "Small number of fixes for clock drivers and a single null pointer
  dereference fix in the framework core code.

  The driver fixes vary from fixing section mismatch warnings to
  preventing machines from hanging (and preventing developers from
  crying)"

* tag 'clk-fixes-for-linus' of git://git.linaro.org/people/mike.turquette/linux:
  clk: fix possible null pointer dereference
  Revert "clk: ppc-corenet: Fix Section mismatch warning"
  clk: rockchip: fix deadlock possibility in cpuclk
  clk: berlin: bg2q: remove non-exist "smemc" gate clock
  clk: at91: keep slow clk enabled to prevent system hang
  clk: rockchip: fix rk3288 cpuclk core dividers
  clk: rockchip: fix rk3066 pll lock bit location
  clk: rockchip: Fix clock gate for rk3188 hclk_emem_peri
  clk: rockchip: add CLK_IGNORE_UNUSED flag to fix rk3066/rk3188 USB Host

9 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sun, 18 Jan 2015 03:26:52 +0000 (15:26 +1200)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "This is one fix for a Multiqueue sleeping in invalid context problem
  and a MAINTAINER file update for Qlogic"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: ->queue_rq can't sleep
  MAINTAINERS: Update maintainer list for qla4xxx

9 years agoclk: fix possible null pointer dereference
Stanimir Varbanov [Mon, 5 Jan 2015 16:04:23 +0000 (18:04 +0200)]
clk: fix possible null pointer dereference

The commit 646cafc6 (clk: Change clk_ops->determine_rate to
return a clk_hw as the best parent) opens a possibility for
null pointer dereference, fix this.

Signed-off-by: Stanimir Varbanov <svarbanov@mm-sol.com>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Michael Turquette <mturquette@linaro.org>
9 years agoRevert "clk: ppc-corenet: Fix Section mismatch warning"
Kevin Hao [Wed, 3 Dec 2014 08:53:51 +0000 (16:53 +0800)]
Revert "clk: ppc-corenet: Fix Section mismatch warning"

This reverts commit da788acb28386aa896224e784954bb73c99ff26c.

That commit tried to fix the section mismatch warning by moving the
ppc_corenet_clk_driver struct to init section. This is definitely wrong
because the kernel would free the memories occupied by this struct
after boot while this driver is still registered in the driver core.
The kernel would panic when accessing this driver struct.

Cc: stable@vger.kernel.org # 3.17
Signed-off-by: Kevin Hao <haokexin@gmail.com>
Acked-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Michael Turquette <mturquette@linaro.org>
9 years agoclk: rockchip: fix deadlock possibility in cpuclk
Heiko Stübner [Fri, 16 Jan 2015 16:52:44 +0000 (17:52 +0100)]
clk: rockchip: fix deadlock possibility in cpuclk

Lockdep reported a possible deadlock between the cpuclk lock and for example
the i2c driver.

       CPU0                    CPU1
       ----                    ----
  lock(clk_lock);
                               local_irq_disable();
                               lock(&(&i2c->lock)->rlock);
                               lock(clk_lock);
  <Interrupt>
    lock(&(&i2c->lock)->rlock);

 *** DEADLOCK ***

The generic clock-types of the core ccf already use spin_lock_irqsave when
touching clock registers, so do the same for the cpuclk.

Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Reviewed-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Michael Turquette <mturquette@linaro.org>
[mturquette@linaro.org: removed initialization of "flags"]

9 years agoMerge branch 'fixes' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Sat, 17 Jan 2015 18:26:24 +0000 (06:26 +1200)]
Merge branch 'fixes' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine fixes from Vinod Koul:
 "Two patches, the first by Andy to fix dw dmac runtime pm and second
  one by me to fix the dmaengine headers in MAINTAINERS"

* 'fixes' of git://git.infradead.org/users/vkoul/slave-dma:
  dmaengine: dw: balance PM runtime calls
  MAINTAINERS: dmaengine: fix the header file for dmaengine

9 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 17 Jan 2015 18:24:30 +0000 (06:24 +1200)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Ingo Molnar:
 "Mostly tooling fixes, but also two PMU driver fixes"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf tools powerpc: Use dwfl_report_elf() instead of offline.
  perf tools: Fix segfault for symbol annotation on TUI
  perf test: Fix dwarf unwind using libunwind.
  perf tools: Avoid build splat for syscall numbers with uclibc
  perf tools: Elide strlcpy warning with uclibc
  perf tools: Fix statfs.f_type data type mismatch build error with uclibc
  tools: Remove bitops/hweight usage of bits in tools/perf
  perf machine: Fix __machine__findnew_thread() error path
  perf tools: Fix building error in x86_64 when dwarf unwind is on
  perf probe: Propagate error code when write(2) failed
  perf/x86/intel: Fix bug for "cycles:p" and "cycles:pp" on SLM
  perf/rapl: Fix sysfs_show() initialization for RAPL PMU

9 years agoMerge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git...
Ingo Molnar [Sat, 17 Jan 2015 10:04:35 +0000 (11:04 +0100)]
Merge tag 'perf-urgent-for-mingo' of git://git./linux/kernel/git/acme/linux into perf/urgent

Pull perf/urgent fixes from Arnaldo Carvalho de Melo:

  - Fix segfault when using both the map symtab viewer and annotation
    in the TUI (Namhyung Kim).

  - uClibc build fixes (Alexey Brodkin, Vineet Gupta).

  - bitops/hweight were moved from tools/perf/ too tools/include, move
    some leftovers (Arnaldo Carvalho de Melo)

  - Fix dwarf unwind x86_64 build error (Namhyung Kim)

  - Fix __machine__findnew_thread() error path (Namhyung Kim)

  - Propagate error code when write(2) failed in 'perf probe' (Namhyung Kim)

  - Use dwfl_report_elf() instead of offline in powerpc bits to
    properly handle non prelinked DSOs (Sukadev Bhattiprolu).

  - Fix dwarf unwind using libunwind in 'perf test' (Wang Nan)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoMerge tag 'samsung-fixes-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene...
Olof Johansson [Thu, 15 Jan 2015 23:45:18 +0000 (15:45 -0800)]
Merge tag 'samsung-fixes-3.19' of git://git./linux/kernel/git/kgene/linux-samsung into fixes

Merge "Samsung fixes for v3.19" from Kukjin Kim:

Samsung fixes for v3.19
- exynos_defconfig: enable LM90 driver and display panel support
   - HWMON
   - SENSORS_LM90
   - Direct Rendering Manager (DRM)
   - DRM bridge registration and lookup framework
   - Parade ps8622/ps8625 eDP/LVDS bridge
   - NXP ptn3460 eDP/LVDS bridge
   - Exynos Fully Interactive Mobile Display controller (FIMD)
   - Panel registration and lookup framework
   - Simple panels
   - Backlight & LCD device support

- use pmu_system_controller phandle for dp phy
  : DP PHY requires pmu_system_controller to handle PMU reg. now

* tag 'samsung-fixes-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung:
  ARM: exynos_defconfig: Enable LM90 driver
  ARM: exynos_defconfig: Enable options for display panel support
  arm: dts: Use pmu_system_controller phandle for dp phy

Signed-off-by: Olof Johansson <olof@lixom.net>
9 years agoreset: sunxi: fix spinlock initialization
Tyler Baker [Mon, 12 Jan 2015 15:54:46 +0000 (07:54 -0800)]
reset: sunxi: fix spinlock initialization

Call spin_lock_init() before the spinlocks are used, both in early init
and probe functions preventing a lockdep splat.

I have been observing lockdep complaining [1] during boot on my a80 optimus [2]
when CONFIG_PROVE_LOCKING has been enabled. This patch resolves the splat,
and has been tested on a few other sunxi platforms without issue.

[1] http://storage.kernelci.org/next/next-20150107/arm-multi_v7_defconfig+CONFIG_PROVE_LOCKING=y/lab-tbaker/boot-sun9i-a80-optimus.html
[2] http://kernelci.org/boot/?a80-optimus

Signed-off-by: Tyler Baker <tyler.baker@linaro.org>
Cc: <stable@vger.kernel.org>
Acked-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kevin Hilman <khilman@linaro.org>
Signed-off-by: Olof Johansson <olof@lixom.net>
9 years agoMerge tag 'renesas-soc-fixes-for-v3.19' of git://git.kernel.org/pub/scm/linux/kernel...
Olof Johansson [Tue, 13 Jan 2015 23:32:43 +0000 (15:32 -0800)]
Merge tag 'renesas-soc-fixes-for-v3.19' of git://git./linux/kernel/git/horms/renesas into fixes

Merge "Renesas ARM Based SoC Fixes for v3.19" from Simon Horman:

Renesas ARM Based SoC Fixes for v3.19

This pull request is based on the last round of SoC updates for v3.19,
Fourth Round of Renesas ARM Based SoC Updates for v3.19, tagged as
renesas-soc3-for-v3.19, merged into your next/soc branch and included in
v3.19-rc1.

- ARM: shmobile: r8a7740: Instantiate GIC from C board code in legacy builds

  Set .control_parent for all irqpin instances for sh73a0 SoC when booting
  using legacy C.

- ARM: shmobile: r8a7740: Instantiate GIC from C board code in legacy builds

  This fixes a long standing problem which has been present since
  the sh73a0 SoC started using the INTC External IRQ pin driver.

  The patch that introduced the problem is 341eb5465f67437a ("ARM:
  shmobile: INTC External IRQ pin driver on sh73a0") which was included
  in v3.10.

* tag 'renesas-soc-fixes-for-v3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas:
  ARM: shmobile: sh73a0 legacy: Set .control_parent for all irqpin instances
  ARM: shmobile: r8a7740: Instantiate GIC from C board code in legacy builds

9 years agoARM: dts: disable CCI on exynos5420 based arndale-octa
Abhilash Kesavan [Sat, 10 Jan 2015 03:11:36 +0000 (08:41 +0530)]
ARM: dts: disable CCI on exynos5420 based arndale-octa

The arndale-octa board was giving "imprecise external aborts" during
boot-up with MCPM enabled. CCI enablement of the boot cluster was found
to be the cause of these aborts (possibly because the secure f/w was not
allowing it). Hence, disable CCI for the arndale-octa board.

Signed-off-by: Abhilash Kesavan <a.kesavan@samsung.com>
Tested-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Tested-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Tyler Baker <tyler.baker@linaro.org>
Signed-off-by: Olof Johansson <olof@lixom.net>
9 years agodrivers: bus: check cci device tree node status
Abhilash Kesavan [Sat, 10 Jan 2015 03:11:35 +0000 (08:41 +0530)]
drivers: bus: check cci device tree node status

The arm-cci driver completes the probe sequence even if the cci node is
marked as disabled. Add a check in the driver to honour the cci status
in the device tree.

Signed-off-by: Abhilash Kesavan <a.kesavan@samsung.com>
Acked-by: Sudeep Holla <sudeep.holla@arm.com>
Acked-by: Nicolas Pitre <nico@linaro.org>
Tested-by: Sudeep Holla <sudeep.holla@arm.com>
Tested-by: Kevin Hilman <khilman@linaro.org>
Signed-off-by: Olof Johansson <olof@lixom.net>