Since there no users of the struct sdhci_host, but the shdci host
drivers themselves, let's move the definition of it to the local sdhci
header.

The exported sdhci header then becomes empty, so let's remove it.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Since commit cd1e6504 ("of/device: Don't register disabled
devices"), the disabled device will not be registered at all. So
we don't need to do the check again in the platform device driver.

And the check in the current code is useless even if we really
run into a disabled device. In this case, it just doesn't parse
the dtb for the infos such as quirks or clock, but it will continue
to try to init the disabled device after that check. So just remove it.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Setup a different set of sdhci_ops for tegra114 and later so that
the write_w callback is only used on tegra114. This allows us to
remove the NVQUIRK_SHADOW_XFER_MODE_REG and simply the logic
in tegra_sdhci_writew.
Suggested-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NRhyland Klein <rklein@nvidia.com>
Acked-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

A command end interrupt should not be processed between command issue
and setting of wait_for flag. It expects already the flag to be set.
Therefore the exclusive control was added.
Signed-off-by: NKouichi Tomita <kouichi.tomita.yn@renesas.com>
Signed-off-by: NYoshihiro Kaneko <ykaneko0929@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

If interruption of command already occurred, mrq pointer in dev_err()
would refer to NULL, because the host-state is changed to STATE_IDLE
and mrq pointer is changed to NULL by interrupt handler.
Therefore dev_err is moved after checking STATE_IDLE.
Signed-off-by: NKouichi Tomita <kouichi.tomita.yn@renesas.com>
Signed-off-by: NYoshihiro Kaneko <ykaneko0929@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

It was brought to my attention that the sdhci-pxav3 driver is needed
on a few more ARM machines than I initially thought. Add the missing
architectures to the dependency list.

Credits to Peter Robinson for noticing my mistake and reporting.
Reported-by: NPeter Robinson <pbrobinson@gmail.com>
Signed-off-by: NJean Delvare <jdelvare@suse.de>
Cc: Chris Ball <chris@printf.net>
Cc: Ulf Hansson <ulf.hansson@linaro.org>
Cc: Eric Miao <eric.y.miao@gmail.com>
Cc: Haojian Zhuang <haojian.zhuang@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The alloc() and free() hooks required each pwrseq implementation to set
host->pwrseq themselves. This is error-prone and could be done at a
higher level if alloc() was changed to return a pointer to a struct
mmc_pwrseq instead of an error code.

This patch performs this change and moves the burden of maintaining
host->pwrseq from the power sequence hooks to the pwrseq code.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The sunxi mmc driver tries to calculate a dma address by using pointer
arithmetic, which causes a warning when dma_addr_t is wider than a pointer:

drivers/mmc/host/sunxi-mmc.c: In function 'sunxi_mmc_init_idma_des':
drivers/mmc/host/sunxi-mmc.c:296:35: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
  struct sunxi_idma_des *pdes_pa = (struct sunxi_idma_des *)host->sg_dma;
                                   ^

To avoid this warning and to simplify the logic, this changes
the code to avoid the cast and calculate the correct address
manually. The behavior should be unchanged.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NDavid Lanzendörfer <david.lanzendoerfer@o2s.ch>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The PPC_OF is a ppc specific option which is used to mean that the
firmware device tree access functions are available. Since all the
ppc platforms have a device tree, it is aways set to 'y' for ppc.
So it makes no sense to keep a such option in the current kernel.
Replace it with PPC.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Add IPROC SDHCI driver for IPROC family of Broadcom devices.
Acked-by: NRay Jui <rjui@broadcom.com>
Signed-off-by: NCorneliu Doban <cdoban@broadcom.com>
Signed-off-by: NScott Branden <sbranden@broadcom.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

For CMD53 in block mode, the host does not need to stop the transfer,
as it stops when the block count (present in CMD53) is reached.
Signed-off-by: NCorneliu Doban <cdoban@broadcom.com>
Signed-off-by: NScott Branden <sbranden@broadcom.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Add quirk to handle broken auto-CMD23.
Some controllers do not respond after the first auto-CMD23 is issued.

This allows CMD23 to still work (mandatory for the faster UHS-I mode)
rather than disabling CMD23 entirely via SDHCI_QUIRK2_HOST_NO_CMD23.

Signed-off by: Corneliu Doban <cdoban@broadcom.com>
Signed-off-by: NScott Branden <sbranden@broadcom.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Static checkers suggest that probably we intended to put curly braces
around the writel() to make it part of the else path.  But, I think
actually the indenting is off and the code works fine as is.

The stray tab was introduced in 0322191e ('mmc: sdhci-esdhc-imx: add
sd3.0 SDR clock tuning support')
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Acked-by: NDong Aisheng <b29396@freescale.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

If CONFIG_DEBUG_SLAB=y:

    sh_mobile_sdhi ee100000.sd: Got CD GPIO
    sh_mobile_sdhi ee100000.sd: Got WP GPIO
    platform ee100000.sd: Driver sh_mobile_sdhi requests probe deferral
    ...
    Slab corruption (Not tainted): kmalloc-1024 start=ed8b3c00, len=1024
    2d0: 00 00 00 00 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b  ....kkkkkkkkkkkk
    Prev obj: start=ed8b3800, len=1024
    000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b  kkkkkkkkkkkkkkkk
    010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b  kkkkkkkkkkkkkkkk

Struct tmio_mmc_host is embedded inside struct mmc_host, and thus is
freed by the call to mmc_free_host(). Hence it must not be written to
afterwards, as that will corrupt freed (and perhaps already reused)
memory.

Fixes: 94b110af ("mmc: tmio: add tmio_mmc_host_alloc/free()")
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Signed-off-by: NFengguang Wu <fengguang.wu@intel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Setting a dev_pm_ops suspend/resume pair but not a set of
hibernation functions means those pm functions will not be
called upon hibernation.
Fix this by using SET_SYSTEM_SLEEP_PM_OPS, which appropriately
assigns the suspend and hibernation handlers and move
omap_hsmmc_x callbacks under CONFIG_PM_SLEEP to avoid build warnings.
Signed-off-by: NRuss Dill <Russ.Dill@ti.com>
[Grygorii.Strashko@linaro.org: rebased on top of K4.0]
Signed-off-by: NGrygorii Strashko <grygorii.strashko@linaro.org>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Mobile phones (some) have no card detect pin, but can detect if the
cover is removed. The purpose is the same; detect if card is being
added/removed, but the details differ.
When the cover is removed, it does not mean the card is gone. But it
might, since it is accessible now. It's like a warning. All the driver
does is to limit write access to the card, see protect_card flag.
In contrast, card detect notifies us after the fact, e.g.
card is gone, card is inserted. We can't take precautions, but we can
rely on those events, -- the card is really gone, or do scan the card.
To summarize there is not much code sharing between cover and card
detect, it only increases confusion. By splitting, both will be
simplified in a followup patch.
Signed-off-by: NAndreas Fenkart <afenkart@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The indirection via omap_hsmmc_get_ro and omap_hsmmc_get_wp is
redundant. Also dropped setting gpio_wp to EINVAL since platform date
is read-only
Untested: no device with ro pin was available, but change is fairly
simple
Signed-off-by: NAndreas Fenkart <afenkart@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

addon to: 09108968b7b72b6083a3bfc8f8259a74ed57255e
    mmc: omap_hsmmc: remove prepare/complete system suspend support
Signed-off-by: NAndreas Fenkart <afenkart@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The reset control for the sunxi mmc controller is optional. Some
newer platforms (sun6i, sun8i, sun9i) have it, while older ones
(sun4i, sun5i, sun7i) don't.

Use the properly stubbed _optional version so the driver does not
fail to compile when RESET_CONTROLLER=n.

This patch also adds a check for deferred probing on the reset
control.
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Cc: <stable@vger.kernel.org> # 3.16+
Acked-by: NDavid Lanzendörfer <david.lanzendoerfer@o2s.ch>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

In these drivers, the driver specific .remove function just a simple
wrapper of function sdhci_pltfm_unregister(). So remove these wrappers
and just set .remove to sdhci_pltfm_unregister().
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

So we can avoid to sprinkle the clk_disable_unprepare() in many
drivers.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Actually we can use the "clk" in the struct sdhci_pltfm_host. Also
change the "external clock" to "core clock" and kill two redundant
private functions in this driver as suggested by Ray Jui.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Reviewed-by: NRay Jui <rjui@broadcom.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Actually we can use the "clk" in the struct sdhci_pltfm_host.
With this change we can also kill the private function for get
max clock in this driver.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Simplify the error and remove path.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Acked-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

There is only one "clk" member in this driver specific private struct.
Actually we can use the "clk" member in the struct sdhci_pltfm_host,
and then kill this struct completely.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The function clk_disable_unprepare() already take care of either error
or null cases.
Signed-off-by: NKevin Hao <haokexin@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

For the original tuning code, delay value is set to SD Bus Clock Delay
Register (SD_CLK_DELAY_SETTING) as (val | (Val << 7) | (val << 16)),
which means CLK_DELAY_IN1, CLK_DELAY_IN2 and CLK_DELAY_OUT are the
same and with 128 steps. This is doubtful. In CSR design specification
documents CS-304575-DR-3H, this issue is clarified, the delay[13:0] in
SD_CLK_DELAY_SETTING is simplied to the concatenation of {CLK_DELAY_IN2,
CLK_DELAY_IN1}.
Besides, for CMD19 tuning, no need to set CLK_DELAY_OUT([22,16]
of SD_CLK_DELAY_SETTING).
Signed-off-by: Nweijun yang <york.yang@csr.com>
Signed-off-by: NBarry Song <baohua.song@csr.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

It's unlikely that this is really needed on any single-slot systems
where we disable card detects until the end of probe, but it still
seems safer to check to make sure that a slot has been initted before
we try to dereference it to find the SDIO interrupt mask.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

We really don't want to get a card detect interrupt during probe time
since it can confuse things.  Let's disable the card detect interrupt
until we're in a really good place: the end of probe.  Let's also
simply avoid enabling the card detect interrupt if it's not used.

It appears that (at least on rk3288) when vqmmc is turned on it can
cause a bogus "card detect" interrupt.  That meant that we were
getting a predictable card detect interrupt while we were in
mmc_add_host().  On the version of the kernel I'm working with at
least (3.14), this is not a great time to get a card detect interrupt
since I think that we don't grab all the needed locks in
mmc_add_host() and children.  I put stack dumps in dw_mci_setup_bus()
and found that I could see two distinct stack crawls that looked like:

Caller one:
* dw_mci_setup_bus
* dw_mci_set_ios
* mmc_power_up
* mmc_start_host
* mmc_add_host

Caller two:
* dw_mci_setup_bus
* dw_mci_set_ios
* mmc_set_chip_select
* mmc_go_idle
* mmc_rescan
* process_one_work
* worker_thread
* kthread
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

We've seen problems on some WiFi modules where we seem to send a CMD53
(which requires the data lines) while the module is asserting busy.
We shouldn't do that.

The Designware Databook says that before issuing a new data transfer
command we should check for busy, so that's what we'll do.

We'll leverage the existing dw_mmc knowledge about whether it should
wait for the previous command to finish to know whether we should
check for busy before sending the command.  This means we won't end up
incorrectly waiting for things like CMD52 (SDIO) or CMD13 (SD) which
don't use the data line.

Note that this also has the advantage of making sure that we don't
change the clock while the card is busy, too.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

We should give dw_mmc a good reset after we apply power.  On some
boards vqmmc may actually be connected to the IP block in the SoC so
it's good to reset after power comes in.

Without this we sometimes see failures enumerating cards on rk3288.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

It appears that we can confuse things if we try to turn on the MMC
clock when the power is off.  Adjust is so that we turn the clock on
(using dw_mci_setup_bus) after power is all the way on and we turn the
clock off before the power goes off.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The STOP command can terminate a data transfer between a memory card and
mmc controller.

As show in Synopsys DesignWare Cores Mobile Storage Host Databook:
Data timeout and Data end-bit error will terminate further data transfer
by mmc controller. So we should not send abort command to terminate a
data transfer again if we got DRTO and EBE interrupt.

After this patch, all mmc_test cases can pass on RK3288-Pink2 board.
Signed-off-by: NAddy Ke <addy.ke@rock-chips.com>
Reviewed-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

To support HS200 and UHS mode, mmc core will call init_card() to
execute tuning:
- sdio: init_card can be executed at runtime resume.
- sd and mmc: init_card can be executed at resume or runtime resume,
  which depends on MMC_CAP_RUNTIME_RESUME capability.

On rk3288 SoC, host will get DRTO interrupt when host send command
to read tuning data. This will spend more than 111ms:
drto_ms = drto_clks * 1000 / bus_hz = 111ms.

And the total tuning time will be more than 400ms.

So we should add MMC_CAP_RUNTIME_RESUME capability to execute tuning
at runtime resume. Only if we do so, can we pass resume test.
Reviewed-by: NDoug Anderson <dianders@chromium.org>
Signed-off-by: NAddy Ke <addy.ke@rock-chips.com>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Implements HS400 mode support for exynos host driver.
This also include some updates as new mode is added.
Signed-off-by: NSeungwon Jeon <tgih.jun@samsung.com>
Signed-off-by: NAlim Akhtar <alim.akhtar@samsung.com>
[Alim: addressed review comments]
Tested-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

If ->run() fails, it can either free the data structures it
allocated, or leave that task to ->free() which will be called
on failures.

However:
  md.c calls ->free() even if ->private_data is NULL, which
     causes problems in some personalities.
  raid0.c frees the data, but doesn't clear ->private_data,
     which will become a problem when we fix md.c

So better fix both these issues at once.
Reported-by: NRichard W.M. Jones <rjones@redhat.com>
Fixes: 5aa61f42
URL: https://bugzilla.kernel.org/show_bug.cgi?id=94381Signed-off-by: NNeilBrown <neilb@suse.de>

Commit b4b55cda (Refine the way to release PCI IRQ resources)
introduced a regression in the PCI IRQ resource management by causing
the IRQ resource of a device, established when pci_enabled_device()
is called on a fully disabled device, to be released when the driver
is unbound from the device, regardless of the enable_cnt.

This leads to the situation that an ill-behaved driver can now make a
device unusable to subsequent drivers by an imbalance in their use of
pci_enable/disable_device().  That is a serious problem for secondary
drivers like vfio-pci, which are innocent of the transgressions of
the previous driver.

Since the solution of this problem is not immediate and requires
further discussion, revert commit b4b55cda and the issue it was
supposed to address (a bug related to xen-pciback) will be taken
care of in a different way going forward.
Reported-by: NAlex Williamson <alex.williamson@redhat.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

A check that rejects a CDB with FUA bit set if no write cache is
emulated was added by the following commit:

  fde9f50f target: Add sanity checks for DPO/FUA bit usage

The condition is as follows:

  if (!dev->dev_attrib.emulate_fua_write ||
      !dev->dev_attrib.emulate_write_cache)

However, this check is wrong if the backend device supports WCE but
"emulate_write_cache" is disabled.

This patch uses se_dev_check_wce() (previously named
spc_check_dev_wce) to invoke transport->get_write_cache() if the
device has a write cache or check the "emulate_write_cache" attribute
otherwise.
Reported-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NChristophe Vu-Brugier <cvubrugier@fastmail.fm>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes a NULL pointer dereference triggered by a late
target_configure_device() -> alloc_workqueue() failure that results
in target_free_device() being called with DF_CONFIGURED already set,
which subsequently OOPses in destroy_workqueue() code.

Currently this only happens at modprobe target_core_mod time when
core_dev_setup_virtual_lun0() -> target_configure_device() fails,
and the explicit target_free_device() gets called.

To address this bug originally introduced by commit 0fd97ccf, go
ahead and move DF_CONFIGURED to end of target_configure_device()
code to handle this special failure case.
Reported-by: NClaudio Fleiner <cmf@daterainc.com>
Cc: Claudio Fleiner <cmf@daterainc.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: <stable@vger.kernel.org> # v3.7+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>