During probe for an always-running watchdog a timer is setup to
constantly ping the watchdog while the device is not open. The gpio to
ping the watchdog is setup to inactive.

For a watchdog with hw_algo = "toggle" this results in a ping depending
on the initial state of the gpio, for hw_algo = "level" no ping is
generated.

Make sure that the first automatic ping is sent immediately and not only
when the timer expires the first time. This makes the machine survive in
case more than half of the watchdog timeout is already elapsed. (Which
is very probable for the chip I'm faced with that has a timeout of one
second.)
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

strncmp(algo, "toggle", 6) doesn't compare the trailing '\0' byte, so
using

	hw_algo = "toggleboggle"

is recognized the same way as

	hw_algo = "toggle"

. While this doesn't introduce any problems for a device tree that
sticks to the documented settings it's still ugly.

Fix this by using strcmp to only match on "toggle" and "level".
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

The watchdog infrastructure checks the maximum timeout for us.
Use it.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Tested on the Nvidia chipset with an SMBus controller PCI ID 0x0AA2
(as shown in the PCI listing during the boot sequence).
Signed-off-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Commit f2147de3 ("watchdog: sunxi: support parameterized compatible
strings") introduced a regression in sunxi_wdt_start(), by which
the system reset function of the watchdog is not enabled upon
starting the watchdog. As a result, the system is not reset when the
watchdog expires. Fix it.

Fixes: f2147de3 ("watchdog: sunxi: support parameterized compatible strings")
Signed-off-by: NFrancesco Lavra <francescolavra.fl@gmail.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>
Cc: stable@vger.kernel.org

Without .shutdown(), watchdog might reset the system during power off.
For example, if watchdog's timeout is set to 30s, then it is reset to
zero by mtk_wdt_ping(). During power off, no app will ping watchdog,
but watchdog is still running and may trigger reset.
Signed-off-by: NGreta Zhang <greta.zhang@mediatek.com>
Signed-off-by: NEddie Huang <eddie.huang@mediatek.com>
Acked-by: NMatthias Brugger <matthias.bgg@gmail.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

add mediatek watchdog driver suspend/resume support
Signed-off-by: NGreta Zhang <greta.zhang@mediatek.com>
Signed-off-by: NRoger Lu <roger.lu@mediatek.com>
Signed-off-by: NEddie Huang <eddie.huang@mediatek.com>
Acked-by: NMatthias Brugger <matthias.bgg@gmail.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Commit c631f200 ("watchdog: imgpdc: Add reboot support") introduced
a restart handler but forgot to unregister it on driver removal. Fix it.

Fixes: c631f200 ("watchdog: imgpdc: Add reboot support")
Reported-by: NAriel D'Alessandro <ariel@vanguardiasur.com.ar>
Signed-off-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Switch to my kernel.org alias instead of a badly named gmail address,
which I rarely use.
Signed-off-by: NViresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add a early_enable module parameter to the omap_wdt that starts the
watchdog on module insertion. The default value is 0 which does not
start the watchdog - which also does not change the behavior if the
parameter is not given.
Signed-off-by: NLars Poeschel <poeschel@lemonage.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

In some situation, mainly when it's not possible to disable a
watchdog, you may want the watchdog driver to be started as soon
as possible.

Adding GPIO_WATCHDOG_ARCH_INITCALL to raise initcall from
module_init to arch_initcall.

This patch require watchdog registration deferral mechanism
Signed-off-by: NJean-Baptiste Theou <jtheou@adeneo-embedded.us>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Currently, watchdog subsystem require the misc subsystem to
register a watchdog. This may not be the case in case of an
early registration of a watchdog, which can be required when
the watchdog cannot be disabled.

This patch introduces a deferral mechanism to remove this requirement.
Signed-off-by: NJean-Baptiste Theou <jtheou@adeneo-embedded.us>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

This patch removes the static watchdog device for a new max63xx_wdt data
structure, and constifies the max63xx_timeout data.

The new structure contains pointers to pin access routines, which
abstracts mmap-specific code. This will ease future accesses like GPIO.
Signed-off-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

If watchdog_register_device() fails we should disable the previously
acquired wdev->clk clock on error path.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

clk_prepare_enable() may fail, so we should better check its return value
and propagate it in the case of error.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

WDIOC_SETOPTIONS makes it possible to disable and re-enable the
watchdog timer while the hpwdt driver is loaded.
Signed-off-by: NJean Delvare <jdelvare@suse.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

The omap watchdog hardware is able to read the watchdog timer counter
register. This implements this functionality in the omap_wdt driver, so
one is can read the time until the watchdog will trigger the reset in
seconds using WDIOC_GETTIMELEFT.
Signed-off-by: NLars Poeschel <poeschel@lemonage.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Add watchdog driver support for DA9062
Signed-off-by: NSteve Twiss <stwiss.opensource@diasemi.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

If on watchdog device registration a parent device is not set, then
the registered watchdog is considered to be a virtual device:

  /sys/devices/virtual/watchdog/watchdog0
  /sys/devices/virtual/watchdog/watchdog1

Setting a correct reference to a platform device allows to
distinguish multiple instances of iMX2+ hardware watchdogs:

  /sys/devices/soc0/soc/2000000.aips-bus/20bc000.wdog/watchdog/watchdog0
  /sys/devices/soc0/soc/2000000.aips-bus/20c0000.wdog/watchdog/watchdog1
Signed-off-by: NVladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

In a21_wdt_remove() we do a watchdog_unregister_device() on struct
a21_wdt_drv->wdt but never assign it.

Also move the dev_set_drvdata() call in front of the watchdog_register_device()
call, so it doesn't look like an error.
Signed-off-by: NJohannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

If you've got code that does this in a tight loop
  1. Open watchdog
  2. Send 'expect close'
  3. Close watchdog
...you'll eventually trigger a watchdog reset.  You can reproduce this
by using daisydog (1) and running:
  while true; do daisydog -c > /dev/null; done

The problem is that each time you write to the watchdog for 'expect
close' it moves the timer .5 seconds out.  The timer thus never fires
and never pats the watchdog for you.

1: http://git.chromium.org/gitweb/?p=chromiumos/third_party/daisydog.gitSigned-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJisheng Zhang <jszhang@marvell.com>
Reviewed-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Right now the dw_wdt uses a spinlock to protect dw_wdt_open().  The
problem is that while holding the spinlock we call:
-> dw_wdt_set_top()
   -> dw_wdt_top_in_seconds()
      -> clk_get_rate()
         -> clk_prepare_lock()
            -> mutex_lock()

Locking a mutex while holding a spinlock is not allowed and leads to
warnings like "BUG: spinlock wrong CPU on CPU#1", among other
problems.

There's no reason to use a spinlock.  Only dw_wdt_open() was protected
and the test_and_set_bit() at the start of that function protects us
anyway.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJisheng Zhang <jszhang@marvell.com>
Reviewed-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Commit faad5de0 ("watchdog: imx2_wdt: convert to watchdog core api")
removes the custom ioctl function. The generic ioctl handler is not
setting the wdog->timeout to the new_timeout but handing this preset
value back to the userspace. This patch sets the new value in the
drivers set_timeout function to fix that problem.
Signed-off-by: NMichael Grzeschik <m.grzeschik@pengutronix.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

The GPIO subsystem provides dummy GPIO consumer functions if GPIOLIB is
not enabled. Hence drivers that depend on GPIOLIB, but use GPIO consumer
functionality only, can still be compiled if GPIOLIB is not enabled.

Relax the dependency on GPIOLIB if COMPILE_TEST is enabled, where
appropriate.
Signed-off-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>
Cc: linux-watchdog@vger.kernel.org

Not all architectures have io memory.

Fixes:
drivers/built-in.o: In function `cdns_wdt_probe':
cadence_wdt.c:(.text+0x33b7c9): undefined reference to `devm_ioremap_resource'
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

The platform_device_id is not modified by the driver and core uses it as
const.
Signed-off-by: NKrzysztof Kozlowski <k.kozlowski.k@gmail.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Remove the ARM Kconfig dependency since the Maxim MAX63xx devices are
architecture independent.
Signed-off-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

This watchdog hardware can be configured in terms of power-of-two
clock cycles. Therefore, the watchdog timeout configured by the user
will be rounded-up to the next possible hardware timeout.

This commit adds a comment explaining this.
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Maximum timeout is currently set in clock cycles, but the watchdog
core expects it to be in seconds. Fix it.
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Register a restart handler that will restart the system by writing
to the watchdog's SOFT_RESET register.
Signed-off-by: NAndrew Bresticker <abrestic@chromium.org>
Reviewed-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Tested-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Cc: James Hogan <james.hogan@imgtec.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Set up the watchdog for the specified timeout before attempting to start it.
Signed-off-by: NNaidu Tellapati <naidu.tellapati@imgtec.com>
Signed-off-by: NAndrew Bresticker <abrestic@chromium.org>
Reviewed-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Tested-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Cc: James Hogan <james.hogan@imgtec.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Since the heartbeat is statically initialized to its default value,
watchdog_init_timeout() will never look in the device-tree for a
timeout-sec value.  Instead of statically initializing heartbeat,
fall back to the default timeout value if watchdog_init_timeout()
fails.
Signed-off-by: NAndrew Bresticker <abrestic@chromium.org>
Reviewed-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Tested-by: NEzequiel Garcia <ezequiel.garcia@imgtec.com>
Cc: James Hogan <james.hogan@imgtec.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

The omap watchdog has the annoying behaviour that writes to most
registers don't have any effect when the watchdog is already running.
Quoting the AM335x reference manual:

	To modify the timer counter value (the WDT_WCRR register),
	prescaler ratio (the WDT_WCLR[4:2] PTV bit field), delay
	configuration value (the WDT_WDLY[31:0] DLY_VALUE bit field), or
	the load value (the WDT_WLDR[31:0] TIMER_LOAD bit field), the
	watchdog timer must be disabled by using the start/stop sequence
	(the WDT_WSPR register).

Currently the timer is stopped in the .probe callback but still there
are possibilities that yield to a situation where omap_wdt_start is
entered with the timer running (e.g. when /dev/watchdog is closed
without stopping and then reopened). In such a case programming the
timeout silently fails!

To circumvent this stop the timer before reprogramming.

Assuming one of the first things the watchdog user does is setting the
timeout explicitly nothing too bad should happen because this explicit
setting works fine.

Fixes: 7768a13c ("[PATCH] OMAP: Add Watchdog driver support")
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Instead of using an over-long expression involving the ?: operator use
an if and instead of an else branch rely on the fact that the data
structure was allocated using devm_kzalloc. This also allows to put the
used helper variable into a more local scope.

There is no functional change.
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: NFelipe Balbi <balbi@ti.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

This way only a single allocation is needed (per device). Also this
simplifies the data structure used by the driver because there is no
need anymore to link from one struct to the other (by means of
watchdog_{set,get}_drvdata).
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Instead of (partly) open coding watchdog_init_timeout to determine the
inital timeout use the core function that exists for exactly this
purpose.

As a side effect the "timeout-sec" device-tree property is recognized now
(though currently unused in the omap device trees).
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

This commit add a driver for the watchdog functionality of the Conexant CX92755
SoC, from the Digicolor series of SoCs. Of 8 system timers provided by the
CX92755, the first one, timer A, can reset the chip when its counter reaches
zero. This driver uses this capability to provide userspace with a standard
watchdog, using the watchdog timer driver core framework. This driver also
implements a reboot handler for the reboot(2) system call.

The watchdog driver shares the timer registers with the CX92755 timer driver
(drivers/clocksource/timer-digicolor.c). The timer driver, however, uses only
timers other than A, so both drivers should coexist.
Signed-off-by: NBaruch Siach <baruch@tkos.co.il>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Use endian agnostic IO functions for the watchdog driver for when it
is enabled on ATSAMA5D36 devices running in big endian.
Signed-off-by: NBen Dooks <ben.dooks@codethink.co.uk>
Acked-by: NBoris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Initial submission adding support for this IP only included Watchdog and
the Real-Time Clock.  Now the third (and final) device is enabled this
trivial patch is required to update the comment in the Watchdog driver
to encompass Clocksource.
Signed-off-by: NLee Jones <lee.jones@linaro.org>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Signed-off-by: NDavid Paris <david.paris@st.com>
Signed-off-by: NLee Jones <lee.jones@linaro.org>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>