Change the Kconfig selection rule by letting the STI arch to select
the timer.
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Acked-by: NMaxime Coquelin <maxime.coquelin@st.com>

In order to be consistent with the rest of the drivers compilation, let's
introduce the COMPILE_TEST option. Unfortunately, the delay.h code is not
portable, so the compilation test coverage will be restricted to the ARM
architecture.
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

The dsb() instruction is pointless in this code.

Remove it.

That also fixes the ARM64 compilation issue.
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Tested-by: NCaesar Wang <wxt@rock-chips.com>

Let's checkstyle to clean up the macros with such trivial details.
Signed-off-by: NCaesar Wang <wxt@rock-chips.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Allow the timer core to change the smp affinity of the broadcast timer
irq by setting CLOCK_EVT_FEAT_DYNIRQ flag.

This reduces interrupt pressure and wakeups on CPU0 as well as vastly
reducing the number of timer broadcast IPIs.
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Add error path to clear evt struct allocated by kzalloc() in the beginning of
function mtk_timer_init().
Acked-by: NMatthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: NAlexey Klimov <alexey.klimov@linaro.org>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

1) Change pr_warn()s to pr_err()s. These messages are actually errors and not
   warnings.
2) Add missing \n.
3) Error message for kzalloc() failure is removed per suggestion by Joe Perches.
   There is generic stack_dump() for allocation issues.
Signed-off-by: NAlexey Klimov <alexey.klimov@linaro.org>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

It's a bit unclear what subsystem/driver emits some messages to dmesg in
the function mtk_init_timer(). Use pr_fmt to auto-prefix the messages
appropriately.
Acked-by: NMatthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: NAlexey Klimov <alexey.klimov@linaro.org>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

We can now select clocksource drivers like ti-32k and CONFIG_OF
on ancient machines that still use gettimeoffset, and the combination
results in a link error.

arch/arm/kernel/built-in.o: In function `time_init':
(.init.text+0xc28): undefined reference to `clocksource_probe'

The reason for this is that the Makefile is hidden behind
CONFIG_ARCH_USES_GETTIMEOFFSET, but the Kconfig file is not, and
it has shown up just now because the ti-32k driver was added
and can be selected using COMPILE_TEST on all platforms.

This patch hides the Kconfig menu in CONFIG_ARCH_USES_GETTIMEOFFSET
as well.

Fixes: dfedaf10 "clocksource: ti-32k: make it depend on GENERIC_CLOCKSOURCE"
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Felipe Balbi <balbi@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: http://lkml.kernel.org/r/7579471.4N90fYPQOK@wuerfelSigned-off-by: NThomas Gleixner <tglx@linutronix.de>

The ftm_clockevent_init passes the value of "~0UL" into a function
that takes a 32-bit argument, which drops the upper 32 bits, as
gcc warns about on ARM64:

clocksource/fsl_ftm_timer.c: In function 'ftm_clockevent_init':
clocksource/fsl_ftm_timer.c:206:13: warning: large integer implicitly truncated to unsigned type [-Woverflow]

This was obviously unintended behavior, and is easily avoided by
using '~0u' as the integer literal, because that is 32-bit wide
on all architectures.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Xiubo Li <Li.Xiubo@freescale.com>
Cc: Shawn Guo <shawnguo@kernel.org>
Cc: Sascha Hauer <kernel@pengutronix.de>
Cc: Stefan Agner <stefan@agner.ch>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: http://lkml.kernel.org/r/3990834.xnjhm37Grs@wuerfelSigned-off-by: NThomas Gleixner <tglx@linutronix.de>

Add user-mode implementations of gettimeofday() and clock_gettime() to
the VDSO. This is currently usable with 2 clocksources: the CP0 count
register, which is accessible to user-mode via RDHWR on R2 and later
cores, or the MIPS Global Interrupt Controller (GIC) timer, which
provides a "user-mode visible" section containing a mirror of its
counter registers. This section must be mapped into user memory, which
is done below the VDSO data page.

When a supported clocksource is not in use, the VDSO functions will
return -ENOSYS, which causes libc to fall back on the standard syscall
path.

When support for neither of these clocksources is compiled into the
kernel at all, the VDSO still provides clock_gettime(), as the coarse
realtime/monotonic clocks can still be implemented. However,
gettimeofday() is not provided in this case as nothing can be done
without a suitable clocksource. This causes the symbol lookup to fail
in libc and it will then always use the standard syscall path.

This patch includes a workaround for a bug in QEMU which results in
RDHWR on the CP0 count register always returning a constant (incorrect)
value. A fix for this has been submitted, and the workaround can be
removed after the fix has been in stable releases for a reasonable
amount of time.

A simple performance test which calls gettimeofday() 1000 times in a
loop and calculates the average execution time gives the following
results on a Malta + I6400 (running at 20MHz):

 - Syscall:    ~31000 ns
 - VDSO (GIC): ~15000 ns
 - VDSO (CP0): ~9500 ns

[markos.chandras@imgtec.com:
- Minor code re-arrangements in order for mappings to be made
in the order they appear to the process' address space.
- Move do_{monotonic, realtime} outside of the MIPS_CLOCK_VSYSCALL ifdef
- Use gic_get_usm_range so we can do the GIC mapping in the
arch/mips/kernel/vdso instead of the GIC irqchip driver]
Signed-off-by: NAlex Smith <alex.smith@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/11338/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

On the r7s72100 Genmai board the MTU2 driver currently triggers a common
clock framework WARN_ON(enable_count) when disabling the clock due to
the MTU2 driver after recent callback rework may call ->set_state_shutdown()
multiple times. A similar issue was spotted for the TMU driver and fixed in:
452b1324 clocksource/drivers/sh_tmu: Fix traceback spotted in -next

On r7s72100 Genmai v4.3-rc7 built with shmobile_defconfig spits out the
following during boot:

sh_mtu2 fcff0000.timer: ch0: used for clock events
------------[ cut here ]------------
WARNING: CPU: 0 PID: 1 at drivers/clk/clk.c:675 clk_core_disable+0x2c/0x6c()
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.3.0-rc7 #1
Hardware name: Generic R7S72100 (Flattened Device Tree)
Backtrace:
[<c00133d4>] (dump_backtrace) from [<c0013570>] (show_stack+0x18/0x1c)
[<c0013558>] (show_stack) from [<c01c7aac>] (dump_stack+0x74/0x90)
[<c01c7a38>] (dump_stack) from [<c00272fc>] (warn_slowpath_common+0x88/0xb4)
[<c0027274>] (warn_slowpath_common) from [<c0027400>] (warn_slowpath_null+0x24/0x2c)
[<c00273dc>] (warn_slowpath_null) from [<c03a9320>] (clk_core_disable+0x2c/0x6c)
[<c03a92f4>] (clk_core_disable) from [<c03aa0a0>] (clk_disable+0x40/0x4c)
[<c03aa060>] (clk_disable) from [<c0395d2c>] (sh_mtu2_disable+0x24/0x50)
[<c0395d08>] (sh_mtu2_disable) from [<c0395d6c>] (sh_mtu2_clock_event_shutdown+0x14/0x1c)
[<c0395d58>] (sh_mtu2_clock_event_shutdown) from [<c007d7d0>] (clockevents_switch_state+0xc8/0x114)
[<c007d708>] (clockevents_switch_state) from [<c007d834>] (clockevents_shutdown+0x18/0x28)
[<c007d81c>] (clockevents_shutdown) from [<c007dd58>] (clockevents_exchange_device+0x70/0x78)
[<c007dce8>] (clockevents_exchange_device) from [<c007e578>] (tick_check_new_device+0x88/0xe0)
[<c007e4f0>] (tick_check_new_device) from [<c007daf0>] (clockevents_register_device+0xac/0x120)
[<c007da44>] (clockevents_register_device) from [<c0395be8>] (sh_mtu2_probe+0x230/0x350)
[<c03959b8>] (sh_mtu2_probe) from [<c028b6f0>] (platform_drv_probe+0x50/0x98)
Reported-by: NChris Brandt <chris.brandt@renesas.com>
Fixes: 19a9ffb3 ("clockevents/drivers/sh_mtu2: Migrate to new 'set-state' interface")
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMagnus Damm <damm+renesas@opensource.se>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: NViresh Kumar <viresh.kumar@linaro.org>

Having a traceable function in the sched_clock() path leads to a recursion
within ftrace and a kernel crash.

We should not trace digicolor_timer_sched_read() function. Fix this by adding
the notrace attribute to this function.
Signed-off-by: NJisheng Zhang <jszhang@marvell.com>
Acked-by: NBaruch Siach <baruch@tkos.co.il>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Having a traceable function in the sched_clock() path leads to a recursion
within ftrace and a kernel crash.

We should not trace the ftm_read_sched_clock() function.

Fix this by adding the notrace attribute to this function.
Signed-off-by: NJisheng Zhang <jszhang@marvell.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Having a traceable function in the sched_clock() path leads to a recursion
within ftrace and a kernel crash.

We should not trace the pit_read_sched_clock() function. Fix this by adding a
notrace attribute to this function.
Signed-off-by: NJisheng Zhang <jszhang@marvell.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Currently prima2 timer can be used as a scheduler clock. We properly
marked sirfsoc_read_sched_clock() as notrace but we then call another
function sirfsoc_timer_read() that _wasn't_ notrace.

Having a traceable function in the sched_clock() path leads to a recursion
within ftrace and a kernel crash.

Fix this by adding notrace attribute to the sirfsoc_timer_read() function.
Signed-off-by: NJisheng Zhang <jszhang@marvell.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Currently samsung_pwm_timer can be used as a scheduler clock. We properly
marked samsung_read_sched_clock() as notrace but we then call another
function samsung_clocksource_read() that _wasn't_ notrace.

Having a traceable function in the sched_clock() path leads to a recursion
within ftrace and a kernel crash.

Fix this by adding notrace attribute to the samsung_clocksource_read()
function.
Signed-off-by: NJisheng Zhang <jszhang@marvell.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Currently pistachio can be used as a scheduler clock. We properly marked
pistachio_read_sched_clock() as notrace but we then call another function
pistachio_clocksource_read_cycles() that _wasn't_ notrace.

Having a traceable function in the sched_clock() path leads to a recursion
within ftrace and a kernel crash.

Fix this by adding notrace attribute to the pistachio_clocksource_read_cycles()
function.
Signed-off-by: NJisheng Zhang <jszhang@marvell.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Currently arm_global_timer can be used as a scheduler clock. We properly
marked gt_sched_clock_read() as notrace but we then call another function
gt_counter_read() that _wasn't_ notrace.

Having a traceable function in the sched_clock() path leads to a recursion
within ftrace and a kernel crash.

Fix this by adding an extra notrace function to keep other users of
gt_counter_read() traceable.
Signed-off-by: NJisheng Zhang <jszhang@marvell.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Implement an ARM delay timer to be used for udelay() on Armada 37x
platforms.  This allows us to skip the delay loop calibration at boot,
saving 180ms on the boot time of the kernel (which is around 10%).

It also means that udelay() will be unaffected by CPU frequency changes
when cpufreq is enabled on these platforms.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Tested-by: NGregory CLEMENT <gregory.clement@free-electrons.com>
Acked-by: NGregory CLEMENT <gregory.clement@free-electrons.com>

the new ti 32k clocksource driver should depend on
GENERIC_CLOCKSOURCE because of its reliance on
sched_clock_register().

Let's enable that to avoid any possible build errors
and/or warnings on randbuilds.
Signed-off-by: NFelipe Balbi <balbi@ti.com>
Signed-off-by: NTony Lindgren <tony@atomide.com>

Introduce a new clocksource driver for Texas
Instruments 32.768 Hz device which is available
on most OMAP-like devices.
Acked-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: NFelipe Balbi <balbi@ti.com>

Sigma Designs Tango platforms provide a 27 MHz crystal oscillator.
Use it for clocksource, sched_clock, and delay_timer.
Signed-off-by: NMarc Gonzalez <marc_gonzalez@sigmadesigns.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Allow the timer core to change the smp affinity of the broadcast timer
irq by setting CLOCK_EVT_FEAT_DYNIRQ flag. For this to work the timer
core needs to be told about the used irq.

This reduces interrupt pressure and wakeups on CPU0 as well as vastly
reducing the number of timer broadcast IPIs.
Signed-off-by: NLucas Stach <l.stach@pengutronix.de>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Since evt structure is embedded in per-CPU mevt structure it's
definitely faster to use container_of() to get access to mevt
if we have evt (for example as incoming function argument) instead
of more expensive approach with this_cpu_ptr(&percpu_mct_tick).
this_cpu_ptr() on per-CPU mevt structure leads to access to cp15
to get cpu id and arithmetic operations.
Container_of() is cheaper since it's just one asm instruction.
This should work if used evt pointer is correct and owned by
local mevt structure.

For example, before this patch set_state_shutdown() looks like:

 4a4:	e92d4010 	push	{r4, lr}
 4a8:	e3004000 	movw	r4, #0
 4ac:	ebfffffe 	bl	0 <debug_smp_processor_id>
 4b0:	e3003000 	movw	r3, #0
 4b4:	e3404000 	movt	r4, #0
 4b8:	e3403000 	movt	r3, #0
 4bc:	e7933100 	ldr	r3, [r3, r0, lsl #2]
 4c0:	e0844003 	add	r4, r4, r3
 4c4:	e59400c0 	ldr	r0, [r4, #192]	; 0xc0
 4c8:	ebffffd4 	bl	420 <exynos4_mct_tick_stop.isra.1>
 4cc:	e3a00000 	mov	r0, #0
 4d0:	e8bd8010 	pop	{r4, pc}

With this patch:

 4a4:	e92d4010 	push	{r4, lr}
 4a8:	e59000c0 	ldr	r0, [r0, #192]	; 0xc0
 4ac:	ebffffdb 	bl	420 <exynos4_mct_tick_stop.isra.1>
 4b0:	e3a00000 	mov	r0, #0
 4b4:	e8bd8010 	pop	{r4, pc}

Also, for me size of exynos_mct.o decreased from 84588 bytes
to 83956.
Signed-off-by: NAlexey Klimov <alexey.klimov@linaro.org>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: NKrzysztof Kozlowski <k.kozlowski@samsung.com>

Memory for timer16_priv, timer8_priv and tpu_priv structs is
allocated by devm_kzalloc() in corresponding probe functions
of drivers.
No need to zero it one more time.
Signed-off-by: NAlexey Klimov <alexey.klimov@linaro.org>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Memory for cmt struct is allocated by kzalloc() in sh_cmt_setup.
Signed-off-by: NAlexey Klimov <alexey.klimov@linaro.org>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Memory for cs and ced fields in struct em_sti_priv is allocated
by devm_kzalloc() in the beginning of em_sti_probe() so they
don't need to be zeroed one more time in
em_sti_register_clocksource() and in em_sti_register_clockevent().
Signed-off-by: NAlexey Klimov <alexey.klimov@linaro.org>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

When cpu is in deep idle, arch timer will stop counting. Setup GPT as
sched clock source so it can keep counting in idle.
Signed-off-by: NYingjoe Chen <yingjoe.chen@mediatek.com>
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Acked-by: NStephen Boyd <sboyd@codeaurora.org>
Acked-by: NMatthias Brugger <matthias.bgg@gmail.com>

After analysis done by Yingjoe Chen, the timer appears to have a pending
interrupt when it is enabled.

Fix this by acknowledging the pending interrupt when enabling the timer
interrupt.
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Tested-by: NYingjoe Chen <yingjoe.chen@mediatek.com>

Commit dca1a4b5 ("clk: at91: keep slow clk enabled to prevent system
hang") added a workaround for the slow clock as it is not properly handled
by its users.

Get and use the slow clock as it is necessary for the timer counters.
Signed-off-by: NBoris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: NAlexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Acked-by: NThierry Reding <thierry.reding@gmail.com>

t2_clk is already disabled before request_irq(), it must not be disabled
again.
Signed-off-by: NBoris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: NAlexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

The current slow clock rate is hardcoded. Properly get the slow clock
and use its rate.
Signed-off-by: NAlexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Seeing the 'of' characters in a symbol that is being called from
ACPI seems to freak out people. So let's do a bit of pointless
renaming so that these folks do feel at home.
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>
Acked-by: NCatalin Marinas <catalin.marinas@arm.com>
Reviewed-by: NHanjun Guo <hanjun.guo@linaro.org>
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Tested-by: NHanjun Guo <hanjun.guo@linaro.org>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

It is now absolutely trivial to convert the arch timer driver to
use ACPI probing, just like its DT counterpart.

Let's enjoy another crapectomy.
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>
Acked-by: NCatalin Marinas <catalin.marinas@arm.com>
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Tested-by: NHanjun Guo <hanjun.guo@linaro.org>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

The clocksource probing infrastructure currently depends on
CONFIG_CLKSRC_OF, which depends on CONFIG_OF. In order to make
this infrastructure selectable even if CONFIG_OF is not selected,
introduce a new CONFIG_CLKSRC_PROBE (which allow the infrastructure
to be compiled in), and CONFIG_CLKSRC_ACPI (which is the pendent
of CONFIG_CLKSRC_OF for ACPI).
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Tested-by: NHanjun Guo <hanjun.guo@linaro.org>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

DT enjoys a rather nice probing infrastructure for clocksources,
while ACPI is so far stuck into a very distant past.

This patch introduces a declarative API, allowing clocksources
to be self-contained and be called when parsing the GTDT table.
Signed-off-by: NMarc Zyngier <marc.zyngier@arm.com>
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Tested-by: NHanjun Guo <hanjun.guo@linaro.org>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

The current code assumes the 'irq_of_parse_and_map' will return NO_IRQ in case
of failure. Unfortunately, the NO_IRQ is not consistent across the different
architectures and we must not rely on it.

NO_IRQ is equal to '-1' on ARM and 'irq_of_parse_and_map' returns '0' in case
of an error. Hence, the latter won't be detected and will lead to a crash.

Fix this by just checking 'irq' is different from zero.
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

The current code assumes the 'irq_of_parse_and_map' will return NO_IRQ in case
of failure. Unfortunately, the NO_IRQ is not consistent across the different
architectures and we must not rely on it.

NO_IRQ is equal to '-1' on ARM and 'irq_of_parse_and_map' returns '0' in case
of an error. Hence, the latter won't be detected and will lead to a crash.

Fix this by just checking 'irq' is different from zero.
Signed-off-by: NDaniel Lezcano <daniel.lezcano@linaro.org>

Fix Numachip build conflict from:
ce2e572c x86/numachip: Introduce Numachip2 timer mechanisms

drivers/built-in.o:(.discard+0x1b): multiple definition of `__pcpu_unique_cpu_ced'
arch/x86/built-in.o:(.discard+0xa0da): first defined here

Ensure cpu_ced is unique by prefixing with 'numachip2'.
Signed-off-by: NDaniel J Blueman <daniel@numascale.com>
Cc: <tipbuild@zytor.com>
Cc: <kbuild-all@01.org>
Cc: Steffen Persvold <sp@numascale.com>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>