platform/kernel/linux-exynos.git
8 years agoclk: ti: dpll: convert DPLL support code to use clk_hw instead of clk ptrs
Tero Kristo [Sat, 20 Feb 2016 11:24:26 +0000 (13:24 +0200)]
clk: ti: dpll: convert DPLL support code to use clk_hw instead of clk ptrs

Convert DPLL support code to use clk_hw pointers for reference and bypass
clocks. This allows us to use clk_hw_* APIs for accessing any required
parameters for these clocks, avoiding some locking problems at least with
DPLL enable code; this used clk_get_rate which uses mutex but isn't
good under clk_enable / clk_disable.

Signed-off-by: Tero Kristo <t-kristo@ti.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoMerge branch 'clk-fixes' into clk-next
Stephen Boyd [Mon, 22 Feb 2016 22:16:24 +0000 (14:16 -0800)]
Merge branch 'clk-fixes' into clk-next

* clk-fixes:
  clk: ti: omap3+: dpll: use non-locking version of clk_get_rate

8 years agoclk: qcom: msm8960: fix ce3_core clk enable register
Srinivas Kandagatla [Mon, 22 Feb 2016 11:43:39 +0000 (11:43 +0000)]
clk: qcom: msm8960: fix ce3_core clk enable register

This patch corrects the enable register offset which is actually 0x36cc
instead of 0x36c4

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Fixes: 5f775498bdc4 ("clk: qcom: Fully support apq8064 global clock control")
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: ti: omap3+: dpll: use non-locking version of clk_get_rate
Tero Kristo [Sat, 20 Feb 2016 11:12:57 +0000 (13:12 +0200)]
clk: ti: omap3+: dpll: use non-locking version of clk_get_rate

As the code in this file is being executed within irq context in some
cases, we must avoid the clk_get_rate which uses mutex internally.
Switch the code to use clk_hw_get_rate instead which is non-locking.

This fixes an issue where PM runtime will hang the system if enabled
with a serial console before a suspend-resume cycle.

Signed-off-by: Tero Kristo <t-kristo@ti.com>
Tested-by: Tony Lindgren <tony@atomide.com>
Fixes: a53ad8ef3dcc ("clk: ti: Convert to clk_hw based provider APIs")
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoMerge branch 'clk-samsung' into clk-next
Michael Turquette [Mon, 22 Feb 2016 18:24:40 +0000 (10:24 -0800)]
Merge branch 'clk-samsung' into clk-next

8 years agoclk: samsung: Don't build ARMv8 clock drivers on ARMv7
Krzysztof Kozlowski [Tue, 16 Feb 2016 06:20:31 +0000 (15:20 +0900)]
clk: samsung: Don't build ARMv8 clock drivers on ARMv7

Currently the Exynos5433 (ARMv8 SoC) clock driver depends on ARCH_EXYNOS
so it is built also on ARMv7. This does not bring any kind of benefit.
There won't be a single kernel image for ARMv7 and ARMv8 SoCs (like
multi_v7 for ARMv7).

Instead build clock drivers only for respective SoC's architecture.

Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Acked-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
8 years agoMerge branch 'clk-shmobile-for-v4.6' of git://git.kernel.org/pub/scm/linux/kernel...
Michael Turquette [Fri, 19 Feb 2016 17:46:31 +0000 (09:46 -0800)]
Merge branch 'clk-shmobile-for-v4.6' of git://git./linux/kernel/git/geert/renesas-drivers into clk-next

8 years agoMerge branch 'clk-fixes' into clk-next
Stephen Boyd [Fri, 19 Feb 2016 03:17:29 +0000 (19:17 -0800)]
Merge branch 'clk-fixes' into clk-next

* clk-fixes:
  clk: gpio: Really allow an optional clock= DT property
  Revert "clk: qcom: Specify LE device endianness"

8 years agoclk: gpio: Really allow an optional clock= DT property
Stephen Boyd [Fri, 19 Feb 2016 03:07:44 +0000 (19:07 -0800)]
clk: gpio: Really allow an optional clock= DT property

We mis-merged the original patch from Russell here and so the
patch went almost all the way, except that we still failed to
probe when there wasn't a clocks property in the DT node. Allow
that case by making a negative value from
of_clk_get_parent_count() into "no parents", like the original
patch did.

Fixes: 7ed88aa2efa5 ("clk: fix clk-gpio.c with optional clock= DT property")
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Michael Turquette <mturquette@baylibre.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: shmobile: r8a7795: Add INTC-EX clock
Magnus Damm [Thu, 18 Feb 2016 07:14:03 +0000 (16:14 +0900)]
clk: shmobile: r8a7795: Add INTC-EX clock

Add the "intc-ex" clock to the r8a7795 CPG MSSR driver.

According to information from the hardware team the INTC-EX
parent clock is CP. The next data sheet version will include
this information.

Signed-off-by: Magnus Damm <damm+renesas@opensource.se>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
8 years agoclk: Make of_clk_get_from_provider() available to modules
Andrew F. Davis [Fri, 12 Feb 2016 18:50:16 +0000 (12:50 -0600)]
clk: Make of_clk_get_from_provider() available to modules

Export symbol of_clk_get_from_provider so it can be used in
loadable kernel modules

Signed-off-by: Andrew F. Davis <afd@ti.com>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
8 years agoclk: samsung: Enable COMPILE_TEST for Samsung clocks
Krzysztof Kozlowski [Tue, 16 Feb 2016 06:20:30 +0000 (15:20 +0900)]
clk: samsung: Enable COMPILE_TEST for Samsung clocks

Enable the COMPILE_TEST to get build coverage of some of Samsung clock
controller drivers. Still some of them will be built only if
appropriate SoC is chosen (like SOC_EXYNOS4415 or ARCH_S3C64XX).

Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Tested-by: Javier Martinez Canillas <javier@osg.samsung.com>
Reviewed-by: Andi Shyti <andi.shyti@samsung.com>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
8 years agoMerge branch 'clk-bcm2835' into clk-next
Michael Turquette [Tue, 16 Feb 2016 20:31:51 +0000 (12:31 -0800)]
Merge branch 'clk-bcm2835' into clk-next

8 years agoclk: bcm2835: Reuse CLK_DIVIDER_MAX_AT_ZERO for recalc_rate()
Eric Anholt [Tue, 16 Feb 2016 03:03:58 +0000 (19:03 -0800)]
clk: bcm2835: Reuse CLK_DIVIDER_MAX_AT_ZERO for recalc_rate()

We were rolling this ourselves, but clk-divider can do it now.

Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
8 years agoclk: bcm2835: Fix setting of PLL divider clock rates
Eric Anholt [Tue, 16 Feb 2016 03:03:57 +0000 (19:03 -0800)]
clk: bcm2835: Fix setting of PLL divider clock rates

Our dividers weren't being set successfully because CM_PASSWORD wasn't
included in the register write.  It looks easier to just compute the
divider to write ourselves than to update clk-divider for the ability
to OR in some arbitrary bits on write.

Fixes about half of the video modes on my HDMI monitor (everything
except 720x400).

Cc: stable@vger.kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
8 years agoclk: shmobile: cpg-mssr: Update serial port clock in example
Geert Uytterhoeven [Tue, 20 Oct 2015 08:21:49 +0000 (10:21 +0200)]
clk: shmobile: cpg-mssr: Update serial port clock in example

Cfr. commit a9ec81f4ed5c05db ("serial: sh-sci: Drop the interface
clock").

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Michael Turquette <mturquette@baylibre.com>
8 years agoMerge tag 'v4.6-rockchip-clk1' of git://git.kernel.org/pub/scm/linux/kernel/git/mmind...
Michael Turquette [Mon, 15 Feb 2016 19:59:45 +0000 (11:59 -0800)]
Merge tag 'v4.6-rockchip-clk1' of git://git./linux/kernel/git/mmind/linux-rockchip into clk-next

Introduction of a factor type and a variant containing a gate
to be able to also declare factor clocks in their correct
place in the clock tree instead of having to register factor
clocks in the init callback separately. And as always some more
clock-ids and non-regression fixes for mistakes introduced in
past kernel releases.

8 years agoRevert "clk: qcom: Specify LE device endianness"
Stephen Boyd [Fri, 12 Feb 2016 17:21:13 +0000 (09:21 -0800)]
Revert "clk: qcom: Specify LE device endianness"

This reverts commit 329cabcecf94d8d7821e729dda284ba9dec44c87.

The commit that caused us to specify LE device endianness here,
29bb45f25ff3 (regmap-mmio: Use native endianness for read/write,
2015-10-29), has been reverted in mainline so now when we specify
LE it actively breaks big endian kernels because the byte
swapping in regmap-mmio is incorrect. Let's revert this change
because it will 1) fix the big endian kernels and 2) be redundant
to specify LE because that will become the default soon.

Cc: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Cc: Mark Brown <broonie@kernel.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: qcom: mmcc8974: Use gdscs .parent and remove genpd calls
Rajendra Nayak [Tue, 1 Dec 2015 16:12:16 +0000 (21:42 +0530)]
clk: qcom: mmcc8974: Use gdscs .parent and remove genpd calls

With gdsc driver capable of handling hierarchical power domains,
specify oxili_gdsc as parent of oxilicx_gdsc.

Remove all direct calls to genpd from the mmcc clock driver. The
adding and removing of subdomains is now handled from within
the gdsc driver.

Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: qcom: gdsc: Add mmcc gdscs for msm8996 family
Rajendra Nayak [Tue, 1 Dec 2015 16:12:15 +0000 (21:42 +0530)]
clk: qcom: gdsc: Add mmcc gdscs for msm8996 family

Add all gdsc data which are part of mmcc on msm8996 family

Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: qcom: gdsc: Add GDSCs in msm8996 GCC
Rajendra Nayak [Tue, 1 Dec 2015 16:12:14 +0000 (21:42 +0530)]
clk: qcom: gdsc: Add GDSCs in msm8996 GCC

Add all data for the GDSCs which are part of msm8996 GCC block

Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: qcom: gdsc: Add support for votable gdscs
Rajendra Nayak [Tue, 1 Dec 2015 16:12:13 +0000 (21:42 +0530)]
clk: qcom: gdsc: Add support for votable gdscs

Some gdscs might be controlled via voting registers and might not
really disable when the kernel intends to disable them (due to other
votes keeping them enabled)
Mark these gdscs with a flag for we do not check/wait on a disable
status for these gdscs within the kernel disable callback.

Also at boot, if these GDSCs are found to be ON, we make sure we
vote for them before we inform the genpd framework about their
status. If genpd gets no users, it then disables (removes the vote)
them as part of genpd_poweroff_unused()

Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: qcom: gdsc: Add support for gdscs with gds hw controller
Rajendra Nayak [Tue, 1 Dec 2015 16:12:12 +0000 (21:42 +0530)]
clk: qcom: gdsc: Add support for gdscs with gds hw controller

Some gdsc power domains can have a gds_hw_controller block inside
to help ensure all slave devices within the power domain are idle
before the gdsc is actually switched off.
This is mainly useful in power domains which host a MMU, in which
case its necessary to make sure there are no outstanding MMU operations
or pending bus transactions before the power domain is turned off.

In gdscs with gds_hw_controller block, its necessary to check the
gds_hw_ctrl status bits instead of the ones in gdscr, to determine
the state of the powerdomain.

While at it, also move away from using jiffies and use ktime APIs
instead for busy looping on status bits.

Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: qcom: gdsc: Add support for hierarchical power domains
Rajendra Nayak [Tue, 1 Dec 2015 16:12:11 +0000 (21:42 +0530)]
clk: qcom: gdsc: Add support for hierarchical power domains

Some qcom SoCs' can have hierarchical power domains. Let the gdsc structs
specify the parents (if any) and the driver add genpd subdomains for them.

Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: lpc32xx: add HCLK PLL output configuration
Sylvain Lemieux [Wed, 10 Feb 2016 18:52:32 +0000 (13:52 -0500)]
clk: lpc32xx: add HCLK PLL output configuration

This patch add the support to setup the HCLK PLL output
using the "assigned-clock-rates" parameter in the device tree.

If the option is not use, the clock setup by the kickstart
and/or bootloader remain unchanged.

The previous kernel version did not change the clock frequency
output setup by the kickstart and/or bootloader;
this version always setup the clock frequency output to 208MHz.

Signed-off-by: Sylvain Lemieux <slemieux@tycoint.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoMerge branch 'clk-fixes' into clk-next
Stephen Boyd [Wed, 10 Feb 2016 20:44:33 +0000 (12:44 -0800)]
Merge branch 'clk-fixes' into clk-next

* clk-fixes:
  clk: versatile: mask VCO bits before writing

8 years agoclk: versatile: mask VCO bits before writing
Linus Walleij [Wed, 3 Feb 2016 13:47:08 +0000 (14:47 +0100)]
clk: versatile: mask VCO bits before writing

The Versatile syscon ICST driver OR:s the bits into place but
forgets to mask the previous value, making the code only work
if the register is zero or giving haphazard results. Mask the
19 bits used by the Versatile syscon interface register.

Regression caused and now fixed by yours truly.

Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: linux-clk@vger.kernel.org
Fixes: 179c8fb3c2a6 ("clk: versatile-icst: convert to use regmap")
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: lpc32xx: do not register clock "0"
Sylvain Lemieux [Tue, 9 Feb 2016 18:29:10 +0000 (13:29 -0500)]
clk: lpc32xx: do not register clock "0"

The following errors are display in the console during the power-on:
[    0.000000] lpc32xx_usb_clk_init: failed to register (null) clock: -12
[    0.000000] lpc32xx_clk_init: failed to register (null) clock: -12

There is no need to register clock "0"; the first clock used is 1;

Signed-off-by: Sylvain Lemieux <slemieux@tycoint.com>
Acked-by: Vladimir Zapolskiy <vz@mleia.com>
[sboyd@codeaurora.org: s/prepare/register/]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: fix __clk_init_parent() for single parent clocks
Masahiro Yamada [Tue, 9 Feb 2016 11:19:14 +0000 (20:19 +0900)]
clk: fix __clk_init_parent() for single parent clocks

Before commit b3d192d5121f ("clk: simplify __clk_init_parent()"),
__clk_init_parent() called .get_parent() only for multi-parent
clocks.  That commit changed the behavior to call .get_parent()
if available even for single-parent clocks and root clocks.

It turned out a problem because there are some single-parent clocks
that implement .get_parent() callback and return non-zero index.
The SOCFPGA clock is the case; the commit broke the SOCFPGA boards.

To keep the original behavior, invoke .get_parent() only when
num_parents is greater than 1.

Fixes: b3d192d5121f ("clk: simplify __clk_init_parent()")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reported-by: Dinh Nguyen <dinguyen@opensource.altera.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: qcom: common: check for failure
Sudip Mukherjee [Wed, 23 Dec 2015 12:27:20 +0000 (17:57 +0530)]
clk: qcom: common: check for failure

We were not checking the return from devm_add_action() which can fail.
Start using the helper and devm_add_action_or_reset() and return
directly as we know that the cleanup has been done by this helper.

Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agodevm: add helper devm_add_action_or_reset()
Sudip Mukherjee [Wed, 23 Dec 2015 12:27:19 +0000 (17:57 +0530)]
devm: add helper devm_add_action_or_reset()

Add a helper function devm_add_action_or_reset() which will internally
call devm_add_action(). But if devm_add_action() fails then it will
execute the action mentioned and return the error code.

Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: meson: Fix meson_clk_register_clks() signature type mismatch
Andreas Färber [Sun, 7 Feb 2016 21:13:03 +0000 (22:13 +0100)]
clk: meson: Fix meson_clk_register_clks() signature type mismatch

As preparation for arm64 based mesongxbb, which pulls in this code once
enabling ARCH_MESON, fix a size_t vs. unsigned int type mismatch.
The loop uses a local unsigned int variable, so adopt that type,
matching the header.

Fixes: 7a29a869434e ("clk: meson: Add support for Meson clock controller")
Signed-off-by: Andreas Färber <afaerber@suse.de>
Acked-by: Carlo Caione <carlo@endlessm.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: socfpga: fix __init annotation
Arnd Bergmann [Mon, 8 Feb 2016 14:39:09 +0000 (15:39 +0100)]
clk: socfpga: fix __init annotation

clang found a bug with the __socfpga_pll_init definition:

drivers/clk/socfpga/clk-pll-a10.c:77:15: error: '__section__' attribute only applies to functions and
      global variables

This moves the __init annotation to the right place so the function
actually gets discarded.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agodrivers/clk/Kconfig: Move the TI CDCE chips close together
Mike Looijmans [Tue, 3 Nov 2015 11:55:54 +0000 (12:55 +0100)]
drivers/clk/Kconfig: Move the TI CDCE chips close together

There are two TI CDCE clock chips in this file. Move them close
together so they're easier to find.

No functional change, just cosmetic.

Signed-off-by: Mike Looijmans <mike.looijmans@topic.nl>
[sboyd@codeaurora.org: Alphabetize]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agodrivers/clk/Kconfig: Fix typo "Sypport" instead of "Support"
Mike Looijmans [Tue, 3 Nov 2015 11:55:53 +0000 (12:55 +0100)]
drivers/clk/Kconfig: Fix typo "Sypport" instead of "Support"

Simple cosmetic fix.

Signed-off-by: Mike Looijmans <mike.looijmans@topic.nl>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoMerge branch 'clk-fixes' into clk-next
Stephen Boyd [Mon, 8 Feb 2016 22:01:10 +0000 (14:01 -0800)]
Merge branch 'clk-fixes' into clk-next

* clk-fixes:
  clk: tegra: super: Fix sparse warnings for functions not declared as static
  clk: tegra: Fix sparse warnings for functions not declared as static
  clk: tegra: Fix sparse warning for pll_m
  clk: tegra: Use definition for pll_u override bit
  clk: tegra: Fix warning caused by pll_u failing to lock
  clk: tegra: Fix clock sources for Tegra210 EMC
  clk: tegra: Add the APB2APE audio clock on Tegra210
  clk: tegra: Add missing of_node_put()
  clk: tegra: Fix PLLE SS coefficients
  clk: tegra: Fix typos around clearing PLLE bits during enable
  clk: tegra: Do not disable PLLE when under hardware control
  clk: tegra: Fix pllx dyn step calculation
  clk: tegra: pll: Fix potential sleeping-while-atomic
  clk: tegra: Fix the misnaming of nvenc from msenc
  clk: tegra: Fix naming of MISC registers
  clk: tegra: Remove improper flags for lock_enable
  clk: tegra: Fix divider on VI_I2C

8 years agoMerge tag 'tegra-for-4.5-clk-fixes' of git://git.kernel.org/pub/scm/linux/kernel...
Stephen Boyd [Mon, 8 Feb 2016 21:50:32 +0000 (13:50 -0800)]
Merge tag 'tegra-for-4.5-clk-fixes' of git://git./linux/kernel/git/tegra/linux into clk-fixes

Pull tegra fixes from Thierry Reding:

clk: tegra: Fixes for v4.5-rc3

This set contains a bunch of miscellaneous fixes that have accumulated
over the past couple of weeks, primarily for the Tegra210 support added
in v4.5-rc1.

* tag 'tegra-for-4.5-clk-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tegra/linux:
  clk: tegra: super: Fix sparse warnings for functions not declared as static
  clk: tegra: Fix sparse warnings for functions not declared as static
  clk: tegra: Fix sparse warning for pll_m
  clk: tegra: Use definition for pll_u override bit
  clk: tegra: Fix warning caused by pll_u failing to lock
  clk: tegra: Fix clock sources for Tegra210 EMC
  clk: tegra: Add the APB2APE audio clock on Tegra210
  clk: tegra: Add missing of_node_put()
  clk: tegra: Fix PLLE SS coefficients
  clk: tegra: Fix typos around clearing PLLE bits during enable
  clk: tegra: Do not disable PLLE when under hardware control
  clk: tegra: Fix pllx dyn step calculation
  clk: tegra: pll: Fix potential sleeping-while-atomic
  clk: tegra: Fix the misnaming of nvenc from msenc
  clk: tegra: Fix naming of MISC registers
  clk: tegra: Remove improper flags for lock_enable
  clk: tegra: Fix divider on VI_I2C

8 years agoclk: shmobile: r8a7795: Add USB-DMAC clocks
Yoshihiro Shimoda [Mon, 1 Feb 2016 11:29:05 +0000 (20:29 +0900)]
clk: shmobile: r8a7795: Add USB-DMAC clocks

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
8 years agoclk: shmobile: r8a7795: Add SD divider support
Dirk Behme [Sat, 30 Jan 2016 06:33:59 +0000 (07:33 +0100)]
clk: shmobile: r8a7795: Add SD divider support

This patch adds SD[0..3] clock divider support for R-Car Gen3 SoC.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>
Signed-off-by: Dirk Behme <dirk.behme@de.bosch.com>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
8 years agoclk: provider: Remove of_gpio_{gate,mux}_clk_setup() prototypes
Stephen Boyd [Sun, 7 Feb 2016 07:34:55 +0000 (23:34 -0800)]
clk: provider: Remove of_gpio_{gate,mux}_clk_setup() prototypes

These functions either never existed or were only used in
OF_CLK_DECLARE() macros. Remove the dead prototypes.

Cc: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: Deprecate CLK_IS_ROOT
Stephen Boyd [Wed, 3 Feb 2016 01:24:56 +0000 (17:24 -0800)]
clk: Deprecate CLK_IS_ROOT

We don't use CLK_IS_ROOT but in a few places in the common clk
framework core. Let's replace those checks with a check for the
number of parents a clk has instead of the flag, freeing up one
flag for something else. We don't remove the flag yet so that
things keep building, but we'll remove it once all drivers have
removed their flag usage.

Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: gpio: Make into a platform driver
Stephen Boyd [Wed, 3 Feb 2016 01:09:26 +0000 (17:09 -0800)]
clk: gpio: Make into a platform driver

clk_get() for DT based clks already returns EPROBE_DEFER when the
OF clk provider is not present. So having all this code in the
clk provider to return EPROBE_DEFER when the gpio isn't ready yet
can be replaced with a platform driver that doesn't add the clk
provider until the gpio can be requested. Get rid of the
OF_CLK_DECLARE and convert this to a platform driver instead.

Tested-by: Jyri Sarha <jsarha@ti.com>
Cc: Sergej Sawazki <ce3a@gmx.de>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Fabio Estevam <fabio.estevam@freescale.com>
Cc: Jon Nettleton <jon@solid-run.com>
Cc: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: rockchip: convert manually created factor clocks to the new type
Heiko Stuebner [Sat, 20 Jun 2015 14:06:02 +0000 (16:06 +0200)]
clk: rockchip: convert manually created factor clocks to the new type

Clean up the init code and move the creation of factor clocks to the
appropriate positions coming from the clock architecture diagrams.

This also unifies the artificial separation of the hclk_vcodec etc clocks
again.

We do keep the separate definition of some watchdog and usb480m pseudo
clocks for now, as they're not real factor clocks from the clock-tree
but placeholders for fixes to come (usb480m gets supplied by the
missing driver for the new usbphy type and the watchdog-gate is sitting
somewhere else together which we cannot model currently).

Signed-off-by: Heiko Stuebner <heiko@sntech.de>
8 years agoclk: rockchip: add a factor clock type
Heiko Stuebner [Sat, 20 Jun 2015 11:08:57 +0000 (13:08 +0200)]
clk: rockchip: add a factor clock type

Add a clock type for fixed factor clocks. This allows us to define fixed
factor clocks where they appear in the clock hierarchy instead of in the
init function.

The additional factor_gate type, finally allows us to model some last
parts of the clock tree correctly.

Signed-off-by: Heiko Stuebner <heiko@sntech.de>
8 years agoclk: rockchip: fix parent of hclk_vcodec on rk3036
Heiko Stuebner [Wed, 27 Jan 2016 20:38:37 +0000 (21:38 +0100)]
clk: rockchip: fix parent of hclk_vcodec on rk3036

hclk_vcodec is a child of aclk_vcodec with the fixed factor clock
hclk_vcodec_pre in between and not a child of hclk_disp_pre.

Signed-off-by: Heiko Stuebner <heiko@sntech.de>
8 years agoclk: rockchip: free memory in error cases when registering clock branches
Shawn Lin [Tue, 2 Feb 2016 03:37:50 +0000 (11:37 +0800)]
clk: rockchip: free memory in error cases when registering clock branches

Add free memeory if rockchip_clk_register_branch fails.

Fixes: a245fecbb806 ("clk: rockchip: add basic infrastructure...")
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
8 years agoclk: unlock for handling unregistered clock
Insu Yun [Sat, 30 Jan 2016 15:12:04 +0000 (10:12 -0500)]
clk: unlock for handling unregistered clock

If clock is already unregistered, it returns with holding lock.
It needs to be unlocked.

Signed-off-by: Insu Yun <wuninsu@gmail.com>
[sboyd@codeaurora.org: Use goto instead]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: vt8500: don't return possibly uninitialized data
Arnd Bergmann [Mon, 1 Feb 2016 10:19:10 +0000 (11:19 +0100)]
clk: vt8500: don't return possibly uninitialized data

The clk-vt8500.c driver would previously enter an endless loop
when invalid settings got requested, this was now fixed. However,
the driver will now return uninitialized data for a subset of those
cases instead, as the gcc correctly warns:

clk/clk-vt8500.c: In function 'wm8650_find_pll_bits':
clk/clk-vt8500.c:423:12: error: 'best_div2' may be used uninitialized in this function [-Werror=maybe-uninitialized]
  *divisor2 = best_div2;
            ^
clk/clk-vt8500.c:422:12: error: 'best_div1' may be used uninitialized in this function [-Werror=maybe-uninitialized]
  *divisor1 = best_div1;
            ^
clk/clk-vt8500.c:421:14: error: 'best_mul' may be used uninitialized in this function [-Werror=maybe-uninitialized]
  *multiplier = best_mul;

This reworks the error handling in the driver so we now return
-EINVAL from clk_round_rate() and clk_set_rate() when we get
impossible inputs.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 090341b0a95d ("clk: vt8500: fix sign of possible PLL values")
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: slightly optimize clk_core_set_parent()
Masahiro Yamada [Mon, 28 Dec 2015 10:23:10 +0000 (19:23 +0900)]
clk: slightly optimize clk_core_set_parent()

If clk_fetch_parent_index() fails, p_rate is unused.  Move the
assignment after the error checking.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: simplify clk_fetch_parent_index() function
Masahiro Yamada [Mon, 28 Dec 2015 10:23:09 +0000 (19:23 +0900)]
clk: simplify clk_fetch_parent_index() function

The clk_core_get_parent_by_index can be used as a helper function
to simplify the implementation of clk_fetch_parent_index().

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: make sure parent is not NULL in clk_fetch_parent_index()
Masahiro Yamada [Mon, 28 Dec 2015 10:23:08 +0000 (19:23 +0900)]
clk: make sure parent is not NULL in clk_fetch_parent_index()

If parent is given with NULL, clk_fetch_parent_index() could return
a positive index value.

Currently, parent is checked by the callers of this function, but
it would be safer to do it in this function.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: walk the orphan clock list more simply
Masahiro Yamada [Mon, 28 Dec 2015 10:23:07 +0000 (19:23 +0900)]
clk: walk the orphan clock list more simply

This loop can be much simpler. If a new parent is available for
orphan clocks, __clk_init_parent(orphan) can detect it.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: avoid circular clock topology
Masahiro Yamada [Mon, 28 Dec 2015 10:23:06 +0000 (19:23 +0900)]
clk: avoid circular clock topology

Currently, clk_register() never checks a circular parent looping,
but clock providers could register such an insane clock topology.
For example, "clk_a" could have "clk_b" as a parent, and vice versa.
In this case, clk_core_reparent() creates a circular parent list
and __clk_recalc_accuracies() calls itself recursively forever.

The core infrastructure should be kind enough to bail out, showing
an appropriate error message in such a case.  This helps to easily
find a bug in clock providers.  (uh, I made such a silly mistake
when I was implementing my clock providers first.  I was upset
because the kernel did not respond, without any error message.)

This commit adds a new helper function, __clk_is_ancestor().  It
returns true if the second argument is a possible ancestor of the
first one.  If a clock core is a possible ancestor of itself, it
would make a loop when it were registered.  That should be detected
as an error.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: simplify __clk_init_parent()
Masahiro Yamada [Mon, 28 Dec 2015 10:23:04 +0000 (19:23 +0900)]
clk: simplify __clk_init_parent()

The translation from the index into clk_core is done by
clk_core_get_parent_by_index().  The if-block for num_parents == 1
case is duplicating the code in the clk_core_get_parent_by_index().

Drop the "if (num_parents == 1)" from the special case.  Instead,
set the index to zero if .get_parent() is missing.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: move checking .get_parent to __clk_core_init()
Masahiro Yamada [Mon, 28 Dec 2015 10:23:04 +0000 (19:23 +0900)]
clk: move checking .get_parent to __clk_core_init()

The .get_parent is mandatory for multi-parent clocks.  Move the check
to __clk_core_init(), like other callback checkings.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
[sboyd@codeaurora.org: Squashed in error path handling, fix typos
in commit message]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: tegra: super: Fix sparse warnings for functions not declared as static
Jon Hunter [Fri, 4 Dec 2015 17:04:25 +0000 (17:04 +0000)]
clk: tegra: super: Fix sparse warnings for functions not declared as static

Sparse reports the following warnings for structures and functions that
should be declared static:

drivers/clk/tegra/clk-tegra-super-gen4.c:70:35: warning: symbol
 'tegra_super_gen_info_gen4' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra-super-gen4.c:96:35: warning: symbol
 'tegra_super_gen_info_gen5' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra-super-gen4.c:174:13: warning: symbol
 'tegra_super_clk_init' was not declared. Should it be static?

Fix this by making the above static.

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix sparse warnings for functions not declared as static
Jon Hunter [Fri, 4 Dec 2015 17:04:24 +0000 (17:04 +0000)]
clk: tegra: Fix sparse warnings for functions not declared as static

Sparse reports the following warnings for functions in clk-tegra210.c
that should be declared as static:

drivers/clk/tegra/clk-tegra210.c:460:6: warning: symbol
 'tegra210_pllcx_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:485:6: warning: symbol
 '_pllc_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:490:6: warning: symbol
 '_pllc2_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:495:6: warning: symbol
 '_pllc3_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:500:6: warning: symbol
 '_plla1_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:510:6: warning: symbol
 'tegra210_plla_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:562:6: warning: symbol
 'tegra210_plld_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:701:6: warning: symbol
 'tegra210_plld2_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:709:6: warning: symbol
 'tegra210_plldp_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:722:6: warning: symbol
 'tegra210_pllc4_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:731:6: warning: symbol
 'tegra210_pllre_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:844:6: warning: symbol
 'tegra210_pllx_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:904:6: warning: symbol
 'tegra210_pllmb_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:963:6: warning: symbol
 'tegra210_pllp_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:1025:6: warning: symbol
 'tegra210_pllu_set_defaults' was not declared. Should it be static?
drivers/clk/tegra/clk-tegra210.c:1215:15: warning: symbol
 'tegra210_clk_adjust_vco_min' was not declared. Should it be static?

Fix this by declaring the above as static.

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix sparse warning for pll_m
Jon Hunter [Fri, 4 Dec 2015 17:04:23 +0000 (17:04 +0000)]
clk: tegra: Fix sparse warning for pll_m

Sparse generates the following warning for the pll_m params structure:

drivers/clk/tegra/clk-tegra210.c:1569:10: warning: Initializer entry
 defined twice
drivers/clk/tegra/clk-tegra210.c:1570:10:   also defined here

Fix this by correcting the index for the MISC1 register.

Fixes: b31eba5ff3f7 ("clk: tegra: Add support for Tegra210 clocks")

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Use definition for pll_u override bit
Jon Hunter [Mon, 21 Dec 2015 12:56:32 +0000 (12:56 +0000)]
clk: tegra: Use definition for pll_u override bit

The definition, PLLU_BASE_OVERRIDE, for the pll_u OVERRIDE bit is defined
but not used and when the OVERRIDE bit is cleared in tegra210_pll_init()
the code directly uses the bit number. Therefore, use the definition,
PLLU_BASE_OVERRIDE when clearing the OVERRIDE bit.

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix warning caused by pll_u failing to lock
Jon Hunter [Mon, 21 Dec 2015 12:56:31 +0000 (12:56 +0000)]
clk: tegra: Fix warning caused by pll_u failing to lock

If the pll_u is not configured by the bootloader, then on kernel boot the
following warning is seen:

 clk_pll_wait_for_lock: Timed out waiting for pll pll_u_vco lock
 tegra_init_from_table: Failed to enable pll_u_out1
 ------------[ cut here ]------------
 WARNING: at drivers/clk/tegra/clk.c:269
 Modules linked in:

 CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.4.0-rc4-next-20151214+ #1
 Hardware name: NVIDIA Tegra210 P2371 reference board (E.1) (DT)
 task: ffffffc0bc0a0000 ti: ffffffc0bc0a8000 task.ti: ffffffc0bc0a8000
 PC is at tegra_init_from_table+0x140/0x164
 LR is at tegra_init_from_table+0x140/0x164
 pc : [<ffffffc0008fee78>] lr : [<ffffffc0008fee78>] pstate: 80000045
 sp : ffffffc0bc0abd50
 x29: ffffffc0bc0abd50 x28: ffffffc00090b8a8
 x27: ffffffc000a06000 x26: ffffffc0bc019780
 x25: ffffffc00086a708 x24: ffffffc00086a790
 x23: ffffffc0006d7188 x22: ffffffc0bc010000
 x21: 000000000000016e x20: ffffffc0bc00d100
 x19: ffffffc000944178 x18: 0000000000000007
 x17: 000000000000000e x16: 0000000000000001
 x15: 0000000000000007 x14: 000000000000000e
 x13: 0000000000000013 x12: 000000000000001a
 x11: 000000000000004d x10: 0000000000000750
 x9 : ffffffc0bc0a8000 x8 : ffffffc0bc0a07b0
 x7 : 0000000000000001 x6 : 0000000002d5f0f8
 x5 : 0000000000000000 x4 : 0000000000000000
 x3 : 0000000000000002 x2 : ffffffc000996724
 x1 : 0000000000000000 x0 : 0000000000000032

 ---[ end trace cbd20ae519e92ced ]---
 Call trace:
 [<ffffffc0008fee78>] tegra_init_from_table+0x140/0x164
 [<ffffffc000900ac8>] tegra210_clock_apply_init_table+0x20/0x28
 [<ffffffc0008fec40>] tegra_clocks_apply_init_table+0x18/0x24
 [<ffffffc00008291c>] do_one_initcall+0x90/0x194
 [<ffffffc0008cfab0>] kernel_init_freeable+0x148/0x1e8
 [<ffffffc000636bb0>] kernel_init+0x10/0xdc
 [<ffffffc000085cd0>] ret_from_fork+0x10/0x40
 clk_pll_wait_for_lock: Timed out waiting for pll pll_u_vco lock
 tegra_init_from_table: Failed to enable pll_u_out2
 ------------[ cut here ]------------

pll_u can be either controlled by software or hardware and this is
selected via the OVERRIDE bit in the pll_u base register. In the function
tegra210_pll_init(), the OVERRIDE bit for pll_u is cleared, which selects
hardware control of the pll. However, at the same time the pll_u clocks
are populated in the init_table for tegra210 and so software will try to
configure the pll_u if it is not already configured and hence, the above
warning is seen when the pll fails to lock. Remove the pll_u clocks from
the init_table so that software does not try to configure this pll on
boot.

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix clock sources for Tegra210 EMC
Jon Hunter [Fri, 18 Dec 2015 13:45:28 +0000 (13:45 +0000)]
clk: tegra: Fix clock sources for Tegra210 EMC

The EMC clock sources for Tegra210 currently incorrectly include pll_c2
and pll_c3. However, both of these should have been pll_mb as shown in
the TRM. If Tegra210 happens to be configured such that the pll_mb is the
default clock for the EMC, as configured by the bootloader, then this will
cause a system hang on boot. This is because the kernel will disable the
pll_mb when disabling unused clock as it appears to be unused when it is
not.

Also add the additional pll_p clock source for the EMC.

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Acked-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Add the APB2APE audio clock on Tegra210
Jon Hunter [Thu, 28 Jan 2016 16:33:50 +0000 (16:33 +0000)]
clk: tegra: Add the APB2APE audio clock on Tegra210

The APB2APE clock for the audio subsystem is required for powering up the
audio power domain and accessing the various modules in this subsystem on
Tegra210 devices. Add this clock for Tegra210.

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Add missing of_node_put()
Amitoj Kaur Chawla [Sun, 24 Jan 2016 15:15:20 +0000 (20:45 +0530)]
clk: tegra: Add missing of_node_put()

for_each_child_of_node() performs an of_node_get() on each iteration, so
before breaking out of the loop an of_node_put() is required.

Found using Coccinelle. The semantic patch used for this is as follows:

// <smpl>
@@
expression e;
local idexpression child;
@@

 for_each_child_of_node(root, child) {
   ... when != of_node_put(child)
       when != e = child
(
   return child;
|
+  of_node_put(child);
?  return ...;
)
   ...
 }
// </smpl>

Signed-off-by: Amitoj Kaur Chawla <amitoj1606@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix PLLE SS coefficients
Mark Kuo [Thu, 14 Jan 2016 19:26:42 +0000 (14:26 -0500)]
clk: tegra: Fix PLLE SS coefficients

The PLLE SS coefficients are different between Tegra210 and Tegra114.
Add SoC generation specific versions for Tegra114 and Tegra210 and use
them in their respective ->enable() callbacks.

Signed-off-by: Mark Kuo <mkuo@nvidia.com>
Signed-off-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix typos around clearing PLLE bits during enable
Rhyland Klein [Thu, 14 Jan 2016 19:24:37 +0000 (14:24 -0500)]
clk: tegra: Fix typos around clearing PLLE bits during enable

While enabling PLLE on both Tegra114 and Tegra210, we should be clearing
PLLE_MISC_VREG_BG_CTRL_MASK and PLLE_MISC_VREG_CTRL_MASK not setting
them. This patch fixes both places where we incorrectly set instead of
cleared those bits.

Signed-off-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Do not disable PLLE when under hardware control
Mark Kuo [Thu, 14 Jan 2016 19:24:36 +0000 (14:24 -0500)]
clk: tegra: Do not disable PLLE when under hardware control

Software should not disable PLLE if PLLE is already put under hardware
control.

Signed-off-by: Mark Kuo <mkuo@nvidia.com>
Signed-off-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix pllx dyn step calculation
Rhyland Klein [Thu, 14 Jan 2016 19:24:35 +0000 (14:24 -0500)]
clk: tegra: Fix pllx dyn step calculation

The logic for calculating the input rate used when figuring out the
proper dynamic steps for pllx was incorrect. It is supposed to be
calculated using parent_rate / m but it was just using the parent rate
directly, therefore using the wrong step values.

Signed-off-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: pll: Fix potential sleeping-while-atomic
Andrew Bresticker [Thu, 14 Jan 2016 19:24:34 +0000 (14:24 -0500)]
clk: tegra: pll: Fix potential sleeping-while-atomic

Since the ->enable() callback is called with a spinlock held, we cannot
call potentially blocking functions such as clk_get_rate() or
clk_get_parent(), so use the unlocked versions instead.

Signed-off-by: Andrew Bresticker <abrestic@chromium.org>
[rklein: Adapted from ChromeOS patch, removing pllu_enable cleanup as
it isn't present upstream]
Signed-off-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix the misnaming of nvenc from msenc
Rhyland Klein [Thu, 14 Jan 2016 19:24:33 +0000 (14:24 -0500)]
clk: tegra: Fix the misnaming of nvenc from msenc

When adding the nvenc clock, it was partially named msenc in the code.
Since the msenc clock isn't present in Tegra210 and has been replaced by
the nvenc clock, its misleading to see it present. Therefore, properly
rename it.

Signed-off-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: tegra: Fix naming of MISC registers
Rhyland Klein [Thu, 14 Jan 2016 19:24:32 +0000 (14:24 -0500)]
clk: tegra: Fix naming of MISC registers

Some register for PLLM and PLLMB were named MISC0 but according to the
TRM, they have different names. Sync up the names to make it easier to
understand which register they are really referring to.

Signed-off-by: Rhyland Klein <rklein@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
8 years agoclk: replace pr_warn() with pr_err() for fatal cases
Masahiro Yamada [Mon, 28 Dec 2015 10:23:03 +0000 (19:23 +0900)]
clk: replace pr_warn() with pr_err() for fatal cases

These three cases let clk_register() fail.  They should be considered
as error messages.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: drop the initial core->parents look-ups from __clk_core_init()
Masahiro Yamada [Mon, 28 Dec 2015 10:23:02 +0000 (19:23 +0900)]
clk: drop the initial core->parents look-ups from __clk_core_init()

The core->parents is a cache to save expensive clock parent look-ups.
It will be filled as needed later.  We do not have to do it here.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: simplify clk_core_get_parent_by_index()
Masahiro Yamada [Mon, 28 Dec 2015 10:23:01 +0000 (19:23 +0900)]
clk: simplify clk_core_get_parent_by_index()

Drop the "if (!core->parents)" case and refactor the function a bit
because core->parents is always allocated.  (Strictly speaking, it is
ZERO_SIZE_PTR if core->num_parents == 0, but such a case is omitted
by the if-conditional above.)

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: move core->parents allocation to clk_register()
Masahiro Yamada [Mon, 28 Dec 2015 10:23:00 +0000 (19:23 +0900)]
clk: move core->parents allocation to clk_register()

Currently, __clk_core_init() allows failure of the kcalloc() for the
core->parents.  So, clk_fetch_parent_index() and __clk_init_parent()
also try to allocate core->parents in case it has not been allocated
yet.  Scattering memory allocation here and there makes things
complicated.

Like other clk_core members, allocate core->parents in clk_register()
and let it fail in case of memory shortage.  If we cannot allocate
such a small piece of memory, the system is already insane.  There is
no point to postpone the memory allocation.

Also, allocate core->parents regardless of core->num_parents.  We want
it even if core->num_parents == 1 because clk_fetch_parent_index()
might be called against the clk_core with a single parent.

If core->num_parents == 0, core->parents is set to ZERO_SIZE_PTR. It
is harmless because no access happens to core->parents in such a case.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: change sizeof(struct clk *) to sizeof(*core->parents)
Masahiro Yamada [Mon, 28 Dec 2015 10:22:59 +0000 (19:22 +0900)]
clk: change sizeof(struct clk *) to sizeof(*core->parents)

Now, the clock parent is not "struct clk *", but "struct clk_core *".
Of course, the size of a pointer is always same, but strictly speaking,
sizeof(struct clk *) should be sizeof(struct clk_core *) here.

This mismatch happened when we split the structure into struct clk
and struct clk_core.  For the potential possibility of future renaming,
sizeof(*core->parents) would be better.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: remove unnecessary !core->parents conditional
Masahiro Yamada [Mon, 28 Dec 2015 10:22:58 +0000 (19:22 +0900)]
clk: remove unnecessary !core->parents conditional

This if-block has been here since the introduction of the common
clock framework.  Now no clock drivers are statically initialized.
core->parent is always NULL at this point.  Drop the redundant
check and the confusing comment.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: rename __clk_init() into __clk_core_init()
Masahiro Yamada [Mon, 28 Dec 2015 10:22:57 +0000 (19:22 +0900)]
clk: rename __clk_init() into __clk_core_init()

Now this function takes clk_core as its argument.  __clk_core_init()
would be more suitable for the name of this function.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: change the argument of __clk_init() into pointer to clk_core
Masahiro Yamada [Mon, 28 Dec 2015 10:22:56 +0000 (19:22 +0900)]
clk: change the argument of __clk_init() into pointer to clk_core

The argument clk_user is used only for the clk_user->core.  The rest
of this function only takes care of clk_core.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: remove unused first argument of __clk_init()
Masahiro Yamada [Mon, 28 Dec 2015 10:22:55 +0000 (19:22 +0900)]
clk: remove unused first argument of __clk_init()

The "struct device *dev" is not used at all in this function.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoMerge branch 'clk-fixes' into clk-next
Stephen Boyd [Sat, 30 Jan 2016 01:26:31 +0000 (17:26 -0800)]
Merge branch 'clk-fixes' into clk-next

* clk-fixes:
  clk: rockchip: rk3368: fix some clock gates
  clk: rockchip: rk3036: rename emac ext source clock
  clk: rockchip: rk3036: fix the div offset for emac clock
  clk: rockchip: rk3036: fix uarts clock error
  clk: rockchip: rk3036: fix the FLAGs for clock mux

8 years agoMerge tag 'v4.5-rockchip-clkfixes1' of git://git.kernel.org/pub/scm/linux/kernel...
Stephen Boyd [Sat, 30 Jan 2016 01:24:28 +0000 (17:24 -0800)]
Merge tag 'v4.5-rockchip-clkfixes1' of git://git./linux/kernel/git/mmind/linux-rockchip into clk-fixes

Pull rockchip fixes from Heiko Stuebner:

Fixes for wrong register offsets in both rk3036 and rk3368.
Also rename the external input for the emac on rk3036, which
should still be ok to do, as that binding was only introduced
during this merge-window.

* tag 'v4.5-rockchip-clkfixes1' of git://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip:
  clk: rockchip: rk3368: fix some clock gates
  clk: rockchip: rk3036: rename emac ext source clock
  clk: rockchip: rk3036: fix the div offset for emac clock
  clk: rockchip: rk3036: fix uarts clock error
  clk: rockchip: rk3036: fix the FLAGs for clock mux

8 years agoclk: st: avoid uninitialized variable use
Arnd Bergmann [Mon, 25 Jan 2016 15:54:04 +0000 (16:54 +0100)]
clk: st: avoid uninitialized variable use

My previous patch fixed some warnings about printing a couple
of variables that are always uninitialized in quadfs_pll_fs660c32_set_rate(),
but I now got a warning that only shows up in some configurations (i.e.
without gcc -Os) about the params.ndiv being used uninitialized in the
error case:

drivers/clk/st/clkgen-fsyn.c: In function 'quadfs_pll_fs660c32_set_rate':
drivers/clk/st/clkgen-fsyn.c:584:75: warning: 'params.ndiv' may be used uninitialized in this function [-Wmaybe-uninitialized]
drivers/clk/st/clkgen-fsyn.c:574:16: note: 'params.ndiv' was declared here

This changes the error handling so we bail for invalid arguments rather
than continuing with uninitialized data.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: axi-clkgen: Remove sometimes impossible check
Stephen Boyd [Sat, 30 Jan 2016 01:09:01 +0000 (17:09 -0800)]
clk: axi-clkgen: Remove sometimes impossible check

The size of unsigned long on 64-bit architectures is equal to the
size of u64, so this check is impossible there. This throws off
static checkers:

drivers/clk/clk-axi-clkgen.c:331 axi_clkgen_recalc_rate() warn:
impossible condition '(tmp > (~0)) => (0-u64max > u64max)'

Let's change this code to use min_t() instead so that we
get the same effect on architectures where sizeof(unsigned long)
doesn't equal sizeof(u64).

Cc: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: axi-clkgen: Add multi-parent support
Lars-Peter Clausen [Mon, 30 Nov 2015 16:54:56 +0000 (17:54 +0100)]
clk: axi-clkgen: Add multi-parent support

The clock generator has two clock inputs that can be used as the reference
clock. Add support for switching between them at runtime.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: axi-clkgen: Remove version 1 support
Lars-Peter Clausen [Mon, 30 Nov 2015 16:54:55 +0000 (17:54 +0100)]
clk: axi-clkgen: Remove version 1 support

Version 1 of the axi-clkgen core has not been used in new designs for over
two years now. This is a soft peripheral used in FPGAs and anybody who has
updated their kernel to the latest version will also have updated the
bitstream containing the clock generator. So it should be safe to drop
support for this now.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: palmas: fix a possible NULL dereference
LABBE Corentin [Wed, 25 Nov 2015 12:52:02 +0000 (13:52 +0100)]
clk: palmas: fix a possible NULL dereference

of_match_device could return NULL, and so cause a NULL pointer
dereference later.
Even if the probability of this case is very low, fixing it made
static analyzers happy.

Solving this with of_device_get_match_data made also code simplier.

Reported-by: coverity (CID 1324137)
Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: palmas: constify the palmas_clks_of_match_data structure
LABBE Corentin [Wed, 25 Nov 2015 12:52:01 +0000 (13:52 +0100)]
clk: palmas: constify the palmas_clks_of_match_data structure

The palmas_clks_of_match_data structures are never modified.
This patch constify them.

Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: optimize the divider walk in clk_divider_bestdiv()
Masahiro Yamada [Tue, 5 Jan 2016 03:43:41 +0000 (12:43 +0900)]
clk: optimize the divider walk in clk_divider_bestdiv()

Because _next_div() returns a valid divider, there is no need to
consult _is_valid_div() for the validity of the divider in every
iteration.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: vt8500: fix sign of possible PLL values
Andrzej Hajda [Fri, 2 Oct 2015 04:49:53 +0000 (06:49 +0200)]
clk: vt8500: fix sign of possible PLL values

With unsigned values underflow in loops can occur resulting in
theoretically infinite loops.

The problem has been detected using proposed semantic patch
scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].

[1]: http://permalink.gmane.org/gmane.linux.kernel/2038576

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: add clk_unregister_fixed_rate()
Masahiro Yamada [Wed, 6 Jan 2016 04:25:10 +0000 (13:25 +0900)]
clk: add clk_unregister_fixed_rate()

Allow to unregister fixed rate clock.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: add clk_unregister_fixed_factor()
Masahiro Yamada [Wed, 6 Jan 2016 04:25:09 +0000 (13:25 +0900)]
clk: add clk_unregister_fixed_factor()

Allow to unregister fixed factor clock.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk:gcc-msm8916: add missing mss_q6_bimc_axi clock
Srinivas Kandagatla [Mon, 4 Jan 2016 11:01:53 +0000 (11:01 +0000)]
clk:gcc-msm8916: add missing mss_q6_bimc_axi clock

This clock is required for loading the qdsp firmware.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoMerge branch 'clk-iproc' into clk-next
Stephen Boyd [Sat, 30 Jan 2016 00:35:02 +0000 (16:35 -0800)]
Merge branch 'clk-iproc' into clk-next

* clk-iproc:
  clk: iproc: Remove __init from header
  clk: iproc: Add support for Cygnus audio clocks
  Documentation: dt-bindings: Add DT bindings for Cygnus audio clock

8 years agoclk: iproc: Remove __init from header
Ray Jui [Wed, 27 Jan 2016 01:18:40 +0000 (17:18 -0800)]
clk: iproc: Remove __init from header

Remove __init macro from all function prototypes in clk-iproc.h

Signed-off-by: Ray Jui <rjui@broadcom.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: iproc: Add support for Cygnus audio clocks
Simran Rai [Wed, 27 Jan 2016 01:18:39 +0000 (17:18 -0800)]
clk: iproc: Add support for Cygnus audio clocks

This patch adds support for Broadcom Cygnus audio PLL and leaf
clocks

Signed-off-by: Simran Rai <ssimran@broadcom.com>
Reviewed-by: Scott Branden <sbranden@broadcom.com>
Signed-off-by: Ray Jui <rjui@broadcom.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoDocumentation: dt-bindings: Add DT bindings for Cygnus audio clock
Simran Rai [Wed, 27 Jan 2016 01:18:38 +0000 (17:18 -0800)]
Documentation: dt-bindings: Add DT bindings for Cygnus audio clock

This patch adds audio clock device tree binding documentation to an
existing Cygnus clock DT bindings document.

Signed-off-by: Simran Rai <ssimran@broadcom.com>
Reviewed-by: Ray Jui <rjui@broadcom.com>
Reviewed-by: Lori Hikichi <lhikichi@broadcom.com>
Reviewed-by: Scott Branden <sbranden@broadcom.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: Move vendor's Kconfig into CCF menu section
James Liao [Thu, 28 Jan 2016 08:58:57 +0000 (16:58 +0800)]
clk: Move vendor's Kconfig into CCF menu section

Move all vendor's Kconfig into CCF menu section to prevent
new drivers putting their Kconfig files in a wrong place.

Some Kconfigs need to be modified at the same time to avoid build
warnings.

Signed-off-by: James Liao <jamesjj.liao@mediatek.com>
Acked-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
[sboyd@codeaurora.org: Fix typos in commit message]
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: mediatek: Fix memory leak on clock init fail
James Liao [Fri, 8 Jan 2016 08:15:33 +0000 (16:15 +0800)]
clk: mediatek: Fix memory leak on clock init fail

mtk_clk_register_composite() may leak memory due to some error
handling path don't free all allocated memory. This patch
free all pointers that may allocate memory before error return.
And it's safe because kfree() can handle NULL pointers.

Signed-off-by: James Liao <jamesjj.liao@mediatek.com>
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: move the common clock's to_clk_*(_hw) macros to clk-provider.h
Geliang Tang [Fri, 8 Jan 2016 15:51:46 +0000 (23:51 +0800)]
clk: move the common clock's to_clk_*(_hw) macros to clk-provider.h

to_clk_*(_hw) macros have been repeatedly defined in many places.
This patch moves all the to_clk_*(_hw) definitions in the common
clock framework to public header clk-provider.h, and drop the local
definitions.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
8 years agoclk: xgene: Remove return from void function
Stephen Boyd [Fri, 29 Jan 2016 20:57:15 +0000 (12:57 -0800)]
clk: xgene: Remove return from void function

This function doesn't return anything because it's void. Drop the
return statement.

Cc: Loc Ho <lho@apm.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>