Add driver for IDT VersaClock 5 5P49V5923 and 5P49V5933 chips. These
chips have two clock inputs, XTAL or CLK, which are muxed into single
PLL/VCO input. In case of 5P49V5923, the XTAL in built into the chip
while the 5P49V5923 requires external XTAL.
The PLL feeds two fractional dividers. Each fractional divider feeds
output mux, which allows selecting between clock from the fractional
divider itself or from output mux on output N-1. In case of output
mux 0, the output N-1 is instead connected to the output from the mux
feeding the PLL.
The driver thus far supports only the 5P49V5923 and 5P49V5933, while
it should be easily extensible to the whole 5P49V59xx family of chips
as they are all pretty similar.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Add bindings for IDT VersaClock 5 5P49V5923 and 5P49V5933 chips.
These are I2C clock generators with optional clock source from
either XTal or dedicated clock generator and, depending on the
model, two or more clock outputs.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Cc: Michael Turquette <mturquette@baylibre.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Rob Herring <robh@kernel.org>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: devicetree@vger.kernel.org
Cc: linux-renesas-soc@vger.kernel.org
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
controlled through the General Register Files, support for the rk3328
clock-controller (including a new pll-type) and the usual clock ids and
some fixes.
-----BEGIN PGP SIGNATURE-----
iQFEBAABCAAuFiEE7v+35S2Q1vLNA3Lx86Z5yZzRHYEFAlh+vFMQHGhlaWtvQHNu
dGVjaC5kZQAKCRDzpnnJnNEdgf7TCACncJmYlvYp+aZEgBlEcyCTulHNSyRZHVPd
c4zBsgZmgxqmgrxe8YsW6DcfPP96MtL/C/fIupxTphxxfAS5HQx2KCeVpOrfffk7
lJ1CPyCP5GdjaZ98hKVuMpKVkQu1u26DTNWSy62hbSQQndbpP0NbGsIJJUFia1vm
JX0POVYt0xSo6GnbVRcKN/5b9k0HJNG9aejL8u/uA3+yr8diiKzYrtnaFzai5kvE
3LgXIDGPUZVfwZ2vrJfDCYqNlij/tF2yLIhEDDoMSl00WolJfmnMKOg1Lt0DzuMz
OQY6ZIexTV1CCV+9BjXMVCLth7DH7K3EztQsvyJtr90wrfYs+4Vd
=4WOJ
-----END PGP SIGNATURE-----
Merge tag 'v4.11-rockchip-clk1' of git://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip into clk-next
Pull Rockchip clk updates from Heiko Stuebner:
A new clock-type for the 1-2 muxes per soc that are for whatever reason
controlled through the General Register Files, support for the rk3328
clock-controller (including a new pll-type) and the usual clock ids and
some fixes.
* tag 'v4.11-rockchip-clk1' of git://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip:
dt-bindings: clk: add rockchip,grf property for RK3399
clk: rockchip: use clock ids for memory controller parts on rk3066/rk3188
clk: rockchip: use rk3288 isp_in clock ids
clk: rockchip: add clock ids for memory controller parts on rk3066/rk3188
clk: rockchip: add rk3288 isp_in clock ids
clk: rockchip: Remove useless init of "grf" to -EPROBE_DEFER
clk: rockchip: add clock controller for rk3328
dt-bindings: add bindings for rk3328 clock controller
clk: rockchip: add dt-binding header for rk3328
clk: rockchip: add new pll-type for rk3328
clk: rockchip: describe aclk_vcodec using the new muxgrf type on rk3288
clk: rockchip: add a clock-type for muxes based in the grf
- addition of the CPU clock configuration data for Exynos4412
Prime SoC variant,
- removal of driver for deprecated Exynos4415 SoC,
- switching from the syscore to regular system sleep PM ops
in the audio subsystem clocks controller driver,
- updates of the definitions of some "Network On Chip" related
clocks.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=mSQz
-----END PGP SIGNATURE-----
Merge tag 'clk-v4.11-samsung' of git://linuxtv.org/snawrocki/samsung into clk-next
Pull Samsung clk updates from Sylwester Nawrocki:
- addition of the CPU clock configuration data for Exynos4412
Prime SoC variant,
- removal of driver for deprecated Exynos4415 SoC,
- switching from the syscore to regular system sleep PM ops
in the audio subsystem clocks controller driver,
- updates of the definitions of some "Network On Chip" related
clocks.
* tag 'clk-v4.11-samsung' of git://linuxtv.org/snawrocki/samsung:
clk: samsung: Remove Exynos4415 driver (SoC not supported anymore)
clk: samsung: exynos-audss: Replace syscore PM with platform device PM
clk: samsung: exynos5433: Set NoC (Network On Chip) clocks as critical
clk: samsung: Add CPU clk configuration data for Exynos4412 Prime
The LDB mux/gate layout has been fixed on QuadPlus, so there is no need
to restrict the LDB mux changes on this hardware, as the erratum
preventing this from working properly is gone.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Support for Exynos4415 is going away because there are no internal nor
external users.
Since commit 46dcf0ff0d ("ARM: dts: exynos: Remove exynos4415.dtsi"),
the platform cannot be instantiated so remove also the drivers.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Kukjin Kim <kgene@kernel.org>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Add support for rockchip,grf property which is used for GRF muxes
on RK3399.
Signed-off-by: Xing Zheng <zhengxing@rock-chips.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Add the newly added clock ids to the clock entries of the rk3066/rk3188
clock driver. We won't be needing them in the kernel for a bit yet
but as they're used in the new u-boot ddr setup code/dts we should make
sure the clock ids stay identical and do not differ.
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Reference the newly added isp clock-ids in the clock-tree.
Signed-off-by: Jacob Chen <jacob-chen@iotwrt.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The failure path in the newly added function tries to free an
uninitialized pointer:
drivers/clk/clk-stm32f4.c: In function 'stm32f4_rcc_init':
drivers/clk/clk-stm32f4.c:1106:4: error: 'gate' may be used uninitialized in this function [-Werror=maybe-uninitialized]
I'm adding an initialization to NULL here to make the kfree()
succeed, and I'm also rearranging the cleanup so that the
same kfree() is used for any error path, making the function
slightly more robust against newly introduced bugs in the
error handling.
Fixes: daf2d117cb ("clk: stm32f4: Add lcd-tft clock")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Gabriel Fernandez <gabriel.fernandez@st.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
The CDCE925 is a member of the CDCE(L)9xx programmable clock generator
family. There are also CDCE913, CDCE937, CDCE949 which have different
number of PLLs and outputs.
The clk-cdce925 driver supports only CDCE925 in the family. This adds
support for the CDCE913, CDCE937, CDCE949, too.
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Acked-by: Rob Herring <robh@kernel.org>
Cc: Mike Looijmans <mike.looijmans@topic.nl>
Cc: Michael Turquette <mturquette@linaro.org>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Add definition of EBI2 clock used by MDM9615 NAND controller.
Cc: Andy Gross <andy.gross@linaro.org>
Cc: David Brown <david.brown@linaro.org>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: linux-arm-msm@vger.kernel.org
Cc: linux-soc@vger.kernel.org
Cc: linux-clk@vger.kernel.org
Cc: devicetree@vger.kernel.org
Signed-off-by: Zoran Markovic <zmarkovic@sierrawireless.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
[sboyd@codeaurora.org: ebi2_clk halt bit is 24 not 23]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
This include was needed to suppress build error when this driver
was initially merged because <linux/regmap.h> did not include
<linux/delay.h> at that time. (developers' headache across
sub-systems)
The root cause has been fixed by commit adf08d481b ("regmap:
include <linux/delay.h> from include/linux/regmap.h"), so this
line can be dropped now.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
This commit adjusts the list of possible "Sample At Reset" values that
define the CPU clock frequency of the AP806 (part of Marvell Armada
7K/8K) to the values that have been validated with the production
chip. Earlier values were preliminary.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Add clock drivers for hi3660 SoC, this driver controls the SoC
registers to supply different clocks to different IPs in the SoC.
Signed-off-by: Zhangfei Gao <zhangfei.gao@linaro.org>
[sboyd@codeaurora.org: Simplify probe with function pointer]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Some parent clocks of the Exynos542x clock blocks, which have separate
power domains (like DISP, MFC, MSC, GSC, FSYS and G2D) must be always
enabled to access any register related to power management unit or devices
connected to it. For the time being, until a proper solution based on
runtime PM is applied, mark those clocks as critical (instead of ignore
unused or even no flags) to prevent disabling them.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Reviewed-by: Chanwoo Choi <cw00.choi@samsung.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Tested-by: Javier Martinez Canillas <javier@osg.samsung.com> [Exynos5800 Peach Pi Chromebook]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
EPROBE_DEFER is not an error, hence printing an error message like
clk: couldn't get clock 0 for /soc/display@feb00000
may confuse the user.
Suppress error messages in case of probe deferral to fix this.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Reviewed-by: Marek Vasut <marek.vasut@gmail.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Currently we add the virtual cpufreq device unconditionally even when
the SCPI DVFS clock provider node is disabled. This will cause cpufreq
driver to throw errors when it gets initailised on boot/modprobe and
also when the CPUs are hot-plugged back in.
This patch fixes the issue by adding the virtual cpufreq device only if
the SCPI DVFS clock provider is available and registered.
Fixes: 9490f01e24 ("clk: scpi: add support for cpufreq virtual device")
Reported-by: Michał Zegan <webczat_200@poczta.onet.pl>
Cc: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Tested-by: Michał Zegan <webczat_200@poczta.onet.pl>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Add support to use reset control framework for resetting MSS
with hexagon v56 1.5.0.
Signed-off-by: Avaneesh Kumar Dwivedi <akdwived@codeaurora.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
The audio related clock support is missing from the existing zx296718
clock driver. Let's add it, so that the upstream ZX SPDIF driver can
work for HDMI audio support.
Signed-off-by: Jun Nie <jun.nie@linaro.org>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
[sboyd@codeaurora.org: Staticize some more structures]
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
It adds the compatible string for zx296718 audio clock controller.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Instead of using panic, we should give an error message and return error
code when of_clk_add_hw_provider() call fails.
Since we have error prompt for failures, the "init over" pr_info output
isn't really necessary but becomes a debug noise. So let's clean it up
along the way.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
The delay here is not in atomic context and does not seem critical with
respect to precision, but usleep_range(min,max) with min==max results in
giving the timer subsystem no room to optimize uncritical delays. Fix
this by setting the range to 2000,3000 us.
Fixes: commit f05259a6ff ("clk: wm831x: Add initial WM831x clock driver")
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
On vf610, PLL1 and PLL2 have registers to configure fractional part of
frequency multiplier.
This patch adds support for these registers.
This fixes "fast system clock" issue on boards where bootloader sets
fractional multiplier for PLL1.
Suggested-by: Andrey Smirnov <andrew.smirnov@gmail.com>
CC: Chris Healy <cphealy@gmail.com>
Signed-off-by: Nikita Yushchenko <nikita.yoush@cogentembedded.com>
Tested-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
When we used to defer setting the "grf" member to
rockchip_clk_get_grf() it was important to init the "grf" member to an
error value in rockchip_clk_init(). With recent changes, we now set
"grf" right in rockchip_clk_init() (two lines below the place where we
initted it). That makes the old init useless. Get rid of it.
Fixes: 6f339dc271 ("clk: rockchip: lookup General Register Files in rockchip_clk_init")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Add the clock tree definition for the new rk3328 SoC.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Add devicetree bindings for Rockchip cru which found on
Rockchip SoCs.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Add the dt-bindings header for the rk3328, that gets shared between
the clock controller and the clock references in the dts.
Add softreset ID for rk3328.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The rk3328's pll and clock are similar with rk3036's,
it different with pll_mode_mask, the rk3328 soc
pll mode only one bit(rk3036 soc have two bits)
so these should be independent and separate from
the series of rk3328s.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
With the newly introduced clk type for muxes in the grf we now can
describe some missing clocks, like the aclk_vcodec that selects between
aclk_vdpu and aclk_vepu based on a bit set in the general register files.
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Rockchip socs often have some tiny number of muxes not controlled from
the core clock controller but through bits set in the general register
files. Add a clock-type that can control these as well, so that we
don't need to work around them being absent.
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Pull DAX updates from Dan Williams:
"The completion of Jan's DAX work for 4.10.
As I mentioned in the libnvdimm-for-4.10 pull request, these are some
final fixes for the DAX dirty-cacheline-tracking invalidation work
that was merged through the -mm, ext4, and xfs trees in -rc1. These
patches were prepared prior to the merge window, but we waited for
4.10-rc1 to have a stable merge base after all the prerequisites were
merged.
Quoting Jan on the overall changes in these patches:
"So I'd like all these 6 patches to go for rc2. The first three
patches fix invalidation of exceptional DAX entries (a bug which
is there for a long time) - without these patches data loss can
occur on power failure even though user called fsync(2). The other
three patches change locking of DAX faults so that ->iomap_begin()
is called in a more relaxed locking context and we are safe to
start a transaction there for ext4"
These have received a build success notification from the kbuild
robot, and pass the latest libnvdimm unit tests. There have not been
any -next releases since -rc1, so they have not appeared there"
* 'libnvdimm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
ext4: Simplify DAX fault path
dax: Call ->iomap_begin without entry lock during dax fault
dax: Finish fault completely when loading holes
dax: Avoid page invalidation races and unnecessary radix tree traversals
mm: Invalidate DAX radix tree entries only if appropriate
ext2: Return BH_New buffers for zeroed blocks
- A merge error on my part broke the DocBook build. I've requisitioned
one of tglx's frozen sharks for appropriate disciplinary action and
resolved to be more careful about testing the DocBook stuff as long as
it's still around.
- Fix an error in unaligned-memory-access.txt
-----BEGIN PGP SIGNATURE-----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=W4e5
-----END PGP SIGNATURE-----
Merge tag 'docs-4.10-rc1-fix' of git://git.lwn.net/linux
Pull documentation fixes from Jonathan Corbet:
"Two small fixes:
- A merge error on my part broke the DocBook build. I've
requisitioned one of tglx's frozen sharks for appropriate
disciplinary action and resolved to be more careful about testing
the DocBook stuff as long as it's still around.
- Fix an error in unaligned-memory-access.txt"
* tag 'docs-4.10-rc1-fix' of git://git.lwn.net/linux:
Documentation/unaligned-memory-access.txt: fix incorrect comparison operator
docs: Fix build failure
Pull crypto fix from Herbert Xu:
"This fixes a boot failure on some platforms when crypto self test is
enabled along with the new acomp interface"
* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
crypto: testmgr - Use heap buffer for acomp test input
mm/filemap.c: In function 'clear_bit_unlock_is_negative_byte':
mm/filemap.c:933:9: error: too few arguments to function 'test_bit'
return test_bit(PG_waiters);
^~~~~~~~
Fixes: b91e1302ad ('mm: optimize PageWaiters bit use for unlock_page()')
Signed-off-by: Olof Johansson <olof@lixom.net>
Brown-paper-bag-by: Linus Torvalds <dummy@duh.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In commit 6290602709 ("mm: add PageWaiters indicating tasks are
waiting for a page bit") Nick Piggin made our page locking no longer
unconditionally touch the hashed page waitqueue, which not only helps
performance in general, but is particularly helpful on NUMA machines
where the hashed wait queues can bounce around a lot.
However, the "clear lock bit atomically and then test the waiters bit"
sequence turns out to be much more expensive than it needs to be,
because you get a nasty stall when trying to access the same word that
just got updated atomically.
On architectures where locking is done with LL/SC, this would be trivial
to fix with a new primitive that clears one bit and tests another
atomically, but that ends up not working on x86, where the only atomic
operations that return the result end up being cmpxchg and xadd. The
atomic bit operations return the old value of the same bit we changed,
not the value of an unrelated bit.
On x86, we could put the lock bit in the high bit of the byte, and use
"xadd" with that bit (where the overflow ends up not touching other
bits), and look at the other bits of the result. However, an even
simpler model is to just use a regular atomic "and" to clear the lock
bit, and then the sign bit in eflags will indicate the resulting state
of the unrelated bit #7.
So by moving the PageWaiters bit up to bit #7, we can atomically clear
the lock bit and test the waiters bit on x86 too. And architectures
with LL/SC (which is all the usual RISC suspects), the particular bit
doesn't matter, so they are fine with this approach too.
This avoids the extra access to the same atomic word, and thus avoids
the costly stall at page unlock time.
The only downside is that the interface ends up being a bit odd and
specialized: clear a bit in a byte, and test the sign bit. Nick doesn't
love the resulting name of the new primitive, but I'd rather make the
name be descriptive and very clear about the limitation imposed by
trying to work across all relevant architectures than make it be some
generic thing that doesn't make the odd semantics explicit.
So this introduces the new architecture primitive
clear_bit_unlock_is_negative_byte();
and adds the trivial implementation for x86. We have a generic
non-optimized fallback (that just does a "clear_bit()"+"test_bit(7)"
combination) which can be overridden by any architecture that can do
better. According to Nick, Power has the same hickup x86 has, for
example, but some other architectures may not even care.
All these optimizations mean that my page locking stress-test (which is
just executing a lot of small short-lived shell scripts: "make test" in
the git source tree) no longer makes our page locking look horribly bad.
Before all these optimizations, just the unlock_page() costs were just
over 3% of all CPU overhead on "make test". After this, it's down to
0.66%, so just a quarter of the cost it used to be.
(The difference on NUMA is bigger, but there this micro-optimization is
likely less noticeable, since the big issue on NUMA was not the accesses
to 'struct page', but the waitqueue accesses that were already removed
by Nick's earlier commit).
Acked-by: Nick Piggin <npiggin@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Bob Peterson <rpeterso@redhat.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Andrew Lutomirski <luto@kernel.org>
Cc: Andreas Gruenbacher <agruenba@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Exynos AUDSS clock driver has been already converted to platform driver,
so remove the dependency on the syscore ops - the last remaining
non-platform driver feature. Platform device's system sleep PM provides
all needed infrastructure for replacing syscore-based PM, so do it now.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com>
Reviewed-by: Chanwoo Choi <cw00.choi@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
The ACLK_BUS0/1/2 are used for NoC (Network on Chip). If NoC's clocks are
disabled, the system halt happens. Following clocks must be always enabled:
- CLK_ACLK_BUS0_400 : NoC's bus clock for PERIC/PERIS/FSYS/MSCL,
- CLK_ACLK_BUS1_400 : NoC's bus clock for MFC/HEVC/G3D,
- CLK_ACLK_BUS2_400 : NoC's bus clock for GSCL/DISP/G2D/CAM0/CAM1/ISP.
This patch also adds the CLK_SET_RATE_PARENT flag to the CLK_SCLK_JPEG_MSCL
because this clock should be used for bus frequency scaling. This clock need
to be changed on the fly with CLK_SET_RATE_PARENT flag.
Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Add cpu clock configuration data for Exynos4412 Prime SoC
(it supports additional PLL rates & CPU frequencies).
Based on Hardkernel's kernel for ODROID-X2/U2/U3 boards.
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
