kernel/git/stable/linux.git/drivers/clk/ti, branch linux-4.7.y

Merge tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux

2016-05-21T03:18:12+00:00

Pull clk updates from Stephen Boyd: "It's the usual big pile of driver updates and additions, but we do have a couple core changes in here as well. Core: - CLK_IS_CRITICAL support has been added. This should allow drivers to properly express that a certain clk should stay on even if their prepare/enable count drops to 0 (and in turn the parents of these clks should stay enabled). - A clk registration API has been added, clk_hw_register(), and an OF clk provider API has been added, of_clk_add_hw_provider(). These APIs have been put in place to further split clk providers from clk consumers, with the goal being to have clk providers never deal with struct clk pointers at all. Conversion of provider drivers is on going. clkdev has also gained support for registering clk_hw pointers directly so we can convert drivers that don't use devicetree. New Drivers: - Marvell ap806 and cp110 system controllers (with clks inside!) - Hisilicon Hi3519 clock and reset controller - Axis ARTPEC-6 clock controllers - Oxford Semiconductor OXNAS clock controllers - AXS10X I2S PLL - Rockchip RK3399 clock and reset controller Updates: - MMC2 and UART2 clks on Samsung Exynos 3250, ACLK on Samsung Exynos 542x SoCs, and some more clk ID exporting for bus frequency scaling - Proper BCM2835 PCM clk support and various other clks - i.MX clk updates for i.MX6SX, i.MX7, and VF610 - Renesas updates for R-Car H3 - Tegra210 got updates for DisplayPort and HDMI 2.0 - Rockchip driver refactorings and fixes due to adding RK3399 support" * tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux: (139 commits) clk: fix critical clock locking clk: qcom: mmcc-8996: Remove clocks that should be controlled by RPM clk: ingenic: Allow divider value to be divided clk: sunxi: Add display and TCON0 clocks driver clk: rockchip: drop old_rate calculation on pll rate changes clk: rockchip: simplify GRF handling in pll clocks clk: rockchip: lookup General Register Files in rockchip_clk_init clk: rockchip: fix the rk3399 sdmmc sample / drv name clk: mvebu: new driver for Armada CP110 system controller dt-bindings: arm: add DT binding for Marvell CP110 system controller clk: mvebu: new driver for Armada AP806 system controller clk: hisilicon: add CRG driver for hi3519 soc clk: hisilicon: export some hisilicon APIs to modules reset: hisilicon: add reset controller driver for hisilicon SOCs clk: bcm/kona: Do not use sizeof on pointer type clk: qcom: msm8916: Fix crypto clock flags clk: nxp: lpc18xx: Initialize clk_init_data::flags to 0 clk/axs10x: Add I2S PLL clock driver clk: imx7d: fix ahb clock mux 1 clk: fix comment of devm_clk_hw_register() ...

clk: ti: dra7-atl-clock: Fix of_node reference counting

2016-04-20T01:56:14+00:00

of_find_node_by_name() will call of_node_put() on the node so we need to get it first to avoid warnings. The cfg_node needs to be put after we have finished processing the properties. Signed-off-by: Peter Ujfalusi Tested-by: Nishanth Menon Signed-off-by: Stephen Boyd

clk: ti: dra7: fix kernel boot with arg 'clocksource=gp_timer'

2016-04-16T00:30:07+00:00

The OMAP Platform code provides possibility to select GP Timer as default clocksource instead of counter_32K by using bootcmd parameter 'clocksource', but the system will crash during early boot when this option is used on dra7 or omap5 platforms, because it will hit BUG() statement: omap2_gptimer_clocksource_init ->BUG_ON(res); This happens because clk_dev alias "sys_clkin_ck" is not registered. Hence, fix it by adding missing "sys_clkin_ck" clk_dev aliases definitions for omap5 and dra7. Acked-by: Tero Kristo Cc: Tony Lindgren Signed-off-by: Grygorii Strashko Signed-off-by: Stephen Boyd

clk: ti: amx3xx: limit the maximum frequency of DPLLs based on spec

2016-04-16T00:26:51+00:00

AM33xx/AM43xx devices use the same DPLL IP blocks, which only support maximum rate of 1GHz [1] for the default and 2GHz for the low-jitter type DPLLs [2]. Reflect this limitation in the DPLL init code by adding the max-rate parameter based on the DPLL types. [1] Functional, integration & test specification for GS70 ADPLLS, Rev 1.0-01 [2] Functional, integration & test specification for GS70 ADPLLLJ, Rev 0.8-02 Signed-off-by: Tero Kristo Cc: Nishanth Menon Cc: Tomi Valkeinen Cc: Lokesh Vutla Signed-off-by: Stephen Boyd

clk: ti: dpll: add support for specifying max rate for DPLLs

2016-04-16T00:26:18+00:00

DPLLs typically have a maximum rate they can support, and this varies from DPLL to DPLL. Add support of the maximum rate value to the DPLL data struct, and also add check for this in the DPLL round_rate function. Signed-off-by: Tero Kristo Reviewed-by: Nishanth Menon Cc: Tomi Valkeinen Cc: Lokesh Vutla Signed-off-by: Stephen Boyd

clk: ti: dflt: remove redundant unlikely

2016-04-15T22:52:48+00:00

Commit 7aba4f5201d1 ("clk: ti: dflt: fix enable_reg validity check") fixed a validation check by using an IS_ERR() macro within the existing unlikely expression, but IS_ERR() macro already has an unlikely inside it, so get rid of the redundant unlikely macro from the validation check. Reported-by: Stephen Boyd Signed-off-by: Suman Anna Signed-off-by: Stephen Boyd

ARM: dts: dra7xx: Correct mcasp8_ahclkx_mux name

2016-04-11T19:48:45+00:00

rename the mcasp8_ahclk_mux to mcasp8_ahclkx_mux. Signed-off-by: Peter Ujfalusi [tony@atomide.com: updated for the unit offsets] Signed-off-by: Tony Lindgren

clk: ti: Remove CLK_IS_ROOT

2016-03-03T01:47:46+00:00

This flag is a no-op now. Remove usage of the flag. Acked-by: Tero Kristo Signed-off-by: Stephen Boyd

Merge branch 'clk-ti' into clk-next

2016-03-02T00:23:40+00:00

Conflicts: drivers/clk/Kconfig

clk: ti: Fix some errors found by static checkers

2016-03-02T00:21:25+00:00

drivers/clk/ti/clk-814x.c:34:12: warning: symbol 'dm814x_adpll_early_init' was not declared. Should it be static? drivers/clk/ti/clk-814x.c:58:12: warning: symbol 'dm814x_adpll_enable_init_clocks' was not declared. Should it be static? drivers/clk/ti/adpll.c:465 ti_adpll_recalc_rate() warn: should '__readw(d->regs + 20) << 18' be a 64 bit type? drivers/clk/ti/adpll.c:945 ti_adpll_probe() error: we previously assumed 'd->clocks' could be null (see line 921) The last one looks like a real bug because we don't return an error on allocation failure. Cc: Tero Kristo Tested-by: Tony Lindgren Signed-off-by: Stephen Boyd