* [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly @ 2013-06-09 10:55 Linus Walleij [not found] ` <1370775348-7440-1-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> 0 siblings, 1 reply; 11+ messages in thread From: Linus Walleij @ 2013-06-09 10:55 UTC (permalink / raw) To: Mike Turquette Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r The Nomadik clock implementation was a stub just using fixed clocks. This implements the clocks properly instead of relying on them all being on at boot and leaving them all on. The PLLs are on the top locking to the main chrystal oscillator, then the HCLK for the peripherals are below PLL2. The gated clocks are implemented with zero cells and given the clock ID as a property of each node, so every gate need to have its own node in the device tree. This is because the gate registers contain both HCLK gates and PCLK gates, where the latter has HCLK as parent. As can be seen from the register layout, this is a complete mixup, which means all these gates need their own node to properly model parent/child relations for PCLKs apart from the HCLKs. This driver also adds a helpful debugfs file to inspect the hardware state of the clock gates. This is the end result in <debugfs>/clk/clk_summary after applying a proper device tree: ulpiclk 0 0 60000000 mxtal 3 3 19200000 pll2 1 1 864000000 clk48 3 3 48000000 rngcclk 1 1 48000000 usbmclk 0 0 48000000 mshcclk 0 0 48000000 mspclk3 0 0 48000000 x3dclk 0 0 48000000 skeclk 0 0 48000000 owmclk 0 0 48000000 mspclk2 0 0 48000000 mspclk1 0 0 48000000 uart2clk 0 0 48000000 ipbmcclk 0 0 48000000 ipi2cclk 0 0 48000000 usbclk 0 0 48000000 mspclk0 0 0 48000000 uart1clk 1 2 48000000 i2c1clk 0 0 48000000 i2c0clk 0 0 48000000 sdiclk 1 1 48000000 uart0clk 0 0 48000000 sspiclk 0 0 48000000 irdaclk 0 0 48000000 clk72 0 0 72000000 difclk 0 0 72000000 clcdclk 0 0 72000000 clk216 0 0 216000000 hsiclkrx 0 0 216000000 clk108 0 0 108000000 hsiclktx 0 0 108000000 clk27 0 0 27000000 pll1 1 1 264000000 hclk 3 3 264000000 hclkrng 1 1 264000000 hclkusbm 0 0 264000000 hclkcryp 0 0 264000000 hclkhash 0 0 264000000 hclk3d 0 0 264000000 hclkhpi 0 0 264000000 hclksva 0 0 264000000 hclksaa 0 0 264000000 hclkdif 0 0 264000000 hclkusb 0 0 264000000 hclkclcd 0 0 264000000 hclkdma1 0 0 264000000 hclksdram 0 0 264000000 hclksmc 1 1 264000000 hclkdma0 0 0 264000000 pclk 7 9 264000000 pclkmsp3 0 0 264000000 pclkmshc 0 0 264000000 pclkhsem 0 0 264000000 pclkske 0 0 264000000 pclkowm 0 0 264000000 pclkmsp2 0 0 264000000 pclkmsp1 0 0 264000000 pclkuart2 0 0 264000000 pclkxti 0 0 264000000 pclkhsi 0 0 264000000 pclkmsp0 0 0 264000000 pclkuart1 1 1 264000000 pclki2c1 0 0 264000000 pclki2c0 0 0 264000000 pclksdi 1 1 264000000 pclkuart0 1 1 264000000 pclkssp 0 0 264000000 pclkirda 0 0 264000000 timclk 1 1 2400000 Signed-off-by: Linus Walleij <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> --- Hi Mike, I'm seeking an ACK to take this into the ARM SoC tree on top of the other pending Nomadik clock patches there. --- .../devicetree/bindings/clock/st,nomadik.txt | 104 ++++ drivers/clk/clk-nomadik.c | 553 ++++++++++++++++++++- 2 files changed, 654 insertions(+), 3 deletions(-) create mode 100644 Documentation/devicetree/bindings/clock/st,nomadik.txt diff --git a/Documentation/devicetree/bindings/clock/st,nomadik.txt b/Documentation/devicetree/bindings/clock/st,nomadik.txt new file mode 100644 index 0000000..7fc0977 --- /dev/null +++ b/Documentation/devicetree/bindings/clock/st,nomadik.txt @@ -0,0 +1,104 @@ +ST Microelectronics Nomadik SRC System Reset and Control + +This binding uses the common clock binding: +Documentation/devicetree/bindings/clock/clock-bindings.txt + +The Nomadik SRC controller is responsible of controlling chrystals, +PLLs and clock gates. + +Required properties for the SRC node: +- compatible: must be "stericsson,nomadik-src" +- reg: must contain the SRC register base and size + +Optional properties for the SRC node: +- disable-sxtalo: if present this will disable the SXTALO + i.e. the driver output for the slow 32kHz chrystal, if the + board has its own circuitry for providing this oscillator +- disable-mxtal: if present this will disable the MXTALO, + i.e. the driver output for the main (~19.2 MHz) chrystal, + if the board has its own circuitry for providing this + osciallator + + +PLL nodes: these nodes represent the two PLLs on the system, +which should both have the main chrystal, represented as a +fixed frequency clock, as parent. + +Required properties for the two PLL nodes: +- compatible: must be "st,nomadik-pll-clock" +- clock-cells: must be 0 +- clock-id: must be 1 or 2 for PLL1 and PLL2 respectively +- clocks: this clock will have main chrystal as parent + + +HCLK nodes: these represent the clock gates on individual +lines from the HCLK clock tree and the gate for individual +lines from the PCLK clock tree. + +Requires properties for the HCLK nodes: +- compatible: must be "st,nomadik-hclk-clock" +- clock-cells: must be 0 +- clock-id: must be the clock ID from 0 to 63 according to + this table: + + 0: HCLKDMA0 + 1: HCLKSMC + 2: HCLKSDRAM + 3: HCLKDMA1 + 4: HCLKCLCD + 5: PCLKIRDA + 6: PCLKSSP + 7: PCLKUART0 + 8: PCLKSDI + 9: PCLKI2C0 + 10: PCLKI2C1 + 11: PCLKUART1 + 12: PCLMSP0 + 13: HCLKUSB + 14: HCLKDIF + 15: HCLKSAA + 16: HCLKSVA + 17: PCLKHSI + 18: PCLKXTI + 19: PCLKUART2 + 20: PCLKMSP1 + 21: PCLKMSP2 + 22: PCLKOWM + 23: HCLKHPI + 24: PCLKSKE + 25: PCLKHSEM + 26: HCLK3D + 27: HCLKHASH + 28: HCLKCRYP + 29: PCLKMSHC + 30: HCLKUSBM + 31: HCLKRNG + (32, 33, 34, 35 RESERVED) + 36: CLDCLK + 37: IRDACLK + 38: SSPICLK + 39: UART0CLK + 40: SDICLK + 41: I2C0CLK + 42: I2C1CLK + 43: UART1CLK + 44: MSPCLK0 + 45: USBCLK + 46: DIFCLK + 47: IPI2CCLK + 48: IPBMCCLK + 49: HSICLKRX + 50: HSICLKTX + 51: UART2CLK + 52: MSPCLK1 + 53: MSPCLK2 + 54: OWMCLK + (55 RESERVED) + 56: SKECLK + (57 RESERVED) + 58: 3DCLK + 59: PCLKMSP3 + 60: MSPCLK3 + 61: MSHCCLK + 62: USBMCLK + 63: RNGCCLK diff --git a/drivers/clk/clk-nomadik.c b/drivers/clk/clk-nomadik.c index 4a1ab27..0cd927f 100644 --- a/drivers/clk/clk-nomadik.c +++ b/drivers/clk/clk-nomadik.c @@ -1,21 +1,568 @@ +/* + * Nomadik clock implementation + * Copyright (C) 2013 ST-Ericsson AB + * License terms: GNU General Public License (GPL) version 2 + * Author: Linus Walleij <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> + */ + +#define pr_fmt(fmt) "Nomadik SRC clocks: " fmt + +#include <linux/bitops.h> #include <linux/clk.h> #include <linux/clkdev.h> #include <linux/err.h> #include <linux/io.h> #include <linux/clk-provider.h> #include <linux/of.h> +#include <linux/of_address.h> +#include <linux/debugfs.h> +#include <linux/seq_file.h> +#include <linux/spinlock.h> +#include <linux/reboot.h> /* * The Nomadik clock tree is described in the STN8815A12 DB V4.2 * reference manual for the chip, page 94 ff. + * Clock IDs are in the STn8815 Reference Manual table 3, page 27. + */ + +#define SRC_CR 0x00U +#define SRC_XTALCR 0x0CU +#define SRC_XTALCR_XTALTIMEN BIT(20) +#define SRC_XTALCR_SXTALDIS BIT(19) +#define SRC_XTALCR_MXTALSTAT BIT(2) +#define SRC_XTALCR_MXTALEN BIT(1) +#define SRC_XTALCR_MXTALOVER BIT(0) +#define SRC_PLLCR 0x10U +#define SRC_PLLCR_PLLTIMEN BIT(29) +#define SRC_PLLCR_PLL2EN BIT(28) +#define SRC_PLLCR_PLL1STAT BIT(2) +#define SRC_PLLCR_PLL1EN BIT(1) +#define SRC_PLLCR_PLL1OVER BIT(0) +#define SRC_PLLFR 0x14U +#define SRC_PCKEN0 0x24U +#define SRC_PCKDIS0 0x28U +#define SRC_PCKENSR0 0x2CU +#define SRC_PCKSR0 0x30U +#define SRC_PCKEN1 0x34U +#define SRC_PCKDIS1 0x38U +#define SRC_PCKENSR1 0x3CU +#define SRC_PCKSR1 0x40U + +/* Lock protecting the SRC_CR register */ +static DEFINE_SPINLOCK(src_lock); +/* Base address of the SRC */ +static void __iomem *src_base; + +/** + * struct clk_pll1 - Nomadik PLL1 clock + * @hw: corresponding clock hardware entry + * @id: PLL instance: 1 or 2 + */ +struct clk_pll { + struct clk_hw hw; + int id; +}; + +/** + * struct clk_src - Nomadik src clock + * @hw: corresponding clock hardware entry + * @id: the clock ID + * @group1: true if the clock is in group1, else it is in group0 + * @clkbit: bit 0...31 corresponding to the clock in each clock register + */ +struct clk_src { + struct clk_hw hw; + int id; + bool group1; + u32 clkbit; +}; + +#define to_pll(_hw) container_of(_hw, struct clk_pll, hw) +#define to_src(_hw) container_of(_hw, struct clk_src, hw) + +static int pll_clk_enable(struct clk_hw *hw) +{ + struct clk_pll *pll = to_pll(hw); + u32 val; + + spin_lock(&src_lock); + val = readl(src_base + SRC_PLLCR); + if (pll->id == 1) { + if (val & SRC_PLLCR_PLL1OVER) { + val |= SRC_PLLCR_PLL1EN; + writel(val, src_base + SRC_PLLCR); + } + } else if (pll->id == 2) { + val |= SRC_PLLCR_PLL2EN; + writel(val, src_base + SRC_PLLCR); + } + spin_unlock(&src_lock); + return 0; +} + +static void pll_clk_disable(struct clk_hw *hw) +{ + struct clk_pll *pll = to_pll(hw); + u32 val; + + return; + + spin_lock(&src_lock); + val = readl(src_base + SRC_PLLCR); + if (pll->id == 1) { + if (val & SRC_PLLCR_PLL1OVER) { + val &= ~SRC_PLLCR_PLL1EN; + writel(val, src_base + SRC_PLLCR); + } + } else if (pll->id == 2) { + val &= ~SRC_PLLCR_PLL2EN; + writel(val, src_base + SRC_PLLCR); + } + spin_unlock(&src_lock); +} + +static int pll_clk_is_enabled(struct clk_hw *hw) +{ + struct clk_pll *pll = to_pll(hw); + u32 val; + + val = readl(src_base + SRC_PLLCR); + if (pll->id == 1) { + if (val & SRC_PLLCR_PLL1OVER) + return !!(val & SRC_PLLCR_PLL1EN); + } else if (pll->id == 2) { + return !!(val & SRC_PLLCR_PLL2EN); + } + return 1; +} + +static unsigned long pll_clk_recalc_rate(struct clk_hw *hw, + unsigned long parent_rate) +{ + struct clk_pll *pll = to_pll(hw); + u32 val; + + val = readl(src_base + SRC_PLLFR); + + if (pll->id == 1) { + u8 mul; + u8 div; + + mul = (val >> 8) & 0x3FU; + mul += 2; + div = val & 0x07U; + return (parent_rate * mul) >> div; + } + + if (pll->id == 2) { + u8 mul; + + mul = (val >> 24) & 0x3FU; + mul += 2; + return (parent_rate * mul); + } + + /* Unknown PLL */ + return 0; +} + + +static const struct clk_ops pll_clk_ops = { + .enable = pll_clk_enable, + .disable = pll_clk_disable, + .is_enabled = pll_clk_is_enabled, + .recalc_rate = pll_clk_recalc_rate, +}; + +static struct clk * __init +pll_clk_register(struct device *dev, const char *name, + const char *parent_name, u32 id) +{ + struct clk *clk; + struct clk_pll *pll; + struct clk_init_data init; + + if (id != 1 && id != 2) { + pr_err("%s: the Nomadik has only PLL 1 & 2\n", __func__); + return ERR_PTR(-EINVAL); + } + + pll = kzalloc(sizeof(*pll), GFP_KERNEL); + if (!pll) { + pr_err("%s: could not allocate PLL clk\n", __func__); + return ERR_PTR(-ENOMEM); + } + + init.name = name; + init.ops = &pll_clk_ops; + init.parent_names = (parent_name ? &parent_name : NULL); + init.num_parents = (parent_name ? 1 : 0); + pll->hw.init = &init; + pll->id = id; + + pr_debug("register PLL1 clock \"%s\"\n", name); + + clk = clk_register(dev, &pll->hw); + if (IS_ERR(clk)) + kfree(pll); + + return clk; +} + +/* + * The Nomadik SRC clocks are gated, but not in the sense that + * you read-modify-write a register. Instead there are separate + * clock enable and clock disable registers. Writing a '1' bit in + * the enable register for a certain clock ungates that clock without + * affecting the other clocks. The disable register works the opposite + * way. */ -static const __initconst struct of_device_id cpu8815_clk_match[] = { - { .compatible = "fixed-clock", .data = of_fixed_clk_setup, }, +static int src_clk_enable(struct clk_hw *hw) +{ + struct clk_src *sclk = to_src(hw); + u32 enreg = sclk->group1 ? SRC_PCKEN1 : SRC_PCKEN0; + u32 sreg = sclk->group1 ? SRC_PCKSR1 : SRC_PCKSR0; + + writel(sclk->clkbit, src_base + enreg); + /* spin until enabled */ + while (!(readl(src_base + sreg) & sclk->clkbit)) + cpu_relax(); + return 0; +} + +static void src_clk_disable(struct clk_hw *hw) +{ + struct clk_src *sclk = to_src(hw); + u32 disreg = sclk->group1 ? SRC_PCKDIS1 : SRC_PCKDIS0; + u32 sreg = sclk->group1 ? SRC_PCKSR1 : SRC_PCKSR0; + + writel(sclk->clkbit, src_base + disreg); + /* spin until disabled */ + while (readl(src_base + sreg) & sclk->clkbit) + cpu_relax(); +} + +static int src_clk_is_enabled(struct clk_hw *hw) +{ + struct clk_src *sclk = to_src(hw); + u32 sreg = sclk->group1 ? SRC_PCKSR1 : SRC_PCKSR0; + u32 val = readl(src_base + sreg); + + return !!(val & sclk->clkbit); +} + +static unsigned long +src_clk_recalc_rate(struct clk_hw *hw, + unsigned long parent_rate) +{ + return parent_rate; +} + +static const struct clk_ops src_clk_ops = { + .enable = src_clk_enable, + .disable = src_clk_disable, + .is_enabled = src_clk_is_enabled, + .recalc_rate = src_clk_recalc_rate, +}; + +static struct clk * __init +src_clk_register(struct device *dev, const char *name, + const char *parent_name, u8 id) +{ + struct clk *clk; + struct clk_src *sclk; + struct clk_init_data init; + + sclk = kzalloc(sizeof(*sclk), GFP_KERNEL); + if (!sclk) { + pr_err("could not allocate SRC clock %s\n", + name); + return ERR_PTR(-ENOMEM); + } + init.name = name; + init.ops = &src_clk_ops; + /* Do not force-disable the static SDRAM controller */ + if (id == 2) + init.flags = CLK_IGNORE_UNUSED; + else + init.flags = 0; + init.parent_names = (parent_name ? &parent_name : NULL); + init.num_parents = (parent_name ? 1 : 0); + sclk->hw.init = &init; + sclk->id = id; + sclk->group1 = (id > 31); + sclk->clkbit = BIT(id & 0x1f); + + pr_debug("register clock \"%s\" ID: %d group: %d bits: %08x\n", + name, id, sclk->group1, sclk->clkbit); + + clk = clk_register(dev, &sclk->hw); + if (IS_ERR(clk)) + kfree(sclk); + + return clk; +} + +#ifdef CONFIG_DEBUG_FS + +static u32 src_pcksr0_boot; +static u32 src_pcksr1_boot; + +static const char * const src_clk_names[] = { + "HCLKDMA0 ", + "HCLKSMC ", + "HCLKSDRAM ", + "HCLKDMA1 ", + "HCLKCLCD ", + "PCLKIRDA ", + "PCLKSSP ", + "PCLKUART0 ", + "PCLKSDI ", + "PCLKI2C0 ", + "PCLKI2C1 ", + "PCLKUART1 ", + "PCLMSP0 ", + "HCLKUSB ", + "HCLKDIF ", + "HCLKSAA ", + "HCLKSVA ", + "PCLKHSI ", + "PCLKXTI ", + "PCLKUART2 ", + "PCLKMSP1 ", + "PCLKMSP2 ", + "PCLKOWM ", + "HCLKHPI ", + "PCLKSKE ", + "PCLKHSEM ", + "HCLK3D ", + "HCLKHASH ", + "HCLKCRYP ", + "PCLKMSHC ", + "HCLKUSBM ", + "HCLKRNG ", + "RESERVED ", + "RESERVED ", + "RESERVED ", + "RESERVED ", + "CLDCLK ", + "IRDACLK ", + "SSPICLK ", + "UART0CLK ", + "SDICLK ", + "I2C0CLK ", + "I2C1CLK ", + "UART1CLK ", + "MSPCLK0 ", + "USBCLK ", + "DIFCLK ", + "IPI2CCLK ", + "IPBMCCLK ", + "HSICLKRX ", + "HSICLKTX ", + "UART2CLK ", + "MSPCLK1 ", + "MSPCLK2 ", + "OWMCLK ", + "RESERVED ", + "SKECLK ", + "RESERVED ", + "3DCLK ", + "PCLKMSP3 ", + "MSPCLK3 ", + "MSHCCLK ", + "USBMCLK ", + "RNGCCLK ", +}; + +static int nomadik_src_clk_show(struct seq_file *s, void *what) +{ + int i; + u32 src_pcksr0 = readl(src_base + SRC_PCKSR0); + u32 src_pcksr1 = readl(src_base + SRC_PCKSR1); + u32 src_pckensr0 = readl(src_base + SRC_PCKENSR0); + u32 src_pckensr1 = readl(src_base + SRC_PCKENSR1); + + seq_printf(s, "Clock: Boot: Now: Request: ASKED:\n"); + for (i = 0; i < ARRAY_SIZE(src_clk_names); i++) { + u32 pcksrb = (i < 0x20) ? src_pcksr0_boot : src_pcksr1_boot; + u32 pcksr = (i < 0x20) ? src_pcksr0 : src_pcksr1; + u32 pckreq = (i < 0x20) ? src_pckensr0 : src_pckensr1; + u32 mask = BIT(i & 0x1f); + + seq_printf(s, "%s %s %s %s\n", + src_clk_names[i], + (pcksrb & mask) ? "on " : "off", + (pcksr & mask) ? "on " : "off", + (pckreq & mask) ? "on " : "off"); + } + return 0; +} + +static int nomadik_src_clk_open(struct inode *inode, struct file *file) +{ + return single_open(file, nomadik_src_clk_show, NULL); +} + +static const struct file_operations nomadik_src_clk_debugfs_ops = { + .open = nomadik_src_clk_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int __init nomadik_src_clk_init_debugfs(void) +{ + src_pcksr0_boot = readl(src_base + SRC_PCKSR0); + src_pcksr1_boot = readl(src_base + SRC_PCKSR1); + debugfs_create_file("nomadik-src-clk", S_IFREG | S_IRUGO, + NULL, NULL, &nomadik_src_clk_debugfs_ops); + return 0; +} + +module_init(nomadik_src_clk_init_debugfs); + +#endif + +static void __init of_nomadik_pll_setup(struct device_node *np) +{ + struct clk *clk = ERR_PTR(-EINVAL); + const char *clk_name = np->name; + const char *parent_name; + u32 pll_id; + + if (of_property_read_u32(np, "pll-id", &pll_id)) { + pr_err("%s: PLL \"%s\" missing pll-id property\n", + __func__, clk_name); + return; + } + parent_name = of_clk_get_parent_name(np, 0); + clk = pll_clk_register(NULL, clk_name, parent_name, pll_id); + if (!IS_ERR(clk)) + of_clk_add_provider(np, of_clk_src_simple_get, clk); +} + +static void __init of_nomadik_hclk_setup(struct device_node *np) +{ + struct clk *clk = ERR_PTR(-EINVAL); + const char *clk_name = np->name; + const char *parent_name; + + parent_name = of_clk_get_parent_name(np, 0); + /* + * The HCLK divides PLL1 with 1 (passthru), 2, 3 or 4. + */ + clk = clk_register_divider(NULL, clk_name, parent_name, + 0, src_base + SRC_CR, + 13, 2, + CLK_DIVIDER_ONE_BASED | CLK_DIVIDER_ALLOW_ZERO, + &src_lock); + if (!IS_ERR(clk)) + of_clk_add_provider(np, of_clk_src_simple_get, clk); +} + +static void __init of_nomadik_src_clk_setup(struct device_node *np) +{ + struct clk *clk = ERR_PTR(-EINVAL); + const char *clk_name = np->name; + const char *parent_name; + u32 clk_id; + + if (of_property_read_u32(np, "clock-id", &clk_id)) { + pr_err("%s: SRC clock \"%s\" missing clock-id property\n", + __func__, clk_name); + return; + } + parent_name = of_clk_get_parent_name(np, 0); + clk = src_clk_register(NULL, clk_name, parent_name, clk_id); + if (!IS_ERR(clk)) + of_clk_add_provider(np, of_clk_src_simple_get, clk); +} + +static const __initconst struct of_device_id nomadik_src_match[] = { + { .compatible = "stericsson,nomadik-src" }, { /* sentinel */ } }; +static const __initconst struct of_device_id nomadik_src_clk_match[] = { + { + .compatible = "fixed-clock", + .data = of_fixed_clk_setup, + }, + { + .compatible = "fixed-factor-clock", + .data = of_fixed_factor_clk_setup, + }, + { + .compatible = "st,nomadik-pll-clock", + .data = of_nomadik_pll_setup, + }, + { + .compatible = "st,nomadik-hclk-clock", + .data = of_nomadik_hclk_setup, + }, + { + .compatible = "st,nomadik-src-clock", + .data = of_nomadik_src_clk_setup, + }, + { /* sentinel */ } +}; + +static int nomadik_clk_reboot_handler(struct notifier_block *this, + unsigned long code, + void *unused) +{ + u32 val; + + /* The main chrystal need to be enabled for reboot to work */ + val = readl(src_base + SRC_XTALCR); + val &= ~SRC_XTALCR_MXTALOVER; + val |= SRC_XTALCR_MXTALEN; + pr_crit("force-enabling MXTALO\n"); + writel(val, src_base + SRC_XTALCR); + return NOTIFY_OK; +} + +static struct notifier_block nomadik_clk_reboot_notifier = { + .notifier_call = nomadik_clk_reboot_handler, +}; + void __init nomadik_clk_init(void) { - of_clk_init(cpu8815_clk_match); + struct device_node *np; + u32 val; + + np = of_find_matching_node(NULL, nomadik_src_match); + if (!np) { + pr_crit("no matching node for SRC, aborting clock init\n"); + return; + } + src_base = of_iomap(np, 0); + if (!src_base) { + pr_err("%s: must have src parent node with REGS (%s)\n", + __func__, np->name); + return; + } + val = readl(src_base + SRC_XTALCR); + pr_info("SXTALO is %s\n", + (val & SRC_XTALCR_SXTALDIS) ? "disabled" : "enabled"); + pr_info("MXTAL is %s\n", + (val & SRC_XTALCR_MXTALSTAT) ? "enabled" : "disabled"); + if (of_property_read_bool(np, "disable-sxtalo")) { + /* The machine uses an external oscillator circuit */ + val |= SRC_XTALCR_SXTALDIS; + pr_info("disabling SXTALO\n"); + } + if (of_property_read_bool(np, "disable-mxtalo")) { + /* Disable this too: also run by external oscillator */ + val |= SRC_XTALCR_MXTALOVER; + val &= ~SRC_XTALCR_MXTALEN; + pr_info("disabling MXTALO\n"); + } + writel(val, src_base + SRC_XTALCR); + register_reboot_notifier(&nomadik_clk_reboot_notifier); + + of_clk_init(nomadik_src_clk_match); } -- 1.8.1.4 ^ permalink raw reply related [flat|nested] 11+ messages in thread
[parent not found: <1370775348-7440-1-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>]
* [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree [not found] ` <1370775348-7440-1-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> @ 2013-06-09 10:55 ` Linus Walleij [not found] ` <1370775348-7440-2-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> 2013-06-16 19:08 ` [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly Linus Walleij 2013-06-20 6:38 ` Mike Turquette 2 siblings, 1 reply; 11+ messages in thread From: Linus Walleij @ 2013-06-09 10:55 UTC (permalink / raw) To: Mike Turquette Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r This revamps the device tree to fit with the new clock implementation and brings it quite a bit closer to how the hardware actually works. After this the clock implementation knows about all clock gates and will gate off all unused clocks at boot time and save a bit of power. Signed-off-by: Linus Walleij <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> --- arch/arm/boot/dts/ste-nomadik-s8815.dts | 6 + arch/arm/boot/dts/ste-nomadik-stn8815.dtsi | 501 ++++++++++++++++++++++++++--- 2 files changed, 470 insertions(+), 37 deletions(-) diff --git a/arch/arm/boot/dts/ste-nomadik-s8815.dts b/arch/arm/boot/dts/ste-nomadik-s8815.dts index 638ec8d..fd573d2 100644 --- a/arch/arm/boot/dts/ste-nomadik-s8815.dts +++ b/arch/arm/boot/dts/ste-nomadik-s8815.dts @@ -14,6 +14,12 @@ bootargs = "root=/dev/ram0 console=ttyAMA1,115200n8 earlyprintk"; }; + src@101e0000 { + /* These chrystal drivers are not used on this board */ + disable-sxtalo; + disable-mxtalo; + }; + pinctrl { /* Hog CD pins */ pinctrl-names = "default"; diff --git a/arch/arm/boot/dts/ste-nomadik-stn8815.dtsi b/arch/arm/boot/dts/ste-nomadik-stn8815.dtsi index dbf476b..a3acfa7 100644 --- a/arch/arm/boot/dts/ste-nomadik-stn8815.dtsi +++ b/arch/arm/boot/dts/ste-nomadik-stn8815.dtsi @@ -168,37 +168,464 @@ src: src@101e0000 { compatible = "stericsson,nomadik-src"; reg = <0x101e0000 0x1000>; - clocks { - /* - * Dummy clock for primecells - */ - pclk: pclk@0 { - #clock-cells = <0>; - compatible = "fixed-clock"; - clock-frequency = <0>; - }; - /* - * The 2.4 MHz TIMCLK reference clock is active at - * boot time, this is actually the MXTALCLK @19.2 MHz - * divided by 8. This clock is used by the timers and - * watchdog. See page 105 ff. - */ - timclk: timclk@2.4M { - #clock-cells = <0>; - compatible = "fixed-clock"; - clock-frequency = <2400000>; - }; - /* - * At boot time, PLL2 is set to generate a set of - * fixed clocks, one of them is CLK48, the 48 MHz - * clock, routed to the UART, MMC/SD, I2C, IrDA, - * USB and SSP blocks. - */ - clk48: clk48@48M { - #clock-cells = <0>; - compatible = "fixed-clock"; - clock-frequency = <48000000>; - }; + disable-sxtalo; + disable-mxtalo; + + /* + * MXTAL "Main Chrystal" is a chrystal oscillator @19.2 MHz + * that is parent of TIMCLK, PLL1 and PLL2 + */ + mxtal: mxtal-QvK0ZaTcdGg@public.gmane.org { + #clock-cells = <0>; + compatible = "fixed-clock"; + clock-frequency = <19200000>; + }; + + /* + * The 2.4 MHz TIMCLK reference clock is active at + * boot time, this is actually the MXTALCLK @19.2 MHz + * divided by 8. This clock is used by the timers and + * watchdog. See page 105 ff. + */ + timclk: timclk@2.4M { + #clock-cells = <0>; + compatible = "fixed-factor-clock"; + clock-div = <8>; + clock-mult = <1>; + clocks = <&mxtal>; + }; + + /* PLL1 is locked to MXTALI and variable from 20.4 to 334 MHz */ + pll1: pll1@0 { + #clock-cells = <0>; + compatible = "st,nomadik-pll-clock"; + pll-id = <1>; + clocks = <&mxtal>; + }; + + /* HCLK divides the PLL1 with 1,2,3 or 4 */ + hclk: hclk@0 { + #clock-cells = <0>; + compatible = "st,nomadik-hclk-clock"; + clocks = <&pll1>; + }; + /* The PCLK domain uses HCLK right off */ + pclk: pclk@0 { + #clock-cells = <0>; + compatible = "fixed-factor-clock"; + clock-div = <1>; + clock-mult = <1>; + clocks = <&hclk>; + }; + + /* PLL2 is usually 864 MHz and divided into a few fixed rates */ + pll2: pll2@0 { + #clock-cells = <0>; + compatible = "st,nomadik-pll-clock"; + pll-id = <2>; + clocks = <&mxtal>; + }; + clk216: clk216@216M { + #clock-cells = <0>; + compatible = "fixed-factor-clock"; + clock-div = <4>; + clock-mult = <1>; + clocks = <&pll2>; + }; + clk108: clk108@108M { + #clock-cells = <0>; + compatible = "fixed-factor-clock"; + clock-div = <2>; + clock-mult = <1>; + clocks = <&clk216>; + }; + clk72: clk72@72M { + #clock-cells = <0>; + compatible = "fixed-factor-clock"; + /* The data sheet does not say how this is derived */ + clock-div = <12>; + clock-mult = <1>; + clocks = <&pll2>; + }; + clk48: clk48@48M { + #clock-cells = <0>; + compatible = "fixed-factor-clock"; + /* The data sheet does not say how this is derived */ + clock-div = <18>; + clock-mult = <1>; + clocks = <&pll2>; + }; + clk27: clk27@27M { + #clock-cells = <0>; + compatible = "fixed-factor-clock"; + clock-div = <4>; + clock-mult = <1>; + clocks = <&clk108>; + }; + + /* This apparently exists as well */ + ulpiclk: ulpiclk@60M { + #clock-cells = <0>; + compatible = "fixed-clock"; + clock-frequency = <60000000>; + }; + + /* + * IP AMBA bus clocks, driving the bus side of the + * peripheral clocking, clock gates. + */ + + hclkdma0: hclkdma0@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <0>; + clocks = <&hclk>; + }; + hclksmc: hclksmc@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <1>; + clocks = <&hclk>; + }; + hclksdram: hclksdram@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <2>; + clocks = <&hclk>; + }; + hclkdma1: hclkdma1@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <3>; + clocks = <&hclk>; + }; + hclkclcd: hclkclcd@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <4>; + clocks = <&hclk>; + }; + pclkirda: pclkirda@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <5>; + clocks = <&pclk>; + }; + pclkssp: pclkssp@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <6>; + clocks = <&pclk>; + }; + pclkuart0: pclkuart0@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <7>; + clocks = <&pclk>; + }; + pclksdi: pclksdi@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <8>; + clocks = <&pclk>; + }; + pclki2c0: pclki2c0@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <9>; + clocks = <&pclk>; + }; + pclki2c1: pclki2c1@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <10>; + clocks = <&pclk>; + }; + pclkuart1: pclkuart1@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <11>; + clocks = <&pclk>; + }; + pclkmsp0: pclkmsp0@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <12>; + clocks = <&pclk>; + }; + hclkusb: hclkusb@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <13>; + clocks = <&hclk>; + }; + hclkdif: hclkdif@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <14>; + clocks = <&hclk>; + }; + hclksaa: hclksaa@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <15>; + clocks = <&hclk>; + }; + hclksva: hclksva@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <16>; + clocks = <&hclk>; + }; + pclkhsi: pclkhsi@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <17>; + clocks = <&pclk>; + }; + pclkxti: pclkxti@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <18>; + clocks = <&pclk>; + }; + pclkuart2: pclkuart2@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <19>; + clocks = <&pclk>; + }; + pclkmsp1: pclkmsp1@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <20>; + clocks = <&pclk>; + }; + pclkmsp2: pclkmsp2@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <21>; + clocks = <&pclk>; + }; + pclkowm: pclkowm@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <22>; + clocks = <&pclk>; + }; + hclkhpi: hclkhpi@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <23>; + clocks = <&hclk>; + }; + pclkske: pclkske@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <24>; + clocks = <&pclk>; + }; + pclkhsem: pclkhsem@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <25>; + clocks = <&pclk>; + }; + hclk3d: hclk3d@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <26>; + clocks = <&hclk>; + }; + hclkhash: hclkhash@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <27>; + clocks = <&hclk>; + }; + hclkcryp: hclkcryp@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <28>; + clocks = <&hclk>; + }; + pclkmshc: pclkmshc@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <29>; + clocks = <&pclk>; + }; + hclkusbm: hclkusbm@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <30>; + clocks = <&hclk>; + }; + hclkrng: hclkrng@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <31>; + clocks = <&hclk>; + }; + + /* IP kernel clocks */ + clcdclk: clcdclk@0 { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <36>; + clocks = <&clk72 &clk48>; + }; + irdaclk: irdaclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <37>; + clocks = <&clk48>; + }; + sspiclk: sspiclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <38>; + clocks = <&clk48>; + }; + uart0clk: uart0clk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <39>; + clocks = <&clk48>; + }; + sdiclk: sdiclk@48M { + /* Also called MCCLK in some documents */ + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <40>; + clocks = <&clk48>; + }; + i2c0clk: i2c0clk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <41>; + clocks = <&clk48>; + }; + i2c1clk: i2c1clk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <42>; + clocks = <&clk48>; + }; + uart1clk: uart1clk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <43>; + clocks = <&clk48>; + }; + mspclk0: mspclk0@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <44>; + clocks = <&clk48>; + }; + usbclk: usbclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <45>; + clocks = <&clk48>; /* 48 MHz not ULPI */ + }; + difclk: difclk@72M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <46>; + clocks = <&clk72>; + }; + ipi2cclk: ipi2cclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <47>; + clocks = <&clk48>; /* Guess */ + }; + ipbmcclk: ipbmcclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <48>; + clocks = <&clk48>; /* Guess */ + }; + hsiclkrx: hsiclkrx@216M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <49>; + clocks = <&clk216>; + }; + hsiclktx: hsiclktx@108M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <50>; + clocks = <&clk108>; + }; + uart2clk: uart2clk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <51>; + clocks = <&clk48>; + }; + mspclk1: mspclk1@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <52>; + clocks = <&clk48>; + }; + mspclk2: mspclk2@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <53>; + clocks = <&clk48>; + }; + owmclk: owmclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <54>; + clocks = <&clk48>; /* Guess */ + }; + skeclk: skeclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <56>; + clocks = <&clk48>; /* Guess */ + }; + x3dclk: x3dclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <58>; + clocks = <&clk48>; /* Guess */ + }; + pclkmsp3: pclkmsp3@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <59>; + clocks = <&pclk>; + }; + mspclk3: mspclk3@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <60>; + clocks = <&clk48>; + }; + mshcclk: mshcclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <61>; + clocks = <&clk48>; /* Guess */ + }; + usbmclk: usbmclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <62>; + /* Stated as "48 MHz not ULPI clock" */ + clocks = <&clk48>; + }; + rngcclk: rngcclk@48M { + #clock-cells = <0>; + compatible = "st,nomadik-src-clock"; + clock-id = <63>; + clocks = <&clk48>; /* Guess */ }; }; @@ -212,7 +639,7 @@ <0x41000000 0x2000>, /* NAND Base ADDR */ <0x40800000 0x2000>; /* NAND Base CMD */ reg-names = "fsmc_regs", "nand_data", "nand_addr", "nand_cmd"; - clocks = <&pclk>; + clocks = <&hclksmc>; status = "okay"; partition@0 { @@ -334,7 +761,7 @@ reg = <0x101fd000 0x1000>; interrupt-parent = <&vica>; interrupts = <12>; - clocks = <&clk48>, <&pclk>; + clocks = <&uart0clk>, <&pclkuart0>; clock-names = "uartclk", "apb_pclk"; pinctrl-names = "default"; pinctrl-0 = <&uart0_default_mux>; @@ -345,7 +772,7 @@ reg = <0x101fb000 0x1000>; interrupt-parent = <&vica>; interrupts = <17>; - clocks = <&clk48>, <&pclk>; + clocks = <&uart1clk>, <&pclkuart1>; clock-names = "uartclk", "apb_pclk"; pinctrl-names = "default"; pinctrl-0 = <&uart1_default_mux>; @@ -356,7 +783,7 @@ reg = <0x101f2000 0x1000>; interrupt-parent = <&vica>; interrupts = <28>; - clocks = <&clk48>, <&pclk>; + clocks = <&uart2clk>, <&pclkuart2>; clock-names = "uartclk", "apb_pclk"; status = "disabled"; }; @@ -364,7 +791,7 @@ rng: rng@101b0000 { compatible = "arm,primecell"; reg = <0x101b0000 0x1000>; - clocks = <&clk48>, <&pclk>; + clocks = <&rngcclk>, <&hclkrng>; clock-names = "rng", "apb_pclk"; }; @@ -380,7 +807,7 @@ mmcsd: sdi@101f6000 { compatible = "arm,pl18x", "arm,primecell"; reg = <0x101f6000 0x1000>; - clocks = <&clk48>, <&pclk>; + clocks = <&sdiclk>, <&pclksdi>; clock-names = "mclk", "apb_pclk"; interrupt-parent = <&vica>; interrupts = <22>; -- 1.8.1.4 ^ permalink raw reply related [flat|nested] 11+ messages in thread
[parent not found: <1370775348-7440-2-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>]
* Re: [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree [not found] ` <1370775348-7440-2-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> @ 2013-06-09 14:21 ` Arnd Bergmann [not found] ` <CAJAp7OiYZx46aWRsXc+CmS+5F0STz5P+rdYnJEoaRRMUnkD1Cg@mail.gmail.com> [not found] ` <201306091621.02506.arnd-r2nGTMty4D4@public.gmane.org> 0 siblings, 2 replies; 11+ messages in thread From: Arnd Bergmann @ 2013-06-09 14:21 UTC (permalink / raw) To: Linus Walleij Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ, Mike Turquette, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r On Sunday 09 June 2013, Linus Walleij wrote: > + /* > + * IP AMBA bus clocks, driving the bus side of the > + * peripheral clocking, clock gates. > + */ > + > + hclkdma0: hclkdma0@48M { > + #clock-cells = <0>; > + compatible = "st,nomadik-src-clock"; > + clock-id = <0>; > + clocks = <&hclk>; > + }; > + hclksmc: hclksmc@48M { > + #clock-cells = <0>; > + compatible = "st,nomadik-src-clock"; > + clock-id = <1>; > + clocks = <&hclk>; > + }; > + hclksdram: hclksdram@48M { > + #clock-cells = <0>; > + compatible = "st,nomadik-src-clock"; > + clock-id = <2>; > + clocks = <&hclk>; > + }; > + hclkdma1: hclkdma1@48M { > + #clock-cells = <0>; > + compatible = "st,nomadik-src-clock"; > + clock-id = <3>; > + clocks = <&hclk>; > + }; Sorry if I'm being slow to understand how the clock bindings work, but if you have 63 identical clocks that only differ in ther clock-id, can't you just have a single DT node for them instead with #clock-cells=1 to pass the number from the device using it? Arnd ^ permalink raw reply [flat|nested] 11+ messages in thread
[parent not found: <CAJAp7OiYZx46aWRsXc+CmS+5F0STz5P+rdYnJEoaRRMUnkD1Cg@mail.gmail.com>]
* Re: [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree [not found] ` <CAJAp7OiYZx46aWRsXc+CmS+5F0STz5P+rdYnJEoaRRMUnkD1Cg@mail.gmail.com> @ 2013-06-10 5:12 ` Bjorn Andersson [not found] ` <CAJAp7Oji=+9PqwGkF84BRaxkFeeO_kWoY378Ur-A5z6G9Z5xSQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> 0 siblings, 1 reply; 11+ messages in thread From: Bjorn Andersson @ 2013-06-10 5:12 UTC (permalink / raw) To: Arnd Bergmann Cc: Linus Walleij, Mike Turquette, devicetree-discuss, linux-arm-kernel While hacking on clock support for msm8960 the other day I did the same thing; I figured that providing clocks with descriptive names would be of great benefit while reading the dts. I had my uart device setup with: clocks = <&gsbi8_uart_clk>, <&gsbi8_p_clk>; However, all currently present clock bindings follow the pattern described by Arnd. Following this pattern my uart would be setup as: clocks = <&clks 42>, <&clks 44>; Needless to say there is no way to make sense of this without consulting the specific binding documentation. So, although your way of setting up the clock binding is not the standard way I do prefer it, but it comes with plenty extra nodes in the tree. Regards, Bjorn On Sun, Jun 9, 2013 at 10:10 PM, Bjorn Andersson <bjorn@kryo.se> wrote: > Linus, > > While hacking on clock support for msm8960 the other day I did the same > thing; I figured that providing clocks with descriptive names would be of > great benefit while reading the dts. > > I had my uart device setup with: > clocks = <&gsbi8_uart_clk>, <&gsbi8_p_clk>; > > However, all currently present clock bindings seem to follow the pattern > described by Arnd. > Following this pattern my uart would be setup as: > clocks = <&clks 42>, <&clks 44>; > > Needless to say there is no way to make sense of this without consulting the > specific binding documentation. So, although your way of setting up the > clock binding is not the standard way I do prefer it, but it comes with > plenty extra nodes in the tree. > > Regards, > Bjorn > > > On Sun, Jun 9, 2013 at 7:21 AM, Arnd Bergmann <arnd@arndb.de> wrote: >> >> On Sunday 09 June 2013, Linus Walleij wrote: >> > + /* >> > + * IP AMBA bus clocks, driving the bus side of the >> > + * peripheral clocking, clock gates. >> > + */ >> > + >> > + hclkdma0: hclkdma0@48M { >> > + #clock-cells = <0>; >> > + compatible = "st,nomadik-src-clock"; >> > + clock-id = <0>; >> > + clocks = <&hclk>; >> > + }; >> > + hclksmc: hclksmc@48M { >> > + #clock-cells = <0>; >> > + compatible = "st,nomadik-src-clock"; >> > + clock-id = <1>; >> > + clocks = <&hclk>; >> > + }; >> > + hclksdram: hclksdram@48M { >> > + #clock-cells = <0>; >> > + compatible = "st,nomadik-src-clock"; >> > + clock-id = <2>; >> > + clocks = <&hclk>; >> > + }; >> > + hclkdma1: hclkdma1@48M { >> > + #clock-cells = <0>; >> > + compatible = "st,nomadik-src-clock"; >> > + clock-id = <3>; >> > + clocks = <&hclk>; >> > + }; >> >> Sorry if I'm being slow to understand how the clock bindings work, but if >> you have 63 identical clocks that only differ in ther clock-id, can't you >> just have a single DT node for them instead with #clock-cells=1 to pass >> the >> number from the device using it? >> >> Arnd >> >> _______________________________________________ >> linux-arm-kernel mailing list >> linux-arm-kernel@lists.infradead.org >> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel > > ^ permalink raw reply [flat|nested] 11+ messages in thread
[parent not found: <CAJAp7Oji=+9PqwGkF84BRaxkFeeO_kWoY378Ur-A5z6G9Z5xSQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>]
* Re: [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree [not found] ` <CAJAp7Oji=+9PqwGkF84BRaxkFeeO_kWoY378Ur-A5z6G9Z5xSQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> @ 2013-06-10 12:29 ` Linus Walleij 0 siblings, 0 replies; 11+ messages in thread From: Linus Walleij @ 2013-06-10 12:29 UTC (permalink / raw) To: Bjorn Andersson Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org, Mike Turquette, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org On Mon, Jun 10, 2013 at 7:12 AM, Bjorn Andersson <bjorn-UYDU3/A3LUY@public.gmane.org> wrote: > While hacking on clock support for msm8960 the other day I did the > same thing; I figured that providing clocks with descriptive names > would be of great benefit while reading the dts. > > I had my uart device setup with: > clocks = <&gsbi8_uart_clk>, <&gsbi8_p_clk>; > > However, all currently present clock bindings follow the pattern > described by Arnd. > Following this pattern my uart would be setup as: > clocks = <&clks 42>, <&clks 44>; > > Needless to say there is no way to make sense of this without > consulting the specific binding documentation. So, although your way > of setting up the clock binding is not the standard way I do prefer > it, but it comes with plenty extra nodes in the tree. I agree. In my case this was just a side-effect of the different bits in the register having different parents, so the common pattern could not be used, but when looking at the end result it's strikingly more readable than the cryptic numbers. However it is possible to mitigate the latter by using the new DTS preprocessor support like so: #include <dt-bindings/clk/my-clocks.h>: /* Clock IDs */ #define GSBI8_UART_CLK 42 #define GSBI8_P_CLK 44 (...) clocks = <&clks GSBI8_UART_CLK>, <&clks GSBI8_P_CLK>; Yours, Linus Walleij ^ permalink raw reply [flat|nested] 11+ messages in thread
[parent not found: <201306091621.02506.arnd-r2nGTMty4D4@public.gmane.org>]
* Re: [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree [not found] ` <201306091621.02506.arnd-r2nGTMty4D4@public.gmane.org> @ 2013-06-10 7:43 ` Linus Walleij [not found] ` <CACRpkdZNZ4_FzjEsstUm0bn8hE7FPTN1y_QnXtWUvGUzCNynBg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> 0 siblings, 1 reply; 11+ messages in thread From: Linus Walleij @ 2013-06-10 7:43 UTC (permalink / raw) To: Arnd Bergmann Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org, Mike Turquette, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org On Sun, Jun 9, 2013 at 4:21 PM, Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org> wrote: > On Sunday 09 June 2013, Linus Walleij wrote: >> + /* >> + * IP AMBA bus clocks, driving the bus side of the >> + * peripheral clocking, clock gates. >> + */ >> + >> + hclkdma0: hclkdma0@48M { >> + #clock-cells = <0>; >> + compatible = "st,nomadik-src-clock"; >> + clock-id = <0>; >> + clocks = <&hclk>; >> + }; >> + hclksmc: hclksmc@48M { >> + #clock-cells = <0>; >> + compatible = "st,nomadik-src-clock"; >> + clock-id = <1>; >> + clocks = <&hclk>; >> + }; >> + hclksdram: hclksdram@48M { >> + #clock-cells = <0>; >> + compatible = "st,nomadik-src-clock"; >> + clock-id = <2>; >> + clocks = <&hclk>; >> + }; >> + hclkdma1: hclkdma1@48M { >> + #clock-cells = <0>; >> + compatible = "st,nomadik-src-clock"; >> + clock-id = <3>; >> + clocks = <&hclk>; >> + }; > > Sorry if I'm being slow to understand how the clock bindings work, but if > you have 63 identical clocks that only differ in ther clock-id, can't you > just have a single DT node for them instead with #clock-cells=1 to pass the > number from the device using it? I first had exactly that implementation but it didn't work. This is explained in the commit message on patch 1/1: The gated clocks are implemented with zero cells and given the clock ID as a property of each node, so every gate need to have its own node in the device tree. This is because the gate registers contain both HCLK gates and PCLK gates, where the latter has HCLK as parent. As can be seen from the register layout, this is a complete mixup, which means all these gates need their own node to properly model parent/child relations for PCLKs apart from the HCLKs. Basically the "shortcoming" in device tree is that modelling this with clock-cells = 1 requires them all to have one and the same parent, but they don't. Yours, Linus Walleij ^ permalink raw reply [flat|nested] 11+ messages in thread
[parent not found: <CACRpkdZNZ4_FzjEsstUm0bn8hE7FPTN1y_QnXtWUvGUzCNynBg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>]
* Re: [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree [not found] ` <CACRpkdZNZ4_FzjEsstUm0bn8hE7FPTN1y_QnXtWUvGUzCNynBg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> @ 2013-06-12 13:24 ` Grant Likely 2013-06-13 8:37 ` Linus Walleij 0 siblings, 1 reply; 11+ messages in thread From: Grant Likely @ 2013-06-12 13:24 UTC (permalink / raw) To: Linus Walleij, Arnd Bergmann Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org, Mike Turquette, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org On Mon, 10 Jun 2013 09:43:30 +0200, Linus Walleij <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote: > On Sun, Jun 9, 2013 at 4:21 PM, Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org> wrote: > > On Sunday 09 June 2013, Linus Walleij wrote: > >> + /* > >> + * IP AMBA bus clocks, driving the bus side of the > >> + * peripheral clocking, clock gates. > >> + */ > >> + > >> + hclkdma0: hclkdma0@48M { > >> + #clock-cells = <0>; > >> + compatible = "st,nomadik-src-clock"; > >> + clock-id = <0>; > >> + clocks = <&hclk>; > >> + }; > >> + hclksmc: hclksmc@48M { > >> + #clock-cells = <0>; > >> + compatible = "st,nomadik-src-clock"; > >> + clock-id = <1>; > >> + clocks = <&hclk>; > >> + }; > >> + hclksdram: hclksdram@48M { > >> + #clock-cells = <0>; > >> + compatible = "st,nomadik-src-clock"; > >> + clock-id = <2>; > >> + clocks = <&hclk>; > >> + }; > >> + hclkdma1: hclkdma1@48M { > >> + #clock-cells = <0>; > >> + compatible = "st,nomadik-src-clock"; > >> + clock-id = <3>; > >> + clocks = <&hclk>; > >> + }; > > > > Sorry if I'm being slow to understand how the clock bindings work, but if > > you have 63 identical clocks that only differ in ther clock-id, can't you > > just have a single DT node for them instead with #clock-cells=1 to pass the > > number from the device using it? > > I first had exactly that implementation but it didn't work. This is explained > in the commit message on patch 1/1: > > The gated clocks are implemented with zero cells and > given the clock ID as a property of each node, so every > gate need to have its own node in the device tree. > This is because the gate registers contain both HCLK > gates and PCLK gates, where the latter has HCLK as > parent. As can be seen from the register layout, this > is a complete mixup, which means all these gates need > their own node to properly model parent/child relations > for PCLKs apart from the HCLKs. > > Basically the "shortcoming" in device tree is that modelling this with > clock-cells = 1 requires them all to have one and the same parent, > but they don't. Are you prevented from increasing the number of clock-cells used here? It is a lot of identical data being added. g. ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree 2013-06-12 13:24 ` Grant Likely @ 2013-06-13 8:37 ` Linus Walleij 0 siblings, 0 replies; 11+ messages in thread From: Linus Walleij @ 2013-06-13 8:37 UTC (permalink / raw) To: Grant Likely Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org, Mike Turquette, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org On Wed, Jun 12, 2013 at 3:24 PM, Grant Likely <grant.likely-s3s/WqlpOiPyB63q8FvJNQ@public.gmane.org> wrote: > On Mon, 10 Jun 2013 09:43:30 +0200, Linus Walleij <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote: >> On Sun, Jun 9, 2013 at 4:21 PM, Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org> wrote: >> > On Sunday 09 June 2013, Linus Walleij wrote: >> >> + /* >> >> + * IP AMBA bus clocks, driving the bus side of the >> >> + * peripheral clocking, clock gates. >> >> + */ >> >> + >> >> + hclkdma0: hclkdma0@48M { >> >> + #clock-cells = <0>; >> >> + compatible = "st,nomadik-src-clock"; >> >> + clock-id = <0>; >> >> + clocks = <&hclk>; >> >> + }; >> >> + hclksmc: hclksmc@48M { >> >> + #clock-cells = <0>; >> >> + compatible = "st,nomadik-src-clock"; >> >> + clock-id = <1>; >> >> + clocks = <&hclk>; >> >> + }; >> >> + hclksdram: hclksdram@48M { >> >> + #clock-cells = <0>; >> >> + compatible = "st,nomadik-src-clock"; >> >> + clock-id = <2>; >> >> + clocks = <&hclk>; >> >> + }; >> >> + hclkdma1: hclkdma1@48M { >> >> + #clock-cells = <0>; >> >> + compatible = "st,nomadik-src-clock"; >> >> + clock-id = <3>; >> >> + clocks = <&hclk>; >> >> + }; >> > >> > Sorry if I'm being slow to understand how the clock bindings work, but if >> > you have 63 identical clocks that only differ in ther clock-id, can't you >> > just have a single DT node for them instead with #clock-cells=1 to pass the >> > number from the device using it? >> >> I first had exactly that implementation but it didn't work. This is explained >> in the commit message on patch 1/1: >> >> The gated clocks are implemented with zero cells and >> given the clock ID as a property of each node, so every >> gate need to have its own node in the device tree. >> This is because the gate registers contain both HCLK >> gates and PCLK gates, where the latter has HCLK as >> parent. As can be seen from the register layout, this >> is a complete mixup, which means all these gates need >> their own node to properly model parent/child relations >> for PCLKs apart from the HCLKs. >> >> Basically the "shortcoming" in device tree is that modelling this with >> clock-cells = 1 requires them all to have one and the same parent, >> but they don't. > > Are you prevented from increasing the number of clock-cells used here? No, I can do this, but it involves loosing HW information from the device tree and the debugfs representation of the clocks will not correspond to the actual clock hierarchy. In short it creates a mess for the understanding of the system for some poor engineer with a datasheet and a device tree. And AFAIK one of the goals of the device tree is to make things easier to understand, not harder. Doing this with additional cells would sacrifice readability and HW representation for fewer lines of DTS, basically. What I wonder is which one is more important for the DT community. We cannot let the kernel community's will to keep down the number of lines in the files (as I'm starting to suspect is behind some of the reasoning here, correct me if wrong) influence everyone else, if we want to keep things reusable for other OS:es. That said I've recently had the feeling that the majority of the civilized world has started to think of DT as a Linux thing only ... so maybe I'm fighting windmills? Yours, Linus Walleij ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly [not found] ` <1370775348-7440-1-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> 2013-06-09 10:55 ` [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree Linus Walleij @ 2013-06-16 19:08 ` Linus Walleij 2013-06-20 6:38 ` Mike Turquette 2 siblings, 0 replies; 11+ messages in thread From: Linus Walleij @ 2013-06-16 19:08 UTC (permalink / raw) To: Mike Turquette Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org On Sun, Jun 9, 2013 at 12:55 PM, Linus Walleij <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote: > The Nomadik clock implementation was a stub just using > fixed clocks. > > This implements the clocks properly instead of relying > on them all being on at boot and leaving them all on. Mike, ping on this. There was some debate on whether to use 0 or 1 cells for the clocks, but I made some argument that this is how they need to be represented. As subsystem maintainer I suspect you will need to be the Judge (Dredd) on this. I think I overheard some kind of positive tone to this style of modelling them in some other review but it might've been imagination ... Yours, Linus Walleij ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly [not found] ` <1370775348-7440-1-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> 2013-06-09 10:55 ` [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree Linus Walleij 2013-06-16 19:08 ` [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly Linus Walleij @ 2013-06-20 6:38 ` Mike Turquette 2013-06-20 8:21 ` Linus Walleij 2 siblings, 1 reply; 11+ messages in thread From: Mike Turquette @ 2013-06-20 6:38 UTC (permalink / raw) To: Linus Walleij Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r Quoting Linus Walleij (2013-06-09 03:55:47) > +static void pll_clk_disable(struct clk_hw *hw) > +{ > + struct clk_pll *pll = to_pll(hw); > + u32 val; > + > + return; Is the early return intentional? Everything else looks good. Acked-by: Mike Turquette <mturquette-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> > + > + spin_lock(&src_lock); > + val = readl(src_base + SRC_PLLCR); > + if (pll->id == 1) { > + if (val & SRC_PLLCR_PLL1OVER) { > + val &= ~SRC_PLLCR_PLL1EN; > + writel(val, src_base + SRC_PLLCR); > + } > + } else if (pll->id == 2) { > + val &= ~SRC_PLLCR_PLL2EN; > + writel(val, src_base + SRC_PLLCR); > + } > + spin_unlock(&src_lock); > +} ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly 2013-06-20 6:38 ` Mike Turquette @ 2013-06-20 8:21 ` Linus Walleij 0 siblings, 0 replies; 11+ messages in thread From: Linus Walleij @ 2013-06-20 8:21 UTC (permalink / raw) To: Mike Turquette Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org On Thu, Jun 20, 2013 at 8:38 AM, Mike Turquette <mturquette-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote: > Quoting Linus Walleij (2013-06-09 03:55:47) >> +static void pll_clk_disable(struct clk_hw *hw) >> +{ >> + struct clk_pll *pll = to_pll(hw); >> + u32 val; >> + >> + return; > > Is the early return intentional? Everything else looks good. Of course not :-) At one time it had no proper children registered so would declock the entire system if I didn't have that in... fixed up. > Acked-by: Mike Turquette <mturquette-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> Thanks, sending this to ARM SoC now. Yours, Linus Walleij ^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2013-06-20 8:21 UTC | newest] Thread overview: 11+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2013-06-09 10:55 [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly Linus Walleij [not found] ` <1370775348-7440-1-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> 2013-06-09 10:55 ` [PATCH 2/2] ARM: nomadik: add the new clocks to the device tree Linus Walleij [not found] ` <1370775348-7440-2-git-send-email-linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> 2013-06-09 14:21 ` Arnd Bergmann [not found] ` <CAJAp7OiYZx46aWRsXc+CmS+5F0STz5P+rdYnJEoaRRMUnkD1Cg@mail.gmail.com> 2013-06-10 5:12 ` Bjorn Andersson [not found] ` <CAJAp7Oji=+9PqwGkF84BRaxkFeeO_kWoY378Ur-A5z6G9Z5xSQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> 2013-06-10 12:29 ` Linus Walleij [not found] ` <201306091621.02506.arnd-r2nGTMty4D4@public.gmane.org> 2013-06-10 7:43 ` Linus Walleij [not found] ` <CACRpkdZNZ4_FzjEsstUm0bn8hE7FPTN1y_QnXtWUvGUzCNynBg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> 2013-06-12 13:24 ` Grant Likely 2013-06-13 8:37 ` Linus Walleij 2013-06-16 19:08 ` [PATCH 1/2] clk: nomadik: implement the Nomadik clocks properly Linus Walleij 2013-06-20 6:38 ` Mike Turquette 2013-06-20 8:21 ` Linus Walleij
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).