Il 31/08/22 14:48, Johnson Wang ha scritto:
To implement frequency hopping and spread spectrum clocking
function, we introduce new clock type and APIs to handle
FHCTL hardware.
Co-developed-by: Edward-JW Yang <edward-jw.yang@xxxxxxxxxxxx>
Signed-off-by: Edward-JW Yang <edward-jw.yang@xxxxxxxxxxxx>
Signed-off-by: Johnson Wang <johnson.wang@xxxxxxxxxxxx>
---
drivers/clk/mediatek/Makefile | 2 +-
drivers/clk/mediatek/clk-fhctl.c | 258 +++++++++++++++++++++++++++++
drivers/clk/mediatek/clk-fhctl.h | 27 +++
drivers/clk/mediatek/clk-pllfh.c | 271 +++++++++++++++++++++++++++++++
drivers/clk/mediatek/clk-pllfh.h | 81 +++++++++
5 files changed, 638 insertions(+), 1 deletion(-)
create mode 100644 drivers/clk/mediatek/clk-fhctl.c
create mode 100644 drivers/clk/mediatek/clk-fhctl.h
create mode 100644 drivers/clk/mediatek/clk-pllfh.c
create mode 100644 drivers/clk/mediatek/clk-pllfh.h
diff --git a/drivers/clk/mediatek/Makefile b/drivers/clk/mediatek/Makefile
index caf2ce93d666..0e674a55e51e 100644
--- a/drivers/clk/mediatek/Makefile
+++ b/drivers/clk/mediatek/Makefile
@@ -1,5 +1,5 @@
# SPDX-License-Identifier: GPL-2.0
-obj-$(CONFIG_COMMON_CLK_MEDIATEK) += clk-mtk.o clk-pll.o clk-gate.o clk-apmixed.o clk-cpumux.o reset.o clk-mux.o
+obj-$(CONFIG_COMMON_CLK_MEDIATEK) += clk-mtk.o clk-pll.o clk-gate.o clk-apmixed.o clk-cpumux.o reset.o clk-mux.o clk-fhctl.o clk-pllfh.o
obj-$(CONFIG_COMMON_CLK_MT6765) += clk-mt6765.o
obj-$(CONFIG_COMMON_CLK_MT6765_AUDIOSYS) += clk-mt6765-audio.o
diff --git a/drivers/clk/mediatek/clk-fhctl.c b/drivers/clk/mediatek/clk-fhctl.c
new file mode 100644
index 000000000000..69bd4ce2bc72
--- /dev/null
+++ b/drivers/clk/mediatek/clk-fhctl.c
@@ -0,0 +1,258 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2022 MediaTek Inc.
+ * Author: Edward-JW Yang <edward-jw.yang@xxxxxxxxxxxx>
+ */
+
#include <linux/io.h>
Please don't rely on implicit header inclusion.
+#include <linux/iopoll.h>
+
+#include "clk-mtk.h"
+#include "clk-pllfh.h"
+#include "clk-fhctl.h"
+
+#define PERCENT_TO_DDSLMT(dds, percent_m10) \
+ ((((dds) * (percent_m10)) >> 5) / 100)
+
+static const struct fhctl_offset fhctl_offset = {
+ .offset_hp_en = 0x0,
+ .offset_clk_con = 0x8,
+ .offset_rst_con = 0xc,
+ .offset_slope0 = 0x10,
+ .offset_slope1 = 0x14,
+ .offset_cfg = 0x0,
+ .offset_updnlmt = 0x4,
+ .offset_dds = 0x8,
+ .offset_dvfs = 0xc,
+ .offset_mon = 0x10,
+};
+
+const struct fhctl_offset *fhctl_get_offset_table(void)
+{
+ return &fhctl_offset;
+}
+
+static void dump_hw(struct mtk_clk_pll *pll, struct fh_pll_regs *regs,
+ const struct fh_pll_data *data)
+{
+ pr_info("hp_en<%x>,clk_con<%x>,slope0<%x>,slope1<%x>\n",
+ readl(regs->reg_hp_en), readl(regs->reg_clk_con),
+ readl(regs->reg_slope0), readl(regs->reg_slope1));
+ pr_info("cfg<%x>,lmt<%x>,dds<%x>,dvfs<%x>,mon<%x>\n",
+ readl(regs->reg_cfg), readl(regs->reg_updnlmt),
+ readl(regs->reg_dds), readl(regs->reg_dvfs),
+ readl(regs->reg_mon));
+ pr_info("pcw<%x>\n", readl(pll->pcw_addr));
+}
+
+static int fhctl_set_ssc_regs(struct mtk_clk_pll *pll, struct fh_pll_regs *regs,
+ const struct fh_pll_data *data, u32 rate)
+{
+ u32 updnlmt_val, r;
+
+ writel((readl(regs->reg_cfg) & ~(data->frddsx_en)), regs->reg_cfg);
+ writel((readl(regs->reg_cfg) & ~(data->sfstrx_en)), regs->reg_cfg);
+ writel((readl(regs->reg_cfg) & ~(data->fhctlx_en)), regs->reg_cfg);
+
+ if (rate > 0) {
+ /* Set the relative parameter registers (dt/df/upbnd/downbnd) */
+ r = readl(regs->reg_cfg);
+ r &= ~(data->msk_frddsx_dys);
+ r |= (data->df_val << (ffs(data->msk_frddsx_dys) - 1));
+ writel(r, regs->reg_cfg);
+
+ r = readl(regs->reg_cfg);
+ r &= ~(data->msk_frddsx_dts);
+ r |= (data->dt_val << (ffs(data->msk_frddsx_dts) - 1));
+ writel(r, regs->reg_cfg);
+
+ writel((readl(pll->pcw_addr) & data->dds_mask) | data->tgl_org,
+ regs->reg_dds);
+
+ /* Calculate UPDNLMT */
+ updnlmt_val = PERCENT_TO_DDSLMT((readl(regs->reg_dds) &
+ data->dds_mask), rate) <<
+ data->updnlmt_shft;
+
+ writel(updnlmt_val, regs->reg_updnlmt);
+ writel(readl(regs->reg_hp_en) | BIT(data->fh_id),
+ regs->reg_hp_en);
+ /* Enable SSC */
+ writel(readl(regs->reg_cfg) | data->frddsx_en, regs->reg_cfg);
+ /* Enable Hopping control */
+ writel(readl(regs->reg_cfg) | data->fhctlx_en, regs->reg_cfg);
+
+ } else {
+ /* Switch to APMIXEDSYS control */
+ writel(readl(regs->reg_hp_en) & ~BIT(data->fh_id),
+ regs->reg_hp_en);
+ /* Wait for DDS to be stable */
+ udelay(30);
+ }
+
+ return 0;
+}
+
+static int hopping_hw_flow(struct mtk_clk_pll *pll, struct fh_pll_regs *regs,
+ const struct fh_pll_data *data,
+ struct fh_pll_state *state, unsigned int new_dds)
+{
+ u32 dds_mask = data->dds_mask;
+ u32 mon_dds = 0;
+ u32 con_pcw_tmp;
+ int ret;
+
+ if (state->ssc_rate)
+ fhctl_set_ssc_regs(pll, regs, data, 0);
+
+ writel((readl(pll->pcw_addr) & dds_mask) | data->tgl_org,
+ regs->reg_dds);
+
+ writel(readl(regs->reg_cfg) | data->sfstrx_en, regs->reg_cfg);
+ writel(readl(regs->reg_cfg) | data->fhctlx_en, regs->reg_cfg);
+ writel(data->slope0_value, regs->reg_slope0);
+ writel(data->slope1_value, regs->reg_slope1);
+
+ writel(readl(regs->reg_hp_en) | BIT(data->fh_id), regs->reg_hp_en);
+ writel((new_dds) | (data->dvfs_tri), regs->reg_dvfs);
+
+ /* Wait 1000 us until DDS stable */
+ ret = readl_poll_timeout_atomic(regs->reg_mon, mon_dds,
+ (mon_dds & dds_mask) == new_dds,
+ 10, 1000);
+ if (ret) {
+ pr_warn("%s: FHCTL hopping timeout\n", pll->data->name);
+ dump_hw(pll, regs, data);
+ }
+
+ con_pcw_tmp = readl(pll->pcw_addr) & (~dds_mask);
+ con_pcw_tmp = (con_pcw_tmp | (readl(regs->reg_mon) & dds_mask) |
+ data->pcwchg);
+
+ writel(con_pcw_tmp, pll->pcw_addr);
+ writel(readl(regs->reg_hp_en) & ~BIT(data->fh_id), regs->reg_hp_en);
+
+ if (state->ssc_rate)
+ fhctl_set_ssc_regs(pll, regs, data, state->ssc_rate);
Does it really make sense to set the SSC registers if the FH operation times out?
+
+ return ret;
+}
+
+static unsigned int __get_postdiv(struct mtk_clk_pll *pll,
+ struct fh_pll_regs *regs,
+ const struct fh_pll_data *fh_data)
You don't need regs, nor fh_data, so this can be as simple as
static unsigned int __get_postdiv(struct mtk_clk_pll *pll)
+{
+ unsigned int regval;
+
+ regval = readl(pll->pd_addr) >> pll->data->pd_shift;
+ regval &= POSTDIV_MASK;
+
return BIT(regval);
+ return (1 << regval);
+}
+
+static void __set_postdiv(struct mtk_clk_pll *pll, struct fh_pll_regs *regs,
+ const struct fh_pll_data *data, int postdiv)
static void __set_postdiv(struct mtk_clk_pll *pll, unsigned int val)
+{
+ unsigned int regval;
+
+ regval = readl(pll->pd_addr);
+ regval &= ~(POSTDIV_MASK << pll->data->pd_shift);
+ regval |= (ffs(postdiv) - 1) << pll->data->pd_shift;
+ writel(regval, pll->pd_addr);
+}
+
+static int fhctl_hopping(struct mtk_fh *fh, unsigned int new_dds, int postdiv)
The postdiv cannot ever be negative, so you can change it to an unsigned type...
static int fhctl_hopping(struct mtk_fh *fh, unsigned int new_dds,
unsigned int postdiv)
+{
+ const struct fh_pll_data *data = &fh->pllfh_data->data;
+ struct fh_pll_state *state = &fh->pllfh_data->state;
+ struct fh_pll_regs *regs = &fh->regs;
+ struct mtk_clk_pll *pll = &fh->clk_pll;
+ spinlock_t *lock = fh->lock;
+ unsigned int pll_postdiv;
+ unsigned long flags = 0;
+ int ret;
+
...so since postdiv is unsigned, we can write this conditional as
if (postdiv)
+ if (postdiv > 0) {
+ pll_postdiv = __get_postdiv(pll, regs, data);
+
+ if (postdiv > pll_postdiv)
+ __set_postdiv(pll, regs, data, postdiv);
+ }
+
+ spin_lock_irqsave(lock, flags);
+
+ ret = hopping_hw_flow(pll, regs, data, state, new_dds);
+
+ spin_unlock_irqrestore(lock, flags);
+
+ if (postdiv > 0) {
+ if (postdiv < pll_postdiv)
...and this one as
if (postdiv && postdiv < pll_postdiv)
+ __set_postdiv(pll, regs, data, postdiv);
+ }
+
+ return ret;
+}
+
+static int __fhctl_ssc_enable(struct mtk_fh *fh, u32 rate)
+{
+ const struct fh_pll_data *data = &fh->pllfh_data->data;
+ struct fh_pll_state *state = &fh->pllfh_data->state;
+ struct fh_pll_regs *regs = &fh->regs;
+ struct mtk_clk_pll *pll = &fh->clk_pll;
+ spinlock_t *lock = fh->lock;
+ unsigned long flags = 0;
+
+ spin_lock_irqsave(lock, flags);
+
+ fhctl_set_ssc_regs(pll, regs, data, rate);
+ state->ssc_rate = rate;
+
+ spin_unlock_irqrestore(lock, flags);
+
+ return 0;
+}
+
+static int fhctl_ssc_enable(struct mtk_fh *fh, u32 rate)
+{
+ return __fhctl_ssc_enable(fh, rate);
+}
+
+static int fhctl_ssc_disable(struct mtk_fh *fh)
+{
+ return __fhctl_ssc_enable(fh, 0);
+}
+
+static const struct fh_operation fhctl_ops = {
+ .hopping = fhctl_hopping,
+ .ssc_enable = fhctl_ssc_enable,
Just one callback for ssc_enable is enough, it's fine to keep the logic as
ssc_enable(fh, >0) to enable and ssc_enable(fh <=0) to disable.
+ .ssc_disable = fhctl_ssc_disable,
+};
+
+const struct fh_operation *fhctl_get_ops(void)
+{
+ return &fhctl_ops;
+}
+
+void fhctl_hw_init(struct mtk_fh *fh)
+{
+ const struct fh_pll_data data = fh->pllfh_data->data;
+ struct fh_pll_state state = fh->pllfh_data->state;
+ struct fh_pll_regs regs = fh->regs;
+ u32 val;
+
+ /* initial hw register */
+ val = readl(regs.reg_clk_con) | BIT(data.fh_id);
+ writel(val, regs.reg_clk_con);
+
+ val = readl(regs.reg_rst_con) & ~BIT(data.fh_id);
+ writel(val, regs.reg_rst_con);
+ val = readl(regs.reg_rst_con) | BIT(data.fh_id);
+ writel(val, regs.reg_rst_con);
+
+ writel(0x0, regs.reg_cfg);
+ writel(0x0, regs.reg_updnlmt);
+ writel(0x0, regs.reg_dds);
+
+ /* enable ssc if needed */
+ if (state.ssc_rate)
+ fh->ops->ssc_enable(fh, state.ssc_rate);
+}
diff --git a/drivers/clk/mediatek/clk-fhctl.h b/drivers/clk/mediatek/clk-fhctl.h
new file mode 100644
index 000000000000..3cd4921c39e9
--- /dev/null
+++ b/drivers/clk/mediatek/clk-fhctl.h
@@ -0,0 +1,27 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2022 MediaTek Inc.
+ * Author: Edward-JW Yang <edward-jw.yang@xxxxxxxxxxxx>
+ */
+
+#ifndef __CLK_FHCTL_H
+#define __CLK_FHCTL_H
+
One blank line is enough.
+
+struct fhctl_offset {
+ u32 offset_hp_en;
+ u32 offset_clk_con;
+ u32 offset_rst_con;
+ u32 offset_slope0;
+ u32 offset_slope1;
+ u32 offset_cfg;
+ u32 offset_updnlmt;
+ u32 offset_dds;
+ u32 offset_dvfs;
+ u32 offset_mon;
+};
+const struct fhctl_offset *fhctl_get_offset_table(void);
+const struct fh_operation *fhctl_get_ops(void);
+void fhctl_hw_init(struct mtk_fh *fh);
+
+#endif
diff --git a/drivers/clk/mediatek/clk-pllfh.c b/drivers/clk/mediatek/clk-pllfh.c
new file mode 100644
index 000000000000..71b35323b526
--- /dev/null
+++ b/drivers/clk/mediatek/clk-pllfh.c
@@ -0,0 +1,271 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2022 MediaTek Inc.
+ * Author: Edward-JW Yang <edward-jw.yang@xxxxxxxxxxxx>
+ */
+
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/clkdev.h>
+#include <linux/delay.h>
+
+#include "clk-mtk.h"
+#include "clk-pllfh.h"
+#include "clk-fhctl.h"
+
+static DEFINE_SPINLOCK(pllfh_lock);
+
+inline struct mtk_fh *to_mtk_fh(struct clk_hw *hw)
+{
+ struct mtk_clk_pll *pll = to_mtk_clk_pll(hw);
+
+ return container_of(pll, struct mtk_fh, clk_pll);
+}
+
+static int mtk_fhctl_set_rate(struct clk_hw *hw, unsigned long rate,
+ unsigned long parent_rate)
+{
+ struct mtk_clk_pll *pll = to_mtk_clk_pll(hw);
+ struct mtk_fh *fh = to_mtk_fh(hw);
+ u32 pcw = 0;
+ u32 postdiv;
+
+ mtk_pll_calc_values(pll, &pcw, &postdiv, rate, parent_rate);
+ fh->ops->hopping(fh, pcw, postdiv);
return fh->ops->hopping(....);
+
+ return 0;
+}
+
..snip..
diff --git a/drivers/clk/mediatek/clk-pllfh.h b/drivers/clk/mediatek/clk-pllfh.h
new file mode 100644
index 000000000000..d9f7c8527548
--- /dev/null
+++ b/drivers/clk/mediatek/clk-pllfh.h
@@ -0,0 +1,81 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2022 MediaTek Inc.
+ * Author: Edward-JW Yang <edward-jw.yang@xxxxxxxxxxxx>
+ */
+
+#ifndef __DRV_CLKFH_H
+#define __DRV_CLKFH_H
+
+#include "clk-pll.h"
+
+struct fh_pll_state {
Please, base as first member.
+ u32 fh_enable;
+ u32 ssc_rate;
+ void __iomem *base;
+};
+
Regards,
Angelo
