Re: [PATCH V3 3/8] clk: qcom: Add A53 PLL support for ipq6018 devices

From: Sivaprakash Murugesan
Date: Wed Apr 22 2020 - 06:45:00 EST


Hi Stephen,

On 4/22/2020 2:30 PM, Stephen Boyd wrote:
Quoting Sivaprakash Murugesan (2020-04-13 19:55:17)
The CPUs on Qualcomm IPQ6018 platform is primarily clocked by A53 PLL.
This patch adds support for the A53 PLL on IPQ6018 devices which can
support CPU frequencies above 1Ghz.

Signed-off-by: Sivaprakash Murugesan <sivaprak@xxxxxxxxxxxxxx>
---
drivers/clk/qcom/a53-pll.c | 136 ++++++++++++++++++++++++++++++++++++---------
1 file changed, 111 insertions(+), 25 deletions(-)

diff --git a/drivers/clk/qcom/a53-pll.c b/drivers/clk/qcom/a53-pll.c
index 45cfc57..a95351c 100644
--- a/drivers/clk/qcom/a53-pll.c
+++ b/drivers/clk/qcom/a53-pll.c
@@ -11,11 +11,40 @@
#include <linux/platform_device.h>
#include <linux/regmap.h>
#include <linux/module.h>
+#include <linux/of_device.h>
Why does this driver need to change to use of_device APIs?
we can use devm APIs and avoid this in next patch.

#include "clk-pll.h"
#include "clk-regmap.h"
+#include "clk-alpha-pll.h"
-static const struct pll_freq_tbl a53pll_freq[] = {
+struct a53_alpha_pll {
+ struct alpha_pll_config *pll_config;
+ struct clk_alpha_pll *pll;
+};
+
+union a53pll {
+ struct clk_pll *pll;
+ struct a53_alpha_pll alpha_pll;
+};
+
+struct a53pll_data {
+#define PLL_IS_ALPHA BIT(0)
+ u8 flags;
+ union a53pll a53pll;
Why is there a union? Can't we have different clk ops for the two types
of PLLs and then use container_of to get it from the clk ops?
ok.

+};
+
+static const u8 ipq_pll_offsets[] = {
+ [PLL_OFF_L_VAL] = 0x08,
+ [PLL_OFF_ALPHA_VAL] = 0x10,
+ [PLL_OFF_USER_CTL] = 0x18,
+ [PLL_OFF_CONFIG_CTL] = 0x20,
+ [PLL_OFF_CONFIG_CTL_U] = 0x24,
+ [PLL_OFF_STATUS] = 0x28,
+ [PLL_OFF_TEST_CTL] = 0x30,
+ [PLL_OFF_TEST_CTL_U] = 0x34,
+};
+
+static const struct pll_freq_tbl msm8996_a53pll_freq[] = {
{ 998400000, 52, 0x0, 0x1, 0 },
{ 1094400000, 57, 0x0, 0x1, 0 },
{ 1152000000, 62, 0x0, 0x1, 0 },
@@ -26,6 +55,64 @@ static const struct pll_freq_tbl a53pll_freq[] = {
{ }
};
+static struct clk_pll msm8996_pll = {
+ .mode_reg = 0x0,
+ .l_reg = 0x04,
+ .m_reg = 0x08,
+ .n_reg = 0x0c,
+ .config_reg = 0x14,
+ .status_reg = 0x1c,
+ .status_bit = 16,
+ .freq_tbl = msm8996_a53pll_freq,
+ .clkr.hw.init = &(struct clk_init_data){
+ .name = "a53pll",
+ .flags = CLK_IS_CRITICAL,
+ .parent_data = &(const struct clk_parent_data){
+ .fw_name = "xo",
+ .name = "xo",
+ },
+ .num_parents = 1,
+ .ops = &clk_pll_sr2_ops,
+ },
+};
+
+static struct clk_alpha_pll ipq6018_pll = {
+ .offset = 0x0,
+ .regs = ipq_pll_offsets,
+ .flags = SUPPORTS_DYNAMIC_UPDATE,
+ .clkr = {
+ .enable_reg = 0x0,
+ .enable_mask = BIT(0),
+ .hw.init = &(struct clk_init_data){
+ .name = "a53pll",
+ .flags = CLK_IS_CRITICAL,
+ .parent_data = &(const struct clk_parent_data){
+ .fw_name = "xo",
+ },
+ .num_parents = 1,
+ .ops = &clk_alpha_pll_huayra_ops,
+ },
+ },
+};
+
+static struct alpha_pll_config ipq6018_pll_config = {
Can this be const?
yeah it can be. will fix up in next patch.

+ .l = 0x37,
+ .config_ctl_val = 0x04141200,
+ .config_ctl_hi_val = 0x0,
+ .early_output_mask = BIT(3),
+ .main_output_mask = BIT(0),
+};
+
+static struct a53pll_data msm8996pll_data = {
+ .a53pll.pll = &msm8996_pll,
+};
+
+static struct a53pll_data ipq6018pll_data = {
+ .flags = PLL_IS_ALPHA,
+ .a53pll.alpha_pll.pll = &ipq6018_pll,
+ .a53pll.alpha_pll.pll_config = &ipq6018_pll_config,
+};
+
static const struct regmap_config a53pll_regmap_config = {
.reg_bits = 32,
.reg_stride = 4,
@@ -39,14 +126,16 @@ static int qcom_a53pll_probe(struct platform_device *pdev)
struct device *dev = &pdev->dev;
struct regmap *regmap;
struct resource *res;
- struct clk_pll *pll;
+ const struct a53pll_data *pll_data;
+ struct clk_regmap *clkr;
void __iomem *base;
- struct clk_init_data init = { };
int ret;
- pll = devm_kzalloc(dev, sizeof(*pll), GFP_KERNEL);
- if (!pll)
- return -ENOMEM;
+ pll_data = of_device_get_match_data(dev);
Use device_get_match_data() please.
ok.

+ if (!pll_data) {
+ dev_err(dev, "failed to get platform data\n");
No error message please.
ok.

+ return -ENODEV;
+ }
res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
base = devm_ioremap_resource(dev, res);
@@ -57,30 +146,26 @@ static int qcom_a53pll_probe(struct platform_device *pdev)
if (IS_ERR(regmap))
return PTR_ERR(regmap);
- pll->l_reg = 0x04;
- pll->m_reg = 0x08;
- pll->n_reg = 0x0c;
- pll->config_reg = 0x14;
- pll->mode_reg = 0x00;
- pll->status_reg = 0x1c;
- pll->status_bit = 16;
- pll->freq_tbl = a53pll_freq;
-
- init.name = "a53pll";
- init.parent_names = (const char *[]){ "xo" };
- init.num_parents = 1;
- init.ops = &clk_pll_sr2_ops;
- init.flags = CLK_IS_CRITICAL;
Please document why a clk is critical.
ok

- pll->clkr.hw.init = &init;
-
- ret = devm_clk_register_regmap(dev, &pll->clkr);
+ if (pll_data->flags & PLL_IS_ALPHA) {
+ struct clk_alpha_pll *alpha_pll =
+ pll_data->a53pll.alpha_pll.pll;
+ struct alpha_pll_config *alpha_pll_config =
+ pll_data->a53pll.alpha_pll.pll_config;
+
+ clk_alpha_pll_configure(alpha_pll, regmap, alpha_pll_config);
+ clkr = &pll_data->a53pll.alpha_pll.pll->clkr;
+ } else {
+ clkr = &pll_data->a53pll.pll->clkr;
+ }
Sorry, the design is confusing.

The basic idea is to add support for various PLLs available to clock the A53 core.

if this messing up the code, can the alpha pll support be moved to a separate file?

It would be very helpful if you provide your input on this.