Re: [v5 2/2] clk: qcom: clk-rpmh: Add QCOM RPMh clock driver
From: Taniya Das
Date: Wed May 02 2018 - 06:22:31 EST
Thanks Stephen for the comments.
On 5/2/2018 2:57 AM, Stephen Boyd wrote:
Quoting Taniya Das (2018-05-01 01:41:33)
Add the RPMh clock driver to control the RPMh managed clock resources on
some of the Qualcomm Technologies, Inc. SoCs.
Signed-off-by: Amit Nischal <anischal@xxxxxxxxxxxxxx>
Signed-off-by: Taniya Das <tdas@xxxxxxxxxxxxxx>
Drop Amit's signoff, or make it "Co-authored-by" please. First signoff
should be the author and there should be a "From:" line either from the
email header of the sender or explicitly here before the commit text
body to indicate the authorship.
Overall this is looking close. Maybe one more round.
diff --git a/drivers/clk/qcom/clk-rpmh.c b/drivers/clk/qcom/clk-rpmh.c
new file mode 100644
index 0000000..9cb7aa4
--- /dev/null
+++ b/drivers/clk/qcom/clk-rpmh.c
@@ -0,0 +1,387 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2018, The Linux Foundation. All rights reserved.
+ */
+
+#include <linux/clk-provider.h>
+#include <linux/err.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+#include <soc/qcom/cmd-db.h>
+#include <soc/qcom/rpmh.h>
+
+#include <dt-bindings/clock/qcom,rpmh.h>
+
+#define CLK_RPMH_ARC_EN_OFFSET 0
+#define CLK_RPMH_VRM_EN_OFFSET 4
+
+/**
+ * struct clk_rpmh - individual rpmh clock data structure
+ * @hw: handle between common and hardware-specific interfaces
+ * @res_name: resource name for the rpmh clock
+ * @div: clock divider to compute the clock rate
+ * @res_addr: base address of the rpmh resource within the RPMh
+ * @res_on_val: rpmh clock enable value
+ * @res_off_val: rpmh clock disable value
It's still unused.
It is unused as the currently the resource off value is '0'. I would
remove, will add it back if any different value is required.
+ * @state: rpmh clock requested state
+ * @aggr_state: rpmh clock aggregated state
+ * @last_sent_aggr_state: rpmh clock last aggr state sent to RPMh
+ * @valid_state_mask: mask to determine the state of the rpmh clock
+ * @rpmh_client: handle used for rpmh communications
+ * @dev: device to which it is attached
+ * @peer: pointer to the clock rpmh sibling
+ */
+struct clk_rpmh {
+ struct clk_hw hw;
+ const char *res_name;
+ u8 div;
Weird space here.
This is no longer required, removed.
+ u32 res_addr;
+ u32 res_on_val;
+ u32 res_off_val;
+ u32 state;
+ u32 aggr_state;
+ u32 last_sent_aggr_state;
+ u32 valid_state_mask;
+ struct rpmh_client *rpmh_client;
+ struct device *dev;
+ struct clk_rpmh *peer;
+};
+
+struct clk_rpmh_desc {
+ struct clk_hw **clks;
+ size_t num_clks;
+};
+
+static DEFINE_MUTEX(rpmh_clk_lock);
+
+#define __DEFINE_CLK_RPMH(_platform, _name, _name_active, _res_name, \
+ _res_en_offset, _res_on, _res_off) \
+ static struct clk_rpmh _platform##_##_name_active; \
+ static struct clk_rpmh _platform##_##_name = { \
+ .res_name = _res_name, \
+ .res_addr = _res_en_offset, \
+ .res_on_val = _res_on, \
+ .res_off_val = _res_off, \
+ .peer = &_platform##_##_name_active, \
+ .valid_state_mask = (BIT(RPMH_WAKE_ONLY_STATE) | \
+ BIT(RPMH_ACTIVE_ONLY_STATE) | \
+ BIT(RPMH_SLEEP_STATE)), \
+ .hw.init = &(struct clk_init_data){ \
+ .ops = &clk_rpmh_ops, \
+ .name = #_name, \
+ .parent_names = (const char *[]){ "xo_board" }, \
This would be different for the xo_board and the xo_board / 2 clk names.
Would add parents "xo_board" & "xo_board_div".
+ .num_parents = 1, \
+ }, \
+ }; \
+ static struct clk_rpmh _platform##_##_name_active = { \
+ .res_name = _res_name, \
+ .res_addr = _res_en_offset, \
+ .res_on_val = _res_on, \
+ .res_off_val = _res_off, \
+ .peer = &_platform##_##_name, \
+ .valid_state_mask = (BIT(RPMH_WAKE_ONLY_STATE) | \
+ BIT(RPMH_ACTIVE_ONLY_STATE)), \
+ .hw.init = &(struct clk_init_data){ \
+ .ops = &clk_rpmh_ops, \
+ .name = #_name_active, \
+ .parent_names = (const char *[]){ "xo_board" }, \
+ .num_parents = 1, \
+ }, \
+ }
+
+#define DEFINE_CLK_RPMH_ARC(_platform, _name, _name_active, _res_name, \
+ _res_on, _res_off) \
+ __DEFINE_CLK_RPMH(_platform, _name, _name_active, _res_name, \
+ CLK_RPMH_ARC_EN_OFFSET, _res_on, _res_off)
+
+#define DEFINE_CLK_RPMH_VRM(_platform, _name, _name_active, _res_name) \
+ __DEFINE_CLK_RPMH(_platform, _name, _name_active, _res_name, \
+ CLK_RPMH_VRM_EN_OFFSET, true, false)
s/true, false/1, 0/ ?
Would take care of this too.
+
+static inline struct clk_rpmh *to_clk_rpmh(struct clk_hw *_hw)
+{
+ return container_of(_hw, struct clk_rpmh, hw);
+}
+
+static inline bool has_state_changed(struct clk_rpmh *c, u32 state)
+{
+ return (c->last_sent_aggr_state & BIT(state))
+ != (c->aggr_state & BIT(state));
+}
+
+static int clk_rpmh_send_aggregate_command(struct clk_rpmh *c)
+{
+ struct tcs_cmd cmd = { 0 };
+ u32 cmd_state, on_val;
+ enum rpmh_state state = RPMH_SLEEP_STATE;
+ int ret;
+
+ cmd.addr = c->res_addr;
+ cmd_state = c->aggr_state;
+ on_val = c->res_on_val;
+
+ for (; state <= RPMH_ACTIVE_ONLY_STATE; state++) {
+ if (has_state_changed(c, state)) {
+ cmd.data = (cmd_state & BIT(state)) ? on_val : 0;
Make this into an "if" please. Default is already 0 so just set bit if
cmd_state & BIT(state).
Would take care of it.
+ ret = rpmh_write_async(c->rpmh_client, state,
+ &cmd, 1);
+ if (ret) {
+ dev_err(c->dev, "set %s state of %s failed: (%d)\n",
+ !state ? "sleep" :
+ state == RPMH_WAKE_ONLY_STATE ?
+ "wake" : "active", c->res_name, ret);
+ return ret;
+ }
+ }
+ }
+
+ c->last_sent_aggr_state = c->aggr_state;
+ c->peer->last_sent_aggr_state = c->last_sent_aggr_state;
+
+ return 0;
+}
+
+/*
+ * Update state and aggregate state values based on enable value.
+ */
+static int clk_rpmh_aggregate_state_send_command(struct clk_rpmh *c,
+ bool enable)
+{
+ int ret;
+
+ c->state = enable ? c->valid_state_mask : 0;
+ c->aggr_state = c->state | c->peer->state;
+ c->peer->aggr_state = c->aggr_state;
+
+ ret = clk_rpmh_send_aggregate_command(c);
+ if (ret && enable) {
+ c->state = 0;
Why don't we print a WARN if it fails to enable?
I would add a WARN for enable failure too.
+ } else if (ret) {
+ c->state = c->valid_state_mask;
+ WARN(1, "clk: %s failed to disable\n", c->res_name);
+ }
+
+ return ret;
+}
+
+static int clk_rpmh_prepare(struct clk_hw *hw)
+{
+ struct clk_rpmh *c = to_clk_rpmh(hw);
+ int ret = 0;
+
+ mutex_lock(&rpmh_clk_lock);
+
+ if (!c->state)
Can we push this into clk_rpmh_aggregate_state_send_command()? Something
like:
/* Nothing to do if already off or on */
if (enable == c->state)
return 0;
Thanks, would add these changes.
+ ret = clk_rpmh_aggregate_state_send_command(c, true);
+
+ mutex_unlock(&rpmh_clk_lock);
+ return ret;
+};
+
+static void clk_rpmh_unprepare(struct clk_hw *hw)
+{
+ struct clk_rpmh *c = to_clk_rpmh(hw);
+
+ mutex_lock(&rpmh_clk_lock);
+
+ if (c->state)
+ clk_rpmh_aggregate_state_send_command(c, false);
+
+ mutex_unlock(&rpmh_clk_lock);
+};
+
+static unsigned long clk_rpmh_recalc_rate(struct clk_hw *hw,
+ unsigned long prate)
+{
+ struct clk_rpmh *r = to_clk_rpmh(hw);
+ unsigned long rate;
+
+ /*
+ * RPMh clocks have a fixed rate.
+ */
+ rate = prate;
+ do_div(rate, r->div);
+
+ return rate;
+}
You shouldn't need any recalc rate in this driver.
Yes, these would be removed.
+
+static int clk_rpmh_probe(struct platform_device *pdev)
+{
+ struct clk *clk;
+ struct clk_hw_onecell_data *hw_data;
+ struct clk_hw **hw_clks;
+ struct clk_rpmh *rpmh_clk;
+ const struct clk_rpmh_desc *desc;
+ struct rpmh_client *rpmh_client;
+ struct device *dev;
+ size_t num_clks, i;
+ int ret;
+
+ dev = &pdev->dev;
Just do this when dev is defined please.
+
+ ret = cmd_db_ready();
+ if (ret) {
+ if (ret != -EPROBE_DEFER)
+ dev_err(dev, "Command DB not available (%d)\n", ret);
+ return ret;
+ }
+
+ rpmh_client = rpmh_get_client(pdev);
+ if (IS_ERR(rpmh_client)) {
+ ret = PTR_ERR(rpmh_client);
+ dev_err(dev, "Failed to request RPMh client, ret=%d\n", ret);
+ return ret;
+ }
+
+ desc = of_device_get_match_data(&pdev->dev);
+ hw_clks = desc->clks;
+ num_clks = desc->num_clks;
+
+ hw_data = devm_kzalloc(dev,
+ sizeof(*hw_data) + num_clks * sizeof(struct clk_hw *),
+ GFP_KERNEL);
+ if (!hw_data)
+ return -ENOMEM;
+
+ hw_data->num = num_clks;
+
+ for (i = 0; i < num_clks; i++) {
+ const char *clk_name;
+ struct property *prop;
+ const __be32 *p;
+ u32 res_addr;
+ int div, j = 0;
+
+ rpmh_clk = to_clk_rpmh(hw_clks[i]);
+ res_addr = cmd_db_read_addr(rpmh_clk->res_name);
+ if (!res_addr) {
+ dev_err(dev, "missing RPMh resource address for %s\n",
+ rpmh_clk->res_name);
+ ret = -ENODEV;
+ goto err;
+ }
+ rpmh_clk->res_addr += res_addr;
Thanks!
+
+ rpmh_clk->rpmh_client = rpmh_client;
+
+ /* default div for all clocks */
+ if (!rpmh_clk->div)
+ rpmh_clk->div = 1;
+
+ of_property_for_each_u32(pdev->dev.of_node, "assigned-clk-divs",
+ prop, p, div) {
+ of_property_read_string_index(pdev->dev.of_node,
+ "clk-output-names", j, &clk_name);
+ if (!strcmp(clk_name, hw_clks[i]->init->name)) {
+ rpmh_clk->div = div;
+ rpmh_clk->peer->div = div;
+ }
+ j++;
+ }
+
+ clk = devm_clk_register(&pdev->dev, hw_clks[i]);
+ if (IS_ERR(clk)) {
+ ret = PTR_ERR(clk);
+ dev_err(dev, "failed to register %s\n",
+ hw_clks[i]->init->name);
+ goto err;
+ }
+
+ rpmh_clk->dev = &pdev->dev;
+ }
+
+ ret = devm_of_clk_add_hw_provider(dev, of_clk_hw_onecell_get,
Why not roll your own "getter" function and then just index directly
into your array of rpmh clks for sdm845? Then we don't have to copy
things from one place to another.
Added a new function.
+ hw_data);
+ if (ret) {
+ dev_err(dev, "Failed to add clock provider\n");
+ goto err;
+ }
+
+ dev_dbg(dev, "Registered RPMh clocks\n");
+
+ return 0;
+err:
+ if (rpmh_client)
+ rpmh_release(rpmh_client);
+
+ return ret;
+}
+
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation.
--