All of lore.kernel.org
 help / color / mirror / Atom feed
From: Samuel Holland <samuel@sholland.org>
To: Stephen Boyd <sboyd@kernel.org>,
	Michael Turquette <mturquette@baylibre.com>,
	Andy Gross <agross@kernel.org>,
	Bjorn Andersson <bjorn.andersson@linaro.org>,
	Maxime Ripard <mripard@kernel.org>
Cc: linux-arm-msm@vger.kernel.org, linux-clk@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Samuel Holland <samuel@sholland.org>
Subject: [PATCH RESEND 1/2] clk: Implement protected-clocks for all OF clock providers
Date: Wed,  2 Sep 2020 23:00:14 -0500	[thread overview]
Message-ID: <20200903040015.5627-2-samuel@sholland.org> (raw)
In-Reply-To: <20200903040015.5627-1-samuel@sholland.org>

This is a generic implementation of the "protected-clocks" property from
the common clock binding. It allows firmware to inform the OS about
clocks that must not be disabled while the OS is running.

This implementation comes with some caveats:

1) Clocks that have CLK_IS_CRITICAL in their init data are prepared/
enabled before they are attached to the clock tree. protected-clocks are
only protected once the clock provider is added, which is generally
after all of the clocks it provides have been registered. This leaves a
window of opportunity where something could disable or modify the clock,
such as a driver running on another CPU, or the clock core itself. There
is a comment to this effect in __clk_core_init():

  /*
   * Enable CLK_IS_CRITICAL clocks so newly added critical clocks
   * don't get accidentally disabled when walking the orphan tree and
   * reparenting clocks
   */

Similarly, these clocks will be enabled after they are first reparented,
unlike other CLK_IS_CRITICAL clocks. See the comment in
clk_core_reparent_orphans_nolock():

  /*
   * We need to use __clk_set_parent_before() and _after() to
   * to properly migrate any prepare/enable count of the orphan
   * clock. This is important for CLK_IS_CRITICAL clocks, which
   * are enabled during init but might not have a parent yet.
   */

Ideally we could detect protected clocks before they are reparented, but
there are two problems with that:

  a) From the clock core's perspective, hw->init is const.

  b) The clock core doesn't see the device_node until __clk_register is
     called on the first clock.

So the only "race-free" way to detect protected-clocks is to do it in
the middle of __clk_register, between when core->flags is initialized
and calling __clk_core_init(). That requires scanning the device tree
again for each clock, which is part of why I didn't do it that way.

2) __clk_protect needs to be idempotent, for two reasons:

  a) Clocks with CLK_IS_CRITICAL in their init data are already
     prepared/enabled, and we don't want to prepare/enable them again.

  b) of_clk_set_defaults() is called twice for (at least some) clock
     controllers registered with CLK_OF_DECLARE. It is called first in
     of_clk_add_provider()/of_clk_add_hw_provider() inside clk_init_cb,
     and again afterward in of_clk_init(). The second call in
     of_clk_init() may be unnecessary, but verifying that would require
     auditing all users of CLK_OF_DECLARE to ensure they called one of
     the of_clk_add{,_hw}_provider functions.

Signed-off-by: Samuel Holland <samuel@sholland.org>
---
 drivers/clk/clk-conf.c | 54 ++++++++++++++++++++++++++++++++++++++++++
 drivers/clk/clk.c      | 31 ++++++++++++++++++++++++
 drivers/clk/clk.h      |  2 ++
 3 files changed, 87 insertions(+)

diff --git a/drivers/clk/clk-conf.c b/drivers/clk/clk-conf.c
index 2ef819606c41..a57d28b0f397 100644
--- a/drivers/clk/clk-conf.c
+++ b/drivers/clk/clk-conf.c
@@ -11,6 +11,54 @@
 #include <linux/of.h>
 #include <linux/printk.h>
 
+#include "clk.h"
+
+static int __set_clk_flags(struct device_node *node)
+{
+	struct of_phandle_args clkspec;
+	struct property *prop;
+	int i, index = 0, rc;
+	const __be32 *cur;
+	struct clk *clk;
+	u32 nr_cells;
+
+	rc = of_property_read_u32(node, "#clock-cells", &nr_cells);
+	if (rc < 0) {
+		pr_err("clk: missing #clock-cells property on %pOF\n", node);
+		return rc;
+	}
+
+	clkspec.np         = node;
+	clkspec.args_count = nr_cells;
+
+	of_property_for_each_u32(node, "protected-clocks", prop, cur, clkspec.args[0]) {
+		/* read the remainder of the clock specifier */
+		for (i = 1; i < nr_cells; ++i) {
+			cur = of_prop_next_u32(prop, cur, &clkspec.args[i]);
+			if (!cur) {
+				pr_err("clk: invalid value of protected-clocks"
+				       " property at %pOF\n", node);
+				return -EINVAL;
+			}
+		}
+		clk = of_clk_get_from_provider(&clkspec);
+		if (IS_ERR(clk)) {
+			if (PTR_ERR(clk) != -EPROBE_DEFER)
+				pr_err("clk: couldn't get protected clock"
+				       " %u for %pOF\n", index, node);
+			return PTR_ERR(clk);
+		}
+
+		rc = __clk_protect(clk);
+		if (rc < 0)
+			pr_warn("clk: failed to protect %s: %d\n",
+				__clk_get_name(clk), rc);
+		clk_put(clk);
+		index++;
+	}
+	return 0;
+}
+
 static int __set_clk_parents(struct device_node *node, bool clk_supplier)
 {
 	struct of_phandle_args clkspec;
@@ -135,6 +183,12 @@ int of_clk_set_defaults(struct device_node *node, bool clk_supplier)
 	if (!node)
 		return 0;
 
+	if (clk_supplier) {
+		rc = __set_clk_flags(node);
+		if (rc < 0)
+			return rc;
+	}
+
 	rc = __set_clk_parents(node, clk_supplier);
 	if (rc < 0)
 		return rc;
diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 0a9261a099bd..31fde39f1bda 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -4187,6 +4187,37 @@ void devm_clk_hw_unregister(struct device *dev, struct clk_hw *hw)
 }
 EXPORT_SYMBOL_GPL(devm_clk_hw_unregister);
 
+/*
+ * clk-conf helpers
+ */
+
+int __clk_protect(struct clk *clk)
+{
+	struct clk_core *core = clk->core;
+	int ret = 0;
+
+	clk_prepare_lock();
+
+	/*
+	 * If CLK_IS_CRITICAL was set in the clock's init data, then
+	 * the clock was already prepared/enabled when it was added.
+	 */
+	if (core->flags & CLK_IS_CRITICAL)
+		goto out;
+
+	core->flags |= CLK_IS_CRITICAL;
+	ret = clk_core_prepare(core);
+	if (ret)
+		goto out;
+
+	ret = clk_core_enable_lock(core);
+
+out:
+	clk_prepare_unlock();
+
+	return ret;
+}
+
 /*
  * clkdev helpers
  */
diff --git a/drivers/clk/clk.h b/drivers/clk/clk.h
index 2d801900cad5..367a0f036b13 100644
--- a/drivers/clk/clk.h
+++ b/drivers/clk/clk.h
@@ -24,6 +24,7 @@ struct clk_hw *clk_find_hw(const char *dev_id, const char *con_id);
 #ifdef CONFIG_COMMON_CLK
 struct clk *clk_hw_create_clk(struct device *dev, struct clk_hw *hw,
 			      const char *dev_id, const char *con_id);
+int __clk_protect(struct clk *clk);
 void __clk_put(struct clk *clk);
 #else
 /* All these casts to avoid ifdefs in clkdev... */
@@ -33,6 +34,7 @@ clk_hw_create_clk(struct device *dev, struct clk_hw *hw, const char *dev_id,
 {
 	return (struct clk *)hw;
 }
+static inline int __clk_protect(struct clk *clk) { return 0; }
 static inline void __clk_put(struct clk *clk) { }
 
 #endif
-- 
2.26.2


  reply	other threads:[~2020-09-03  4:00 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-09-03  4:00 [PATCH RESEND 0/2] Common protected-clocks implementation Samuel Holland
2020-09-03  4:00 ` Samuel Holland [this message]
2020-09-03  7:31   ` [PATCH RESEND 1/2] clk: Implement protected-clocks for all OF clock providers Maxime Ripard
2020-09-03  4:00 ` [PATCH RESEND 2/2] Revert "clk: qcom: Support 'protected-clocks' property" Samuel Holland
2021-03-09  8:03 ` [PATCH RESEND 0/2] Common protected-clocks implementation Rasmus Villemoes
2021-03-10  8:56   ` Maxime Ripard
2021-03-10 14:12     ` Samuel Holland
2021-03-10 20:13   ` Bjorn Andersson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200903040015.5627-2-samuel@sholland.org \
    --to=samuel@sholland.org \
    --cc=agross@kernel.org \
    --cc=bjorn.andersson@linaro.org \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-clk@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mripard@kernel.org \
    --cc=mturquette@baylibre.com \
    --cc=sboyd@kernel.org \
    --subject='Re: [PATCH RESEND 1/2] clk: Implement protected-clocks for all OF clock providers' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.