Skip to contentSkip to conversation

clk: Some Clock Range Fixes

PATCH Applied Maxime Ripard 5 participants 37 messages Mar 25, 2022 – Apr 8, 2022

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v2 cover clk: Some Clock Range Fixes

Hi,

This fixes one regression for Tegra30 reported by Dmitry, and another issue I came across while looking at it.

We're also adding a bunch of unit tests to cover those cases.

Let me know what you think,
Maxime

Changes from v1:
- Removed duplicate documentation

Maxime Ripard (3):
clk: Initialize orphan req_rate
clk: test: Test clk_set_rate_range on orphan mux
clk: Drop the rate range on clk_put

 drivers/clk/clk.c      |  55 ++++++++---
 drivers/clk/clk_test.c | 213 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 254 insertions(+), 14 deletions(-)
Signature · 1 line
2.35.1
#1 lore raw

PATCH v2 1/3 clk: Initialize orphan req_rate

When registering a clock that doesn't have a recalc_rate implementation, and doesn't have its parent registered yet, we initialize the clk_core rate and req_rate fields to 0.

The rate field is later updated when the parent is registered in clk_core_reparent_orphans_nolock() using __clk_recalc_rates(), but the req_rate field is never updated.

This leads to an issue in clk_set_rate_range() and clk_put(), since those functions will call clk_set_rate with the content of req_rate to provide drivers with the opportunity to change the rate based on the new boundaries. In this case, we would call clk_set_rate() with a rate of 0, effectively enforcing the minimum allowed for this clock whenever we would call one of those two functions, even though the actual rate might be within range.

Let's fix this by setting req_rate in clk_core_reparent_orphans_nolock() with the rate field content just updated by the call to __clk_recalc_rates().

Fixes 1c8e600440c7 ("clk: Add rate constraints to clocks")
---
 drivers/clk/clk.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)
drivers/clk/clk.c +13 -0
index 9bc8bf434b94..915a2fa363b1 100644
@@ -3479,6 +3479,19 @@ static void clk_core_reparent_orphans_nolock(void)
 			__clk_set_parent_after(orphan, parent, NULL);
 			__clk_recalc_accuracies(orphan);
 			__clk_recalc_rates(orphan, 0);
+
+			/*
+			 * If the clock doesn't have .recalc_rate (so
+			 * wouldn't affect the parent rate) and is
+			 * orphan when it's registered, the
+			 * __clk_init_parent will set the initial
+			 * req_rate to 0.
+			 *
+			 * req_rate is then used by clk_set_rate_range
+			 * and clk_put to trigger a clk_set_rate call
+			 * whenever the boundaries are modified.
+			 */
+			orphan->req_rate = orphan->rate;
 		}
 	}
 }
Signature · 1 line
2.35.1
maintainer Applied
#2 lore raw

Re: [PATCH v2 1/3] clk: Initialize orphan req_rate

Quoting Maxime Ripard (2022-03-25 09:11:42)

Quoted 20 lines

When registering a clock that doesn't have a recalc_rate implementation, and doesn't have its parent registered yet, we initialize the clk_core rate and req_rate fields to 0.

The rate field is later updated when the parent is registered in clk_core_reparent_orphans_nolock() using __clk_recalc_rates(), but the req_rate field is never updated.

This leads to an issue in clk_set_rate_range() and clk_put(), since those functions will call clk_set_rate with the content of req_rate to provide drivers with the opportunity to change the rate based on the new boundaries. In this case, we would call clk_set_rate() with a rate of 0, effectively enforcing the minimum allowed for this clock whenever we would call one of those two functions, even though the actual rate might be within range.

Let's fix this by setting req_rate in clk_core_reparent_orphans_nolock() with the rate field content just updated by the call to __clk_recalc_rates().

Quoted 5 lines
Fixes 1c8e600440c7 ("clk: Add rate constraints to clocks")

---

Applied to clk-next

#3 lore raw

PATCH v2 2/3 clk: test: Test clk_set_rate_range on orphan mux

In reply to Maxime Ripard

A bug recently affected the Tegra30 where calling clk_set_rate_range() on a clock would make it change its rate to the minimum.

This was due to the clock in question being a mux that was orphan at registration, which lead to the clk_core req_rate being 0, and the clk_set_rate_range() function then calling clk_set_rate() with req_rate, effectively making that clock running at the minimum rate allowed, even though the initial rate was within that range.

Make a test suite to create a mux initially orphan, and then make sure that if our clock rate was initially within a given range, then enforcing that range won't affect it.

---
 drivers/clk/clk_test.c | 105 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 105 insertions(+)
drivers/clk/clk_test.c +105 -0
index a92600311506..146b1759798e 100644
@@ -72,6 +72,19 @@ static int clk_dummy_set_rate(struct clk_hw *hw,
 	return 0;
 }
 
+static int clk_dummy_single_set_parent(struct clk_hw *hw, u8 index)
+{
+	if (index >= clk_hw_get_num_parents(hw))
+		return -EINVAL;
+
+	return 0;
+}
+
+static u8 clk_dummy_single_get_parent(struct clk_hw *hw)
+{
+	return 0;
+}
+
 static const struct clk_ops clk_dummy_rate_ops = {
 	.recalc_rate = clk_dummy_recalc_rate,
 	.determine_rate = clk_dummy_determine_rate,
@@ -90,6 +103,11 @@ static const struct clk_ops clk_dummy_minimize_rate_ops = {
 	.set_rate = clk_dummy_set_rate,
 };
 
+static const struct clk_ops clk_dummy_single_parent_ops = {
+	.set_parent = clk_dummy_single_set_parent,
+	.get_parent = clk_dummy_single_get_parent,
+};
+
 static int clk_test_init_with_ops(struct kunit *test, const struct clk_ops *ops)
 {
 	struct clk_dummy_context *ctx;
@@ -239,6 +257,92 @@ static struct kunit_suite clk_test_suite = {
 	.test_cases = clk_test_cases,
 };
 
+struct clk_single_parent_ctx {
+	struct clk_dummy_context parent_ctx;
+	struct clk_hw hw;
+};
+
+static int clk_orphan_transparent_single_parent_mux_test_init(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx;
+	struct clk_init_data init = { };
+	const char *parents[] = { "orphan_parent" };
+	int ret;
+
+	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	test->priv = ctx;
+
+	init.name = "test_orphan_dummy_parent";
+	init.ops = &clk_dummy_single_parent_ops;
+	init.parent_names = parents;
+	init.num_parents = ARRAY_SIZE(parents);
+	init.flags = CLK_SET_RATE_PARENT;
+	ctx->hw.init = &init;
+
+	ret = clk_hw_register(NULL, &ctx->hw);
+	if (ret)
+		return ret;
+
+	memset(&init, 0, sizeof(init));
+	init.name = "orphan_parent";
+	init.ops = &clk_dummy_rate_ops;
+	ctx->parent_ctx.hw.init = &init;
+	ctx->parent_ctx.rate = DUMMY_CLOCK_INIT_RATE;
+
+	ret = clk_hw_register(NULL, &ctx->parent_ctx.hw);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void clk_orphan_transparent_single_parent_mux_test_exit(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+
+	clk_hw_unregister(&ctx->hw);
+	clk_hw_unregister(&ctx->parent_ctx.hw);
+}
+
+/*
+ * Test that a mux-only clock, with an initial rate within a range,
+ * will still have the same rate after the range has been enforced.
+ */
+static void clk_test_orphan_transparent_parent_mux_set_range(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = hw->clk;
+	unsigned long rate, new_rate;
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate_range(clk,
+					   ctx->parent_ctx.rate - 1000,
+					   ctx->parent_ctx.rate + 1000),
+			0);
+
+	new_rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, new_rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, new_rate);
+}
+
+static struct kunit_case clk_orphan_transparent_single_parent_mux_test_cases[] = {
+	KUNIT_CASE(clk_test_orphan_transparent_parent_mux_set_range),
+	{}
+};
+
+static struct kunit_suite clk_orphan_transparent_single_parent_test_suite = {
+	.name = "clk-orphan-transparent-single-parent-test",
+	.init = clk_orphan_transparent_single_parent_mux_test_init,
+	.exit = clk_orphan_transparent_single_parent_mux_test_exit,
+	.test_cases = clk_orphan_transparent_single_parent_mux_test_cases,
+};
+
 /*
  * Test that clk_set_rate_range won't return an error for a valid range
  * and that it will make sure the rate of the clock is within the
@@ -788,6 +892,7 @@ static struct kunit_suite clk_range_minimize_test_suite = {
 
 kunit_test_suites(
 	&clk_test_suite,
+	&clk_orphan_transparent_single_parent_test_suite,
 	&clk_range_test_suite,
 	&clk_range_maximize_test_suite,
 	&clk_range_minimize_test_suite
Signature · 1 line
2.35.1
maintainer Applied
#4 lore raw

Re: [PATCH v2 2/3] clk: test: Test clk_set_rate_range on orphan mux

Quoting Maxime Ripard (2022-03-25 09:11:43)

Quoted 9 lines

A bug recently affected the Tegra30 where calling clk_set_rate_range() on a clock would make it change its rate to the minimum.

This was due to the clock in question being a mux that was orphan at registration, which lead to the clk_core req_rate being 0, and the clk_set_rate_range() function then calling clk_set_rate() with req_rate, effectively making that clock running at the minimum rate allowed, even though the initial rate was within that range.

Quoted 6 lines

Make a test suite to create a mux initially orphan, and then make sure that if our clock rate was initially within a given range, then enforcing that range won't affect it.

---

Applied to clk-next

#5 lore raw

PATCH v2 3/3 clk: Drop the rate range on clk_put

In reply to Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")
---
 drivers/clk/clk.c      |  42 ++++++++++------
 drivers/clk/clk_test.c | 108 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 136 insertions(+), 14 deletions(-)
drivers/clk/clk.c +28 -14
index 915a2fa363b1..91f863b7a824 100644
@@ -2332,19 +2332,15 @@ int clk_set_rate_exclusive(struct clk *clk, unsigned long rate)
 }
 EXPORT_SYMBOL_GPL(clk_set_rate_exclusive);
 
-/**
- * clk_set_rate_range - set a rate range for a clock source
- * @clk: clock source
- * @min: desired minimum clock rate in Hz, inclusive
- * @max: desired maximum clock rate in Hz, inclusive
- *
- * Returns success (0) or negative errno.
- */
-int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
+static int clk_set_rate_range_nolock(struct clk *clk,
+				     unsigned long min,
+				     unsigned long max)
 {
 	int ret = 0;
 	unsigned long old_min, old_max, rate;
 
+	lockdep_assert_held(&prepare_lock);
+
 	if (!clk)
 		return 0;
 
@@ -2357,8 +2353,6 @@ int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
 		return -EINVAL;
 	}
 
-	clk_prepare_lock();
-
 	if (clk->exclusive_count)
 		clk_core_rate_unprotect(clk->core);
 
@@ -2402,6 +2396,28 @@ int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
 	if (clk->exclusive_count)
 		clk_core_rate_protect(clk->core);
 
+	return ret;
+}
+
+/**
+ * clk_set_rate_range - set a rate range for a clock source
+ * @clk: clock source
+ * @min: desired minimum clock rate in Hz, inclusive
+ * @max: desired maximum clock rate in Hz, inclusive
+ *
+ * Returns success (0) or negative errno.
+ */
+int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
+{
+	int ret;
+
+	if (!clk)
+		return 0;
+
+	clk_prepare_lock();
+
+	ret = clk_set_rate_range_nolock(clk, min, max);
+
 	clk_prepare_unlock();
 
 	return ret;
@@ -4403,9 +4419,7 @@ void __clk_put(struct clk *clk)
 	}
 
 	hlist_del(&clk->clks_node);
-	if (clk->min_rate > clk->core->req_rate ||
-	    clk->max_rate < clk->core->req_rate)
-		clk_core_set_rate_nolock(clk->core, clk->core->req_rate);
+	clk_set_rate_range_nolock(clk, 0, ULONG_MAX);
 
 	owner = clk->core->owner;
 	kref_put(&clk->core->ref, __clk_release);
drivers/clk/clk_test.c +108 -0

126 lines not on this page · raw message

Signature · 1 line
2.35.1
maintainer Applied
#6 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoting Maxime Ripard (2022-03-25 09:11:44)

Quoted 7 lines

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 8 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

---

Applied to clk-next

#7 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

In reply to Maxime Ripard

Hi,

Quoted 7 lines · Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 7 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

This patch landed recently in linux-next 20220328 as commit 7dabfa2bc480 ("clk: Drop the rate range on clk_put()"). Sadly it breaks booting of the few of my test systems: Samsung ARM 32bit Exynos3250 based Rinato board and all Amlogic Meson G12B/SM1 based boards (Odroid C4, N2, Khadas VIM3/VIM3l). Rinato hangs always with the following oops:

--->8---

Kernel panic - not syncing: MCT hangs after writing 4 (offset:0x420)
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.17.0-rc1-00014-g7dabfa2bc480 
#11551
Hardware name: Samsung Exynos (Flattened Device Tree)
  unwind_backtrace from show_stack+0x10/0x14
  show_stack from dump_stack_lvl+0x58/0x70
  dump_stack_lvl from panic+0x10c/0x328
  panic from exynos4_mct_tick_stop+0x0/0x2c
---[ end Kernel panic - not syncing: MCT hangs after writing 4 

(offset:0x420) ]---

--->8---

Amlogic boards hang randomly during early userspace init, usually just after loading the driver modules.

Reverting $subject on top of linux-next fixes all those problems.

I will try to analyze it a bit more and if possible provide some more useful/meaning full logs later.

Quoted 205 lines
---
  drivers/clk/clk.c      |  42 ++++++++++------
  drivers/clk/clk_test.c | 108 +++++++++++++++++++++++++++++++++++++++++
  2 files changed, 136 insertions(+), 14 deletions(-)
drivers/clk/clk.c +28 -14
index 915a2fa363b1..91f863b7a824 100644
@@ -2332,19 +2332,15 @@ int clk_set_rate_exclusive(struct clk *clk, unsigned long rate)
  }
  EXPORT_SYMBOL_GPL(clk_set_rate_exclusive);
 
-/**
- * clk_set_rate_range - set a rate range for a clock source
- * @clk: clock source
- * @min: desired minimum clock rate in Hz, inclusive
- * @max: desired maximum clock rate in Hz, inclusive
- *
- * Returns success (0) or negative errno.
- */
-int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
+static int clk_set_rate_range_nolock(struct clk *clk,
+				     unsigned long min,
+				     unsigned long max)
  {
  	int ret = 0;
  	unsigned long old_min, old_max, rate;
 
+	lockdep_assert_held(&prepare_lock);
+
  	if (!clk)
  		return 0;
 
@@ -2357,8 +2353,6 @@ int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
  		return -EINVAL;
  	}
 
-	clk_prepare_lock();
-
  	if (clk->exclusive_count)
  		clk_core_rate_unprotect(clk->core);
 
@@ -2402,6 +2396,28 @@ int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
  	if (clk->exclusive_count)
  		clk_core_rate_protect(clk->core);
 
+	return ret;
+}
+
+/**
+ * clk_set_rate_range - set a rate range for a clock source
+ * @clk: clock source
+ * @min: desired minimum clock rate in Hz, inclusive
+ * @max: desired maximum clock rate in Hz, inclusive
+ *
+ * Returns success (0) or negative errno.
+ */
+int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
+{
+	int ret;
+
+	if (!clk)
+		return 0;
+
+	clk_prepare_lock();
+
+	ret = clk_set_rate_range_nolock(clk, min, max);
+
  	clk_prepare_unlock();
 
  	return ret;
@@ -4403,9 +4419,7 @@ void __clk_put(struct clk *clk)
  	}
 
  	hlist_del(&clk->clks_node);
-	if (clk->min_rate > clk->core->req_rate ||
-	    clk->max_rate < clk->core->req_rate)
-		clk_core_set_rate_nolock(clk->core, clk->core->req_rate);
+	clk_set_rate_range_nolock(clk, 0, ULONG_MAX);
 
  	owner = clk->core->owner;
  	kref_put(&clk->core->ref, __clk_release);
drivers/clk/clk_test.c +104 -0
index 146b1759798e..b205c329cf32 100644
@@ -760,9 +760,65 @@ static void clk_range_test_multiple_set_range_rate_maximized(struct kunit *test)
  	clk_put(user1);
  }
 
+/*
+ * Test that if we have several subsequent calls to
+ * clk_set_rate_range(), across multiple users, the core will reevaluate
+ * whether a new rate is needed, including when a user drop its clock.
+ *
+ * With clk_dummy_maximize_rate_ops, this means that the the rate will
+ * trail along the maximum as it evolves.
+ */
+static void clk_range_test_multiple_set_range_rate_put_maximized(struct kunit *test)
+{
+	struct clk_dummy_context *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = hw->clk;
+	struct clk *user1, *user2;
+	unsigned long rate;
+
+	user1 = clk_hw_get_clk(hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, user1);
+
+	user2 = clk_hw_get_clk(hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, user2);
+
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate(clk, DUMMY_CLOCK_RATE_2 + 1000),
+			0);
+
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate_range(user1,
+					   0,
+					   DUMMY_CLOCK_RATE_2),
+			0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_2);
+
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate_range(user2,
+					   0,
+					   DUMMY_CLOCK_RATE_1),
+			0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+
+	clk_put(user2);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(user1);
+}
+
  static struct kunit_case clk_range_maximize_test_cases[] = {
  	KUNIT_CASE(clk_range_test_set_range_rate_maximized),
  	KUNIT_CASE(clk_range_test_multiple_set_range_rate_maximized),
+	KUNIT_CASE(clk_range_test_multiple_set_range_rate_put_maximized),
  	{}
  };
 
@@ -877,9 +933,61 @@ static void clk_range_test_multiple_set_range_rate_minimized(struct kunit *test)
  	clk_put(user1);
  }
 
+/*
+ * Test that if we have several subsequent calls to
+ * clk_set_rate_range(), across multiple users, the core will reevaluate
+ * whether a new rate is needed, including when a user drop its clock.
+ *
+ * With clk_dummy_minimize_rate_ops, this means that the the rate will
+ * trail along the minimum as it evolves.
+ */
+static void clk_range_test_multiple_set_range_rate_put_minimized(struct kunit *test)
+{
+	struct clk_dummy_context *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = hw->clk;
+	struct clk *user1, *user2;
+	unsigned long rate;
+
+	user1 = clk_hw_get_clk(hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, user1);
+
+	user2 = clk_hw_get_clk(hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, user2);
+
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate_range(user1,
+					   DUMMY_CLOCK_RATE_1,
+					   ULONG_MAX),
+			0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate_range(user2,
+					   DUMMY_CLOCK_RATE_2,
+					   ULONG_MAX),
+			0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(user2);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+

drivers/clk/clk_test.c static void clk_range_test_multiple_set_range_rate_minimized(struct kunit *test) · line 993

Quoted 10 lines
drivers/clk/clk_test.c +4 -0
+	clk_put(user1);
+}
+
  static struct kunit_case clk_range_minimize_test_cases[] = {
  	KUNIT_CASE(clk_range_test_set_range_rate_minimized),
  	KUNIT_CASE(clk_range_test_multiple_set_range_rate_minimized),
+	KUNIT_CASE(clk_range_test_multiple_set_range_rate_put_minimized),
  	{}
  };
 

Best regards

Signature · 2 lines
Marek Szyprowski, PhD
Samsung R&D Institute Poland
#8 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi Marek,

Quoted 37 lines · Marek Szyprowski
Quoted 7 lines · Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 7 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

This patch landed recently in linux-next 20220328 as commit 7dabfa2bc480 ("clk: Drop the rate range on clk_put()"). Sadly it breaks booting of the few of my test systems: Samsung ARM 32bit Exynos3250 based Rinato board and all Amlogic Meson G12B/SM1 based boards (Odroid C4, N2, Khadas VIM3/VIM3l). Rinato hangs always with the following oops:

--->8---

Kernel panic - not syncing: MCT hangs after writing 4 (offset:0x420)
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.17.0-rc1-00014-g7dabfa2bc480 
#11551
Hardware name: Samsung Exynos (Flattened Device Tree)
  unwind_backtrace from show_stack+0x10/0x14
  show_stack from dump_stack_lvl+0x58/0x70
  dump_stack_lvl from panic+0x10c/0x328
  panic from exynos4_mct_tick_stop+0x0/0x2c
---[ end Kernel panic - not syncing: MCT hangs after writing 4 

(offset:0x420) ]---

--->8---

Quoted 7 lines

Amlogic boards hang randomly during early userspace init, usually just after loading the driver modules.

Reverting $subject on top of linux-next fixes all those problems.

I will try to analyze it a bit more and if possible provide some more useful/meaning full logs later.

I'm not sure what could go wrong there, but if you can figure out the clock, if it tries to set a new rate and what rate it is, it would be awesome :)

Maxime

#9 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi,

Quoted 43 lines · Maxime Ripard

Hi Marek,

Quoted 40 lines · Marek Szyprowski
Quoted 7 lines · Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 7 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

This patch landed recently in linux-next 20220328 as commit 7dabfa2bc480 ("clk: Drop the rate range on clk_put()"). Sadly it breaks booting of the few of my test systems: Samsung ARM 32bit Exynos3250 based Rinato board and all Amlogic Meson G12B/SM1 based boards (Odroid C4, N2, Khadas VIM3/VIM3l). Rinato hangs always with the following oops:

--->8---

Kernel panic - not syncing: MCT hangs after writing 4 (offset:0x420)
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.17.0-rc1-00014-g7dabfa2bc480 
#11551
Hardware name: Samsung Exynos (Flattened Device Tree)
  unwind_backtrace from show_stack+0x10/0x14
  show_stack from dump_stack_lvl+0x58/0x70
  dump_stack_lvl from panic+0x10c/0x328
  panic from exynos4_mct_tick_stop+0x0/0x2c
---[ end Kernel panic - not syncing: MCT hangs after writing 4 

(offset:0x420) ]---

--->8---

Amlogic boards hang randomly during early userspace init, usually just after loading the driver modules.

Quoted 8 lines
Quoted 4 lines

Reverting $subject on top of linux-next fixes all those problems.

I will try to analyze it a bit more and if possible provide some more useful/meaning full logs later.

I'm not sure what could go wrong there, but if you can figure out the clock, if it tries to set a new rate and what rate it is, it would be awesome :)

I'm also seeing clockevent break on omaps as a wrong source clock gets picked.

It seems the dts assigned-clock-parents no longer works now?

So the following no longer sets omap_32k_fck as the clockevent source:

timer@0 {
	assigned-clocks = <&gpt1_fck>;
	assigned-clock-parents = <&omap_32k_fck>;
};

Regards,

Tony

#10 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi,

Quoted 49 lines · Tony Lindgren
Quoted 48 lines · Maxime Ripard

Hi Marek,

Quoted 44 lines · Marek Szyprowski
Quoted 7 lines · Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 7 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

This patch landed recently in linux-next 20220328 as commit 7dabfa2bc480 ("clk: Drop the rate range on clk_put()"). Sadly it breaks booting of the few of my test systems: Samsung ARM 32bit Exynos3250 based Rinato board and all Amlogic Meson G12B/SM1 based boards (Odroid C4, N2, Khadas VIM3/VIM3l). Rinato hangs always with the following oops:

--->8---

Kernel panic - not syncing: MCT hangs after writing 4 (offset:0x420)
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.17.0-rc1-00014-g7dabfa2bc480 
#11551
Hardware name: Samsung Exynos (Flattened Device Tree)
  unwind_backtrace from show_stack+0x10/0x14
  show_stack from dump_stack_lvl+0x58/0x70
  dump_stack_lvl from panic+0x10c/0x328
  panic from exynos4_mct_tick_stop+0x0/0x2c
---[ end Kernel panic - not syncing: MCT hangs after writing 4 

(offset:0x420) ]---

--->8---

Amlogic boards hang randomly during early userspace init, usually just after loading the driver modules.

Reverting $subject on top of linux-next fixes all those problems.

I will try to analyze it a bit more and if possible provide some more useful/meaning full logs later.

Quoted 8 lines
Quoted 3 lines

I'm not sure what could go wrong there, but if you can figure out the clock, if it tries to set a new rate and what rate it is, it would be awesome :)

I'm also seeing clockevent break on omaps as a wrong source clock gets picked.

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Maxime

#11 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 7 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

Regards,

Tony

#12 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi Tony,

Quoted 13 lines · Tony Lindgren
Quoted 7 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 5 lines
Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

-- >8 --

drivers/clk/clk.c +25 -0
index 91f863b7a824..c2ab1be62f96 100644
@@ -552,6 +552,12 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	unsigned long best = 0;
 	struct clk_rate_request parent_req = *req;
 
+	pr_crit("%s: requested rate %lu\n", __func__, req->rate);
+
+	parent = core->parent;
+	pr_crit("%s: current parent %s\n", __func__, parent->name);
+	pr_crit("%s: current parent rate %lu\n", __func__, clk_core_get_rate_nolock(parent));
+
 	/* if NO_REPARENT flag set, pass through to current parent */
 	if (core->flags & CLK_SET_RATE_NO_REPARENT) {
 		parent = core->parent;
@@ -578,6 +584,11 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 		if (!parent)
 			continue;
 
+		pr_crit("%s: Trying parent %s (%lu)\n",
+			__func__,
+			parent->name,
+			clk_core_get_rate_nolock(parent));
+
 		if (core->flags & CLK_SET_RATE_PARENT) {
 			parent_req = *req;
 			ret = __clk_determine_rate(parent->hw, &parent_req);
@@ -603,6 +614,11 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	req->best_parent_rate = best;
 	req->rate = best;
 
+	pr_crit("%s: Best parent %s (%lu)\n",
+		__func__,
+		best_parent->name,
+		best);
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(clk_mux_determine_rate_flags);
@@ -2201,8 +2217,12 @@ static int clk_core_set_rate_nolock(struct clk_core *core,
 	if (!core)
 		return 0;
 
+	pr_crit("%s: rate %lu\n", __func__, req_rate);
+
 	rate = clk_core_req_round_rate_nolock(core, req_rate);
 
+	pr_crit("%s: rounded rate %lu\n", __func__, req_rate);
+
 	/* bail early if nothing to do */
 	if (rate == clk_core_get_rate_nolock(core))
 		return 0;
@@ -2367,6 +2387,8 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 		goto out;
 	}
 
+	pr_crit("%s: core req rate %lu\n", __func__, clk->core->req_rate);
+
 	/*
 	 * Since the boundaries have been changed, let's give the
 	 * opportunity to the provider to adjust the clock rate based on
@@ -2385,6 +2407,9 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 	 *   this corner case when determining the rate
 	 */
 	rate = clamp(clk->core->req_rate, min, max);
+
+	pr_crit("%s: clamped rate %lu\n", __func__, rate);
+
 	ret = clk_core_set_rate_nolock(clk->core, rate);
 	if (ret) {
 		/* rollback the changes */

-- >8 --

Thanks!
Maxime

_______________________________________________
linux-amlogic mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-amlogic

#13 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 14 lines · Maxime Ripard
Quoted 13 lines · Tony Lindgren
Quoted 7 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

Thanks that produces some interesting output. In the working case with the $subject patch reverted we have:

[    0.000000] Clocking rate (Crystal/Core/MPU): 26.0/400/600 MHz
[    0.000000] clk_core_set_rate_nolock: rate 960000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 960000000
[    0.000000] clk_core_set_rate_nolock: rate 120000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 120000000
[    0.000000] OMAP clocksource: 32k_counter at 32768 Hz
[    0.000000] clocksource: 32k_counter: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 58327039986419 ns
[    0.000000] sched_clock: 32 bits at 32kHz, resolution 30517ns, wraps every 65535999984741ns
[    0.011779] TI gptimer clockevent: always-on 32768 Hz at /ocp@68000000/target-module@48318000

In the failing case With the $subject patch not reverted, the debug output goes a bit crazy, see below :)

Regards,

Tony

8< ----------------

[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 0
[    0.000000] clk_set_rate_range_nolock: clamped rate 0
[    0.000000] clk_core_set_rate_nolock: rate 0
[    0.000000] clk_core_set_rate_nolock: rounded rate 0
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 32768
[    0.000000] clk_set_rate_range_nolock: clamped rate 32768
[    0.000000] clk_core_set_rate_nolock: rate 32768
[    0.000000] clk_core_set_rate_nolock: rounded rate 32768
[    0.000000] clk_set_rate_range_nolock: core req rate 12000000
[    0.000000] clk_set_rate_range_nolock: clamped rate 12000000
[    0.000000] clk_core_set_rate_nolock: rate 12000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 12000000
[    0.000000] clk_set_rate_range_nolock: core req rate 13000000
[    0.000000] clk_set_rate_range_nolock: clamped rate 13000000
[    0.000000] clk_core_set_rate_nolock: rate 13000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 13000000
[    0.000000] clk_set_rate_range_nolock: core req rate 19200000
[    0.000000] clk_set_rate_range_nolock: clamped rate 19200000
[    0.000000] clk_core_set_rate_nolock: rate 19200000
[    0.000000] clk_core_set_rate_nolock: rounded rate 19200000
[    0.000000] clk_set_rate_range_nolock: core req rate 26000000
[    0.000000] clk_set_rate_range_nolock: clamped rate 26000000
[    0.000000] clk_core_set_rate_nolock: rate 26000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 26000000
[    0.000000] clk_set_rate_range_nolock: core req rate 38400000
[    0.000000] clk_set_rate_range_nolock: clamped rate 38400000
[    0.000000] clk_core_set_rate_nolock: rate 38400000
[    0.000000] clk_core_set_rate_nolock: rounded rate 38400000
[    0.000000] clk_set_rate_range_nolock: core req rate 16800000
[    0.000000] clk_set_rate_range_nolock: clamped rate 16800000
[    0.000000] clk_core_set_rate_nolock: rate 16800000
[    0.000000] clk_core_set_rate_nolock: rounded rate 16800000
[    0.000000] clk_set_rate_range_nolock: core req rate 12000000
[    0.000000] clk_set_rate_range_nolock: clamped rate 12000000
[    0.000000] clk_core_set_rate_nolock: rate 12000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 12000000
[    0.000000] clk_set_rate_range_nolock: core req rate 13000000
[    0.000000] clk_set_rate_range_nolock: clamped rate 13000000
[    0.000000] clk_core_set_rate_nolock: rate 13000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 13000000
[    0.000000] clk_set_rate_range_nolock: core req rate 19200000
[    0.000000] clk_set_rate_range_nolock: clamped rate 19200000
[    0.000000] clk_core_set_rate_nolock: rate 19200000
[    0.000000] clk_core_set_rate_nolock: rounded rate 19200000
[    0.000000] clk_set_rate_range_nolock: core req rate 26000000
[    0.000000] clk_set_rate_range_nolock: clamped rate 26000000
[    0.000000] clk_core_set_rate_nolock: rate 26000000
[    0.000000] clk_core_set_rate_nolock: rounded rate 26000000
[    0.000000] clk_set_rate_range_nolock: core req rate 38400000
[    0.000000] clk_set_rate_range_nolock: clamped rate 38400000
[    0.000000] clk_core_set_rate_nolock: rate 38400000
[    0.000000] clk_core_set_rate_nolock: rounded rate 38400000
[    0.000000] clk_set_rate_range_nolock: core req rate 16800000
[    0.000000] clk_set_rate_range_nolock: clamped rate 16800000
[    0.000000] clk_core_set_rate_nolock: rate 16800000
[    0.000000] clk_core_set_rate_nolock: rounded rate 16800000
[    0.000000] clk_set_rate_range_nolock: core req rate 26000000
[    0.000000] clk_set_rate_range_nolock: clamped rate 26000000
[    0.000000] clk_core_set_rate_nolock: rate 26000000
[    0.000000] clk_mux_determine_rate_flags: requested rate 26000000
[    0.000000] clk_mux_determine_rate_flags: current parent virt_26000000_ck
[    0.000000] clk_mux_determine_rate_flags: current parent rate 26000000
[    0.000000] 8<--- cut here ---
[    0.000000] Unable to handle kernel NULL pointer dereference at virtual address 00000000
[    0.000000] [00000000] *pgd=00000000
[    0.000000] Internal error: Oops: 5 [#1] SMP ARM
[    0.000000] Modules linked in:
[    0.000000] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.17.0-13433-g14667a708872 #193
[    0.000000] Hardware name: Generic OMAP36xx (Flattened Device Tree)
[    0.000000] PC is at clk_mux_determine_rate_flags+0x284/0x2dc
[    0.000000] LR is at clk_mux_determine_rate_flags+0x9c/0x2dc
[    0.000000] pc : [<c0a3e28c>]    lr : [<c0a3e0a4>]    psr: 600000d3
[    0.000000] sp : c0f01e68  ip : 3ffff7ff  fp : c0b49f04
[    0.000000] r10: c0b49ef8  r9 : c18843c0  r8 : c18843c0
[    0.000000] r7 : c0f01eb0  r6 : 018cba80  r5 : 00000000  r4 : 018cba80
[    0.000000] r3 : 00000000  r2 : 00000000  r1 : c0f01d38  r0 : c0c923e8
[    0.000000] Flags: nZCv  IRQs off  FIQs off  Mode SVC_32  ISA ARM  Segment none
[    0.000000] Control: 10c5387d  Table: 80004019  DAC: 00000051
[    0.000000] Register r0 information: non-slab/vmalloc memory
[    0.000000] Register r1 information: non-slab/vmalloc memory
[    0.000000] Register r2 information: NULL pointer
[    0.000000] Register r3 information: NULL pointer
[    0.000000] Register r4 information: non-paged memory
[    0.000000] Register r5 information: NULL pointer
[    0.000000] Register r6 information: non-paged memory
[    0.000000] Register r7 information: non-slab/vmalloc memory
[    0.000000] Register r8 information: slab kmalloc-192 start c18843c0 pointer offset 0 size 192
[    0.000000] Register r9 information: slab kmalloc-192 start c18843c0 pointer offset 0 size 192
[    0.000000] Register r10 information: non-slab/vmalloc memory
[    0.000000] Register r11 information: non-slab/vmalloc memory
[    0.000000] Register r12 information: non-paged memory
[    0.000000] Process swapper/0 (pid: 0, stack limit = 0x(ptrval))
[    0.000000] Stack: (0xc0f01e68 to 0xc0f02000)
[    0.000000] 1e60:                   00000000 ffffffff 018cba80 00000000 ffffffff 018cba80
[    0.000000] 1e80: c180adc0 c0f051c8 c18843c0 c18843c0 00000000 018cba80 00000000 ffffffff
[    0.000000] 1ea0: c18843c0 c7c9fae8 c7c9fb54 c0a3e378 018cba80 00000000 ffffffff 018cba80
[    0.000000] 1ec0: c180adc0 c0f051c8 c1028c80 c180aa40 018cba80 018cba80 00000000 c065349c
[    0.000000] 1ee0: 00000000 00000000 c7c9fb54 00000000 c180aa40 c180aa40 c180aa80 c0f01f24
[    0.000000] 1f00: c180aac0 00000000 00000001 c0653720 00000000 c180aa80 c0f01f24 c0e2c1a0
[    0.000000] 1f20: 00000000 c180a788 c1880b48 c7c9f940 00000000 00000000 00000000 c7cd0d98
[    0.000000] 1f40: c0e69b50 c0f01f70 a00000d3 c082c740 c7ca0c00 c0e69b50 c0f051c0 c7dffa40
[    0.000000] 1f60: ffffffff 00000000 10c5387d c0e0f6dc 00000000 c0f051c8 ffffffff c100843c
[    0.000000] 1f80: c0e5ba60 c0f051c0 c7dffa40 ffffffff 00000000 10c5387d 00000000 c0e0ba4c
[    0.000000] 1fa0: c1008000 c0e10e7c c1008000 c0e00f80 ffffffff ffffffff 00000000 c0e00728
[    0.000000] 1fc0: 00000000 c0e5ba60 a5aa33e0 c0f051c8 00000000 c0e004bc 00000051 10c0387d
[    0.000000] 1fe0: ffffffff 86feb000 413fc082 10c5387d 00000000 00000000 00000000 00000000
[    0.000000]  clk_mux_determine_rate_flags from clk_core_set_rate_nolock.part.0+0x84/0x1a8
[    0.000000]  clk_core_set_rate_nolock.part.0 from clk_set_rate_range_nolock.part.0+0x278/0x2a0
[    0.000000]  clk_set_rate_range_nolock.part.0 from __clk_put+0x58/0x160
[    0.000000]  __clk_put from of_clk_init+0x1b4/0x268
[    0.000000]  of_clk_init from omap_clk_init+0x38/0x58
[    0.000000]  omap_clk_init from omap_init_time_of+0x8/0x10
[    0.000000]  omap_init_time_of from start_kernel+0x480/0x6b0
[    0.000000]  start_kernel from 0x0
[    0.000000] Code: 0a000008 e587400c e5874000 e59f0050 (e5952000) 
[    0.000000] ---[ end trace 0000000000000000 ]---
[    0.000000] Kernel panic - not syncing: Attempted to kill the idle task!
[    0.000000] ---[ end Kernel panic - not syncing: Attempted to kill the idle task! ]---
maintainer
#14 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoting Tony Lindgren (2022-03-31 10:00:09)

Quoted 18 lines · Maxime Ripard
Quoted 17 lines · Tony Lindgren
Quoted 16 lines · Maxime Ripard
Quoted 7 lines · Tony Lindgren
Quoted 1 line

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

Thanks that produces some interesting output. In the working case with the $subject patch reverted we have:

I don't think clk_put() dropping a range request is very important right now. If this isn't fixed tomorrow then we should revert out this patch so systems can boot -rc1 and try to fix it in parallel.

#15 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi,

Quoted 25 lines · Stephen Boyd

Quoting Tony Lindgren (2022-03-31 10:00:09)

Quoted 24 lines · Maxime Ripard
Quoted 22 lines · Tony Lindgren
Quoted 13 lines · Maxime Ripard
Quoted 7 lines · Tony Lindgren
Quoted 1 line

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 5 lines
Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

Quoted 6 lines
Quoted 2 lines

Thanks that produces some interesting output. In the working case with the $subject patch reverted we have:

I don't think clk_put() dropping a range request is very important right now. If this isn't fixed tomorrow then we should revert out this patch so systems can boot -rc1 and try to fix it in parallel.

Yeah, it can definitely be reverted. I'm not so sure that the issue is with this patch itself though but more that it now triggers a fault reliably.

Maxime

maintainer
#16 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoting Maxime Ripard (2022-04-01 05:28:39)

Quoted 4 lines

Hi,

Quoted 1 line · Stephen Boyd
Quoted 8 lines
Quoted 3 lines

I don't think clk_put() dropping a range request is very important right now. If this isn't fixed tomorrow then we should revert out this patch so systems can boot -rc1 and try to fix it in parallel.

Yeah, it can definitely be reverted. I'm not so sure that the issue is with this patch itself though but more that it now triggers a fault reliably.

I don't see a revert sent yet so I'll send one now.

#17 lore raw

Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

In reply to Maxime Ripard

Hello,

Am Donnerstag, 31. März 2022, 17:31:34 CEST schrieb Maxime Ripard:

Quoted 18 lines

* PGP Signed by an unknown key

Hi Tony,

Quoted 13 lines · Tony Lindgren
Quoted 7 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

I'm affected by this patch as well on an imx8mp platform (see [1] for some details)

In the failing case with with your patch applied I get the following error
---

[    0.661064] clk_set_rate_range_nolock: core req rate 500000000                                                                    
[    0.664084] clk_set_rate_range_nolock: clamped rate 500000000                                                                     
[    0.669851] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.674843] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.680536] clk_set_rate_range_nolock: core req rate 800000000                                                                    
[    0.686389] clk_set_rate_range_nolock: clamped rate 800000000                                                                     
[    0.692164] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.697153] clk_mux_determine_rate_flags: requested rate 800000000                                                                
[    0.703363] clk_mux_determine_rate_flags: current parent sys_pll1                                                                 
[    0.709487] clk_mux_determine_rate_flags: current parent rate 800000000                                                           
[    0.716147] Unable to handle kernel NULL pointer dereference at virtual 
address 0000000000000000
[    0.724977] Mem abort info:                                                                                                       
[    0.727775]   ESR = 0x96000004                                                                                                    
[    0.730835]   EC = 0x25: DABT (current EL), IL = 32 bits                                                                          
[    0.736177]   SET = 0, FnV = 0                                                                                                    
[    0.739239]   EA = 0, S1PTW = 0                                                                                                   
[    0.742382]   FSC = 0x04: level 0 translation fault                                                                               
[    0.747287] Data abort info:                                                                                                      
[    0.750172]   ISV = 0, ISS = 0x00000004                                                                                           
[    0.754027]   CM = 0, WnR = 0                                                                                                     
[    0.757002] [0000000000000000] user address but active_mm is swapper                                                              
[    0.763394] Internal error: Oops: 96000004 [#1] PREEMPT SMP                                                                       
[    0.768985] Modules linked in:                                                                                                    
[    0.772049] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.17.0-next-20220331+ 
#48 8e9d24095c7f6b15f85bc2ad57a5609e219130b9          
[    0.782984] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)                                                       
[    0.789985] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)                                                       
[    0.796985] pc : clk_mux_determine_rate_flags+0x280/0x2cc                                                                         
[    0.802407] lr : clk_mux_determine_rate_flags+0xf4/0x2cc                                                                          
[    0.807747] sp : ffff800009ceb590                                                                                                 
[    0.811072] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27: 
ffff800008eaa038                                                     
[    0.818245] x26: ffff8000092fe0b0 x25: ffff000000090000 x24: 
ffff000000090000                                                     
[    0.825420] x23: 0000000000000000 x22: ffff800008ea95d8 x21: 
ffff0000028f4700                                                     
[    0.832595] x20: 000000002faf0800 x19: 0000000000000000 x18: 
0000000000004590                                                     
[    0.839770] x17: 0000000000004570 x16: 0000000000004560 x15: 
ffff8000092ff250                                                     
[    0.846945] x14: 0000000000000000 x13: 3030303030303030 x12: 
3820657461722074                                                     
[    0.854120] x11: 6e6572617020746e x10: 6572727563203a73 x9 : 
7563203a7367616c                                                     
[    0.861295] x8 : 665f657461725f65 x7 : 205d373834393037 x6 : 
ffff800009a947c8                                                     
[    0.868472] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 : 
000000002faf0800                                                     
[    0.875645] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 : 
ffff8000092fd5b8                                                     
[    0.882822] Call trace:                                                                                                           
[    0.885273]  clk_mux_determine_rate_flags+0x280/0x2cc                                                                             
[    0.890347]  clk_mux_determine_rate+0x10/0x20                                                                                     
[    0.894720]  clk_core_determine_round_nolock+0x4c/0xb4                                                                            
[    0.899882]  clk_core_round_rate_nolock+0x30/0x80                                                                                 
[    0.904607]  clk_core_round_rate_nolock+0x70/0x80                                                                                 
[    0.909334]  clk_hw_round_rate+0x44/0x74                                                                                          
[    0.913270]  clk_factor_round_rate+0x60/0x80                                                                                      
[    0.917557]  clk_core_determine_round_nolock+0x88/0xb4                                                                            
[    0.922720]  clk_core_round_rate_nolock+0x30/0x80                                                                                 
[    0.927445]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0                                                                           
[    0.932695]  clk_set_rate_range_nolock+0x234/0x244                                                                                
[    0.937507]  __clk_put+0x60/0x12c                                                                                                 
[    0.940834]  clk_put+0xc/0x1c                                                                                                     
[    0.943809]  __set_clk_parents+0x12c/0x244                                                                                        
[    0.947920]  of_clk_set_defaults+0x20/0x50                                                                                        
[    0.952032]  of_clk_add_hw_provider.part.0+0x94/0x120                                                                             
[    0.957107]  of_clk_add_hw_provider+0x10/0x20                                                                                     
[    0.961482]  imx8mp_clocks_probe+0x3458/0x34d0                                                                                    
[    0.965945]  platform_probe+0x64/0x100                                                                                            
[    0.969707]  call_driver_probe+0x28/0x130                                                                                         
[    0.973732]  really_probe+0x178/0x310                                                                                             
[    0.977409]  __driver_probe_device+0xfc/0x144                                                                                     
[    0.981782]  driver_probe_device+0x38/0x12c                                                                                       
[    0.985982]  __driver_attach+0xcc/0x220                                                                                           
[    0.989834]  bus_for_each_dev+0x6c/0xc0                                                                                           
[    0.993682]  driver_attach+0x20/0x2c                                                                                              
[    0.997270]  bus_add_driver+0x140/0x230                                                                                           
[    1.001120]  driver_register+0x74/0x120                                                                                           
[    1.004970]  __platform_driver_register+0x24/0x30                                                                                 
[    1.009697]  imx8mp_clk_driver_init+0x18/0x20                                                                                     
[    1.014070]  do_one_initcall+0x58/0x200                                                                                           
[    1.017920]  do_initcalls+0x164/0x19c                                                                                             
[    1.021597]  kernel_init_freeable+0x134/0x17c                                                                                     
[    1.025970]  kernel_init+0x2c/0x150                                                                                               
[    1.029470]  ret_from_fork+0x10/0x20                                                                                              
[    1.033065] Code: f9000f94 912982c1 b0002900 9116e000 (f9400262)                                                                  
[    1.039188] ---[ end trace 0000000000000000 ]---                                                                                  
[    1.043869] Kernel panic - not syncing: Attempted to kill init! 
exitcode=0x0000000b                                               
[    1.051523] SMP: stopping secondary CPUs                                                                                          
[    1.055467] Kernel Offset: disabled                                                                                               
[    1.058960] CPU features: 0x000,00020009,00001082                                                                                 
[    1.063684] Memory Limit: none                                                                                                    
[    1.066748] ---[ end Kernel panic - not syncing: Attempted to kill init! 

exitcode=0x0000000b ]---
---

With the $subject patch reverted and bootable system:
---

[    0.659922] clk_core_set_rate_nolock: rate 1000000000                                                                             
[    0.662154] clk_core_set_rate_nolock: rounded rate 1000000000                                                                     
[    0.667932] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.672918] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.678601] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.683592] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.689276] clk_core_set_rate_nolock: rate 400000000                                                                              
[    0.694267] clk_core_set_rate_nolock: rounded rate 400000000                                                                      
[    0.699980] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.704942] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.710627] clk_core_set_rate_nolock: rate 393216000                                                                              
[    0.715611] clk_core_set_rate_nolock: rounded rate 393216000                                                                      
[    0.721815] clk_core_set_rate_nolock: rate 361267200                                                                              
[    0.726284] clk_core_set_rate_nolock: rounded rate 361267200                                                                      
[    0.734097] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.736977] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.742652] clk_core_set_rate_nolock: rate 400000000                                                                              
[    0.747645] clk_core_set_rate_nolock: rounded rate 400000000                                                                      
[    0.754565] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.758331] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.764688] SoC: i.MX8MP revision 1.1                                                                                             
[    0.767931] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.772675] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.778354] clk_core_set_rate_nolock: rate 200000000                                                                              
[    0.783351] clk_core_set_rate_nolock: rounded rate 200000000                                                                      
[    0.793748] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled                                                               
[    0.798952] 30860000.serial: ttymxc0 at MMIO 0x30860000 (irq = 34, 
base_baud = 5000000) is a IMX                                  
[    0.806527] 30880000.serial: ttymxc2 at MMIO 0x30880000 (irq = 35, 
base_baud = 5000000) is a IMX                                  
[    0.815329] 30890000.serial: ttymxc1 at MMIO 0x30890000 (irq = 36, 
base_baud = 5000000) is a IMX                                  
[    0.824176] 30a60000.serial: ttymxc3 at MMIO 0x30a60000 (irq = 43, 
base_baud = 1500000) is a IMX                                  
[    0.832588] printk: console [ttymxc3] enabled                                                                                     
[    0.832588] printk: console [ttymxc3] enabled                                                                                     
[    0.841244] printk: bootconsole [ec_imx6q0] disabled                                                                              
[    0.841244] printk: bootconsole [ec_imx6q0] disabled                                                                              
[    0.857871] clk_core_set_rate_nolock: rate 80000000                                                                               
[    0.862796] clk_core_set_rate_nolock: rounded rate 80000000                                                                       
[    0.868469] clk_core_set_rate_nolock: rate 20000000                                                                               
[    0.873364] clk_core_set_rate_nolock: rounded rate 20000000                                                                       
[    0.879258] clk_core_set_rate_nolock: rate 80000000                                                                               
[    0.884154] clk_core_set_rate_nolock: rounded rate 80000000

[...]
---

The 500000000 and 800000000 look a bit like the assigned-clock-rates for clock-controller@30380000 in arch/arm64/boot/dts/freescale/imx8mp.dtsi

If you need some more information, do not hesitate to ask

Best regards,
Alexander

[1] https://lore.kernel.org/all/1911426.usQuhbGJ8B@steina-w/

#18 lore raw

Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi Alexander, Tony, Marek,

Quoted 178 lines · Alexander Stein

Am Donnerstag, 31. März 2022, 17:31:34 CEST schrieb Maxime Ripard:

Quoted 18 lines

* PGP Signed by an unknown key

Hi Tony,

Quoted 13 lines · Tony Lindgren
Quoted 7 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

I'm affected by this patch as well on an imx8mp platform (see [1] for some details)

In the failing case with with your patch applied I get the following error
---

[    0.661064] clk_set_rate_range_nolock: core req rate 500000000                                                                    
[    0.664084] clk_set_rate_range_nolock: clamped rate 500000000                                                                     
[    0.669851] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.674843] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.680536] clk_set_rate_range_nolock: core req rate 800000000                                                                    
[    0.686389] clk_set_rate_range_nolock: clamped rate 800000000                                                                     
[    0.692164] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.697153] clk_mux_determine_rate_flags: requested rate 800000000                                                                
[    0.703363] clk_mux_determine_rate_flags: current parent sys_pll1                                                                 
[    0.709487] clk_mux_determine_rate_flags: current parent rate 800000000                                                           
[    0.716147] Unable to handle kernel NULL pointer dereference at virtual 
address 0000000000000000
[    0.724977] Mem abort info:                                                                                                       
[    0.727775]   ESR = 0x96000004                                                                                                    
[    0.730835]   EC = 0x25: DABT (current EL), IL = 32 bits                                                                          
[    0.736177]   SET = 0, FnV = 0                                                                                                    
[    0.739239]   EA = 0, S1PTW = 0                                                                                                   
[    0.742382]   FSC = 0x04: level 0 translation fault                                                                               
[    0.747287] Data abort info:                                                                                                      
[    0.750172]   ISV = 0, ISS = 0x00000004                                                                                           
[    0.754027]   CM = 0, WnR = 0                                                                                                     
[    0.757002] [0000000000000000] user address but active_mm is swapper                                                              
[    0.763394] Internal error: Oops: 96000004 [#1] PREEMPT SMP                                                                       
[    0.768985] Modules linked in:                                                                                                    
[    0.772049] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.17.0-next-20220331+ 
#48 8e9d24095c7f6b15f85bc2ad57a5609e219130b9          
[    0.782984] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)                                                       
[    0.789985] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)                                                       
[    0.796985] pc : clk_mux_determine_rate_flags+0x280/0x2cc                                                                         
[    0.802407] lr : clk_mux_determine_rate_flags+0xf4/0x2cc                                                                          
[    0.807747] sp : ffff800009ceb590                                                                                                 
[    0.811072] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27: 
ffff800008eaa038                                                     
[    0.818245] x26: ffff8000092fe0b0 x25: ffff000000090000 x24: 
ffff000000090000                                                     
[    0.825420] x23: 0000000000000000 x22: ffff800008ea95d8 x21: 
ffff0000028f4700                                                     
[    0.832595] x20: 000000002faf0800 x19: 0000000000000000 x18: 
0000000000004590                                                     
[    0.839770] x17: 0000000000004570 x16: 0000000000004560 x15: 
ffff8000092ff250                                                     
[    0.846945] x14: 0000000000000000 x13: 3030303030303030 x12: 
3820657461722074                                                     
[    0.854120] x11: 6e6572617020746e x10: 6572727563203a73 x9 : 
7563203a7367616c                                                     
[    0.861295] x8 : 665f657461725f65 x7 : 205d373834393037 x6 : 
ffff800009a947c8                                                     
[    0.868472] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 : 
000000002faf0800                                                     
[    0.875645] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 : 
ffff8000092fd5b8                                                     
[    0.882822] Call trace:                                                                                                           
[    0.885273]  clk_mux_determine_rate_flags+0x280/0x2cc                                                                             
[    0.890347]  clk_mux_determine_rate+0x10/0x20                                                                                     
[    0.894720]  clk_core_determine_round_nolock+0x4c/0xb4                                                                            
[    0.899882]  clk_core_round_rate_nolock+0x30/0x80                                                                                 
[    0.904607]  clk_core_round_rate_nolock+0x70/0x80                                                                                 
[    0.909334]  clk_hw_round_rate+0x44/0x74                                                                                          
[    0.913270]  clk_factor_round_rate+0x60/0x80                                                                                      
[    0.917557]  clk_core_determine_round_nolock+0x88/0xb4                                                                            
[    0.922720]  clk_core_round_rate_nolock+0x30/0x80                                                                                 
[    0.927445]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0                                                                           
[    0.932695]  clk_set_rate_range_nolock+0x234/0x244                                                                                
[    0.937507]  __clk_put+0x60/0x12c                                                                                                 
[    0.940834]  clk_put+0xc/0x1c                                                                                                     
[    0.943809]  __set_clk_parents+0x12c/0x244                                                                                        
[    0.947920]  of_clk_set_defaults+0x20/0x50                                                                                        
[    0.952032]  of_clk_add_hw_provider.part.0+0x94/0x120                                                                             
[    0.957107]  of_clk_add_hw_provider+0x10/0x20                                                                                     
[    0.961482]  imx8mp_clocks_probe+0x3458/0x34d0                                                                                    
[    0.965945]  platform_probe+0x64/0x100                                                                                            
[    0.969707]  call_driver_probe+0x28/0x130                                                                                         
[    0.973732]  really_probe+0x178/0x310                                                                                             
[    0.977409]  __driver_probe_device+0xfc/0x144                                                                                     
[    0.981782]  driver_probe_device+0x38/0x12c                                                                                       
[    0.985982]  __driver_attach+0xcc/0x220                                                                                           
[    0.989834]  bus_for_each_dev+0x6c/0xc0                                                                                           
[    0.993682]  driver_attach+0x20/0x2c                                                                                              
[    0.997270]  bus_add_driver+0x140/0x230                                                                                           
[    1.001120]  driver_register+0x74/0x120                                                                                           
[    1.004970]  __platform_driver_register+0x24/0x30                                                                                 
[    1.009697]  imx8mp_clk_driver_init+0x18/0x20                                                                                     
[    1.014070]  do_one_initcall+0x58/0x200                                                                                           
[    1.017920]  do_initcalls+0x164/0x19c                                                                                             
[    1.021597]  kernel_init_freeable+0x134/0x17c                                                                                     
[    1.025970]  kernel_init+0x2c/0x150                                                                                               
[    1.029470]  ret_from_fork+0x10/0x20                                                                                              
[    1.033065] Code: f9000f94 912982c1 b0002900 9116e000 (f9400262)                                                                  
[    1.039188] ---[ end trace 0000000000000000 ]---                                                                                  
[    1.043869] Kernel panic - not syncing: Attempted to kill init! 
exitcode=0x0000000b                                               
[    1.051523] SMP: stopping secondary CPUs                                                                                          
[    1.055467] Kernel Offset: disabled                                                                                               
[    1.058960] CPU features: 0x000,00020009,00001082                                                                                 
[    1.063684] Memory Limit: none                                                                                                    
[    1.066748] ---[ end Kernel panic - not syncing: Attempted to kill init! 

exitcode=0x0000000b ]---
---

With the $subject patch reverted and bootable system:
---

[    0.659922] clk_core_set_rate_nolock: rate 1000000000                                                                             
[    0.662154] clk_core_set_rate_nolock: rounded rate 1000000000                                                                     
[    0.667932] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.672918] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.678601] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.683592] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.689276] clk_core_set_rate_nolock: rate 400000000                                                                              
[    0.694267] clk_core_set_rate_nolock: rounded rate 400000000                                                                      
[    0.699980] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.704942] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.710627] clk_core_set_rate_nolock: rate 393216000                                                                              
[    0.715611] clk_core_set_rate_nolock: rounded rate 393216000                                                                      
[    0.721815] clk_core_set_rate_nolock: rate 361267200                                                                              
[    0.726284] clk_core_set_rate_nolock: rounded rate 361267200                                                                      
[    0.734097] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.736977] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.742652] clk_core_set_rate_nolock: rate 400000000                                                                              
[    0.747645] clk_core_set_rate_nolock: rounded rate 400000000                                                                      
[    0.754565] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.758331] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.764688] SoC: i.MX8MP revision 1.1                                                                                             
[    0.767931] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.772675] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.778354] clk_core_set_rate_nolock: rate 200000000                                                                              
[    0.783351] clk_core_set_rate_nolock: rounded rate 200000000                                                                      
[    0.793748] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled                                                               
[    0.798952] 30860000.serial: ttymxc0 at MMIO 0x30860000 (irq = 34, 
base_baud = 5000000) is a IMX                                  
[    0.806527] 30880000.serial: ttymxc2 at MMIO 0x30880000 (irq = 35, 
base_baud = 5000000) is a IMX                                  
[    0.815329] 30890000.serial: ttymxc1 at MMIO 0x30890000 (irq = 36, 
base_baud = 5000000) is a IMX                                  
[    0.824176] 30a60000.serial: ttymxc3 at MMIO 0x30a60000 (irq = 43, 
base_baud = 1500000) is a IMX                                  
[    0.832588] printk: console [ttymxc3] enabled                                                                                     
[    0.832588] printk: console [ttymxc3] enabled                                                                                     
[    0.841244] printk: bootconsole [ec_imx6q0] disabled                                                                              
[    0.841244] printk: bootconsole [ec_imx6q0] disabled                                                                              
[    0.857871] clk_core_set_rate_nolock: rate 80000000                                                                               
[    0.862796] clk_core_set_rate_nolock: rounded rate 80000000                                                                       
[    0.868469] clk_core_set_rate_nolock: rate 20000000                                                                               
[    0.873364] clk_core_set_rate_nolock: rounded rate 20000000                                                                       
[    0.879258] clk_core_set_rate_nolock: rate 80000000                                                                               
[    0.884154] clk_core_set_rate_nolock: rounded rate 80000000
Quoted 7 lines

[...]
---

The 500000000 and 800000000 look a bit like the assigned-clock-rates for clock-controller@30380000 in arch/arm64/boot/dts/freescale/imx8mp.dtsi

If you need some more information, do not hesitate to ask

Thanks a lot to you three for all your testing. I think I know what might be going on:

We use the last requested rate on clk_set_rate_range (clk_core.req_rate), and that requested rate if the clock is orphan will be set to 0, so if we were to call clk_set_rate_range before the parent clock is registered, we would effectively call a clk_set_rate to 0

And the assigned-clocks stuff is handled by __set_clk_parents and __set_clk_rates, called by of_clk_set_defaults(), in turn called by of_clk_init and of_clk_add_provider. Both __set_clk_parents and __set_clk_rates will call clk_put once done with the clock, and we will with this patch trigger the clk_set_rate to 0 I mentioned before.

So we just became very good at triggering the underlying issue :)

And I think it's that while we update the requested rate when the missing parent is registered, we never do when we mux away from it using clk_set_parent.

Could you test the following patch and let me know if it works?

--->8---

drivers/clk/clk.c +2 -0
index 91f863b7a824..ee5a0223e47d 100644
@@ -2599,6 +2599,8 @@ static int clk_core_set_parent_nolock(struct clk_core *core,
 	} else {
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
+
+		core->req_rate = core->rate;
 	}
 
 runtime_put:

--->8---

Thanks!
Maxime

#19 lore raw

Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi Maxime,

Am Freitag, 1. April 2022, 14:27:36 CEST schrieb Maxime Ripard:

Quoted 220 lines · Alexander Stein
Quoted 182 lines

Am Donnerstag, 31. März 2022, 17:31:34 CEST schrieb Maxime Ripard:

Quoted 24 lines
Quoted 1 line

Old Signed by an unknown key

Hi Tony,

Quoted 19 lines · Tony Lindgren
Quoted 8 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on
both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 4 lines

It looks like the gpt1_fck driver might favor another parent for that
rate, which, if it's an invalid configuration, shouldn't really happen?

Quoted 5 lines
Quoted 2 lines

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

I'm affected by this patch as well on an imx8mp platform (see [1] for some details)

In the failing case with with your patch applied I get the following error
---

[    0.661064] clk_set_rate_range_nolock: core req rate 500000000
[    0.664084] clk_set_rate_range_nolock: clamped rate 500000000
[    0.669851] clk_core_set_rate_nolock: rate 500000000
[    0.674843] clk_core_set_rate_nolock: rounded rate 500000000
[    0.680536] clk_set_rate_range_nolock: core req rate 800000000
[    0.686389] clk_set_rate_range_nolock: clamped rate 800000000
[    0.692164] clk_core_set_rate_nolock: rate 800000000
[    0.697153] clk_mux_determine_rate_flags: requested rate 800000000
[    0.703363] clk_mux_determine_rate_flags: current parent sys_pll1
[    0.709487] clk_mux_determine_rate_flags: current parent rate 800000000
[    0.716147] Unable to handle kernel NULL pointer dereference at virtual
address 0000000000000000
[    0.724977] Mem abort info:
[    0.727775]   ESR = 0x96000004
[    0.730835]   EC = 0x25: DABT (current EL), IL = 32 bits
[    0.736177]   SET = 0, FnV = 0
[    0.739239]   EA = 0, S1PTW = 0
[    0.742382]   FSC = 0x04: level 0 translation fault
[    0.747287] Data abort info:
[    0.750172]   ISV = 0, ISS = 0x00000004
[    0.754027]   CM = 0, WnR = 0
[    0.757002] [0000000000000000] user address but active_mm is swapper
[    0.763394] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.768985] Modules linked in:
[    0.772049] CPU: 1 PID: 1 Comm: swapper/0 Not tainted
5.17.0-next-20220331+ #48 8e9d24095c7f6b15f85bc2ad57a5609e219130b9
[    0.782984] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL
(DT) [    0.789985] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS
BTYPE=--) [    0.796985] pc : clk_mux_determine_rate_flags+0x280/0x2cc
[    0.802407] lr : clk_mux_determine_rate_flags+0xf4/0x2cc
[    0.807747] sp : ffff800009ceb590
[    0.811072] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27:
ffff800008eaa038
[    0.818245] x26: ffff8000092fe0b0 x25: ffff000000090000 x24:
ffff000000090000
[    0.825420] x23: 0000000000000000 x22: ffff800008ea95d8 x21:
ffff0000028f4700
[    0.832595] x20: 000000002faf0800 x19: 0000000000000000 x18:
0000000000004590
[    0.839770] x17: 0000000000004570 x16: 0000000000004560 x15:
ffff8000092ff250
[    0.846945] x14: 0000000000000000 x13: 3030303030303030 x12:
3820657461722074
[    0.854120] x11: 6e6572617020746e x10: 6572727563203a73 x9 :
7563203a7367616c
[    0.861295] x8 : 665f657461725f65 x7 : 205d373834393037 x6 :
ffff800009a947c8
[    0.868472] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 :
000000002faf0800
[    0.875645] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 :
ffff8000092fd5b8
[    0.882822] Call trace:
[    0.885273]  clk_mux_determine_rate_flags+0x280/0x2cc
[    0.890347]  clk_mux_determine_rate+0x10/0x20
[    0.894720]  clk_core_determine_round_nolock+0x4c/0xb4
[    0.899882]  clk_core_round_rate_nolock+0x30/0x80
[    0.904607]  clk_core_round_rate_nolock+0x70/0x80
[    0.909334]  clk_hw_round_rate+0x44/0x74
[    0.913270]  clk_factor_round_rate+0x60/0x80
[    0.917557]  clk_core_determine_round_nolock+0x88/0xb4
[    0.922720]  clk_core_round_rate_nolock+0x30/0x80
[    0.927445]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0
[    0.932695]  clk_set_rate_range_nolock+0x234/0x244
[    0.937507]  __clk_put+0x60/0x12c
[    0.940834]  clk_put+0xc/0x1c
[    0.943809]  __set_clk_parents+0x12c/0x244
[    0.947920]  of_clk_set_defaults+0x20/0x50
[    0.952032]  of_clk_add_hw_provider.part.0+0x94/0x120
[    0.957107]  of_clk_add_hw_provider+0x10/0x20
[    0.961482]  imx8mp_clocks_probe+0x3458/0x34d0
[    0.965945]  platform_probe+0x64/0x100
[    0.969707]  call_driver_probe+0x28/0x130
[    0.973732]  really_probe+0x178/0x310
[    0.977409]  __driver_probe_device+0xfc/0x144
[    0.981782]  driver_probe_device+0x38/0x12c
[    0.985982]  __driver_attach+0xcc/0x220
[    0.989834]  bus_for_each_dev+0x6c/0xc0
[    0.993682]  driver_attach+0x20/0x2c
[    0.997270]  bus_add_driver+0x140/0x230
[    1.001120]  driver_register+0x74/0x120
[    1.004970]  __platform_driver_register+0x24/0x30
[    1.009697]  imx8mp_clk_driver_init+0x18/0x20
[    1.014070]  do_one_initcall+0x58/0x200
[    1.017920]  do_initcalls+0x164/0x19c
[    1.021597]  kernel_init_freeable+0x134/0x17c
[    1.025970]  kernel_init+0x2c/0x150
[    1.029470]  ret_from_fork+0x10/0x20
[    1.033065] Code: f9000f94 912982c1 b0002900 9116e000 (f9400262)
[    1.039188] ---[ end trace 0000000000000000 ]---
[    1.043869] Kernel panic - not syncing: Attempted to kill init!
exitcode=0x0000000b
[    1.051523] SMP: stopping secondary CPUs
[    1.055467] Kernel Offset: disabled
[    1.058960] CPU features: 0x000,00020009,00001082
[    1.063684] Memory Limit: none
[    1.066748] ---[ end Kernel panic - not syncing: Attempted to kill

init!
exitcode=0x0000000b ]---
---

With the $subject patch reverted and bootable system:
---

[    0.659922] clk_core_set_rate_nolock: rate 1000000000
[    0.662154] clk_core_set_rate_nolock: rounded rate 1000000000
[    0.667932] clk_core_set_rate_nolock: rate 800000000
[    0.672918] clk_core_set_rate_nolock: rounded rate 800000000
[    0.678601] clk_core_set_rate_nolock: rate 500000000
[    0.683592] clk_core_set_rate_nolock: rounded rate 500000000
[    0.689276] clk_core_set_rate_nolock: rate 400000000
[    0.694267] clk_core_set_rate_nolock: rounded rate 400000000
[    0.699980] clk_core_set_rate_nolock: rate 800000000
[    0.704942] clk_core_set_rate_nolock: rounded rate 800000000
[    0.710627] clk_core_set_rate_nolock: rate 393216000
[    0.715611] clk_core_set_rate_nolock: rounded rate 393216000
[    0.721815] clk_core_set_rate_nolock: rate 361267200
[    0.726284] clk_core_set_rate_nolock: rounded rate 361267200
[    0.734097] clk_core_set_rate_nolock: rate 800000000
[    0.736977] clk_core_set_rate_nolock: rounded rate 800000000
[    0.742652] clk_core_set_rate_nolock: rate 400000000
[    0.747645] clk_core_set_rate_nolock: rounded rate 400000000
[    0.754565] clk_core_set_rate_nolock: rate 500000000
[    0.758331] clk_core_set_rate_nolock: rounded rate 500000000
[    0.764688] SoC: i.MX8MP revision 1.1
[    0.767931] clk_core_set_rate_nolock: rate 500000000
[    0.772675] clk_core_set_rate_nolock: rounded rate 500000000
[    0.778354] clk_core_set_rate_nolock: rate 200000000
[    0.783351] clk_core_set_rate_nolock: rounded rate 200000000
[    0.793748] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[    0.798952] 30860000.serial: ttymxc0 at MMIO 0x30860000 (irq = 34,
base_baud = 5000000) is a IMX
[    0.806527] 30880000.serial: ttymxc2 at MMIO 0x30880000 (irq = 35,
base_baud = 5000000) is a IMX
[    0.815329] 30890000.serial: ttymxc1 at MMIO 0x30890000 (irq = 36,
base_baud = 5000000) is a IMX
[    0.824176] 30a60000.serial: ttymxc3 at MMIO 0x30a60000 (irq = 43,
base_baud = 1500000) is a IMX
[    0.832588] printk: console [ttymxc3] enabled
[    0.832588] printk: console [ttymxc3] enabled
[    0.841244] printk: bootconsole [ec_imx6q0] disabled
[    0.841244] printk: bootconsole [ec_imx6q0] disabled
[    0.857871] clk_core_set_rate_nolock: rate 80000000
[    0.862796] clk_core_set_rate_nolock: rounded rate 80000000
[    0.868469] clk_core_set_rate_nolock: rate 20000000
[    0.873364] clk_core_set_rate_nolock: rounded rate 20000000
[    0.879258] clk_core_set_rate_nolock: rate 80000000
[    0.884154] clk_core_set_rate_nolock: rounded rate 80000000
Quoted 7 lines

[...]
---

The 500000000 and 800000000 look a bit like the assigned-clock-rates for clock-controller@30380000 in arch/arm64/boot/dts/freescale/imx8mp.dtsi

If you need some more information, do not hesitate to ask

Thanks a lot to you three for all your testing. I think I know what might be going on:

We use the last requested rate on clk_set_rate_range (clk_core.req_rate), and that requested rate if the clock is orphan will be set to 0, so if we were to call clk_set_rate_range before the parent clock is registered, we would effectively call a clk_set_rate to 0

And the assigned-clocks stuff is handled by __set_clk_parents and __set_clk_rates, called by of_clk_set_defaults(), in turn called by of_clk_init and of_clk_add_provider. Both __set_clk_parents and __set_clk_rates will call clk_put once done with the clock, and we will with this patch trigger the clk_set_rate to 0 I mentioned before.

So we just became very good at triggering the underlying issue :)

And I think it's that while we update the requested rate when the missing parent is registered, we never do when we mux away from it using clk_set_parent.

Could you test the following patch and let me know if it works?

--->8---

drivers/clk/clk.c +0 -0
index 91f863b7a824..ee5a0223e47d 100644

drivers/clk/clk.c static int clk_core_set_parent_nolock(struct clk_core

Quoted 10 lines
drivers/clk/clk.c +2 -1
@@ -2599,6 +2599,8 @@ static int clk_core_set_parent_nolock(struct clk_core
 *core, } else {
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
+
+		core->req_rate = core->rate;
 	}
 
 runtime_put:
 
--->8---

Thanks for the patch. Unfortunately it does not help in my case. Here is the output form the other patch. It's the same clock (sys_pll1) but a different rate for the first calls.

---

[    0.658706] clk_set_rate_range_nolock: core req rate 800000000                                                                    
[    0.661715] clk_set_rate_range_nolock: clamped rate 800000000                                                                     
[    0.667496] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.672485] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.678171] clk_set_rate_range_nolock: core req rate 800000000                                                                    
[    0.684033] clk_set_rate_range_nolock: clamped rate 800000000                                                                     
[    0.689804] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.694795] clk_mux_determine_rate_flags: requested rate 800000000                                                                
[    0.701005] clk_mux_determine_rate_flags: current parent sys_pll1                                                                 
[    0.707130] clk_mux_determine_rate_flags: current parent rate 800000000                                                           
[    0.713789] Unable to handle kernel NULL pointer dereference at virtual 
address 0000000000000000                                  
[    0.722618] Mem abort info:                                                                                                       
[    0.725411]   ESR = 0x96000004                                                                                                    
[    0.728483]   EC = 0x25: DABT (current EL), IL = 32 bits                                                                          
[    0.733822]   SET = 0, FnV = 0                                                                                                    
[    0.736879]   EA = 0, S1PTW = 0                                                                                                   
[    0.740032]   FSC = 0x04: level 0 translation fault
[    0.744930] Data abort info:
[    0.747820]   ISV = 0, ISS = 0x00000004
[    0.751666]   CM = 0, WnR = 0
[    0.754645] [0000000000000000] user address but active_mm is swapper
[    0.761034] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.766627] Modules linked in:
[    0.769690] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.17.0-next-20220331+ 
#51 62b66a4e908a3493b6d37735830c8ba1462e0fc9
[    0.780625] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)
[    0.787627] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    0.794625] pc : clk_mux_determine_rate_flags+0x280/0x2cc
[    0.800049] lr : clk_mux_determine_rate_flags+0xf4/0x2cc
[    0.805388] sp : ffff800009ceb590
[    0.808713] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27: 
ffff800008eaa038
[    0.815886] x26: ffff8000092fe090 x25: ffff000000090000 x24: 
ffff000000090000
[    0.823061] x23: 0000000000000000 x22: ffff800008ea95d8 x21: 
ffff000002aab700
[    0.830236] x20: 000000002faf0800 x19: 0000000000000000 x18: 
0000000000004590
[    0.837411] x17: 0000000000004570 x16: 0000000000004560 x15: 
ffff8000092ff230
[    0.844586] x14: 0000000000000000 x13: 3030303030303030 x12: 
3820657461722074
[    0.851761] x11: 6e6572617020746e x10: 6572727563203a73 x9 : 
7563203a7367616c
[    0.858936] x8 : 665f657461725f65 x7 : 205d303331373037 x6 : 
ffff800009a947c8
[    0.866113] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 : 
000000002faf0800
[    0.873286] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 : 
ffff8000092fd598
[    0.880464] Call trace:
[    0.882914]  clk_mux_determine_rate_flags+0x280/0x2cc
[    0.887989]  clk_mux_determine_rate+0x10/0x20
[    0.892361]  clk_core_determine_round_nolock+0x4c/0xb4
[    0.897524]  clk_core_round_rate_nolock+0x30/0x80
[    0.902249]  clk_core_round_rate_nolock+0x70/0x80
[    0.906976]  clk_hw_round_rate+0x44/0x74
[    0.910911]  clk_factor_round_rate+0x60/0x80
[    0.915199]  clk_core_determine_round_nolock+0x88/0xb4
[    0.920361]  clk_core_round_rate_nolock+0x30/0x80
[    0.925086]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0
[    0.930336]  clk_set_rate_range_nolock+0x234/0x244
[    0.935149]  __clk_put+0x60/0x12c
[    0.938474]  clk_put+0xc/0x1c
[    0.941451]  __set_clk_parents+0x12c/0x244
[    0.945561]  of_clk_set_defaults+0x20/0x50
[    0.949674]  of_clk_add_hw_provider.part.0+0x94/0x120
[    0.954750]  of_clk_add_hw_provider+0x10/0x20
[    0.959124]  imx8mp_clocks_probe+0x3458/0x34d0
[    0.963586]  platform_probe+0x64/0x100
[    0.967349]  call_driver_probe+0x28/0x130
[    0.971374]  really_probe+0x178/0x310
[    0.975051]  __driver_probe_device+0xfc/0x144
[    0.979424]  driver_probe_device+0x38/0x12c
[    0.983624]  __driver_attach+0xcc/0x220
[    0.987476]  bus_for_each_dev+0x6c/0xc0
[    0.991324]  driver_attach+0x20/0x2c
[    0.994911]  bus_add_driver+0x140/0x230
[    0.998761]  driver_register+0x74/0x120
[    1.002611]  __platform_driver_register+0x24/0x30
[    1.007338]  imx8mp_clk_driver_init+0x18/0x20
[    1.011711]  do_one_initcall+0x58/0x200
[    1.015561]  do_initcalls+0x164/0x19c
[    1.019238]  kernel_init_freeable+0x134/0x17c
[    1.023613]  kernel_init+0x2c/0x150
[    1.027111]  ret_from_fork+0x10/0x20
[    1.030705] Code: f9000f94 912982c1 d0002900 91166000 (f9400262) 
[    1.036829] ---[ end trace 0000000000000000 ]---
[    1.041512] Kernel panic - not syncing: Attempted to kill init! 
exitcode=0x0000000b
[    1.049164] SMP: stopping secondary CPUs
[    1.053108] Kernel Offset: disabled
[    1.056600] CPU features: 0x000,00020009,00001082
[    1.061326] Memory Limit: none
[    1.064390] ---[ end Kernel panic - not syncing: Attempted to kill init! 

exitcode=0x0000000b ]---

Best regards,
Alexander

#20 lore raw

Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 335 lines · Alexander Stein

Hi Maxime,

Am Freitag, 1. April 2022, 14:27:36 CEST schrieb Maxime Ripard:

Quoted 220 lines · Alexander Stein
Quoted 182 lines

Am Donnerstag, 31. März 2022, 17:31:34 CEST schrieb Maxime Ripard:

Quoted 24 lines
Quoted 1 line

Old Signed by an unknown key

Hi Tony,

Quoted 19 lines · Tony Lindgren
Quoted 8 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on
both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 4 lines

It looks like the gpt1_fck driver might favor another parent for that
rate, which, if it's an invalid configuration, shouldn't really happen?

Quoted 5 lines
Quoted 2 lines

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

I'm affected by this patch as well on an imx8mp platform (see [1] for some details)

In the failing case with with your patch applied I get the following error
---

[    0.661064] clk_set_rate_range_nolock: core req rate 500000000
[    0.664084] clk_set_rate_range_nolock: clamped rate 500000000
[    0.669851] clk_core_set_rate_nolock: rate 500000000
[    0.674843] clk_core_set_rate_nolock: rounded rate 500000000
[    0.680536] clk_set_rate_range_nolock: core req rate 800000000
[    0.686389] clk_set_rate_range_nolock: clamped rate 800000000
[    0.692164] clk_core_set_rate_nolock: rate 800000000
[    0.697153] clk_mux_determine_rate_flags: requested rate 800000000
[    0.703363] clk_mux_determine_rate_flags: current parent sys_pll1
[    0.709487] clk_mux_determine_rate_flags: current parent rate 800000000
[    0.716147] Unable to handle kernel NULL pointer dereference at virtual
address 0000000000000000
[    0.724977] Mem abort info:
[    0.727775]   ESR = 0x96000004
[    0.730835]   EC = 0x25: DABT (current EL), IL = 32 bits
[    0.736177]   SET = 0, FnV = 0
[    0.739239]   EA = 0, S1PTW = 0
[    0.742382]   FSC = 0x04: level 0 translation fault
[    0.747287] Data abort info:
[    0.750172]   ISV = 0, ISS = 0x00000004
[    0.754027]   CM = 0, WnR = 0
[    0.757002] [0000000000000000] user address but active_mm is swapper
[    0.763394] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.768985] Modules linked in:
[    0.772049] CPU: 1 PID: 1 Comm: swapper/0 Not tainted
5.17.0-next-20220331+ #48 8e9d24095c7f6b15f85bc2ad57a5609e219130b9
[    0.782984] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL
(DT) [    0.789985] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS
BTYPE=--) [    0.796985] pc : clk_mux_determine_rate_flags+0x280/0x2cc
[    0.802407] lr : clk_mux_determine_rate_flags+0xf4/0x2cc
[    0.807747] sp : ffff800009ceb590
[    0.811072] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27:
ffff800008eaa038
[    0.818245] x26: ffff8000092fe0b0 x25: ffff000000090000 x24:
ffff000000090000
[    0.825420] x23: 0000000000000000 x22: ffff800008ea95d8 x21:
ffff0000028f4700
[    0.832595] x20: 000000002faf0800 x19: 0000000000000000 x18:
0000000000004590
[    0.839770] x17: 0000000000004570 x16: 0000000000004560 x15:
ffff8000092ff250
[    0.846945] x14: 0000000000000000 x13: 3030303030303030 x12:
3820657461722074
[    0.854120] x11: 6e6572617020746e x10: 6572727563203a73 x9 :
7563203a7367616c
[    0.861295] x8 : 665f657461725f65 x7 : 205d373834393037 x6 :
ffff800009a947c8
[    0.868472] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 :
000000002faf0800
[    0.875645] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 :
ffff8000092fd5b8
[    0.882822] Call trace:
[    0.885273]  clk_mux_determine_rate_flags+0x280/0x2cc
[    0.890347]  clk_mux_determine_rate+0x10/0x20
[    0.894720]  clk_core_determine_round_nolock+0x4c/0xb4
[    0.899882]  clk_core_round_rate_nolock+0x30/0x80
[    0.904607]  clk_core_round_rate_nolock+0x70/0x80
[    0.909334]  clk_hw_round_rate+0x44/0x74
[    0.913270]  clk_factor_round_rate+0x60/0x80
[    0.917557]  clk_core_determine_round_nolock+0x88/0xb4
[    0.922720]  clk_core_round_rate_nolock+0x30/0x80
[    0.927445]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0
[    0.932695]  clk_set_rate_range_nolock+0x234/0x244
[    0.937507]  __clk_put+0x60/0x12c
[    0.940834]  clk_put+0xc/0x1c
[    0.943809]  __set_clk_parents+0x12c/0x244
[    0.947920]  of_clk_set_defaults+0x20/0x50
[    0.952032]  of_clk_add_hw_provider.part.0+0x94/0x120
[    0.957107]  of_clk_add_hw_provider+0x10/0x20
[    0.961482]  imx8mp_clocks_probe+0x3458/0x34d0
[    0.965945]  platform_probe+0x64/0x100
[    0.969707]  call_driver_probe+0x28/0x130
[    0.973732]  really_probe+0x178/0x310
[    0.977409]  __driver_probe_device+0xfc/0x144
[    0.981782]  driver_probe_device+0x38/0x12c
[    0.985982]  __driver_attach+0xcc/0x220
[    0.989834]  bus_for_each_dev+0x6c/0xc0
[    0.993682]  driver_attach+0x20/0x2c
[    0.997270]  bus_add_driver+0x140/0x230
[    1.001120]  driver_register+0x74/0x120
[    1.004970]  __platform_driver_register+0x24/0x30
[    1.009697]  imx8mp_clk_driver_init+0x18/0x20
[    1.014070]  do_one_initcall+0x58/0x200
[    1.017920]  do_initcalls+0x164/0x19c
[    1.021597]  kernel_init_freeable+0x134/0x17c
[    1.025970]  kernel_init+0x2c/0x150
[    1.029470]  ret_from_fork+0x10/0x20
[    1.033065] Code: f9000f94 912982c1 b0002900 9116e000 (f9400262)
[    1.039188] ---[ end trace 0000000000000000 ]---
[    1.043869] Kernel panic - not syncing: Attempted to kill init!
exitcode=0x0000000b
[    1.051523] SMP: stopping secondary CPUs
[    1.055467] Kernel Offset: disabled
[    1.058960] CPU features: 0x000,00020009,00001082
[    1.063684] Memory Limit: none
[    1.066748] ---[ end Kernel panic - not syncing: Attempted to kill

init!
exitcode=0x0000000b ]---
---

With the $subject patch reverted and bootable system:
---

[    0.659922] clk_core_set_rate_nolock: rate 1000000000
[    0.662154] clk_core_set_rate_nolock: rounded rate 1000000000
[    0.667932] clk_core_set_rate_nolock: rate 800000000
[    0.672918] clk_core_set_rate_nolock: rounded rate 800000000
[    0.678601] clk_core_set_rate_nolock: rate 500000000
[    0.683592] clk_core_set_rate_nolock: rounded rate 500000000
[    0.689276] clk_core_set_rate_nolock: rate 400000000
[    0.694267] clk_core_set_rate_nolock: rounded rate 400000000
[    0.699980] clk_core_set_rate_nolock: rate 800000000
[    0.704942] clk_core_set_rate_nolock: rounded rate 800000000
[    0.710627] clk_core_set_rate_nolock: rate 393216000
[    0.715611] clk_core_set_rate_nolock: rounded rate 393216000
[    0.721815] clk_core_set_rate_nolock: rate 361267200
[    0.726284] clk_core_set_rate_nolock: rounded rate 361267200
[    0.734097] clk_core_set_rate_nolock: rate 800000000
[    0.736977] clk_core_set_rate_nolock: rounded rate 800000000
[    0.742652] clk_core_set_rate_nolock: rate 400000000
[    0.747645] clk_core_set_rate_nolock: rounded rate 400000000
[    0.754565] clk_core_set_rate_nolock: rate 500000000
[    0.758331] clk_core_set_rate_nolock: rounded rate 500000000
[    0.764688] SoC: i.MX8MP revision 1.1
[    0.767931] clk_core_set_rate_nolock: rate 500000000
[    0.772675] clk_core_set_rate_nolock: rounded rate 500000000
[    0.778354] clk_core_set_rate_nolock: rate 200000000
[    0.783351] clk_core_set_rate_nolock: rounded rate 200000000
[    0.793748] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[    0.798952] 30860000.serial: ttymxc0 at MMIO 0x30860000 (irq = 34,
base_baud = 5000000) is a IMX
[    0.806527] 30880000.serial: ttymxc2 at MMIO 0x30880000 (irq = 35,
base_baud = 5000000) is a IMX
[    0.815329] 30890000.serial: ttymxc1 at MMIO 0x30890000 (irq = 36,
base_baud = 5000000) is a IMX
[    0.824176] 30a60000.serial: ttymxc3 at MMIO 0x30a60000 (irq = 43,
base_baud = 1500000) is a IMX
[    0.832588] printk: console [ttymxc3] enabled
[    0.832588] printk: console [ttymxc3] enabled
[    0.841244] printk: bootconsole [ec_imx6q0] disabled
[    0.841244] printk: bootconsole [ec_imx6q0] disabled
[    0.857871] clk_core_set_rate_nolock: rate 80000000
[    0.862796] clk_core_set_rate_nolock: rounded rate 80000000
[    0.868469] clk_core_set_rate_nolock: rate 20000000
[    0.873364] clk_core_set_rate_nolock: rounded rate 20000000
[    0.879258] clk_core_set_rate_nolock: rate 80000000
[    0.884154] clk_core_set_rate_nolock: rounded rate 80000000
Quoted 7 lines

[...]
---

The 500000000 and 800000000 look a bit like the assigned-clock-rates for clock-controller@30380000 in arch/arm64/boot/dts/freescale/imx8mp.dtsi

If you need some more information, do not hesitate to ask

Thanks a lot to you three for all your testing. I think I know what might be going on:

We use the last requested rate on clk_set_rate_range (clk_core.req_rate), and that requested rate if the clock is orphan will be set to 0, so if we were to call clk_set_rate_range before the parent clock is registered, we would effectively call a clk_set_rate to 0

And the assigned-clocks stuff is handled by __set_clk_parents and __set_clk_rates, called by of_clk_set_defaults(), in turn called by of_clk_init and of_clk_add_provider. Both __set_clk_parents and __set_clk_rates will call clk_put once done with the clock, and we will with this patch trigger the clk_set_rate to 0 I mentioned before.

So we just became very good at triggering the underlying issue :)

And I think it's that while we update the requested rate when the missing parent is registered, we never do when we mux away from it using clk_set_parent.

Could you test the following patch and let me know if it works?

--->8---

drivers/clk/clk.c +0 -0
index 91f863b7a824..ee5a0223e47d 100644

drivers/clk/clk.c static int clk_core_set_parent_nolock(struct clk_core

Quoted 10 lines
drivers/clk/clk.c +2 -1
@@ -2599,6 +2599,8 @@ static int clk_core_set_parent_nolock(struct clk_core
 *core, } else {
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
+
+		core->req_rate = core->rate;
 	}
 
 runtime_put:
 
--->8---

Thanks for the patch. Unfortunately it does not help in my case. Here is the output form the other patch. It's the same clock (sys_pll1) but a different rate for the first calls.

---

[    0.658706] clk_set_rate_range_nolock: core req rate 800000000                                                                    
[    0.661715] clk_set_rate_range_nolock: clamped rate 800000000                                                                     
[    0.667496] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.672485] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.678171] clk_set_rate_range_nolock: core req rate 800000000                                                                    
[    0.684033] clk_set_rate_range_nolock: clamped rate 800000000                                                                     
[    0.689804] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.694795] clk_mux_determine_rate_flags: requested rate 800000000                                                                
[    0.701005] clk_mux_determine_rate_flags: current parent sys_pll1                                                                 
[    0.707130] clk_mux_determine_rate_flags: current parent rate 800000000                                                           
[    0.713789] Unable to handle kernel NULL pointer dereference at virtual 
address 0000000000000000                                  
[    0.722618] Mem abort info:                                                                                                       
[    0.725411]   ESR = 0x96000004                                                                                                    
[    0.728483]   EC = 0x25: DABT (current EL), IL = 32 bits                                                                          
[    0.733822]   SET = 0, FnV = 0                                                                                                    
[    0.736879]   EA = 0, S1PTW = 0                                                                                                   
[    0.740032]   FSC = 0x04: level 0 translation fault
[    0.744930] Data abort info:
[    0.747820]   ISV = 0, ISS = 0x00000004
[    0.751666]   CM = 0, WnR = 0
[    0.754645] [0000000000000000] user address but active_mm is swapper
[    0.761034] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.766627] Modules linked in:
[    0.769690] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.17.0-next-20220331+ 
#51 62b66a4e908a3493b6d37735830c8ba1462e0fc9
[    0.780625] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)
[    0.787627] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    0.794625] pc : clk_mux_determine_rate_flags+0x280/0x2cc
[    0.800049] lr : clk_mux_determine_rate_flags+0xf4/0x2cc
[    0.805388] sp : ffff800009ceb590
[    0.808713] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27: 
ffff800008eaa038
[    0.815886] x26: ffff8000092fe090 x25: ffff000000090000 x24: 
ffff000000090000
[    0.823061] x23: 0000000000000000 x22: ffff800008ea95d8 x21: 
ffff000002aab700
[    0.830236] x20: 000000002faf0800 x19: 0000000000000000 x18: 
0000000000004590
[    0.837411] x17: 0000000000004570 x16: 0000000000004560 x15: 
ffff8000092ff230
[    0.844586] x14: 0000000000000000 x13: 3030303030303030 x12: 
3820657461722074
[    0.851761] x11: 6e6572617020746e x10: 6572727563203a73 x9 : 
7563203a7367616c
[    0.858936] x8 : 665f657461725f65 x7 : 205d303331373037 x6 : 
ffff800009a947c8
[    0.866113] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 : 
000000002faf0800
[    0.873286] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 : 
ffff8000092fd598
[    0.880464] Call trace:
[    0.882914]  clk_mux_determine_rate_flags+0x280/0x2cc
[    0.887989]  clk_mux_determine_rate+0x10/0x20
[    0.892361]  clk_core_determine_round_nolock+0x4c/0xb4
[    0.897524]  clk_core_round_rate_nolock+0x30/0x80
[    0.902249]  clk_core_round_rate_nolock+0x70/0x80
[    0.906976]  clk_hw_round_rate+0x44/0x74
[    0.910911]  clk_factor_round_rate+0x60/0x80
[    0.915199]  clk_core_determine_round_nolock+0x88/0xb4
[    0.920361]  clk_core_round_rate_nolock+0x30/0x80
[    0.925086]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0
[    0.930336]  clk_set_rate_range_nolock+0x234/0x244
[    0.935149]  __clk_put+0x60/0x12c
[    0.938474]  clk_put+0xc/0x1c
[    0.941451]  __set_clk_parents+0x12c/0x244
[    0.945561]  of_clk_set_defaults+0x20/0x50
[    0.949674]  of_clk_add_hw_provider.part.0+0x94/0x120
[    0.954750]  of_clk_add_hw_provider+0x10/0x20
[    0.959124]  imx8mp_clocks_probe+0x3458/0x34d0
[    0.963586]  platform_probe+0x64/0x100
[    0.967349]  call_driver_probe+0x28/0x130
[    0.971374]  really_probe+0x178/0x310
[    0.975051]  __driver_probe_device+0xfc/0x144
[    0.979424]  driver_probe_device+0x38/0x12c
[    0.983624]  __driver_attach+0xcc/0x220
[    0.987476]  bus_for_each_dev+0x6c/0xc0
[    0.991324]  driver_attach+0x20/0x2c
[    0.994911]  bus_add_driver+0x140/0x230
[    0.998761]  driver_register+0x74/0x120
[    1.002611]  __platform_driver_register+0x24/0x30
[    1.007338]  imx8mp_clk_driver_init+0x18/0x20
[    1.011711]  do_one_initcall+0x58/0x200
[    1.015561]  do_initcalls+0x164/0x19c
[    1.019238]  kernel_init_freeable+0x134/0x17c
[    1.023613]  kernel_init+0x2c/0x150
[    1.027111]  ret_from_fork+0x10/0x20
[    1.030705] Code: f9000f94 912982c1 d0002900 91166000 (f9400262) 
[    1.036829] ---[ end trace 0000000000000000 ]---
[    1.041512] Kernel panic - not syncing: Attempted to kill init! 
exitcode=0x0000000b
[    1.049164] SMP: stopping secondary CPUs
[    1.053108] Kernel Offset: disabled
[    1.056600] CPU features: 0x000,00020009,00001082
[    1.061326] Memory Limit: none
[    1.064390] ---[ end Kernel panic - not syncing: Attempted to kill init! 
Quoted 1 line

exitcode=0x0000000b ]---

Does it also happen if you only apply the patch I had above, and not all the debugging?

Maxime

#21 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Am Freitag, 1. April 2022, 15:04:42 CEST schrieb Maxime Ripard:

Quoted 357 lines

* PGP Signed by an unknown key

Quoted 351 lines · Alexander Stein

Hi Maxime,

Am Freitag, 1. April 2022, 14:27:36 CEST schrieb Maxime Ripard:

Quoted 236 lines · Alexander Stein
Quoted 194 lines

Am Donnerstag, 31. März 2022, 17:31:34 CEST schrieb Maxime Ripard:

Quoted 27 lines
Quoted 1 line

Old Signed by an unknown key

Hi Tony,

Quoted 22 lines · Tony Lindgren
Quoted 9 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put
on
both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has
two options. No idea why, but if you have some debug patch in mind I
can give it a try.

Quoted 4 lines

It looks like the gpt1_fck driver might favor another parent for that
rate, which, if it's an invalid configuration, shouldn't really happen?

Quoted 7 lines
Quoted 3 lines

Hmm there's a gate clock and a mux clock, there's not really a rate
selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is
doing the heavy lifting, could you run your test with

I'm affected by this patch as well on an imx8mp platform (see [1] for some
details)

In the failing case with with your patch applied I get the following error
---

[    0.661064] clk_set_rate_range_nolock: core req rate 500000000
[    0.664084] clk_set_rate_range_nolock: clamped rate 500000000
[    0.669851] clk_core_set_rate_nolock: rate 500000000
[    0.674843] clk_core_set_rate_nolock: rounded rate 500000000
[    0.680536] clk_set_rate_range_nolock: core req rate 800000000
[    0.686389] clk_set_rate_range_nolock: clamped rate 800000000
[    0.692164] clk_core_set_rate_nolock: rate 800000000
[    0.697153] clk_mux_determine_rate_flags: requested rate 800000000
[    0.703363] clk_mux_determine_rate_flags: current parent sys_pll1
[    0.709487] clk_mux_determine_rate_flags: current parent rate
800000000
[    0.716147] Unable to handle kernel NULL pointer dereference at
virtual
address 0000000000000000
[    0.724977] Mem abort info:
[    0.727775]   ESR = 0x96000004
[    0.730835]   EC = 0x25: DABT (current EL), IL = 32 bits
[    0.736177]   SET = 0, FnV = 0
[    0.739239]   EA = 0, S1PTW = 0
[    0.742382]   FSC = 0x04: level 0 translation fault
[    0.747287] Data abort info:
[    0.750172]   ISV = 0, ISS = 0x00000004
[    0.754027]   CM = 0, WnR = 0
[    0.757002] [0000000000000000] user address but active_mm is
swapper
[    0.763394] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.768985] Modules linked in:
[    0.772049] CPU: 1 PID: 1 Comm: swapper/0 Not tainted
5.17.0-next-20220331+ #48 8e9d24095c7f6b15f85bc2ad57a5609e219130b9
[    0.782984] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on

MBa8MPxL
(DT) [ 0.789985] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT
-SSBS
BTYPE=--) [ 0.796985] pc : clk_mux_determine_rate_flags+0x280/0x2cc

[    0.802407] lr : clk_mux_determine_rate_flags+0xf4/0x2cc
[    0.807747] sp : ffff800009ceb590
[    0.811072] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27:
ffff800008eaa038
[    0.818245] x26: ffff8000092fe0b0 x25: ffff000000090000 x24:
ffff000000090000
[    0.825420] x23: 0000000000000000 x22: ffff800008ea95d8 x21:
ffff0000028f4700
[    0.832595] x20: 000000002faf0800 x19: 0000000000000000 x18:
0000000000004590
[    0.839770] x17: 0000000000004570 x16: 0000000000004560 x15:
ffff8000092ff250
[    0.846945] x14: 0000000000000000 x13: 3030303030303030 x12:
3820657461722074
[    0.854120] x11: 6e6572617020746e x10: 6572727563203a73 x9 :
7563203a7367616c
[    0.861295] x8 : 665f657461725f65 x7 : 205d373834393037 x6 :
ffff800009a947c8
[    0.868472] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 :
000000002faf0800
[    0.875645] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 :
ffff8000092fd5b8
[    0.882822] Call trace:
[    0.885273]  clk_mux_determine_rate_flags+0x280/0x2cc
[    0.890347]  clk_mux_determine_rate+0x10/0x20
[    0.894720]  clk_core_determine_round_nolock+0x4c/0xb4
[    0.899882]  clk_core_round_rate_nolock+0x30/0x80
[    0.904607]  clk_core_round_rate_nolock+0x70/0x80
[    0.909334]  clk_hw_round_rate+0x44/0x74
[    0.913270]  clk_factor_round_rate+0x60/0x80
[    0.917557]  clk_core_determine_round_nolock+0x88/0xb4
[    0.922720]  clk_core_round_rate_nolock+0x30/0x80
[    0.927445]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0
[    0.932695]  clk_set_rate_range_nolock+0x234/0x244
[    0.937507]  __clk_put+0x60/0x12c
[    0.940834]  clk_put+0xc/0x1c
[    0.943809]  __set_clk_parents+0x12c/0x244
[    0.947920]  of_clk_set_defaults+0x20/0x50
[    0.952032]  of_clk_add_hw_provider.part.0+0x94/0x120
[    0.957107]  of_clk_add_hw_provider+0x10/0x20
[    0.961482]  imx8mp_clocks_probe+0x3458/0x34d0
[    0.965945]  platform_probe+0x64/0x100
[    0.969707]  call_driver_probe+0x28/0x130
[    0.973732]  really_probe+0x178/0x310
[    0.977409]  __driver_probe_device+0xfc/0x144
[    0.981782]  driver_probe_device+0x38/0x12c
[    0.985982]  __driver_attach+0xcc/0x220
[    0.989834]  bus_for_each_dev+0x6c/0xc0
[    0.993682]  driver_attach+0x20/0x2c
[    0.997270]  bus_add_driver+0x140/0x230
[    1.001120]  driver_register+0x74/0x120
[    1.004970]  __platform_driver_register+0x24/0x30
[    1.009697]  imx8mp_clk_driver_init+0x18/0x20
[    1.014070]  do_one_initcall+0x58/0x200
[    1.017920]  do_initcalls+0x164/0x19c
[    1.021597]  kernel_init_freeable+0x134/0x17c
[    1.025970]  kernel_init+0x2c/0x150
[    1.029470]  ret_from_fork+0x10/0x20
[    1.033065] Code: f9000f94 912982c1 b0002900 9116e000 (f9400262)
[    1.039188] ---[ end trace 0000000000000000 ]---
[    1.043869] Kernel panic - not syncing: Attempted to kill init!
exitcode=0x0000000b
[    1.051523] SMP: stopping secondary CPUs
[    1.055467] Kernel Offset: disabled
[    1.058960] CPU features: 0x000,00020009,00001082
[    1.063684] Memory Limit: none
[    1.066748] ---[ end Kernel panic - not syncing: Attempted to kill

init!
exitcode=0x0000000b ]---
---

With the $subject patch reverted and bootable system:
---

[    0.659922] clk_core_set_rate_nolock: rate 1000000000
[    0.662154] clk_core_set_rate_nolock: rounded rate 1000000000
[    0.667932] clk_core_set_rate_nolock: rate 800000000
[    0.672918] clk_core_set_rate_nolock: rounded rate 800000000
[    0.678601] clk_core_set_rate_nolock: rate 500000000
[    0.683592] clk_core_set_rate_nolock: rounded rate 500000000
[    0.689276] clk_core_set_rate_nolock: rate 400000000
[    0.694267] clk_core_set_rate_nolock: rounded rate 400000000
[    0.699980] clk_core_set_rate_nolock: rate 800000000
[    0.704942] clk_core_set_rate_nolock: rounded rate 800000000
[    0.710627] clk_core_set_rate_nolock: rate 393216000
[    0.715611] clk_core_set_rate_nolock: rounded rate 393216000
[    0.721815] clk_core_set_rate_nolock: rate 361267200
[    0.726284] clk_core_set_rate_nolock: rounded rate 361267200
[    0.734097] clk_core_set_rate_nolock: rate 800000000
[    0.736977] clk_core_set_rate_nolock: rounded rate 800000000
[    0.742652] clk_core_set_rate_nolock: rate 400000000
[    0.747645] clk_core_set_rate_nolock: rounded rate 400000000
[    0.754565] clk_core_set_rate_nolock: rate 500000000
[    0.758331] clk_core_set_rate_nolock: rounded rate 500000000
[    0.764688] SoC: i.MX8MP revision 1.1
[    0.767931] clk_core_set_rate_nolock: rate 500000000
[    0.772675] clk_core_set_rate_nolock: rounded rate 500000000
[    0.778354] clk_core_set_rate_nolock: rate 200000000
[    0.783351] clk_core_set_rate_nolock: rounded rate 200000000
[    0.793748] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[    0.798952] 30860000.serial: ttymxc0 at MMIO 0x30860000 (irq = 34,
base_baud = 5000000) is a IMX
[    0.806527] 30880000.serial: ttymxc2 at MMIO 0x30880000 (irq = 35,
base_baud = 5000000) is a IMX
[    0.815329] 30890000.serial: ttymxc1 at MMIO 0x30890000 (irq = 36,
base_baud = 5000000) is a IMX
[    0.824176] 30a60000.serial: ttymxc3 at MMIO 0x30a60000 (irq = 43,
base_baud = 1500000) is a IMX
[    0.832588] printk: console [ttymxc3] enabled
[    0.832588] printk: console [ttymxc3] enabled
[    0.841244] printk: bootconsole [ec_imx6q0] disabled
[    0.841244] printk: bootconsole [ec_imx6q0] disabled
[    0.857871] clk_core_set_rate_nolock: rate 80000000
[    0.862796] clk_core_set_rate_nolock: rounded rate 80000000
[    0.868469] clk_core_set_rate_nolock: rate 20000000
[    0.873364] clk_core_set_rate_nolock: rounded rate 20000000
[    0.879258] clk_core_set_rate_nolock: rate 80000000
[    0.884154] clk_core_set_rate_nolock: rounded rate 80000000
Quoted 8 lines

[...]
---

The 500000000 and 800000000 look a bit like the assigned-clock-rates for
clock-controller@30380000 in arch/arm64/boot/dts/freescale/imx8mp.dtsi

If you need some more information, do not hesitate to ask

Thanks a lot to you three for all your testing. I think I know what might be going on:

We use the last requested rate on clk_set_rate_range (clk_core.req_rate), and that requested rate if the clock is orphan will be set to 0, so if we were to call clk_set_rate_range before the parent clock is registered, we would effectively call a clk_set_rate to 0

And the assigned-clocks stuff is handled by __set_clk_parents and __set_clk_rates, called by of_clk_set_defaults(), in turn called by of_clk_init and of_clk_add_provider. Both __set_clk_parents and __set_clk_rates will call clk_put once done with the clock, and we will with this patch trigger the clk_set_rate to 0 I mentioned before.

So we just became very good at triggering the underlying issue :)

And I think it's that while we update the requested rate when the missing parent is registered, we never do when we mux away from it using clk_set_parent.

Could you test the following patch and let me know if it works?

--->8---

drivers/clk/clk.c +0 -0
index 91f863b7a824..ee5a0223e47d 100644
@@ -2599,6 +2599,8 @@ static int clk_core_set_parent_nolock(struct

clk_core
*core, } else {

drivers/clk/clk.c static int clk_core_set_parent_nolock(struct clk_core *core,

Quoted 10 lines
drivers/clk/clk.c +2 -1
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
 
+
+		core->req_rate = core->rate;
 
 	}
 
 runtime_put:
--->8---

Thanks for the patch. Unfortunately it does not help in my case. Here is the output form the other patch. It's the same clock (sys_pll1) but a different rate for the first calls.

---

[    0.658706] clk_set_rate_range_nolock: core req rate 800000000
[    0.661715] clk_set_rate_range_nolock: clamped rate 800000000
[    0.667496] clk_core_set_rate_nolock: rate 800000000
[    0.672485] clk_core_set_rate_nolock: rounded rate 800000000
[    0.678171] clk_set_rate_range_nolock: core req rate 800000000
[    0.684033] clk_set_rate_range_nolock: clamped rate 800000000
[    0.689804] clk_core_set_rate_nolock: rate 800000000
[    0.694795] clk_mux_determine_rate_flags: requested rate 800000000
[    0.701005] clk_mux_determine_rate_flags: current parent sys_pll1
[    0.707130] clk_mux_determine_rate_flags: current parent rate 800000000
[    0.713789] Unable to handle kernel NULL pointer dereference at virtual
address 0000000000000000
[    0.722618] Mem abort info:
[    0.725411]   ESR = 0x96000004
[    0.728483]   EC = 0x25: DABT (current EL), IL = 32 bits
[    0.733822]   SET = 0, FnV = 0
[    0.736879]   EA = 0, S1PTW = 0
[    0.740032]   FSC = 0x04: level 0 translation fault
[    0.744930] Data abort info:
[    0.747820]   ISV = 0, ISS = 0x00000004
[    0.751666]   CM = 0, WnR = 0
[    0.754645] [0000000000000000] user address but active_mm is swapper
[    0.761034] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.766627] Modules linked in:
[    0.769690] CPU: 2 PID: 1 Comm: swapper/0 Not tainted
5.17.0-next-20220331+ #51 62b66a4e908a3493b6d37735830c8ba1462e0fc9
[    0.780625] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL
(DT) [    0.787627] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS
BTYPE=--) [    0.794625] pc : clk_mux_determine_rate_flags+0x280/0x2cc
[    0.800049] lr : clk_mux_determine_rate_flags+0xf4/0x2cc
[    0.805388] sp : ffff800009ceb590
[    0.808713] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27:
ffff800008eaa038
[    0.815886] x26: ffff8000092fe090 x25: ffff000000090000 x24:
ffff000000090000
[    0.823061] x23: 0000000000000000 x22: ffff800008ea95d8 x21:
ffff000002aab700
[    0.830236] x20: 000000002faf0800 x19: 0000000000000000 x18:
0000000000004590
[    0.837411] x17: 0000000000004570 x16: 0000000000004560 x15:
ffff8000092ff230
[    0.844586] x14: 0000000000000000 x13: 3030303030303030 x12:
3820657461722074
[    0.851761] x11: 6e6572617020746e x10: 6572727563203a73 x9 :
7563203a7367616c
[    0.858936] x8 : 665f657461725f65 x7 : 205d303331373037 x6 :
ffff800009a947c8
[    0.866113] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 :
000000002faf0800
[    0.873286] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 :
ffff8000092fd598
[    0.880464] Call trace:
[    0.882914]  clk_mux_determine_rate_flags+0x280/0x2cc
[    0.887989]  clk_mux_determine_rate+0x10/0x20
[    0.892361]  clk_core_determine_round_nolock+0x4c/0xb4
[    0.897524]  clk_core_round_rate_nolock+0x30/0x80
[    0.902249]  clk_core_round_rate_nolock+0x70/0x80
[    0.906976]  clk_hw_round_rate+0x44/0x74
[    0.910911]  clk_factor_round_rate+0x60/0x80
[    0.915199]  clk_core_determine_round_nolock+0x88/0xb4
[    0.920361]  clk_core_round_rate_nolock+0x30/0x80
[    0.925086]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0
[    0.930336]  clk_set_rate_range_nolock+0x234/0x244
[    0.935149]  __clk_put+0x60/0x12c
[    0.938474]  clk_put+0xc/0x1c
[    0.941451]  __set_clk_parents+0x12c/0x244
[    0.945561]  of_clk_set_defaults+0x20/0x50
[    0.949674]  of_clk_add_hw_provider.part.0+0x94/0x120
[    0.954750]  of_clk_add_hw_provider+0x10/0x20
[    0.959124]  imx8mp_clocks_probe+0x3458/0x34d0
[    0.963586]  platform_probe+0x64/0x100
[    0.967349]  call_driver_probe+0x28/0x130
[    0.971374]  really_probe+0x178/0x310
[    0.975051]  __driver_probe_device+0xfc/0x144
[    0.979424]  driver_probe_device+0x38/0x12c
[    0.983624]  __driver_attach+0xcc/0x220
[    0.987476]  bus_for_each_dev+0x6c/0xc0
[    0.991324]  driver_attach+0x20/0x2c
[    0.994911]  bus_add_driver+0x140/0x230
[    0.998761]  driver_register+0x74/0x120
[    1.002611]  __platform_driver_register+0x24/0x30
[    1.007338]  imx8mp_clk_driver_init+0x18/0x20
[    1.011711]  do_one_initcall+0x58/0x200
[    1.015561]  do_initcalls+0x164/0x19c
[    1.019238]  kernel_init_freeable+0x134/0x17c
[    1.023613]  kernel_init+0x2c/0x150
[    1.027111]  ret_from_fork+0x10/0x20
[    1.030705] Code: f9000f94 912982c1 d0002900 91166000 (f9400262)
[    1.036829] ---[ end trace 0000000000000000 ]---
[    1.041512] Kernel panic - not syncing: Attempted to kill init!
exitcode=0x0000000b
[    1.049164] SMP: stopping secondary CPUs
[    1.053108] Kernel Offset: disabled
[    1.056600] CPU features: 0x000,00020009,00001082
[    1.061326] Memory Limit: none
[    1.064390] ---[ end Kernel panic - not syncing: Attempted to kill
Quoted 2 lines

init!
exitcode=0x0000000b ]---

Quoted 2 lines

Does it also happen if you only apply the patch I had above, and not all the debugging?

Yes, these are the last lines I see:
---

[    1.236306] mmcblk0rpmb: mmc0:0001 DA6016 4.00 MiB, chardev (235:0)                                                               
[    1.241031] i2c i2c-1: IMX I2C adapter registered                                                                                 
[    1.251771] i2c i2c-3: IMX I2C adapter registered                                                                                 
[    1.256957] i2c i2c-5: IMX I2C adapter registered

---

Alexander

#22 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 9 lines · Alexander Stein
Quoted 2 lines

Does it also happen if you only apply the patch I had above, and not all the debugging?

Yes, these are the last lines I see:
---

[    1.236306] mmcblk0rpmb: mmc0:0001 DA6016 4.00 MiB, chardev (235:0)                                                               
[    1.241031] i2c i2c-1: IMX I2C adapter registered                                                                                 
[    1.251771] i2c i2c-3: IMX I2C adapter registered                                                                                 
[    1.256957] i2c i2c-5: IMX I2C adapter registered

Could you add on top of next (so dropping everything we did so far)

---- >8 -----

drivers/clk/clk.c +72 -3
index 91f863b7a824..552b1e16a82d 100644
@@ -540,6 +540,8 @@ static bool mux_is_better_rate(unsigned long rate, unsigned long now,
 	if (flags & CLK_MUX_ROUND_CLOSEST)
 		return abs(now - rate) < abs(best - rate);
 
+	pr_crit("%s +%d rate %lu now %lu best %lu\n", __func__, __LINE__, rate, now, best);
+
 	return now <= rate && now > best;
 }
 
@@ -552,6 +554,12 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	unsigned long best = 0;
 	struct clk_rate_request parent_req = *req;
 
+	pr_crit("%s: %s: requested rate %lu\n", __func__, core->name, req->rate);
+
+	parent = core->parent;
+	pr_crit("%s: %s: current parent %s\n", __func__, core->name, parent ? parent->name : "(null)");
+	pr_crit("%s: %s: current parent rate %lu\n", __func__, core->name, clk_core_get_rate_nolock(parent));
+
 	/* if NO_REPARENT flag set, pass through to current parent */
 	if (core->flags & CLK_SET_RATE_NO_REPARENT) {
 		parent = core->parent;
@@ -578,24 +586,37 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 		if (!parent)
 			continue;
 
+		pr_crit("%s: Trying parent %s (%lu)\n",
+			__func__,
+			parent->name,
+			clk_core_get_rate_nolock(parent));
+
 		if (core->flags & CLK_SET_RATE_PARENT) {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			parent_req = *req;
 			ret = __clk_determine_rate(parent->hw, &parent_req);
+			pr_crit("%s +%d %d\n", __func__, __LINE__, ret);
 			if (ret)
 				continue;
 		} else {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			parent_req.rate = clk_core_get_rate_nolock(parent);
 		}
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
+
 		if (mux_is_better_rate(req->rate, parent_req.rate,
 				       best, flags)) {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			best_parent = parent;
 			best = parent_req.rate;
 		}
 	}
 
-	if (!best_parent)
+	if (!best_parent) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return -EINVAL;
+	}
 
 out:
 	if (best_parent)
@@ -603,6 +624,11 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	req->best_parent_rate = best;
 	req->rate = best;
 
+	pr_crit("%s: Best parent %s (%lu)\n",
+		__func__,
+		best_parent->name,
+		best);
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(clk_mux_determine_rate_flags);
@@ -1345,11 +1371,15 @@ static int clk_core_determine_round_nolock(struct clk_core *core,
 
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d %s\n", __func__, __LINE__, core->name);
 	if (!core)
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
 	req->rate = clamp(req->rate, req->min_rate, req->max_rate);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/*
 	 * At this point, core protection will be disabled
 	 * - if the provider is not protected at all
@@ -1357,10 +1387,13 @@ static int clk_core_determine_round_nolock(struct clk_core *core,
 	 *   over the provider
 	 */
 	if (clk_core_rate_is_protected(core)) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		req->rate = core->rate;
 	} else if (core->ops->determine_rate) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return core->ops->determine_rate(core->hw, req);
 	} else if (core->ops->round_rate) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		rate = core->ops->round_rate(core->hw, req->rate,
 					     &req->best_parent_rate);
 		if (rate < 0)
@@ -1368,6 +1401,7 @@ static int clk_core_determine_round_nolock(struct clk_core *core,
 
 		req->rate = rate;
 	} else {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return -EINVAL;
 	}
 
@@ -1402,17 +1436,26 @@ static int clk_core_round_rate_nolock(struct clk_core *core,
 {
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	if (!core) {
 		req->rate = 0;
 		return 0;
 	}
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	clk_core_init_rate_req(core, req);
 
-	if (clk_core_can_round(core))
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
+	if (clk_core_can_round(core)) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return clk_core_determine_round_nolock(core, req);
-	else if (core->flags & CLK_SET_RATE_PARENT)
+	} else if (core->flags & CLK_SET_RATE_PARENT) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return clk_core_round_rate_nolock(core->parent, req);
+	}
 
 	req->rate = core->rate;
 	return 0;
@@ -2201,21 +2244,31 @@ static int clk_core_set_rate_nolock(struct clk_core *core,
 	if (!core)
 		return 0;
 
+	pr_crit("%s: %s: rate %lu\n", __func__, core->name, req_rate);
+
 	rate = clk_core_req_round_rate_nolock(core, req_rate);
 
+	pr_crit("%s: %s: rounded rate %lu\n", __func__, core->name, req_rate);
+
 	/* bail early if nothing to do */
 	if (rate == clk_core_get_rate_nolock(core))
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/* fail on a direct rate set of a protected provider */
 	if (clk_core_rate_is_protected(core))
 		return -EBUSY;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/* calculate new rates and get the topmost changed clock */
 	top = clk_calc_new_rates(core, req_rate);
 	if (!top)
 		return -EINVAL;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	ret = clk_pm_runtime_get(core);
 	if (ret)
 		return ret;
@@ -2367,6 +2420,16 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 		goto out;
 	}
 
+	pr_crit("%s: %s: orphan ? %c\n",
+		__func__,
+		clk->core->name,
+		clk->core->orphan ? 'y' : 'n');
+
+	pr_crit("%s: %s: core req rate %lu\n",
+		__func__,
+		clk->core->name,
+		clk->core->req_rate);
+
 	/*
 	 * Since the boundaries have been changed, let's give the
 	 * opportunity to the provider to adjust the clock rate based on
@@ -2384,7 +2447,11 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 	 * - the determine_rate() callback does not really check for
 	 *   this corner case when determining the rate
 	 */
+
 	rate = clamp(clk->core->req_rate, min, max);
+
+	pr_crit("%s: %s: clamped rate %lu\n", __func__, clk->core->name, rate);
+
 	ret = clk_core_set_rate_nolock(clk->core, rate);
 	if (ret) {
 		/* rollback the changes */
@@ -2599,6 +2666,8 @@ static int clk_core_set_parent_nolock(struct clk_core *core,
 	} else {
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
+
+		core->req_rate = core->rate;
 	}
 
 runtime_put:

---- >8 -----

Thanks!
Maxime

#23 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Am Freitag, 1. April 2022, 15:34:09 CEST schrieb Maxime Ripard:

Quoted 21 lines

* PGP Signed by an unknown key

Quoted 9 lines · Alexander Stein
Quoted 2 lines

Does it also happen if you only apply the patch I had above, and not all the debugging?

Yes, these are the last lines I see:
---

[    1.236306] mmcblk0rpmb: mmc0:0001 DA6016 4.00 MiB, chardev (235:0)
[    1.241031] i2c i2c-1: IMX I2C adapter registered
[    1.251771] i2c i2c-3: IMX I2C adapter registered
[    1.256957] i2c i2c-5: IMX I2C adapter registered

Could you add on top of next (so dropping everything we did so far)

---- >8 -----

drivers/clk/clk.c +0 -0
index 91f863b7a824..552b1e16a82d 100644

drivers/clk/clk.c static bool mux_is_better_rate(unsigned long rate, · line 543

Quoted 4 lines
drivers/clk/clk.c +1 -0
@@ -540,6 +540,8 @@ static bool mux_is_better_rate(unsigned long rate,
 unsigned long now, if (flags & CLK_MUX_ROUND_CLOSEST)
 		return abs(now - rate) < abs(best - rate);
 
+	pr_crit("%s +%d rate %lu now %lu best %lu\n", __func__, __LINE__, 

rate,

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw, · line 561

Quoted 13 lines

now, best); +

drivers/clk/clk.c +4 -0
 	return now <= rate && now > best;
 }
@@ -552,6 +554,12 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	unsigned long best = 0;
 	struct clk_rate_request parent_req = *req;
 
+	pr_crit("%s: %s: requested rate %lu\n", __func__, core->name, req-
 rate);
+
+	parent = core->parent;
+	pr_crit("%s: %s: current parent %s\n", __func__, core->name, parent 

?

Quoted 8 lines
drivers/clk/clk.c +0 -0
 parent->name : "(null)"); +	pr_crit("%s: %s: current parent rate %lu\n",

__func__, core->name, clk_core_get_rate_nolock(parent)); +

drivers/clk/clk.c +0 -0
 	/* if NO_REPARENT flag set, pass through to current parent */
 	if (core->flags & CLK_SET_RATE_NO_REPARENT) {
 		parent = core->parent;
@@ -578,24 +586,37 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 		if (!parent)
 			continue;

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw, · line 597

Quoted 10 lines
drivers/clk/clk.c +6 -0
 
+		pr_crit("%s: Trying parent %s (%lu)\n",
+			__func__,
+			parent->name,
+			clk_core_get_rate_nolock(parent));
+
 		if (core->flags & CLK_SET_RATE_PARENT) {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			parent_req = *req;
 			ret = __clk_determine_rate(parent->hw, 

&parent_req);

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw, · line 598

Quoted 1 line
drivers/clk/clk.c +1 -0
+			pr_crit("%s +%d %d\n", __func__, __LINE__, 

ret);

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw,

Quoted 5 lines
drivers/clk/clk.c +1 -0
 			if (ret)
 				continue;
 		} else {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			parent_req.rate = 

clk_core_get_rate_nolock(parent);

Quoted 103 lines
drivers/clk/clk.c +15 -1
 		}
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
+
 		if (mux_is_better_rate(req->rate, parent_req.rate,
 				       best, flags)) {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			best_parent = parent;
 			best = parent_req.rate;
 		}
 	}
 
-	if (!best_parent)
+	if (!best_parent) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return -EINVAL;
+	}
 
 out:
 	if (best_parent)
@@ -603,6 +624,11 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	req->best_parent_rate = best;
 	req->rate = best;
 
+	pr_crit("%s: Best parent %s (%lu)\n",
+		__func__,
+		best_parent->name,
+		best);
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(clk_mux_determine_rate_flags);
@@ -1345,11 +1371,15 @@ static int clk_core_determine_round_nolock(struct
 clk_core *core,
 
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d %s\n", __func__, __LINE__, core->name);
 	if (!core)
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
 	req->rate = clamp(req->rate, req->min_rate, req->max_rate);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/*
 	 * At this point, core protection will be disabled
drivers/clk/clk.c +0 -0
 	 * - if the provider is not protected at all
@@ -1357,10 +1387,13 @@ static int clk_core_determine_round_nolock(struct

clk_core *core, * over the provider

drivers/clk/clk.c +3 -0
 	 */
 	if (clk_core_rate_is_protected(core)) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		req->rate = core->rate;
 	} else if (core->ops->determine_rate) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return core->ops->determine_rate(core->hw, req);
 	} else if (core->ops->round_rate) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		rate = core->ops->round_rate(core->hw, req->rate,
 					     &req-

best_parent_rate);

drivers/clk/clk.c +12 -2
 		if (rate < 0)
@@ -1368,6 +1401,7 @@ static int clk_core_determine_round_nolock(struct
 clk_core *core,
 
 		req->rate = rate;
 	} else {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return -EINVAL;
 	}
@@ -1402,17 +1436,26 @@ static int clk_core_round_rate_nolock(struct
 clk_core *core, {
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	if (!core) {
 		req->rate = 0;
 		return 0;
 	}
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	clk_core_init_rate_req(core, req);
 
-	if (clk_core_can_round(core))
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
+	if (clk_core_can_round(core)) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return clk_core_determine_round_nolock(core, req);
-	else if (core->flags & CLK_SET_RATE_PARENT)
+	} else if (core->flags & CLK_SET_RATE_PARENT) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return clk_core_round_rate_nolock(core->parent, req);
+	}
 
 	req->rate = core->rate;
 	return 0;

drivers/clk/clk.c static int clk_core_set_rate_nolock(struct clk_core · line 2251

Quoted 8 lines
drivers/clk/clk.c +3 -0
@@ -2201,21 +2244,31 @@ static int clk_core_set_rate_nolock(struct clk_core
 *core, if (!core)
 		return 0;
 
+	pr_crit("%s: %s: rate %lu\n", __func__, core->name, req_rate);
+
 	rate = clk_core_req_round_rate_nolock(core, req_rate);
 
+	pr_crit("%s: %s: rounded rate %lu\n", __func__, core->name, 

req_rate);

Quoted 41 lines
drivers/clk/clk.c +17 -0
+
 	/* bail early if nothing to do */
 	if (rate == clk_core_get_rate_nolock(core))
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/* fail on a direct rate set of a protected provider */
 	if (clk_core_rate_is_protected(core))
 		return -EBUSY;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/* calculate new rates and get the topmost changed clock */
 	top = clk_calc_new_rates(core, req_rate);
 	if (!top)
 		return -EINVAL;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	ret = clk_pm_runtime_get(core);
 	if (ret)
 		return ret;
@@ -2367,6 +2420,16 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 goto out;
 	}
 
+	pr_crit("%s: %s: orphan ? %c\n",
+		__func__,
+		clk->core->name,
+		clk->core->orphan ? 'y' : 'n');
+
+	pr_crit("%s: %s: core req rate %lu\n",
+		__func__,
+		clk->core->name,
+		clk->core->req_rate);
+
 	/*
 	 * Since the boundaries have been changed, let's give the
 	 * opportunity to the provider to adjust the clock rate based on

drivers/clk/clk.c static int clk_set_rate_range_nolock(struct clk *clk, · line 2453

Quoted 7 lines
drivers/clk/clk.c +3 -0
@@ -2384,7 +2447,11 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 * - the determine_rate() callback does not really check for
 	 *   this corner case when determining the rate
 	 */
+
 	rate = clamp(clk->core->req_rate, min, max);
+
+	pr_crit("%s: %s: clamped rate %lu\n", __func__, clk->core->name, 

rate);

Quoted 5 lines
drivers/clk/clk.c +1 -0
+
 	ret = clk_core_set_rate_nolock(clk->core, rate);
 	if (ret) {
 		/* rollback the changes */

drivers/clk/clk.c static int clk_core_set_parent_nolock(struct clk_core

Quoted 9 lines
drivers/clk/clk.c +2 -1
@@ -2599,6 +2666,8 @@ static int clk_core_set_parent_nolock(struct clk_core
 *core, } else {
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
+
+		core->req_rate = core->rate;
 	}
 
 runtime_put:
---- >8 -----

Sure, here we go
---

[    0.630873] Asymmetric key parser 'x509' registered                                                                               
[    0.635802] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 
243)                                                  
[    0.643210] io scheduler mq-deadline registered                                                                                   
[    0.647758] io scheduler kyber registered                                                                                         
[    0.658708] clk_set_rate_range_nolock: arm_a53_div: orphan ? n                                                                    
[    0.661717] clk_set_rate_range_nolock: arm_a53_div: core req rate 800000000                                                       
[    0.668724] clk_set_rate_range_nolock: arm_a53_div: clamped rate 800000000                                                        
[    0.675633] clk_core_set_rate_nolock: arm_a53_div: rate 800000000                                                                 
[    0.681761] clk_core_round_rate_nolock +1439                                                                                      
[    0.686048] clk_core_round_rate_nolock +1446                                                                                      
[    0.690333] clk_core_round_rate_nolock +1450                                                                                      
[    0.694619] clk_core_round_rate_nolock +1453                                                                                      
[    0.698908] clk_core_determine_round_nolock +1374 arm_a53_div                                                                     
[    0.704681] clk_core_determine_round_nolock +1378                                                                                 
[    0.709408] clk_core_determine_round_nolock +1381                                                                                 
[    0.714133] clk_core_determine_round_nolock +1393                                                                                 
[    0.718860] clk_core_set_rate_nolock: arm_a53_div: rounded rate 800000000                                                         
[    0.725684] clk_set_rate_range_nolock: sys_pll1_800m: orphan ? n                                                                  
[    0.731719] clk_set_rate_range_nolock: sys_pll1_800m: core req rate 
800000000                                                     
[    0.738894] clk_set_rate_range_nolock: sys_pll1_800m: clamped rate 
800000000                                                      
[    0.745983] clk_core_set_rate_nolock: sys_pll1_800m: rate 800000000                                                               
[    0.752281] clk_core_round_rate_nolock +1439                                                                                      
[    0.756569] clk_core_round_rate_nolock +1446                                                                                      
[    0.760862] clk_core_round_rate_nolock +1450                                                                                      
[    0.765152] clk_core_round_rate_nolock +1453                                                                                      
[    0.769435] clk_core_determine_round_nolock +1374 sys_pll1_800m                                                                   
[    0.775385] clk_core_determine_round_nolock +1378                                                                                 
[    0.780114] clk_core_determine_round_nolock +1381                                                                                 
[    0.784833] clk_core_determine_round_nolock +1396                                                                                 
[    0.789559] clk_core_round_rate_nolock +1439                                                                                      
[    0.793844] clk_core_round_rate_nolock +1446                                                                                      
[    0.798133] clk_core_round_rate_nolock +1450                                                                                      
[    0.802423] clk_core_round_rate_nolock +1456                                                                                      
[    0.806708] clk_core_round_rate_nolock +1439                                                                                      
[    0.810994] clk_core_round_rate_nolock +1446                                                                                      
[    0.815284] clk_core_round_rate_nolock +1450                                                                                      
[    0.819570] clk_core_round_rate_nolock +1453                                                                                      
[    0.823856] clk_core_determine_round_nolock +1374 sys_pll1_bypass                                                                 
[    0.829981] clk_core_determine_round_nolock +1378                                                                                 
[    0.834706] clk_core_determine_round_nolock +1381                                                                                 
[    0.839431] clk_core_determine_round_nolock +1393                                                                                 
[    0.844159] clk_mux_determine_rate_flags: sys_pll1_bypass: requested rate 
800000000                                               
[    0.851856] clk_mux_determine_rate_flags: sys_pll1_bypass: current parent 
sys_pll1                                                
[    0.859471] clk_mux_determine_rate_flags: sys_pll1_bypass: current parent 
rate 800000000                                          
[    0.867608] clk_core_round_rate_nolock +1439                                                                                      
[    0.871894] clk_core_round_rate_nolock +1446                                                                                      
[    0.876182] clk_core_round_rate_nolock +1450                                                                                      
[    0.880477] clk_core_round_rate_nolock +1453                                                                                      
[    0.884758] clk_core_determine_round_nolock +1374 sys_pll1                                                                        
[    0.890273] clk_core_determine_round_nolock +1378                                                                                 
[    0.894996] clk_core_determine_round_nolock +1381                                                                                 
[    0.899721] clk_core_determine_round_nolock +1396                                                                                 
[    0.904457] Unable to handle kernel NULL pointer dereference at virtual 
address 0000000000000000                                  
[    0.913285] Mem abort info:                                                                                                       
[    0.916083]   ESR = 0x96000004                                                                                                    
[    0.919147]   EC = 0x25: DABT (current EL), IL = 32 bits                                                                          
[    0.924484]   SET = 0, FnV = 0                                                                                                    
[    0.927547]   EA = 0, S1PTW = 0                                                                                                   
[    0.930697]   FSC = 0x04: level 0 translation fault
[    0.935595] Data abort info:
[    0.938487]   ISV = 0, ISS = 0x00000004
[    0.942334]   CM = 0, WnR = 0
[    0.945304] [0000000000000000] user address but active_mm is swapper
[    0.951696] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.957292] Modules linked in:
[    0.960355] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.17.0-next-20220331+ 
#53 da834fe2485dc10e4c2f50265323ce628a30bc5e
[    0.971291] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)
[    0.978292] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    0.985291] pc : clk_mux_determine_rate_flags+0x33c/0x380
[    0.990714] lr : clk_mux_determine_rate_flags+0x10c/0x380
[    0.996141] sp : ffff800009ceb4a0
[    0.999464] x29: ffff800009ceb4a0 x28: ffff000002cf4700 x27: 
0000000000000001
[    1.006639] x26: ffff8000092fe728 x25: ffff800008eaa028 x24: 
ffff800008ea95d8
[    1.013816] x23: ffff800008ea95d8 x22: ffff000002aab700 x21: 
000000002faf0800
[    1.020989] x20: ffff800009ceb640 x19: 0000000000000000 x18: 
0000000000004590
[    1.028164] x17: 617220746e657261 x16: 7020746e65727275 x15: 
63203a7373617079
[    1.035339] x14: 0000000000000000 x13: 363933312b206b63 x12: 
6f6c6f6e5f646e75
[    1.042514] x11: 6f725f656e696d72 x10: 657465645f65726f x9 : 
206b636f6c6f6e5f
[    1.049689] x8 : 646e756f725f656e x7 : 205d313237393938 x6 : 
ffff800009a947c8
[    1.056864] x5 : ffff800008eb0310 x4 : 0000000000000009 x3 : 
000000002faf0800
[    1.064039] x2 : ffff800008eb039c x1 : ffff800008eaa028 x0 : 
ffff8000092fd8b8
[    1.071217] Call trace:
[    1.073667]  clk_mux_determine_rate_flags+0x33c/0x380
[    1.078741]  clk_mux_determine_rate+0x10/0x20
[    1.083115]  clk_core_determine_round_nolock+0xd4/0x140
[    1.088364]  clk_core_round_rate_nolock+0xac/0xf8
[    1.093090]  clk_core_round_rate_nolock+0xd4/0xf8
[    1.097814]  clk_hw_round_rate+0x44/0x7c
[    1.101751]  clk_factor_round_rate+0x60/0x80
[    1.106041]  clk_core_determine_round_nolock+0x104/0x140
[    1.111376]  clk_core_round_rate_nolock+0xac/0xf8
[    1.116101]  clk_core_set_rate_nolock.part.0+0xac/0x21c
[    1.121351]  clk_set_rate_range_nolock+0x294/0x2b0
[    1.126164]  __clk_put+0x60/0x12c
[    1.129489]  clk_put+0xc/0x1c
[    1.132464]  __set_clk_parents+0x12c/0x244
[    1.136576]  of_clk_set_defaults+0x20/0x50
[    1.140691]  of_clk_add_hw_provider.part.0+0x94/0x120
[    1.145764]  of_clk_add_hw_provider+0x10/0x20
[    1.150139]  imx8mp_clocks_probe+0x3458/0x34d0
[    1.154601]  platform_probe+0x64/0x100
[    1.158364]  call_driver_probe+0x28/0x130
[    1.162389]  really_probe+0x178/0x310
[    1.166064]  __driver_probe_device+0xfc/0x144
[    1.170439]  driver_probe_device+0x38/0x12c
[    1.174641]  __driver_attach+0xcc/0x220
[    1.178489]  bus_for_each_dev+0x6c/0xc0
[    1.182339]  driver_attach+0x20/0x2c
[    1.185926]  bus_add_driver+0x140/0x230
[    1.189776]  driver_register+0x74/0x120
[    1.193626]  __platform_driver_register+0x24/0x30
[    1.198351]  imx8mp_clk_driver_init+0x18/0x20
[    1.202726]  do_one_initcall+0x58/0x200
[    1.206578]  do_initcalls+0x164/0x19c
[    1.210251]  kernel_init_freeable+0x134/0x17c
[    1.214628]  kernel_init+0x2c/0x150
[    1.218126]  ret_from_fork+0x10/0x20
[    1.221720] Code: f9000e95 91294301 d0002900 9122e000 (f9400262) 
[    1.227845] ---[ end trace 0000000000000000 ]---
[    1.232501] Kernel panic - not syncing: Attempted to kill init! 
exitcode=0x0000000b
[    1.240179] SMP: stopping secondary CPUs
[    1.244122] Kernel Offset: disabled
[    1.247616] CPU features: 0x000,00020009,00001082
[    1.252341] Memory Limit: none
[    1.255406] ---[ end Kernel panic - not syncing: Attempted to kill init! 

exitcode=0x0000000b ]---
---

HTH
Alexander

#24 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 241 lines · Alexander Stein

Am Freitag, 1. April 2022, 15:34:09 CEST schrieb Maxime Ripard:

Quoted 21 lines

* PGP Signed by an unknown key

Quoted 9 lines · Alexander Stein
Quoted 2 lines

Does it also happen if you only apply the patch I had above, and not all the debugging?

Yes, these are the last lines I see:
---

[    1.236306] mmcblk0rpmb: mmc0:0001 DA6016 4.00 MiB, chardev (235:0)
[    1.241031] i2c i2c-1: IMX I2C adapter registered
[    1.251771] i2c i2c-3: IMX I2C adapter registered
[    1.256957] i2c i2c-5: IMX I2C adapter registered

Could you add on top of next (so dropping everything we did so far)

---- >8 -----

drivers/clk/clk.c +0 -0
index 91f863b7a824..552b1e16a82d 100644

drivers/clk/clk.c static bool mux_is_better_rate(unsigned long rate, · line 543

Quoted 4 lines
drivers/clk/clk.c +1 -0
@@ -540,6 +540,8 @@ static bool mux_is_better_rate(unsigned long rate,
 unsigned long now, if (flags & CLK_MUX_ROUND_CLOSEST)
 		return abs(now - rate) < abs(best - rate);
 
+	pr_crit("%s +%d rate %lu now %lu best %lu\n", __func__, __LINE__, 

rate,

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw, · line 561

Quoted 13 lines

now, best); +

drivers/clk/clk.c +4 -0
 	return now <= rate && now > best;
 }
@@ -552,6 +554,12 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	unsigned long best = 0;
 	struct clk_rate_request parent_req = *req;
 
+	pr_crit("%s: %s: requested rate %lu\n", __func__, core->name, req-
 rate);
+
+	parent = core->parent;
+	pr_crit("%s: %s: current parent %s\n", __func__, core->name, parent 

?

Quoted 8 lines
drivers/clk/clk.c +0 -0
 parent->name : "(null)"); +	pr_crit("%s: %s: current parent rate %lu\n",

__func__, core->name, clk_core_get_rate_nolock(parent)); +

drivers/clk/clk.c +0 -0
 	/* if NO_REPARENT flag set, pass through to current parent */
 	if (core->flags & CLK_SET_RATE_NO_REPARENT) {
 		parent = core->parent;
@@ -578,24 +586,37 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 		if (!parent)
 			continue;

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw, · line 597

Quoted 10 lines
drivers/clk/clk.c +6 -0
 
+		pr_crit("%s: Trying parent %s (%lu)\n",
+			__func__,
+			parent->name,
+			clk_core_get_rate_nolock(parent));
+
 		if (core->flags & CLK_SET_RATE_PARENT) {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			parent_req = *req;
 			ret = __clk_determine_rate(parent->hw, 

&parent_req);

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw, · line 598

Quoted 1 line
drivers/clk/clk.c +1 -0
+			pr_crit("%s +%d %d\n", __func__, __LINE__, 

ret);

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw,

Quoted 5 lines
drivers/clk/clk.c +1 -0
 			if (ret)
 				continue;
 		} else {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			parent_req.rate = 

clk_core_get_rate_nolock(parent);

Quoted 103 lines
drivers/clk/clk.c +15 -1
 		}
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
+
 		if (mux_is_better_rate(req->rate, parent_req.rate,
 				       best, flags)) {
+			pr_crit("%s +%d\n", __func__, __LINE__);
 			best_parent = parent;
 			best = parent_req.rate;
 		}
 	}
 
-	if (!best_parent)
+	if (!best_parent) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return -EINVAL;
+	}
 
 out:
 	if (best_parent)
@@ -603,6 +624,11 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	req->best_parent_rate = best;
 	req->rate = best;
 
+	pr_crit("%s: Best parent %s (%lu)\n",
+		__func__,
+		best_parent->name,
+		best);
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(clk_mux_determine_rate_flags);
@@ -1345,11 +1371,15 @@ static int clk_core_determine_round_nolock(struct
 clk_core *core,
 
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d %s\n", __func__, __LINE__, core->name);
 	if (!core)
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
 	req->rate = clamp(req->rate, req->min_rate, req->max_rate);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/*
 	 * At this point, core protection will be disabled
drivers/clk/clk.c +0 -0
 	 * - if the provider is not protected at all
@@ -1357,10 +1387,13 @@ static int clk_core_determine_round_nolock(struct

clk_core *core, * over the provider

drivers/clk/clk.c +3 -0
 	 */
 	if (clk_core_rate_is_protected(core)) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		req->rate = core->rate;
 	} else if (core->ops->determine_rate) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return core->ops->determine_rate(core->hw, req);
 	} else if (core->ops->round_rate) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		rate = core->ops->round_rate(core->hw, req->rate,
 					     &req-

best_parent_rate);

drivers/clk/clk.c +12 -2
 		if (rate < 0)
@@ -1368,6 +1401,7 @@ static int clk_core_determine_round_nolock(struct
 clk_core *core,
 
 		req->rate = rate;
 	} else {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return -EINVAL;
 	}
@@ -1402,17 +1436,26 @@ static int clk_core_round_rate_nolock(struct
 clk_core *core, {
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	if (!core) {
 		req->rate = 0;
 		return 0;
 	}
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	clk_core_init_rate_req(core, req);
 
-	if (clk_core_can_round(core))
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
+	if (clk_core_can_round(core)) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return clk_core_determine_round_nolock(core, req);
-	else if (core->flags & CLK_SET_RATE_PARENT)
+	} else if (core->flags & CLK_SET_RATE_PARENT) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 		return clk_core_round_rate_nolock(core->parent, req);
+	}
 
 	req->rate = core->rate;
 	return 0;

drivers/clk/clk.c static int clk_core_set_rate_nolock(struct clk_core · line 2251

Quoted 8 lines
drivers/clk/clk.c +3 -0
@@ -2201,21 +2244,31 @@ static int clk_core_set_rate_nolock(struct clk_core
 *core, if (!core)
 		return 0;
 
+	pr_crit("%s: %s: rate %lu\n", __func__, core->name, req_rate);
+
 	rate = clk_core_req_round_rate_nolock(core, req_rate);
 
+	pr_crit("%s: %s: rounded rate %lu\n", __func__, core->name, 

req_rate);

Quoted 41 lines
drivers/clk/clk.c +17 -0
+
 	/* bail early if nothing to do */
 	if (rate == clk_core_get_rate_nolock(core))
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/* fail on a direct rate set of a protected provider */
 	if (clk_core_rate_is_protected(core))
 		return -EBUSY;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	/* calculate new rates and get the topmost changed clock */
 	top = clk_calc_new_rates(core, req_rate);
 	if (!top)
 		return -EINVAL;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 	ret = clk_pm_runtime_get(core);
 	if (ret)
 		return ret;
@@ -2367,6 +2420,16 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 goto out;
 	}
 
+	pr_crit("%s: %s: orphan ? %c\n",
+		__func__,
+		clk->core->name,
+		clk->core->orphan ? 'y' : 'n');
+
+	pr_crit("%s: %s: core req rate %lu\n",
+		__func__,
+		clk->core->name,
+		clk->core->req_rate);
+
 	/*
 	 * Since the boundaries have been changed, let's give the
 	 * opportunity to the provider to adjust the clock rate based on

drivers/clk/clk.c static int clk_set_rate_range_nolock(struct clk *clk, · line 2453

Quoted 7 lines
drivers/clk/clk.c +3 -0
@@ -2384,7 +2447,11 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 * - the determine_rate() callback does not really check for
 	 *   this corner case when determining the rate
 	 */
+
 	rate = clamp(clk->core->req_rate, min, max);
+
+	pr_crit("%s: %s: clamped rate %lu\n", __func__, clk->core->name, 

rate);

Quoted 12 lines
drivers/clk/clk.c +3 -0
+
 	ret = clk_core_set_rate_nolock(clk->core, rate);
 	if (ret) {
 		/* rollback the changes */
@@ -2599,6 +2666,8 @@ static int clk_core_set_parent_nolock(struct clk_core
 *core, } else {
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
+
+		core->req_rate = core->rate;
 	}
 
Quoted 2 lines

drivers/clk/clk.c static int clk_core_set_parent_nolock(struct clk_core

Quoted 2 lines
drivers/clk/clk.c +0 -1
 runtime_put:
---- >8 -----

So, let's try to follow this through:

Quoted 9 lines

Sure, here we go
---

[    0.630873] Asymmetric key parser 'x509' registered
[    0.635802] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 243)
[    0.643210] io scheduler mq-deadline registered
[    0.647758] io scheduler kyber registered
[    0.658708] clk_set_rate_range_nolock: arm_a53_div: orphan ? n
[    0.661717] clk_set_rate_range_nolock: arm_a53_div: core req rate 800000000
[    0.668724] clk_set_rate_range_nolock: arm_a53_div: clamped rate 800000000

I'm assuming we hit the assigned-clock-parents in the clocks node, and we try to reparent arm_a53_div / IMX8MP_CLK_A53_SRC to sys_pll1_800m

I'm not entirely sure, but it looks like the arm_a53_div is a gate + divider, so that it has the same rate than its parent makes sens, and 800MHz for a CPU clock also makes sense.

It's also not an orphan, so it's likely to be a separate issue from Tony (and thus the fix doesn't help, sorry).

Quoted 1 line
[    0.675633] clk_core_set_rate_nolock: arm_a53_div: rate 800000000

Now, we set the rate to the same rate, this still makes sense.

Quoted 5 lines
[    0.681761] clk_core_round_rate_nolock +1439
[    0.686048] clk_core_round_rate_nolock +1446
[    0.690333] clk_core_round_rate_nolock +1450
[    0.694619] clk_core_round_rate_nolock +1453
[    0.698908] clk_core_determine_round_nolock +1374 arm_a53_div

The clock has a round_rate / determine_rate implementation (clk_divider_round_rate, most likely), thus we call clk_core_determine_round_nolock()

Quoted 3 lines
[    0.704681] clk_core_determine_round_nolock +1378
[    0.709408] clk_core_determine_round_nolock +1381
[    0.714133] clk_core_determine_round_nolock +1393

Still on the right path, we use clk_divider_determine_rate (too bad :)), it updates the rate

Quoted 1 line
[    0.718860] clk_core_set_rate_nolock: arm_a53_div: rounded rate 800000000

But it didn't change, good. The rounded clock hasn't changed, clk_core_set_rate_nolock returns, everything's great.

Quoted 4 lines
[    0.725684] clk_set_rate_range_nolock: sys_pll1_800m: orphan ? n
[    0.731719] clk_set_rate_range_nolock: sys_pll1_800m: core req rate 800000000
[    0.738894] clk_set_rate_range_nolock: sys_pll1_800m: clamped rate 800000000
[    0.745983] clk_core_set_rate_nolock: sys_pll1_800m: rate 800000000

Then, __set_clk_parents calls clk_put() on the new parent, sys_pll1_800m, still not an orphan, still with a rate that makes sense.

Quoted 5 lines
[    0.752281] clk_core_round_rate_nolock +1439
[    0.756569] clk_core_round_rate_nolock +1446
[    0.760862] clk_core_round_rate_nolock +1450
[    0.765152] clk_core_round_rate_nolock +1453
[    0.769435] clk_core_determine_round_nolock +1374 sys_pll1_800m

We still can round the rate, so we go to
clk_core_determine_round_nolock()

Quoted 3 lines
[    0.775385] clk_core_determine_round_nolock +1378
[    0.780114] clk_core_determine_round_nolock +1381
[    0.784833] clk_core_determine_round_nolock +1396

But this time using a round_rate implementation: clk_factor_round_rate (since sys_pll1_800m is a "pure" fixed factor clock). It has the flag CLK_SET_RATE_PARENT (set in imx_clk_hw_fixed_factor), so clk_factor_round_rate calls clk_hw_round_rate on its parent (sys_pll1_out) for the same rate since it has a factor of 1.

Quoted 4 lines
[    0.789559] clk_core_round_rate_nolock +1439
[    0.793844] clk_core_round_rate_nolock +1446
[    0.798133] clk_core_round_rate_nolock +1450
[    0.802423] clk_core_round_rate_nolock +1456

We go through another round_rate cycle here, for sys_pll1_out. It can't modify the rate (since it's a gate) but it has CLK_SET_RATE_PARENT, so the rate rounding is forwarded to its parent: sys_pll1_bypass.

Quoted 4 lines
[    0.806708] clk_core_round_rate_nolock +1439
[    0.810994] clk_core_round_rate_nolock +1446
[    0.815284] clk_core_round_rate_nolock +1450
[    0.819570] clk_core_round_rate_nolock +1453

We go through it, and call clk_core_determine_round_nolock again for sys_pll1_bypass.

Quoted 1 line
[    0.823856] clk_core_determine_round_nolock +1374 sys_pll1_bypass

Makes total sense so far.

Quoted 4 lines
[    0.829981] clk_core_determine_round_nolock +1378
[    0.834706] clk_core_determine_round_nolock +1381
[    0.839431] clk_core_determine_round_nolock +1393
[    0.844159] clk_mux_determine_rate_flags: sys_pll1_bypass: requested rate 800000000

The requested rate does too. We still have our 800MHz.

Quoted 2 lines
[    0.851856] clk_mux_determine_rate_flags: sys_pll1_bypass: current parent sys_pll1
[    0.859471] clk_mux_determine_rate_flags: sys_pll1_bypass: current parent rate 800000000

sys_pll1_bypass has CLK_SET_RATE_NO_REPARENT (set by __imx_clk_hw_mux) and CLK_SET_RATE_PARENT (set by the driver when registering the clock), so clk_mux_determine_rate_flags will call __clk_determine_rate on its
parent: sys_pll1. __clk_determine_rate then calls clk_core_round_rate_nolock.

Quoted 4 lines
[    0.867608] clk_core_round_rate_nolock +1439
[    0.871894] clk_core_round_rate_nolock +1446
[    0.876182] clk_core_round_rate_nolock +1450
[    0.880477] clk_core_round_rate_nolock +1453

We call clk_core_determine_round_nolock on sys_pll1

Quoted 4 lines
[    0.884758] clk_core_determine_round_nolock +1374 sys_pll1
[    0.890273] clk_core_determine_round_nolock +1378
[    0.894996] clk_core_determine_round_nolock +1381
[    0.899721] clk_core_determine_round_nolock +1396

sys_pll1 is a clk_pll14xx driver, it has a PLL_1416X type and a rate table, so it will use clk_pll1416x_ops. It has a round_rate implementation, clk_pll14xx_round_rate, that doesn't seem to be doing anything out of the ordinary. My assumption would be that it succeeds and returns a proper rate.

Quoted 45 lines
[    0.904457] Unable to handle kernel NULL pointer dereference at virtual 
address 0000000000000000
[    0.913285] Mem abort info:
[    0.916083]   ESR = 0x96000004
[    0.919147]   EC = 0x25: DABT (current EL), IL = 32 bits
[    0.924484]   SET = 0, FnV = 0
[    0.927547]   EA = 0, S1PTW = 0
[    0.930697]   FSC = 0x04: level 0 translation fault
[    0.935595] Data abort info:
[    0.938487]   ISV = 0, ISS = 0x00000004
[    0.942334]   CM = 0, WnR = 0
[    0.945304] [0000000000000000] user address but active_mm is swapper
[    0.951696] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.957292] Modules linked in:
[    0.960355] CPU: 2 PID: 1 Comm: swapper/0 Not tainted 5.17.0-next-20220331+
#53 da834fe2485dc10e4c2f50265323ce628a30bc5e
[    0.971291] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)
[    0.978292] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    0.985291] pc : clk_mux_determine_rate_flags+0x33c/0x380
[    0.990714] lr : clk_mux_determine_rate_flags+0x10c/0x380
[    0.996141] sp : ffff800009ceb4a0
[    0.999464] x29: ffff800009ceb4a0 x28: ffff000002cf4700 x27:
0000000000000001
[    1.006639] x26: ffff8000092fe728 x25: ffff800008eaa028 x24:
ffff800008ea95d8
[    1.013816] x23: ffff800008ea95d8 x22: ffff000002aab700 x21:
000000002faf0800
[    1.020989] x20: ffff800009ceb640 x19: 0000000000000000 x18:
0000000000004590
[    1.028164] x17: 617220746e657261 x16: 7020746e65727275 x15:
63203a7373617079
[    1.035339] x14: 0000000000000000 x13: 363933312b206b63 x12:
6f6c6f6e5f646e75
[    1.042514] x11: 6f725f656e696d72 x10: 657465645f65726f x9 :
206b636f6c6f6e5f
[    1.049689] x8 : 646e756f725f656e x7 : 205d313237393938 x6 :
ffff800009a947c8
[    1.056864] x5 : ffff800008eb0310 x4 : 0000000000000009 x3 :
000000002faf0800
[    1.064039] x2 : ffff800008eb039c x1 : ffff800008eaa028 x0 :
ffff8000092fd8b8
[    1.071217] Call trace:
[    1.073667]  clk_mux_determine_rate_flags+0x33c/0x380
[    1.078741]  clk_mux_determine_rate+0x10/0x20
[    1.083115]  clk_core_determine_round_nolock+0xd4/0x140
Quoted 8 lines
[    1.088364]  clk_core_round_rate_nolock+0xac/0xf8
[    1.093090]  clk_core_round_rate_nolock+0xd4/0xf8
[    1.097814]  clk_hw_round_rate+0x44/0x7c
[    1.101751]  clk_factor_round_rate+0x60/0x80
[    1.106041]  clk_core_determine_round_nolock+0x104/0x140
[    1.111376]  clk_core_round_rate_nolock+0xac/0xf8
[    1.116101]  clk_core_set_rate_nolock.part.0+0xac/0x21c
[    1.121351]  clk_set_rate_range_nolock+0x294/0x2b0

But then, where does this come from?

I'm not entirely sure, but the walk up the clock tree is sane to me. Could you run

./scripts/faddr2line vmlinux 'clk_mux_determine_rate_flags+0x33c/0x380'

in your kernel compilation directory? (with ARCH and CROSS_COMPILE set if you're doing cross-compilation)?

My guess would be that we uncovered some other bug there, but I'm not sure what exactly.

Maxime

_______________________________________________
linux-amlogic mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-amlogic

#25 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hello Maxime,

Am Freitag, 1. April 2022, 16:55:02 CEST schrieb Maxime Ripard:

Quoted 545 lines

* PGP Signed by an unknown key

Quoted 346 lines · Alexander Stein

Am Freitag, 1. April 2022, 15:34:09 CEST schrieb Maxime Ripard:

Quoted 22 lines
Quoted 1 line

Old Signed by an unknown key

Quoted 10 lines · Alexander Stein
Quoted 3 lines

Does it also happen if you only apply the patch I had above, and not all
the debugging?

Yes, these are the last lines I see:
---

[    1.236306] mmcblk0rpmb: mmc0:0001 DA6016 4.00 MiB, chardev (235:0)
[    1.241031] i2c i2c-1: IMX I2C adapter registered
[    1.251771] i2c i2c-3: IMX I2C adapter registered
[    1.256957] i2c i2c-5: IMX I2C adapter registered

Could you add on top of next (so dropping everything we did so far)

---- >8 -----

drivers/clk/clk.c +0 -0
index 91f863b7a824..552b1e16a82d 100644

drivers/clk/clk.c static bool mux_is_better_rate(unsigned long rate, · line 544

Quoted 5 lines
drivers/clk/clk.c +1 -0
@@ -540,6 +540,8 @@ static bool mux_is_better_rate(unsigned long rate,
 unsigned long now, if (flags & CLK_MUX_ROUND_CLOSEST)
 
 		return abs(now - rate) < abs(best - rate);
 
+	pr_crit("%s +%d rate %lu now %lu best %lu\n", __func__, __LINE__,

rate,

Quoted 8 lines

now, best); +

drivers/clk/clk.c +0 -0
 	return now <= rate && now > best;
 
 }
@@ -552,6 +554,12 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw *hw,

Quoted 10 lines
drivers/clk/clk.c +4 -0
 	unsigned long best = 0;
 	struct clk_rate_request parent_req = *req;
 
+	pr_crit("%s: %s: requested rate %lu\n", __func__, core->name, req-
 
 rate);
 
+
+	parent = core->parent;
+	pr_crit("%s: %s: current parent %s\n", __func__, core->name, parent

?

Quoted 17 lines

parent->name : "(null)"); + pr_crit("%s: %s: current parent rate %lu\n",
__func__, core->name, clk_core_get_rate_nolock(parent)); +

drivers/clk/clk.c +2 -0
 	/* if NO_REPARENT flag set, pass through to current parent */
 	if (core->flags & CLK_SET_RATE_NO_REPARENT) {
 
 		parent = core->parent;
@@ -578,24 +586,37 @@ int clk_mux_determine_rate_flags(struct clk_hw
 *hw,
 
 		if (!parent)
 
 			continue;
 
+		pr_crit("%s: Trying parent %s (%lu)\n",
+			__func__,

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw · line 603

Quoted 10 lines
drivers/clk/clk.c +4 -0
+			parent->name,
+			clk_core_get_rate_nolock(parent));
+
 
 		if (core->flags & CLK_SET_RATE_PARENT) {
 
+			pr_crit("%s +%d\n", __func__, __LINE__);
 
 			parent_req = *req;
 			ret = __clk_determine_rate(parent->hw,

&parent_req);

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw · line 604

Quoted 1 line
drivers/clk/clk.c +1 -0
+			pr_crit("%s +%d %d\n", __func__, __LINE__,

ret);

drivers/clk/clk.c int clk_mux_determine_rate_flags(struct clk_hw · line 613

Quoted 9 lines
drivers/clk/clk.c +1 -0
 			if (ret)
 
 				continue;
 
 		} else {
 
+			pr_crit("%s +%d\n", __func__, __LINE__);
 
 			parent_req.rate =

clk_core_get_rate_nolock(parent);

Quoted 155 lines
drivers/clk/clk.c +11 -1
 		}
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
+
 
 		if (mux_is_better_rate(req->rate, parent_req.rate,
 
 				       best, flags)) {
 
+			pr_crit("%s +%d\n", __func__, __LINE__);
 
 			best_parent = parent;
 			best = parent_req.rate;
 
 		}
 
 	}
 
-	if (!best_parent)
+	if (!best_parent) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 
 		return -EINVAL;
 
+	}
 
 out:
 	if (best_parent)
@@ -603,6 +624,11 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 
 	req->best_parent_rate = best;
 	req->rate = best;
 
+	pr_crit("%s: Best parent %s (%lu)\n",
+		__func__,
+		best_parent->name,
+		best);
+
 
 	return 0;

}
EXPORT_SYMBOL_GPL(clk_mux_determine_rate_flags);

drivers/clk/clk.c +0 -0
@@ -1345,11 +1371,15 @@ static int

clk_core_determine_round_nolock(struct
clk_core *core,

drivers/clk/clk.c +4 -0
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d %s\n", __func__, __LINE__, core->name);
 
 	if (!core)
 
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
 
 	req->rate = clamp(req->rate, req->min_rate, req->max_rate);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 
 	/*
 
 	 * At this point, core protection will be disabled
 	 * - if the provider is not protected at all
@@ -1357,10 +1387,13 @@ static int

clk_core_determine_round_nolock(struct
clk_core *core, * over the provider

drivers/clk/clk.c +3 -0
 	 */
 
 	if (clk_core_rate_is_protected(core)) {
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
 
 		req->rate = core->rate;
 
 	} else if (core->ops->determine_rate) {
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
 
 		return core->ops->determine_rate(core->hw, req);
 
 	} else if (core->ops->round_rate) {
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
 
 		rate = core->ops->round_rate(core->hw, req->rate,
 
 					     &req-

best_parent_rate);

drivers/clk/clk.c +1 -0
 		if (rate < 0)
@@ -1368,6 +1401,7 @@ static int clk_core_determine_round_nolock(struct
 clk_core *core,
 
 		req->rate = rate;
 
 	} else {
 
+		pr_crit("%s +%d\n", __func__, __LINE__);
drivers/clk/clk.c +11 -2
 		return -EINVAL;
 
 	}
@@ -1402,17 +1436,26 @@ static int clk_core_round_rate_nolock(struct
 clk_core *core, {
 
 	lockdep_assert_held(&prepare_lock);
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 
 	if (!core) {
 
 		req->rate = 0;
 		return 0;
 
 	}
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 
 	clk_core_init_rate_req(core, req);
 
-	if (clk_core_can_round(core))
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
+	if (clk_core_can_round(core)) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 
 		return clk_core_determine_round_nolock(core, req);
 
-	else if (core->flags & CLK_SET_RATE_PARENT)
+	} else if (core->flags & CLK_SET_RATE_PARENT) {
+		pr_crit("%s +%d\n", __func__, __LINE__);
 
 		return clk_core_round_rate_nolock(core->parent, req);
 
+	}
 
 	req->rate = core->rate;
 	return 0;
@@ -2201,21 +2244,31 @@ static int clk_core_set_rate_nolock(struct

clk_core
*core, if (!core)

drivers/clk/clk.c

Quoted 8 lines
drivers/clk/clk.c +3 -0
 		return 0;
 
+	pr_crit("%s: %s: rate %lu\n", __func__, core->name, req_rate);
+
 
 	rate = clk_core_req_round_rate_nolock(core, req_rate);
 
+	pr_crit("%s: %s: rounded rate %lu\n", __func__, core->name,

req_rate);

Quoted 55 lines
drivers/clk/clk.c +17 -0
+
 
 	/* bail early if nothing to do */
 	if (rate == clk_core_get_rate_nolock(core))
 
 		return 0;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 
 	/* fail on a direct rate set of a protected provider */
 	if (clk_core_rate_is_protected(core))
 
 		return -EBUSY;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 
 	/* calculate new rates and get the topmost changed clock */
 	top = clk_calc_new_rates(core, req_rate);
 	if (!top)
 
 		return -EINVAL;
 
+	pr_crit("%s +%d\n", __func__, __LINE__);
+
 
 	ret = clk_pm_runtime_get(core);
 	if (ret)
 
 		return ret;
@@ -2367,6 +2420,16 @@ static int clk_set_rate_range_nolock(struct clk
 *clk, goto out;
 
 	}
 
+	pr_crit("%s: %s: orphan ? %c\n",
+		__func__,
+		clk->core->name,
+		clk->core->orphan ? 'y' : 'n');
+
+	pr_crit("%s: %s: core req rate %lu\n",
+		__func__,
+		clk->core->name,
+		clk->core->req_rate);
+
 
 	/*
drivers/clk/clk.c +0 -0
 	 * Since the boundaries have been changed, let's give the
 	 * opportunity to the provider to adjust the clock rate based on
@@ -2384,7 +2447,11 @@ static int clk_set_rate_range_nolock(struct clk
 *clk, * - the determine_rate() callback does not really check for

drivers/clk/clk.c static int clk_set_rate_range_nolock(struct clk

Quoted 10 lines
drivers/clk/clk.c +3 -0
 
 	 *   this corner case when determining the rate
 	 */
 
+
 
 	rate = clamp(clk->core->req_rate, min, max);
 
+
+	pr_crit("%s: %s: clamped rate %lu\n", __func__, clk->core->name,

rate);

Quoted 14 lines
drivers/clk/clk.c +1 -0
+
 
 	ret = clk_core_set_rate_nolock(clk->core, rate);
 	if (ret) {
 
 		/* rollback the changes */
@@ -2599,6 +2666,8 @@ static int clk_core_set_parent_nolock(struct

clk_core
*core, } else {

drivers/clk/clk.c +0 -0
 		__clk_recalc_rates(core, POST_RATE_CHANGE);
 		__clk_recalc_accuracies(core);
Quoted 7 lines

drivers/clk/clk.c

Quoted 7 lines
drivers/clk/clk.c +2 -1
+
+		core->req_rate = core->rate;
 
 	}
 
 runtime_put:
---- >8 -----

So, let's try to follow this through:

Quoted 10 lines

Sure, here we go
---

[    0.630873] Asymmetric key parser 'x509' registered
[    0.635802] Block layer SCSI generic (bsg) driver version 0.4 loaded
(major 243) [    0.643210] io scheduler mq-deadline registered
[    0.647758] io scheduler kyber registered
[    0.658708] clk_set_rate_range_nolock: arm_a53_div: orphan ? n
[    0.661717] clk_set_rate_range_nolock: arm_a53_div: core req rate

800000000 [ 0.668724] clk_set_rate_range_nolock: arm_a53_div: clamped
rate 800000000

I'm assuming we hit the assigned-clock-parents in the clocks node, and we try to reparent arm_a53_div / IMX8MP_CLK_A53_SRC to sys_pll1_800m

I'm not entirely sure, but it looks like the arm_a53_div is a gate + divider, so that it has the same rate than its parent makes sens, and 800MHz for a CPU clock also makes sense.

It's also not an orphan, so it's likely to be a separate issue from Tony (and thus the fix doesn't help, sorry).

Quoted 1 line
[    0.675633] clk_core_set_rate_nolock: arm_a53_div: rate 800000000

Now, we set the rate to the same rate, this still makes sense.

Quoted 5 lines
[    0.681761] clk_core_round_rate_nolock +1439
[    0.686048] clk_core_round_rate_nolock +1446
[    0.690333] clk_core_round_rate_nolock +1450
[    0.694619] clk_core_round_rate_nolock +1453
[    0.698908] clk_core_determine_round_nolock +1374 arm_a53_div

The clock has a round_rate / determine_rate implementation (clk_divider_round_rate, most likely), thus we call clk_core_determine_round_nolock()

Quoted 3 lines
[    0.704681] clk_core_determine_round_nolock +1378
[    0.709408] clk_core_determine_round_nolock +1381
[    0.714133] clk_core_determine_round_nolock +1393

Still on the right path, we use clk_divider_determine_rate (too bad :)), it updates the rate

Quoted 2 lines
[    0.718860] clk_core_set_rate_nolock: arm_a53_div: rounded rate

800000000

But it didn't change, good. The rounded clock hasn't changed, clk_core_set_rate_nolock returns, everything's great.

Quoted 5 lines
[    0.725684] clk_set_rate_range_nolock: sys_pll1_800m: orphan ? n
[    0.731719] clk_set_rate_range_nolock: sys_pll1_800m: core req rate

800000000 [ 0.738894] clk_set_rate_range_nolock: sys_pll1_800m:
clamped rate 800000000 [ 0.745983] clk_core_set_rate_nolock:
sys_pll1_800m: rate 800000000

Then, __set_clk_parents calls clk_put() on the new parent, sys_pll1_800m, still not an orphan, still with a rate that makes sense.

Quoted 5 lines
[    0.752281] clk_core_round_rate_nolock +1439
[    0.756569] clk_core_round_rate_nolock +1446
[    0.760862] clk_core_round_rate_nolock +1450
[    0.765152] clk_core_round_rate_nolock +1453
[    0.769435] clk_core_determine_round_nolock +1374 sys_pll1_800m

We still can round the rate, so we go to
clk_core_determine_round_nolock()

Quoted 3 lines
[    0.775385] clk_core_determine_round_nolock +1378
[    0.780114] clk_core_determine_round_nolock +1381
[    0.784833] clk_core_determine_round_nolock +1396

But this time using a round_rate implementation: clk_factor_round_rate (since sys_pll1_800m is a "pure" fixed factor clock). It has the flag CLK_SET_RATE_PARENT (set in imx_clk_hw_fixed_factor), so clk_factor_round_rate calls clk_hw_round_rate on its parent (sys_pll1_out) for the same rate since it has a factor of 1.

Quoted 4 lines
[    0.789559] clk_core_round_rate_nolock +1439
[    0.793844] clk_core_round_rate_nolock +1446
[    0.798133] clk_core_round_rate_nolock +1450
[    0.802423] clk_core_round_rate_nolock +1456

We go through another round_rate cycle here, for sys_pll1_out. It can't modify the rate (since it's a gate) but it has CLK_SET_RATE_PARENT, so the rate rounding is forwarded to its parent: sys_pll1_bypass.

Quoted 4 lines
[    0.806708] clk_core_round_rate_nolock +1439
[    0.810994] clk_core_round_rate_nolock +1446
[    0.815284] clk_core_round_rate_nolock +1450
[    0.819570] clk_core_round_rate_nolock +1453

We go through it, and call clk_core_determine_round_nolock again for sys_pll1_bypass.

Quoted 1 line
[    0.823856] clk_core_determine_round_nolock +1374 sys_pll1_bypass

Makes total sense so far.

Quoted 5 lines
[    0.829981] clk_core_determine_round_nolock +1378
[    0.834706] clk_core_determine_round_nolock +1381
[    0.839431] clk_core_determine_round_nolock +1393
[    0.844159] clk_mux_determine_rate_flags: sys_pll1_bypass: requested

rate 800000000

The requested rate does too. We still have our 800MHz.

Quoted 3 lines
[    0.851856] clk_mux_determine_rate_flags: sys_pll1_bypass: current

parent sys_pll1 [ 0.859471] clk_mux_determine_rate_flags:
sys_pll1_bypass: current parent rate 800000000

sys_pll1_bypass has CLK_SET_RATE_NO_REPARENT (set by __imx_clk_hw_mux) and CLK_SET_RATE_PARENT (set by the driver when registering the clock), so clk_mux_determine_rate_flags will call __clk_determine_rate on its
parent: sys_pll1. __clk_determine_rate then calls clk_core_round_rate_nolock.

Quoted 4 lines
[    0.867608] clk_core_round_rate_nolock +1439
[    0.871894] clk_core_round_rate_nolock +1446
[    0.876182] clk_core_round_rate_nolock +1450
[    0.880477] clk_core_round_rate_nolock +1453

We call clk_core_determine_round_nolock on sys_pll1

Quoted 4 lines
[    0.884758] clk_core_determine_round_nolock +1374 sys_pll1
[    0.890273] clk_core_determine_round_nolock +1378
[    0.894996] clk_core_determine_round_nolock +1381
[    0.899721] clk_core_determine_round_nolock +1396

sys_pll1 is a clk_pll14xx driver, it has a PLL_1416X type and a rate table, so it will use clk_pll1416x_ops. It has a round_rate implementation, clk_pll14xx_round_rate, that doesn't seem to be doing anything out of the ordinary. My assumption would be that it succeeds and returns a proper rate.

Quoted 45 lines
[    0.904457] Unable to handle kernel NULL pointer dereference at virtual
address 0000000000000000
[    0.913285] Mem abort info:
[    0.916083]   ESR = 0x96000004
[    0.919147]   EC = 0x25: DABT (current EL), IL = 32 bits
[    0.924484]   SET = 0, FnV = 0
[    0.927547]   EA = 0, S1PTW = 0
[    0.930697]   FSC = 0x04: level 0 translation fault
[    0.935595] Data abort info:
[    0.938487]   ISV = 0, ISS = 0x00000004
[    0.942334]   CM = 0, WnR = 0
[    0.945304] [0000000000000000] user address but active_mm is swapper
[    0.951696] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.957292] Modules linked in:
[    0.960355] CPU: 2 PID: 1 Comm: swapper/0 Not tainted
5.17.0-next-20220331+ #53 da834fe2485dc10e4c2f50265323ce628a30bc5e
[    0.971291] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL
(DT) [    0.978292] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS
BTYPE=--) [    0.985291] pc : clk_mux_determine_rate_flags+0x33c/0x380
[    0.990714] lr : clk_mux_determine_rate_flags+0x10c/0x380
[    0.996141] sp : ffff800009ceb4a0
[    0.999464] x29: ffff800009ceb4a0 x28: ffff000002cf4700 x27:
0000000000000001
[    1.006639] x26: ffff8000092fe728 x25: ffff800008eaa028 x24:
ffff800008ea95d8
[    1.013816] x23: ffff800008ea95d8 x22: ffff000002aab700 x21:
000000002faf0800
[    1.020989] x20: ffff800009ceb640 x19: 0000000000000000 x18:
0000000000004590
[    1.028164] x17: 617220746e657261 x16: 7020746e65727275 x15:
63203a7373617079
[    1.035339] x14: 0000000000000000 x13: 363933312b206b63 x12:
6f6c6f6e5f646e75
[    1.042514] x11: 6f725f656e696d72 x10: 657465645f65726f x9 :
206b636f6c6f6e5f
[    1.049689] x8 : 646e756f725f656e x7 : 205d313237393938 x6 :
ffff800009a947c8
[    1.056864] x5 : ffff800008eb0310 x4 : 0000000000000009 x3 :
000000002faf0800
[    1.064039] x2 : ffff800008eb039c x1 : ffff800008eaa028 x0 :
ffff8000092fd8b8
[    1.071217] Call trace:
[    1.073667]  clk_mux_determine_rate_flags+0x33c/0x380
[    1.078741]  clk_mux_determine_rate+0x10/0x20
[    1.083115]  clk_core_determine_round_nolock+0xd4/0x140
Quoted 8 lines
[    1.088364]  clk_core_round_rate_nolock+0xac/0xf8
[    1.093090]  clk_core_round_rate_nolock+0xd4/0xf8
[    1.097814]  clk_hw_round_rate+0x44/0x7c
[    1.101751]  clk_factor_round_rate+0x60/0x80
[    1.106041]  clk_core_determine_round_nolock+0x104/0x140
[    1.111376]  clk_core_round_rate_nolock+0xac/0xf8
[    1.116101]  clk_core_set_rate_nolock.part.0+0xac/0x21c
[    1.121351]  clk_set_rate_range_nolock+0x294/0x2b0

But then, where does this come from?

I'm not entirely sure, but the walk up the clock tree is sane to me. Could you run

Quoted 7 lines

./scripts/faddr2line vmlinux 'clk_mux_determine_rate_flags+0x33c/0x380'

in your kernel compilation directory? (with ARCH and CROSS_COMPILE set if you're doing cross-compilation)?

My guess would be that we uncovered some other bug there, but I'm not sure what exactly.

Thanks for that lengthy analysis.

Here is the requested output:
---

$ ./scripts/faddr2line build_arm64/vmlinux 
'clk_mux_determine_rate_flags+0x33c/0x380'
clk_mux_determine_rate_flags+0x33c/0x380:
clk_mux_determine_rate_flags at drivers/clk/clk.c:627
---
From a first look it seems that 'best_parent' is just a NULL-pointer here.
With this small fix
--->8---
drivers/clk/clk.c +1 -1
index 071857ef381a..45e081330fac 100644
@@ -626,7 +626,7 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 
        pr_crit("%s: Best parent %s (%lu)\n",
                __func__,
-               best_parent->name,
+               best_parent? best_parent->name : "unknown",
                best);
 
        return 0;

--->8---

The boot eventually get stuck, but at a later point.Which is probably why your analysis found nothing strange. Due to the size of the output I put it on a gist on github [1]. Please note that this is still based on a next-20220331 based tree without the revert.

Best regards,
Alexander

[1] https://gist.github.com/tq-steina/f90c095d141575eaf75395f26671841a

#26 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 25 lines · Alexander Stein

Here is the requested output:
---

$ ./scripts/faddr2line build_arm64/vmlinux 
'clk_mux_determine_rate_flags+0x33c/0x380'
clk_mux_determine_rate_flags+0x33c/0x380:
clk_mux_determine_rate_flags at drivers/clk/clk.c:627
---
From a first look it seems that 'best_parent' is just a NULL-pointer here.
With this small fix
--->8---
drivers/clk/clk.c +1 -2
index 071857ef381a..45e081330fac 100644
@@ -626,7 +626,7 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 
        pr_crit("%s: Best parent %s (%lu)\n",
                __func__,
-               best_parent->name,
+               best_parent? best_parent->name : "unknown",
                best);
 
        return 0;
--->8---
Quoted 4 lines

The boot eventually get stuck, but at a later point.Which is probably why your analysis found nothing strange. Due to the size of the output I put it on a gist on github [1]. Please note that this is still based on a next-20220331 based tree without the revert.

I've looked into it over the weekend, and ran qemu on an imx6 to try to see if it was any similar

I believe the issue comes from the fact that the core will forward rate requests structure to the parent clock as is, and if the parent clock changes the parent it wants, we end up trying to use that parent in the initial clock which doesn't work really well.

I've fixed it in my branch here:
https://github.com/mripard/linux/commits/rpi/clk-improvements-more-fixes

Could you give it a try?
Maxime

#27 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hello Maxime,

Am Montag, 4. April 2022, 09:27:12 CEST schrieb Maxime Ripard:

Quoted 38 lines · Alexander Stein
Quoted 27 lines

Here is the requested output:
---

$ ./scripts/faddr2line build_arm64/vmlinux
'clk_mux_determine_rate_flags+0x33c/0x380'
clk_mux_determine_rate_flags+0x33c/0x380:
clk_mux_determine_rate_flags at drivers/clk/clk.c:627
---
From a first look it seems that 'best_parent' is just a NULL-pointer here.
With this small fix
--->8---
drivers/clk/clk.c +1 -1
index 071857ef381a..45e081330fac 100644
@@ -626,7 +626,7 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 
        pr_crit("%s: Best parent %s (%lu)\n",
 
                __func__,
 
-               best_parent->name,
+               best_parent? best_parent->name : "unknown",
 
                best);
        
        return 0;
Quoted 6 lines

--->8---

The boot eventually get stuck, but at a later point.Which is probably why your analysis found nothing strange. Due to the size of the output I put it on a gist on github [1]. Please note that this is still based on a next-20220331 based tree without the revert.

I've looked into it over the weekend, and ran qemu on an imx6 to try to see if it was any similar

Quoted 7 lines

I believe the issue comes from the fact that the core will forward rate requests structure to the parent clock as is, and if the parent clock changes the parent it wants, we end up trying to use that parent in the initial clock which doesn't work really well.

I've fixed it in my branch here:
https://github.com/mripard/linux/commits/rpi/clk-improvements-more-fixes

Thanks for providing another patchset. Unfortunately, my board hangs still at the same location. For reference I put a branch based on next-20220401 on [1]. Reverting still does the jobs, a branch is shown on [2]

next-20220404 has the offending patch already reverted, so this should work again, I did not test it on that base on purpose.

Best regards,
Alexander

[1] https://github.com/tq-steina/linux/tree/clk-fix
[2] https://github.com/tq-steina/linux/tree/clk-revert

#28 lore raw

Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi Alexander,

Quoted 47 lines · Alexander Stein

Am Montag, 4. April 2022, 09:27:12 CEST schrieb Maxime Ripard:

Quoted 38 lines · Alexander Stein
Quoted 27 lines

Here is the requested output:
---

$ ./scripts/faddr2line build_arm64/vmlinux
'clk_mux_determine_rate_flags+0x33c/0x380'
clk_mux_determine_rate_flags+0x33c/0x380:
clk_mux_determine_rate_flags at drivers/clk/clk.c:627
---
From a first look it seems that 'best_parent' is just a NULL-pointer here.
With this small fix
--->8---
drivers/clk/clk.c +1 -1
index 071857ef381a..45e081330fac 100644
@@ -626,7 +626,7 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 
        pr_crit("%s: Best parent %s (%lu)\n",
 
                __func__,
 
-               best_parent->name,
+               best_parent? best_parent->name : "unknown",
 
                best);
        
        return 0;
Quoted 6 lines

--->8---

The boot eventually get stuck, but at a later point.Which is probably why your analysis found nothing strange. Due to the size of the output I put it on a gist on github [1]. Please note that this is still based on a next-20220331 based tree without the revert.

I've looked into it over the weekend, and ran qemu on an imx6 to try to see if it was any similar

Quoted 7 lines

I believe the issue comes from the fact that the core will forward rate requests structure to the parent clock as is, and if the parent clock changes the parent it wants, we end up trying to use that parent in the initial clock which doesn't work really well.

I've fixed it in my branch here:
https://github.com/mripard/linux/commits/rpi/clk-improvements-more-fixes

Quoted 6 lines

Thanks for providing another patchset. Unfortunately, my board hangs still at the same location. For reference I put a branch based on next-20220401 on [1]. Reverting still does the jobs, a branch is shown on [2]

next-20220404 has the offending patch already reverted, so this should work again, I did not test it on that base on purpose.

Could you give it another try with the branch I pushed today?

I've fixed some issues for exynos4 that could explain why it doesn't boot on imx8 as well.

Maxime

#29 lore raw

Re: (EXT) Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

In reply to Maxime Ripard

Hi Alexandre, Tony, Mareek, Naresh,

Quoted 201 lines · Maxime Ripard
Quoted 178 lines · Alexander Stein

Am Donnerstag, 31. März 2022, 17:31:34 CEST schrieb Maxime Ripard:

Quoted 18 lines

* PGP Signed by an unknown key

Hi Tony,

Quoted 13 lines · Tony Lindgren
Quoted 7 lines · Maxime Ripard
Quoted 1 line · Tony Lindgren

It seems the dts assigned-clock-parents no longer works now?

That would make some kind of sense, __set_clk_parents calls clk_put on both the assigned clock and its parent.

Could you see what parent (and why?) it tries to enforce then?

It picks the other option available for the mux clock that only has two options. No idea why, but if you have some debug patch in mind I can give it a try.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

It looks like the gpt1_fck driver might favor another parent for that rate, which, if it's an invalid configuration, shouldn't really happen?

Hmm there's a gate clock and a mux clock, there's not really a rate selection available here for the sources.

If I followed the OMAP driver properly, clk_mux_determine_rate_flags is doing the heavy lifting, could you run your test with

I'm affected by this patch as well on an imx8mp platform (see [1] for some details)

In the failing case with with your patch applied I get the following error
---

[    0.661064] clk_set_rate_range_nolock: core req rate 500000000                                                                    
[    0.664084] clk_set_rate_range_nolock: clamped rate 500000000                                                                     
[    0.669851] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.674843] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.680536] clk_set_rate_range_nolock: core req rate 800000000                                                                    
[    0.686389] clk_set_rate_range_nolock: clamped rate 800000000                                                                     
[    0.692164] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.697153] clk_mux_determine_rate_flags: requested rate 800000000                                                                
[    0.703363] clk_mux_determine_rate_flags: current parent sys_pll1                                                                 
[    0.709487] clk_mux_determine_rate_flags: current parent rate 800000000                                                           
[    0.716147] Unable to handle kernel NULL pointer dereference at virtual 
address 0000000000000000
[    0.724977] Mem abort info:                                                                                                       
[    0.727775]   ESR = 0x96000004                                                                                                    
[    0.730835]   EC = 0x25: DABT (current EL), IL = 32 bits                                                                          
[    0.736177]   SET = 0, FnV = 0                                                                                                    
[    0.739239]   EA = 0, S1PTW = 0                                                                                                   
[    0.742382]   FSC = 0x04: level 0 translation fault                                                                               
[    0.747287] Data abort info:                                                                                                      
[    0.750172]   ISV = 0, ISS = 0x00000004                                                                                           
[    0.754027]   CM = 0, WnR = 0                                                                                                     
[    0.757002] [0000000000000000] user address but active_mm is swapper                                                              
[    0.763394] Internal error: Oops: 96000004 [#1] PREEMPT SMP                                                                       
[    0.768985] Modules linked in:                                                                                                    
[    0.772049] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.17.0-next-20220331+ 
#48 8e9d24095c7f6b15f85bc2ad57a5609e219130b9          
[    0.782984] Hardware name: TQ-Systems i.MX8MPlus TQMa8MPxL on MBa8MPxL (DT)                                                       
[    0.789985] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)                                                       
[    0.796985] pc : clk_mux_determine_rate_flags+0x280/0x2cc                                                                         
[    0.802407] lr : clk_mux_determine_rate_flags+0xf4/0x2cc                                                                          
[    0.807747] sp : ffff800009ceb590                                                                                                 
[    0.811072] x29: ffff800009ceb590 x28: ffff800009ceb6a0 x27: 
ffff800008eaa038                                                     
[    0.818245] x26: ffff8000092fe0b0 x25: ffff000000090000 x24: 
ffff000000090000                                                     
[    0.825420] x23: 0000000000000000 x22: ffff800008ea95d8 x21: 
ffff0000028f4700                                                     
[    0.832595] x20: 000000002faf0800 x19: 0000000000000000 x18: 
0000000000004590                                                     
[    0.839770] x17: 0000000000004570 x16: 0000000000004560 x15: 
ffff8000092ff250                                                     
[    0.846945] x14: 0000000000000000 x13: 3030303030303030 x12: 
3820657461722074                                                     
[    0.854120] x11: 6e6572617020746e x10: 6572727563203a73 x9 : 
7563203a7367616c                                                     
[    0.861295] x8 : 665f657461725f65 x7 : 205d373834393037 x6 : 
ffff800009a947c8                                                     
[    0.868472] x5 : ffff800008eafe68 x4 : 0000000000000009 x3 : 
000000002faf0800                                                     
[    0.875645] x2 : ffff800008eafef4 x1 : ffff800008eaa038 x0 : 
ffff8000092fd5b8                                                     
[    0.882822] Call trace:                                                                                                           
[    0.885273]  clk_mux_determine_rate_flags+0x280/0x2cc                                                                             
[    0.890347]  clk_mux_determine_rate+0x10/0x20                                                                                     
[    0.894720]  clk_core_determine_round_nolock+0x4c/0xb4                                                                            
[    0.899882]  clk_core_round_rate_nolock+0x30/0x80                                                                                 
[    0.904607]  clk_core_round_rate_nolock+0x70/0x80                                                                                 
[    0.909334]  clk_hw_round_rate+0x44/0x74                                                                                          
[    0.913270]  clk_factor_round_rate+0x60/0x80                                                                                      
[    0.917557]  clk_core_determine_round_nolock+0x88/0xb4                                                                            
[    0.922720]  clk_core_round_rate_nolock+0x30/0x80                                                                                 
[    0.927445]  clk_core_set_rate_nolock.part.0+0xa4/0x1d0                                                                           
[    0.932695]  clk_set_rate_range_nolock+0x234/0x244                                                                                
[    0.937507]  __clk_put+0x60/0x12c                                                                                                 
[    0.940834]  clk_put+0xc/0x1c                                                                                                     
[    0.943809]  __set_clk_parents+0x12c/0x244                                                                                        
[    0.947920]  of_clk_set_defaults+0x20/0x50                                                                                        
[    0.952032]  of_clk_add_hw_provider.part.0+0x94/0x120                                                                             
[    0.957107]  of_clk_add_hw_provider+0x10/0x20                                                                                     
[    0.961482]  imx8mp_clocks_probe+0x3458/0x34d0                                                                                    
[    0.965945]  platform_probe+0x64/0x100                                                                                            
[    0.969707]  call_driver_probe+0x28/0x130                                                                                         
[    0.973732]  really_probe+0x178/0x310                                                                                             
[    0.977409]  __driver_probe_device+0xfc/0x144                                                                                     
[    0.981782]  driver_probe_device+0x38/0x12c                                                                                       
[    0.985982]  __driver_attach+0xcc/0x220                                                                                           
[    0.989834]  bus_for_each_dev+0x6c/0xc0                                                                                           
[    0.993682]  driver_attach+0x20/0x2c                                                                                              
[    0.997270]  bus_add_driver+0x140/0x230                                                                                           
[    1.001120]  driver_register+0x74/0x120                                                                                           
[    1.004970]  __platform_driver_register+0x24/0x30                                                                                 
[    1.009697]  imx8mp_clk_driver_init+0x18/0x20                                                                                     
[    1.014070]  do_one_initcall+0x58/0x200                                                                                           
[    1.017920]  do_initcalls+0x164/0x19c                                                                                             
[    1.021597]  kernel_init_freeable+0x134/0x17c                                                                                     
[    1.025970]  kernel_init+0x2c/0x150                                                                                               
[    1.029470]  ret_from_fork+0x10/0x20                                                                                              
[    1.033065] Code: f9000f94 912982c1 b0002900 9116e000 (f9400262)                                                                  
[    1.039188] ---[ end trace 0000000000000000 ]---                                                                                  
[    1.043869] Kernel panic - not syncing: Attempted to kill init! 
exitcode=0x0000000b                                               
[    1.051523] SMP: stopping secondary CPUs                                                                                          
[    1.055467] Kernel Offset: disabled                                                                                               
[    1.058960] CPU features: 0x000,00020009,00001082                                                                                 
[    1.063684] Memory Limit: none                                                                                                    
[    1.066748] ---[ end Kernel panic - not syncing: Attempted to kill init! 

exitcode=0x0000000b ]---
---

With the $subject patch reverted and bootable system:
---

[    0.659922] clk_core_set_rate_nolock: rate 1000000000                                                                             
[    0.662154] clk_core_set_rate_nolock: rounded rate 1000000000                                                                     
[    0.667932] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.672918] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.678601] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.683592] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.689276] clk_core_set_rate_nolock: rate 400000000                                                                              
[    0.694267] clk_core_set_rate_nolock: rounded rate 400000000                                                                      
[    0.699980] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.704942] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.710627] clk_core_set_rate_nolock: rate 393216000                                                                              
[    0.715611] clk_core_set_rate_nolock: rounded rate 393216000                                                                      
[    0.721815] clk_core_set_rate_nolock: rate 361267200                                                                              
[    0.726284] clk_core_set_rate_nolock: rounded rate 361267200                                                                      
[    0.734097] clk_core_set_rate_nolock: rate 800000000                                                                              
[    0.736977] clk_core_set_rate_nolock: rounded rate 800000000                                                                      
[    0.742652] clk_core_set_rate_nolock: rate 400000000                                                                              
[    0.747645] clk_core_set_rate_nolock: rounded rate 400000000                                                                      
[    0.754565] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.758331] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.764688] SoC: i.MX8MP revision 1.1                                                                                             
[    0.767931] clk_core_set_rate_nolock: rate 500000000                                                                              
[    0.772675] clk_core_set_rate_nolock: rounded rate 500000000                                                                      
[    0.778354] clk_core_set_rate_nolock: rate 200000000                                                                              
[    0.783351] clk_core_set_rate_nolock: rounded rate 200000000                                                                      
[    0.793748] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled                                                               
[    0.798952] 30860000.serial: ttymxc0 at MMIO 0x30860000 (irq = 34, 
base_baud = 5000000) is a IMX                                  
[    0.806527] 30880000.serial: ttymxc2 at MMIO 0x30880000 (irq = 35, 
base_baud = 5000000) is a IMX                                  
[    0.815329] 30890000.serial: ttymxc1 at MMIO 0x30890000 (irq = 36, 
base_baud = 5000000) is a IMX                                  
[    0.824176] 30a60000.serial: ttymxc3 at MMIO 0x30a60000 (irq = 43, 
base_baud = 1500000) is a IMX                                  
[    0.832588] printk: console [ttymxc3] enabled                                                                                     
[    0.832588] printk: console [ttymxc3] enabled                                                                                     
[    0.841244] printk: bootconsole [ec_imx6q0] disabled                                                                              
[    0.841244] printk: bootconsole [ec_imx6q0] disabled                                                                              
[    0.857871] clk_core_set_rate_nolock: rate 80000000                                                                               
[    0.862796] clk_core_set_rate_nolock: rounded rate 80000000                                                                       
[    0.868469] clk_core_set_rate_nolock: rate 20000000                                                                               
[    0.873364] clk_core_set_rate_nolock: rounded rate 20000000                                                                       
[    0.879258] clk_core_set_rate_nolock: rate 80000000                                                                               
[    0.884154] clk_core_set_rate_nolock: rounded rate 80000000
Quoted 7 lines

[...]
---

The 500000000 and 800000000 look a bit like the assigned-clock-rates for clock-controller@30380000 in arch/arm64/boot/dts/freescale/imx8mp.dtsi

If you need some more information, do not hesitate to ask

Thanks a lot to you three for all your testing. I think I know what might be going on:

We use the last requested rate on clk_set_rate_range (clk_core.req_rate), and that requested rate if the clock is orphan will be set to 0, so if we were to call clk_set_rate_range before the parent clock is registered, we would effectively call a clk_set_rate to 0

And the assigned-clocks stuff is handled by __set_clk_parents and __set_clk_rates, called by of_clk_set_defaults(), in turn called by of_clk_init and of_clk_add_provider. Both __set_clk_parents and __set_clk_rates will call clk_put once done with the clock, and we will with this patch trigger the clk_set_rate to 0 I mentioned before.

Quoted 5 lines

So we just became very good at triggering the underlying issue :)

And I think it's that while we update the requested rate when the missing parent is registered, we never do when we mux away from it using clk_set_parent.

I've worked on a similar setup than Alexander today using qemu and figured out a fairly significant bug in the rate setting logic in the clock framework, that could explain why OMAP is going crazy as well.

I've pushed a branch here with multiple fixes:
https://github.com/mripard/linux/tree/rpi/clk-improvements-more-fixes

Let me know if it fixes your issues.

If it doesn't, could you point me to a setup (using qemu?) that could reproduce this issue, or alternatively to a cheap, available, board I could buy to debug this further?

Thanks!
Maxime

#30 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

In reply to Tony Lindgren

Hi Tony,

Quoted 56 lines · Tony Lindgren
Quoted 43 lines · Maxime Ripard

Hi Marek,

Quoted 40 lines · Marek Szyprowski
Quoted 7 lines · Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 7 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

This patch landed recently in linux-next 20220328 as commit 7dabfa2bc480 ("clk: Drop the rate range on clk_put()"). Sadly it breaks booting of the few of my test systems: Samsung ARM 32bit Exynos3250 based Rinato board and all Amlogic Meson G12B/SM1 based boards (Odroid C4, N2, Khadas VIM3/VIM3l). Rinato hangs always with the following oops:

--->8---

Kernel panic - not syncing: MCT hangs after writing 4 (offset:0x420)
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.17.0-rc1-00014-g7dabfa2bc480 
#11551
Hardware name: Samsung Exynos (Flattened Device Tree)
  unwind_backtrace from show_stack+0x10/0x14
  show_stack from dump_stack_lvl+0x58/0x70
  dump_stack_lvl from panic+0x10c/0x328
  panic from exynos4_mct_tick_stop+0x0/0x2c
---[ end Kernel panic - not syncing: MCT hangs after writing 4 

(offset:0x420) ]---

--->8---

Amlogic boards hang randomly during early userspace init, usually just after loading the driver modules.

Quoted 8 lines
Quoted 4 lines

Reverting $subject on top of linux-next fixes all those problems.

I will try to analyze it a bit more and if possible provide some more useful/meaning full logs later.

I'm not sure what could go wrong there, but if you can figure out the clock, if it tries to set a new rate and what rate it is, it would be awesome :)

I'm also seeing clockevent break on omaps as a wrong source clock gets picked.

Quoted 8 lines

It seems the dts assigned-clock-parents no longer works now?

So the following no longer sets omap_32k_fck as the clockevent source:

timer@0 {
	assigned-clocks = <&gpt1_fck>;
	assigned-clock-parents = <&omap_32k_fck>;
};

I haven't been able to find an omap3 board or a qemu target that could help me debug this, but I fixed a few issues already that could fix omap as well.

Could you test today's
https://github.com/mripard/linux/tree/rpi/clk-improvements-more-fixes

And let me know if it works?

Thanks!
Maxime

#31 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi,

Quoted 8 lines · Maxime Ripard

I haven't been able to find an omap3 board or a qemu target that could help me debug this, but I fixed a few issues already that could fix omap as well.

Could you test today's
https://github.com/mripard/linux/tree/rpi/clk-improvements-more-fixes

And let me know if it works?

Yes sorry I've been meaning to try your fixes but had some file system issues on my build box after a power cut while updating the system. All good now though, I should be able to give it a try this afternoon.

Regards,

Tony

#32 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 7 lines · Tony Lindgren

Hi,

Quoted 4 lines · Maxime Ripard

I haven't been able to find an omap3 board or a qemu target that could help me debug this, but I fixed a few issues already that could fix omap as well.

Quoted 8 lines
Quoted 4 lines

Could you test today's
https://github.com/mripard/linux/tree/rpi/clk-improvements-more-fixes

And let me know if it works?

Yes sorry I've been meaning to try your fixes but had some file system issues on my build box after a power cut while updating the system. All good now though, I should be able to give it a try this afternoon.

It now boots, but does a lot of checks on the clocks before the timers get initialized compared to v5.18-rc1. And then there's this:

[    2.532501] clk_core_set_rate_nolock +2293: ssi_ssr_fck_3430es2 affected!
...
[    2.554443]  unwind_backtrace from show_stack+0x10/0x14
[    2.559875]  show_stack from dump_stack_lvl+0x40/0x4c
[    2.565093]  dump_stack_lvl from clk_core_set_rate_nolock+0x278/0x2c4
[    2.571777]  clk_core_set_rate_nolock from clk_set_rate_range_nolock.part.0+0x154/0x384
[    2.580047]  clk_set_rate_range_nolock.part.0 from __clk_put+0x64/0x174
[    2.586853]  __clk_put from clk_add_alias+0x48/0x5c
[    2.591918]  clk_add_alias from _add_clkdev.part.0+0x94/0x154
[    2.597869]  _add_clkdev.part.0 from omap_device_alloc+0x88/0x114
[    2.604156]  omap_device_alloc from _omap_device_notifier_call+0x25c/0x3b4
[    2.611236]  _omap_device_notifier_call from blocking_notifier_call_chain+0x6c/0x90
[    2.619140]  blocking_notifier_call_chain from device_add+0x360/0x894
[    2.625823]  device_add from of_platform_device_create_pdata+0x8c/0xb8
[    2.632568]  of_platform_device_create_pdata from of_platform_bus_create+0x194/0x22c
[    2.640563]  of_platform_bus_create from of_platform_bus_create+0x1e0/0x22c
[    2.647735]  of_platform_bus_create from of_platform_populate+0x60/0xb8
[    2.654571]  of_platform_populate from pdata_quirks_init+0xb4/0xe0
[    2.660980]  pdata_quirks_init from omap_generic_init+0xc/0x18
[    2.666992]  omap_generic_init from customize_machine+0x1c/0x30
[    2.673126]  customize_machine from do_one_initcall+0x44/0x24c
[    2.679138]  do_one_initcall from kernel_init_freeable+0x1e8/0x298
[    2.685546]  kernel_init_freeable from kernel_init+0x14/0x140
[    2.691467]  kernel_init from ret_from_fork+0x14/0x24

I'll email you the full log separately to look at.

Regards,

Tony

#33 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Hi Tony,

Quoted 11 lines · Tony Lindgren
Quoted 10 lines · Tony Lindgren

Hi,

Quoted 7 lines · Maxime Ripard

I haven't been able to find an omap3 board or a qemu target that could help me debug this, but I fixed a few issues already that could fix omap as well.

Could you test today's
https://github.com/mripard/linux/tree/rpi/clk-improvements-more-fixes

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

And let me know if it works?

Yes sorry I've been meaning to try your fixes but had some file system issues on my build box after a power cut while updating the system. All good now though, I should be able to give it a try this afternoon.

It now boots, but does a lot of checks on the clocks before the timers get initialized compared to v5.18-rc1.

I was about to say that this is fairly normal with the new behaviour, but I've reworked the initial patch in that discussion to only call into clk_set_rate_range if there was a range on that clock to begin with.

It should remove the huge majority of the checks you mentioned (and hopefully get rid of most of the side effects as well).

It's now pushed to my branch, so it would be awesome if you could test again.

Quoted 18 lines

And then there's this:

[    2.532501] clk_core_set_rate_nolock +2293: ssi_ssr_fck_3430es2 affected!
...
[    2.554443]  unwind_backtrace from show_stack+0x10/0x14
[    2.559875]  show_stack from dump_stack_lvl+0x40/0x4c
[    2.565093]  dump_stack_lvl from clk_core_set_rate_nolock+0x278/0x2c4
[    2.571777]  clk_core_set_rate_nolock from clk_set_rate_range_nolock.part.0+0x154/0x384
[    2.580047]  clk_set_rate_range_nolock.part.0 from __clk_put+0x64/0x174
[    2.586853]  __clk_put from clk_add_alias+0x48/0x5c
[    2.591918]  clk_add_alias from _add_clkdev.part.0+0x94/0x154
[    2.597869]  _add_clkdev.part.0 from omap_device_alloc+0x88/0x114
[    2.604156]  omap_device_alloc from _omap_device_notifier_call+0x25c/0x3b4
[    2.611236]  _omap_device_notifier_call from blocking_notifier_call_chain+0x6c/0x90
[    2.619140]  blocking_notifier_call_chain from device_add+0x360/0x894
[    2.625823]  device_add from of_platform_device_create_pdata+0x8c/0xb8
[    2.632568]  of_platform_device_create_pdata from of_platform_bus_create+0x194/0x22c
[    2.640563]  of_platform_bus_create from of_platform_bus_create+0x1e0/0x22c
Quoted 8 lines
[    2.647735]  of_platform_bus_create from of_platform_populate+0x60/0xb8
[    2.654571]  of_platform_populate from pdata_quirks_init+0xb4/0xe0
[    2.660980]  pdata_quirks_init from omap_generic_init+0xc/0x18
[    2.666992]  omap_generic_init from customize_machine+0x1c/0x30
[    2.673126]  customize_machine from do_one_initcall+0x44/0x24c
[    2.679138]  do_one_initcall from kernel_init_freeable+0x1e8/0x298
[    2.685546]  kernel_init_freeable from kernel_init+0x14/0x140
[    2.691467]  kernel_init from ret_from_fork+0x14/0x24

It shouldn't be there anymore after that rework, but I couldn't find wher the ssi_ssr_fck clock was defined? The only relevant driver seems to be omap_ssi_core.c but I don't see any clock driver registered there either.

Thanks!
Maxime

_______________________________________________
linux-amlogic mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-amlogic

#34 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 10 lines · Maxime Ripard
Quoted 2 lines · Tony Lindgren

It now boots, but does a lot of checks on the clocks before the timers get initialized compared to v5.18-rc1.

I was about to say that this is fairly normal with the new behaviour, but I've reworked the initial patch in that discussion to only call into clk_set_rate_range if there was a range on that clock to begin with.

It should remove the huge majority of the checks you mentioned (and hopefully get rid of most of the side effects as well).

OK yeah thanks, looks good to me now. Boot time looks normal, timer clocks are right, and runtime PM still works too.

Quoted 4 lines

It shouldn't be there anymore after that rework, but I couldn't find wher the ssi_ssr_fck clock was defined? The only relevant driver seems to be omap_ssi_core.c but I don't see any clock driver registered there either.

I'm not seeing this warning any longer :)

FYI, this clock is defined here:

$ git grep ssi_ssr_fck_3430es2
arch/arm/boot/dts/omap36xx-omap3430es2plus-clocks.dtsi: ssi_ssr_fck: ssi_ssr_fck_3430es2 {
drivers/clk/ti/clk-3xxx.c:      DT_CLK(NULL, "ssi_ssr_fck", "ssi_ssr_fck_3430es2"),

Yeah it's confusing, the clock is still created based on the node name. I have some clean-up patches coming to fix most of the related make dtbs checks warnings now that the clock driver changes got merged.

Regards,

Tony

#35 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

In reply to Maxime Ripard

Hi,

Quoted 40 lines · Maxime Ripard
Quoted 39 lines · Marek Szyprowski
Quoted 7 lines · Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 7 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

This patch landed recently in linux-next 20220328 as commit 7dabfa2bc480 ("clk: Drop the rate range on clk_put()"). Sadly it breaks booting of the few of my test systems: Samsung ARM 32bit Exynos3250 based Rinato board and all Amlogic Meson G12B/SM1 based boards (Odroid C4, N2, Khadas VIM3/VIM3l). Rinato hangs always with the following oops:

--->8---

Kernel panic - not syncing: MCT hangs after writing 4 (offset:0x420)
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.17.0-rc1-00014-g7dabfa2bc480
#11551
Hardware name: Samsung Exynos (Flattened Device Tree)
   unwind_backtrace from show_stack+0x10/0x14
   show_stack from dump_stack_lvl+0x58/0x70
   dump_stack_lvl from panic+0x10c/0x328
   panic from exynos4_mct_tick_stop+0x0/0x2c
---[ end Kernel panic - not syncing: MCT hangs after writing 4

(offset:0x420) ]---

--->8---

Amlogic boards hang randomly during early userspace init, usually just after loading the driver modules.

Quoted 7 lines
Quoted 4 lines

Reverting $subject on top of linux-next fixes all those problems.

I will try to analyze it a bit more and if possible provide some more useful/meaning full logs later.

I'm not sure what could go wrong there, but if you can figure out the clock, if it tries to set a new rate and what rate it is, it would be awesome :)

So far I've noticed that the problem is caused by setting rate of some clocks to zero. The following patch fixes my issues:

drivers/clk/clk.c +0 -0
index 32a9eaf35c6b..39cab08dbecb 100644
@@ -2201,6 +2201,9 @@ static int clk_core_set_rate_nolock(struct 

clk_core *core,
        if (!core)
                return 0;

+       if (req_rate == 0)
+               return 0;
+
        rate = clk_core_req_round_rate_nolock(core, req_rate);

        /* bail early if nothing to do */
--

I will soon grab the call stack and relevant clock topology show how the rate is set to zero.

Best regards

Signature · 8 lines
Marek Szyprowski, PhD
Samsung R&D Institute Poland


_______________________________________________
linux-amlogic mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-amlogic
#36 lore raw

Re: [PATCH v2 3/3] clk: Drop the rate range on clk_put

Quoted 71 lines · Marek Szyprowski

Hi,

Quoted 40 lines · Maxime Ripard
Quoted 39 lines · Marek Szyprowski
Quoted 7 lines · Maxime Ripard

While the current code will trigger a new clk_set_rate call whenever the rate boundaries are changed through clk_set_rate_range, this doesn't occur when clk_put() is called.

However, this is essentially equivalent since, after clk_put() completes, those boundaries won't be enforced anymore.

Quoted 7 lines

Let's add a call to clk_set_rate_range in clk_put to make sure those rate boundaries are dropped and the clock drivers can react.

Let's also add a few tests to make sure this case is covered.

Fixes c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")

This patch landed recently in linux-next 20220328 as commit 7dabfa2bc480 ("clk: Drop the rate range on clk_put()"). Sadly it breaks booting of the few of my test systems: Samsung ARM 32bit Exynos3250 based Rinato board and all Amlogic Meson G12B/SM1 based boards (Odroid C4, N2, Khadas VIM3/VIM3l). Rinato hangs always with the following oops:

--->8---

Kernel panic - not syncing: MCT hangs after writing 4 (offset:0x420)
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.17.0-rc1-00014-g7dabfa2bc480
#11551
Hardware name: Samsung Exynos (Flattened Device Tree)
   unwind_backtrace from show_stack+0x10/0x14
   show_stack from dump_stack_lvl+0x58/0x70
   dump_stack_lvl from panic+0x10c/0x328
   panic from exynos4_mct_tick_stop+0x0/0x2c
---[ end Kernel panic - not syncing: MCT hangs after writing 4

(offset:0x420) ]---

--->8---

Amlogic boards hang randomly during early userspace init, usually just after loading the driver modules.

Quoted 7 lines
Quoted 4 lines

Reverting $subject on top of linux-next fixes all those problems.

I will try to analyze it a bit more and if possible provide some more useful/meaning full logs later.

I'm not sure what could go wrong there, but if you can figure out the clock, if it tries to set a new rate and what rate it is, it would be awesome :)

So far I've noticed that the problem is caused by setting rate of some clocks to zero. The following patch fixes my issues:

drivers/clk/clk.c +3 -0
index 32a9eaf35c6b..39cab08dbecb 100644
@@ -2201,6 +2201,9 @@ static int clk_core_set_rate_nolock(struct 
 clk_core *core,
         if (!core)
                 return 0;
 
+       if (req_rate == 0)
+               return 0;
+
         rate = clk_core_req_round_rate_nolock(core, req_rate);
 

        /* bail early if nothing to do */
--

Quoted 2 lines

I will soon grab the call stack and relevant clock topology show how the rate is set to zero.

The most likely thing to happen is that clk_set_rate_range will call clk_core_set_rate_nolock with clk_core->req_rate, and at the time req_rate is at 0.

And I'm a bit puzzled at this point, the only reason I could spot for req_rate to be at 0 is that it's an orphan clock that doesn't have its parent yet, but during userspace init I'd expect all the clocks to have been registered.

Can you check if that clock is still orphan?

Maxime