[PATCH v9 00/25] clk: More clock rate fixes and tests

[PATCH v9 00/25] clk: More clock rate fixes and tests

[Maxime Ripard]

Hi,

Thanks to the feedback I got on the previous series, I found and fixed a
number of bugs in the clock framework and how it deals with rates,
especially when it comes to orphan clocks.

In order to make sure this doesn't pop up again as a regression, I've
extended the number of tests.

The first patch reintroduces the clk_set_rate_range call on clk_put, but
this time will only do so if there was a range set on that clock to
begin with. It should be less intrusive, and reduce the number of
potential side effects considerably.

We then have a fix for the qcom rcg2 issue that has been reported
recently.

All the other patches should be probably be flagged as fixes, but
they've never seem to have shown any real-world issues until now, and
they aren't all really trivial to backport either, so I'm not sure it's
worth it.

There's also some documentation improvements for recalc_rate and
clk_get_rate to hopefully make the documentation less ambiguous and
acknowledge that recalc_rate() returning 0 on error is fine.

Let me know what you think,
Maxime

Changes from v8:
  - Fixed a regression when probing a clock driver backed by a device accessed
    through a bus that might sleep

Changes from v7:
  - Dropped the RPi fixes
  - Rebased on 6.0-rc1

Changes from v6:
  - Fixed a kernel build bot warning

Changes from v5:
  - Rebased on current next (next-20220711)
  - Dropped clk_get_rate_range, and used a custom function instead
  - Switched all tests to use clk_hw_get_clk() instead of struct clk_hw->clk
  - Removed some intermediate variables
  - Added some comments
  - Dropped clk_get_parent() changes
  - Dropped test on clk_hw pointer non-NULL in clk_hw_get_name
  - Made clk_has_parent more const

Changes from v4:
  - Fix build breakage on SAM9x60

Changes from v3:
  - constness warning fix in clk_core_forward_rate_req

Changes from v2:
  - Rebased on top of current next
  - Fixed locking issue in clk_get_rate_range

Changes from v1:
  - Rebased on top of next-20220428
  - Dropped the patch to prevent non-orphan clocks from registering if
    their recalc_rate hook returns 0
  - Added some patches to clarify the clk_get_rate and recalc_rate
    documentation
  - Dropped the patch to skip the range setup on an orphan clock that
    was introducing a regression on RaspberryPi3 when a monitor wasn't
    connected at boot
  - Added a patch to skip the rate clamping in clk_round_rate() when
    min_rate == max_rate == 0
  - Added a new set of functions to query the clk boundaries and fix a
    regression with the RaspberryPi4
  - Fixed all the drivers hand-crafting their clk_rate_request
  - Reworded the test suite descriptions
  - Reordered a few patches to ease the review
  - Reworded some commit logs to better explain the issues they address
  - Collected the Tested-by of Alexander and Marek
  - More tests

Maxime Ripard (25):
  clk: test: Switch to clk_hw_get_clk
  clk: Drop the rate range on clk_put()
  clk: Skip clamping when rounding if there's no boundaries
  clk: Mention that .recalc_rate can return 0 on error
  clk: Clarify clk_get_rate() expectations
  clk: tests: Add test suites description
  clk: tests: Add reference to the orphan mux bug report
  clk: tests: Add tests for uncached clock
  clk: tests: Add tests for single parent mux
  clk: tests: Add tests for mux with multiple parents
  clk: tests: Add some tests for orphan with multiple parents
  clk: Take into account uncached clocks in clk_set_rate_range()
  clk: Set req_rate on reparenting
  clk: Change clk_core_init_rate_req prototype
  clk: Move clk_core_init_rate_req() from clk_core_round_rate_nolock()
    to its caller
  clk: Introduce clk_hw_init_rate_request()
  clk: Add our request boundaries in clk_core_init_rate_req
  clk: Switch from __clk_determine_rate to clk_core_round_rate_nolock
  clk: Introduce clk_core_has_parent()
  clk: Constify clk_has_parent()
  clk: Stop forwarding clk_rate_requests to the parent
  clk: Zero the clk_rate_request structure
  clk: Introduce the clk_hw_get_rate_range function
  clk: qcom: clk-rcg2: Take clock boundaries into consideration for
    gfx3d
  clk: tests: Add missing test case for ranges

 drivers/clk/at91/clk-generated.c  |    5 +-
 drivers/clk/at91/clk-master.c     |    9 +-
 drivers/clk/at91/clk-peripheral.c |    4 +-
 drivers/clk/clk-composite.c       |    6 +-
 drivers/clk/clk-divider.c         |   20 +-
 drivers/clk/clk.c                 |  288 ++++--
 drivers/clk/clk_test.c            | 1413 ++++++++++++++++++++++++++++-
 drivers/clk/qcom/clk-rcg2.c       |    9 +
 include/linux/clk-provider.h      |   18 +-
 include/linux/clk.h               |    2 +-
 10 files changed, 1665 insertions(+), 109 deletions(-)

-- 
2.37.1

[PATCH v9 01/25] clk: test: Switch to clk_hw_get_clk

[Maxime Ripard]

Following the clk_hw->clk pointer is equivalent to calling
clk_hw_get_clk(), but will make the job harder if we need to rework that
part in the future.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 74 +++++++++++++++++++++++++++++++-----------
 1 file changed, 55 insertions(+), 19 deletions(-)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 6731a822f4e3..7646356f30cb 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -160,12 +160,14 @@ static void clk_test_get_rate(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, ctx->rate);
+
+	clk_put(clk);
 }
 
 /*
@@ -179,7 +181,7 @@ static void clk_test_set_get_rate(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -189,6 +191,8 @@ static void clk_test_set_get_rate(struct kunit *test)
 	rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+
+	clk_put(clk);
 }
 
 /*
@@ -202,7 +206,7 @@ static void clk_test_set_set_get_rate(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -216,6 +220,8 @@ static void clk_test_set_set_get_rate(struct kunit *test)
 	rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 /*
@@ -226,7 +232,7 @@ static void clk_test_round_set_get_rate(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rounded_rate, set_rate;
 
 	rounded_rate = clk_round_rate(clk, DUMMY_CLOCK_RATE_1);
@@ -240,6 +246,8 @@ static void clk_test_round_set_get_rate(struct kunit *test)
 	set_rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, set_rate, 0);
 	KUNIT_EXPECT_EQ(test, rounded_rate, set_rate);
+
+	clk_put(clk);
 }
 
 static struct kunit_case clk_test_cases[] = {
@@ -314,7 +322,7 @@ 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;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate, new_rate;
 
 	rate = clk_get_rate(clk);
@@ -329,6 +337,8 @@ static void clk_test_orphan_transparent_parent_mux_set_range(struct kunit *test)
 	new_rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, new_rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, new_rate);
+
+	clk_put(clk);
 }
 
 static struct kunit_case clk_orphan_transparent_single_parent_mux_test_cases[] = {
@@ -352,7 +362,7 @@ static void clk_range_test_set_range(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -365,6 +375,8 @@ static void clk_range_test_set_range(struct kunit *test)
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
 	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 /*
@@ -375,13 +387,15 @@ static void clk_range_test_set_range_invalid(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 
 	KUNIT_EXPECT_LT(test,
 			clk_set_rate_range(clk,
 					   DUMMY_CLOCK_RATE_1 + 1000,
 					   DUMMY_CLOCK_RATE_1),
 			0);
+
+	clk_put(clk);
 }
 
 /*
@@ -420,7 +434,7 @@ static void clk_range_test_set_range_round_rate_lower(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -433,6 +447,8 @@ static void clk_range_test_set_range_round_rate_lower(struct kunit *test)
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
 	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 /*
@@ -443,7 +459,7 @@ static void clk_range_test_set_range_set_rate_lower(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -460,6 +476,8 @@ static void clk_range_test_set_range_set_rate_lower(struct kunit *test)
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
 	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 /*
@@ -472,7 +490,7 @@ static void clk_range_test_set_range_set_round_rate_consistent_lower(struct kuni
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	long rounded;
 
 	KUNIT_ASSERT_EQ(test,
@@ -489,6 +507,8 @@ static void clk_range_test_set_range_set_round_rate_consistent_lower(struct kuni
 			0);
 
 	KUNIT_EXPECT_EQ(test, rounded, clk_get_rate(clk));
+
+	clk_put(clk);
 }
 
 /*
@@ -499,7 +519,7 @@ static void clk_range_test_set_range_round_rate_higher(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -512,6 +532,8 @@ static void clk_range_test_set_range_round_rate_higher(struct kunit *test)
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
 	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 /*
@@ -522,7 +544,7 @@ static void clk_range_test_set_range_set_rate_higher(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -539,6 +561,8 @@ static void clk_range_test_set_range_set_rate_higher(struct kunit *test)
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
 	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 /*
@@ -551,7 +575,7 @@ static void clk_range_test_set_range_set_round_rate_consistent_higher(struct kun
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	long rounded;
 
 	KUNIT_ASSERT_EQ(test,
@@ -568,6 +592,8 @@ static void clk_range_test_set_range_set_round_rate_consistent_higher(struct kun
 			0);
 
 	KUNIT_EXPECT_EQ(test, rounded, clk_get_rate(clk));
+
+	clk_put(clk);
 }
 
 /*
@@ -582,7 +608,7 @@ static void clk_range_test_set_range_get_rate_raised(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -598,6 +624,8 @@ static void clk_range_test_set_range_get_rate_raised(struct kunit *test)
 	rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+
+	clk_put(clk);
 }
 
 /*
@@ -612,7 +640,7 @@ static void clk_range_test_set_range_get_rate_lowered(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -628,6 +656,8 @@ static void clk_range_test_set_range_get_rate_lowered(struct kunit *test)
 	rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 static struct kunit_case clk_range_test_cases[] = {
@@ -664,7 +694,7 @@ static void clk_range_test_set_range_rate_maximized(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -700,6 +730,8 @@ static void clk_range_test_set_range_rate_maximized(struct kunit *test)
 	rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
 }
 
 /*
@@ -714,7 +746,7 @@ static void clk_range_test_multiple_set_range_rate_maximized(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	struct clk *user1, *user2;
 	unsigned long rate;
 
@@ -758,6 +790,7 @@ static void clk_range_test_multiple_set_range_rate_maximized(struct kunit *test)
 
 	clk_put(user2);
 	clk_put(user1);
+	clk_put(clk);
 }
 
 static struct kunit_case clk_range_maximize_test_cases[] = {
@@ -785,7 +818,7 @@ static void clk_range_test_set_range_rate_minimized(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	unsigned long rate;
 
 	KUNIT_ASSERT_EQ(test,
@@ -821,6 +854,8 @@ static void clk_range_test_set_range_rate_minimized(struct kunit *test)
 	rate = clk_get_rate(clk);
 	KUNIT_ASSERT_GT(test, rate, 0);
 	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+
+	clk_put(clk);
 }
 
 /*
@@ -835,7 +870,7 @@ static void clk_range_test_multiple_set_range_rate_minimized(struct kunit *test)
 {
 	struct clk_dummy_context *ctx = test->priv;
 	struct clk_hw *hw = &ctx->hw;
-	struct clk *clk = hw->clk;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
 	struct clk *user1, *user2;
 	unsigned long rate;
 
@@ -875,6 +910,7 @@ static void clk_range_test_multiple_set_range_rate_minimized(struct kunit *test)
 
 	clk_put(user2);
 	clk_put(user1);
+	clk_put(clk);
 }
 
 static struct kunit_case clk_range_minimize_test_cases[] = {
-- 
2.37.1

[PATCH v9 02/25] clk: Drop the rate range on clk_put()

[Maxime Ripard]

When clk_put() is called we don't make another clk_set_rate() call to
re-evaluate the rate boundaries. This is unlike clk_set_rate_range()
that evaluates the rate again each time it is called.

However, clk_put() is essentially equivalent to clk_set_rate_range()
since after clk_put() completes the consumer's boundaries shouldn'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 provider drivers can react. In
order to be as non-intrusive as possible, we'll just make that call if
the clock had non-default boundaries.

Also add a few tests to make sure this case is covered.

Fixes: c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c      |  45 +++++++++++------
 drivers/clk/clk_test.c | 110 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 141 insertions(+), 14 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 7fc191c15507..a5e0ab8bd6be 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -2325,19 +2325,15 @@ 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;
 
@@ -2350,8 +2346,6 @@ 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);
 
@@ -2395,6 +2389,28 @@ 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
+ *
+ * Return: 0 for success or negative errno on failure.
+ */
+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;
@@ -4348,9 +4364,10 @@ 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);
+
+	/* If we had any boundaries on that clock, let's drop them. */
+	if (clk->min_rate > 0 || clk->max_rate < ULONG_MAX)
+		clk_set_rate_range_nolock(clk, 0, ULONG_MAX);
 
 	owner = clk->core->owner;
 	kref_put(&clk->core->ref, __clk_release);
diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 7646356f30cb..7d9da88c39ee 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -793,9 +793,66 @@ static void clk_range_test_multiple_set_range_rate_maximized(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * 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 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 = clk_hw_get_clk(hw, NULL);
+	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);
+	clk_put(clk);
+}
+
 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),
 	{}
 };
 
@@ -913,9 +970,62 @@ static void clk_range_test_multiple_set_range_rate_minimized(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * 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 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 = clk_hw_get_clk(hw, NULL);
+	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);
+
+	clk_put(user1);
+	clk_put(clk);
+}
+
 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),
 	{}
 };
 
-- 
2.37.1

[PATCH v9 03/25] clk: Skip clamping when rounding if there's no boundaries

[Maxime Ripard]

Commit 948fb0969eae ("clk: Always clamp the rounded rate") recently
started to clamp the request rate in the clk_rate_request passed as an
argument of clk_core_determine_round_nolock() with the min_rate and
max_rate fields of that same request.

While the clk_rate_requests created by the framework itself always have
those fields set, some drivers will create it themselves and don't
always fill min_rate and max_rate.

In such a case, we end up clamping the rate with a minimum and maximum
of 0, thus always rounding the rate to 0.

Let's skip the clamping if both min_rate and max_rate are set to 0 and
complain so that it gets fixed.

Fixes: 948fb0969eae ("clk: Always clamp the rounded rate")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index a5e0ab8bd6be..9d63163244d4 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1341,7 +1341,19 @@ static int clk_core_determine_round_nolock(struct clk_core *core,
 	if (!core)
 		return 0;
 
-	req->rate = clamp(req->rate, req->min_rate, req->max_rate);
+	/*
+	 * Some clock providers hand-craft their clk_rate_requests and
+	 * might not fill min_rate and max_rate.
+	 *
+	 * If it's the case, clamping the rate is equivalent to setting
+	 * the rate to 0 which is bad. Skip the clamping but complain so
+	 * that it gets fixed, hopefully.
+	 */
+	if (!req->min_rate && !req->max_rate)
+		pr_warn("%s: %s: clk_rate_request has initialized min or max rate.\n",
+			__func__, core->name);
+	else
+		req->rate = clamp(req->rate, req->min_rate, req->max_rate);
 
 	/*
 	 * At this point, core protection will be disabled
-- 
2.37.1

[PATCH v9 04/25] clk: Mention that .recalc_rate can return 0 on error

[Maxime Ripard]

Multiple platforms (amlogic, imx8) return 0 when the clock rate cannot
be determined properly by the recalc_rate hook. Mention in the
documentation that the framework is ok with that.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 include/linux/clk-provider.h | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/include/linux/clk-provider.h b/include/linux/clk-provider.h
index 1615010aa0ec..9a14cfa0d201 100644
--- a/include/linux/clk-provider.h
+++ b/include/linux/clk-provider.h
@@ -118,8 +118,9 @@ struct clk_duty {
  *
  * @recalc_rate	Recalculate the rate of this clock, by querying hardware. The
  *		parent rate is an input parameter.  It is up to the caller to
- *		ensure that the prepare_mutex is held across this call.
- *		Returns the calculated rate.  Optional, but recommended - if
+ *		ensure that the prepare_mutex is held across this call. If the
+ *		driver cannot figure out a rate for this clock, it must return
+ *		0. Returns the calculated rate. Optional, but recommended - if
  *		this op is not set then clock rate will be initialized to 0.
  *
  * @round_rate:	Given a target rate as input, returns the closest rate actually
-- 
2.37.1

[PATCH v9 05/25] clk: Clarify clk_get_rate() expectations

[Maxime Ripard]

As shown by a number of clock users already, clk_get_rate() can be
called whether or not the clock is enabled.

Similarly, a number of clock drivers will return a rate of 0 whenever
the rate cannot be figured out.

Since it was a bit ambiguous before, let's make it clear in the
clk_get_rate() documentation.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 9d63163244d4..caa2eb640441 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1672,8 +1672,9 @@ static unsigned long clk_core_get_rate_recalc(struct clk_core *core)
  * @clk: the clk whose rate is being returned
  *
  * Simply returns the cached rate of the clk, unless CLK_GET_RATE_NOCACHE flag
- * is set, which means a recalc_rate will be issued.
- * If clk is NULL then returns 0.
+ * is set, which means a recalc_rate will be issued. Can be called regardless of
+ * the clock enabledness. If clk is NULL, or if an error occurred, then returns
+ * 0.
  */
 unsigned long clk_get_rate(struct clk *clk)
 {
-- 
2.37.1

[PATCH v9 06/25] clk: tests: Add test suites description

[Maxime Ripard]

We start to have a few test suites, and we'll add more, so it will get
pretty confusing to figure out what is supposed to be tested in what
suite.

Let's add some comments to explain what setup they create, and what we
should be testing in every suite.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 7d9da88c39ee..1a7cb482ec58 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -258,6 +258,11 @@ static struct kunit_case clk_test_cases[] = {
 	{}
 };
 
+/*
+ * Test suite for a basic rate clock, without any parent.
+ *
+ * These tests exercise the rate API with simple scenarios
+ */
 static struct kunit_suite clk_test_suite = {
 	.name = "clk-test",
 	.init = clk_test_init,
@@ -346,6 +351,14 @@ static struct kunit_case clk_orphan_transparent_single_parent_mux_test_cases[] =
 	{}
 };
 
+/*
+ * Test suite for a basic mux clock with one parent. The parent is
+ * registered after its child. The clock will thus be an orphan when
+ * registered, but will no longer be when the tests run.
+ *
+ * These tests make sure a clock that used to be orphan has a sane,
+ * consistent, behaviour.
+ */
 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,
@@ -675,6 +688,12 @@ static struct kunit_case clk_range_test_cases[] = {
 	{}
 };
 
+/*
+ * Test suite for a basic rate clock, without any parent.
+ *
+ * These tests exercise the rate range API: clk_set_rate_range(),
+ * clk_set_min_rate(), clk_set_max_rate(), clk_drop_range().
+ */
 static struct kunit_suite clk_range_test_suite = {
 	.name = "clk-range-test",
 	.init = clk_test_init,
@@ -856,6 +875,13 @@ static struct kunit_case clk_range_maximize_test_cases[] = {
 	{}
 };
 
+/*
+ * Test suite for a basic rate clock, without any parent.
+ *
+ * These tests exercise the rate range API: clk_set_rate_range(),
+ * clk_set_min_rate(), clk_set_max_rate(), clk_drop_range(), with a
+ * driver that will always try to run at the highest possible rate.
+ */
 static struct kunit_suite clk_range_maximize_test_suite = {
 	.name = "clk-range-maximize-test",
 	.init = clk_maximize_test_init,
@@ -1029,6 +1055,13 @@ static struct kunit_case clk_range_minimize_test_cases[] = {
 	{}
 };
 
+/*
+ * Test suite for a basic rate clock, without any parent.
+ *
+ * These tests exercise the rate range API: clk_set_rate_range(),
+ * clk_set_min_rate(), clk_set_max_rate(), clk_drop_range(), with a
+ * driver that will always try to run at the lowest possible rate.
+ */
 static struct kunit_suite clk_range_minimize_test_suite = {
 	.name = "clk-range-minimize-test",
 	.init = clk_minimize_test_init,
-- 
2.37.1

[PATCH v9 07/25] clk: tests: Add reference to the orphan mux bug report

[Maxime Ripard]

Some more context might be useful for unit-tests covering a previously
reported bug, so let's add a link to the discussion for that bug.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 1a7cb482ec58..b8e32406a6e4 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -322,6 +322,9 @@ static void clk_orphan_transparent_single_parent_mux_test_exit(struct kunit *tes
 /*
  * 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.
+ *
+ * See:
+ * https://lore.kernel.org/linux-clk/7720158d-10a7-a17b-73a4-a8615c9c6d5c@collabora.com/
  */
 static void clk_test_orphan_transparent_parent_mux_set_range(struct kunit *test)
 {
-- 
2.37.1

[PATCH v9 08/25] clk: tests: Add tests for uncached clock

[Maxime Ripard]

The clock framework supports clocks that can have their rate changed
without the kernel knowing about it using the CLK_GET_RATE_NOCACHE flag.

As its name suggests, this flag turns off the rate caching in the clock
framework, reading out the rate from the hardware any time we need to
read it.

Let's add a couple of tests to make sure it works as intended.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 93 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 92 insertions(+), 1 deletion(-)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index b8e32406a6e4..b269420dafcc 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -270,6 +270,96 @@ static struct kunit_suite clk_test_suite = {
 	.test_cases = clk_test_cases,
 };
 
+static int clk_uncached_test_init(struct kunit *test)
+{
+	struct clk_dummy_context *ctx;
+	int ret;
+
+	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	test->priv = ctx;
+
+	ctx->rate = DUMMY_CLOCK_INIT_RATE;
+	ctx->hw.init = CLK_HW_INIT_NO_PARENT("test-clk",
+					     &clk_dummy_rate_ops,
+					     CLK_GET_RATE_NOCACHE);
+
+	ret = clk_hw_register(NULL, &ctx->hw);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+/*
+ * Test that for an uncached clock, the clock framework doesn't cache
+ * the rate and clk_get_rate() will return the underlying clock rate
+ * even if it changed.
+ */
+static void clk_test_uncached_get_rate(struct kunit *test)
+{
+	struct clk_dummy_context *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	unsigned long rate;
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_INIT_RATE);
+
+	/* We change the rate behind the clock framework's back */
+	ctx->rate = DUMMY_CLOCK_RATE_1;
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+
+	clk_put(clk);
+}
+
+/*
+ * Test that for an uncached clock, clk_set_rate_range() will work
+ * properly if the rate hasn't changed.
+ */
+static void clk_test_uncached_set_range(struct kunit *test)
+{
+	struct clk_dummy_context *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	unsigned long rate;
+
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate_range(clk,
+					   DUMMY_CLOCK_RATE_1,
+					   DUMMY_CLOCK_RATE_2),
+			0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
+}
+
+static struct kunit_case clk_uncached_test_cases[] = {
+	KUNIT_CASE(clk_test_uncached_get_rate),
+	KUNIT_CASE(clk_test_uncached_set_range),
+	{}
+};
+
+/*
+ * Test suite for a basic, uncached, rate clock, without any parent.
+ *
+ * These tests exercise the rate API with simple scenarios
+ */
+static struct kunit_suite clk_uncached_test_suite = {
+	.name = "clk-uncached-test",
+	.init = clk_uncached_test_init,
+	.exit = clk_test_exit,
+	.test_cases = clk_uncached_test_cases,
+};
+
 struct clk_single_parent_ctx {
 	struct clk_dummy_context parent_ctx;
 	struct clk_hw hw;
@@ -1077,6 +1167,7 @@ kunit_test_suites(
 	&clk_orphan_transparent_single_parent_test_suite,
 	&clk_range_test_suite,
 	&clk_range_maximize_test_suite,
-	&clk_range_minimize_test_suite
+	&clk_range_minimize_test_suite,
+	&clk_uncached_test_suite
 );
 MODULE_LICENSE("GPL v2");
-- 
2.37.1

[PATCH v9 09/25] clk: tests: Add tests for single parent mux

[Maxime Ripard]

We have a few tests for a mux with a single parent, testing the case
where it used to be orphan.

Let's leverage most of the code but register the clock properly to test
a few trivial things.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 194 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 185 insertions(+), 9 deletions(-)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index b269420dafcc..06c9220873bb 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -365,6 +365,189 @@ struct clk_single_parent_ctx {
 	struct clk_hw hw;
 };
 
+static int clk_single_parent_mux_test_init(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx;
+	int ret;
+
+	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	test->priv = ctx;
+
+	ctx->parent_ctx.rate = DUMMY_CLOCK_INIT_RATE;
+	ctx->parent_ctx.hw.init =
+		CLK_HW_INIT_NO_PARENT("parent-clk",
+				      &clk_dummy_rate_ops,
+				      0);
+
+	ret = clk_hw_register(NULL, &ctx->parent_ctx.hw);
+	if (ret)
+		return ret;
+
+	ctx->hw.init = CLK_HW_INIT("test-clk", "parent-clk",
+				   &clk_dummy_single_parent_ops,
+				   CLK_SET_RATE_PARENT);
+
+	ret = clk_hw_register(NULL, &ctx->hw);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void
+clk_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 for a clock with a single parent, clk_get_parent() actually
+ * returns the parent.
+ */
+static void
+clk_test_single_parent_mux_get_parent(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent = clk_hw_get_clk(&ctx->parent_ctx.hw, NULL);
+
+	KUNIT_EXPECT_TRUE(test, clk_is_match(clk_get_parent(clk), parent));
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
+/*
+ * Test that for a clock that can't modify its rate and with a single
+ * parent, if we set disjoints range on the parent and then the child,
+ * the second will return an error.
+ *
+ * FIXME: clk_set_rate_range() only considers the current clock when
+ * evaluating whether ranges are disjoints and not the upstream clocks
+ * ranges.
+ */
+static void
+clk_test_single_parent_mux_set_range_disjoint_child_last(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	int ret;
+
+	kunit_skip(test, "This needs to be fixed in the core.");
+
+	parent = clk_get_parent(clk);
+	KUNIT_ASSERT_PTR_NE(test, parent, NULL);
+
+	ret = clk_set_rate_range(parent, 1000, 2000);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate_range(clk, 3000, 4000);
+	KUNIT_EXPECT_LT(test, ret, 0);
+
+	clk_put(clk);
+}
+
+/*
+ * Test that for a clock that can't modify its rate and with a single
+ * parent, if we set disjoints range on the child and then the parent,
+ * the second will return an error.
+ *
+ * FIXME: clk_set_rate_range() only considers the current clock when
+ * evaluating whether ranges are disjoints and not the downstream clocks
+ * ranges.
+ */
+static void
+clk_test_single_parent_mux_set_range_disjoint_parent_last(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	int ret;
+
+	kunit_skip(test, "This needs to be fixed in the core.");
+
+	parent = clk_get_parent(clk);
+	KUNIT_ASSERT_PTR_NE(test, parent, NULL);
+
+	ret = clk_set_rate_range(clk, 1000, 2000);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate_range(parent, 3000, 4000);
+	KUNIT_EXPECT_LT(test, ret, 0);
+
+	clk_put(clk);
+}
+
+/*
+ * Test that for a clock that can't modify its rate and with a single
+ * parent, if we set a range on the parent and a more restrictive one on
+ * the child, and then call clk_round_rate(), the boundaries of the
+ * two clocks are taken into account.
+ */
+static void
+clk_test_single_parent_mux_set_range_round_rate_child_smaller(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long rate;
+	int ret;
+
+	parent = clk_get_parent(clk);
+	KUNIT_ASSERT_PTR_NE(test, parent, NULL);
+
+	ret = clk_set_rate_range(parent, DUMMY_CLOCK_RATE_1, DUMMY_CLOCK_RATE_2);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate_range(clk, DUMMY_CLOCK_RATE_1 + 1000, DUMMY_CLOCK_RATE_2 - 1000);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_round_rate(clk, DUMMY_CLOCK_RATE_1 - 1000);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1 + 1000);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2 - 1000);
+
+	rate = clk_round_rate(clk, DUMMY_CLOCK_RATE_2 + 1000);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1 + 1000);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2 - 1000);
+
+	clk_put(clk);
+}
+
+static struct kunit_case clk_single_parent_mux_test_cases[] = {
+	KUNIT_CASE(clk_test_single_parent_mux_get_parent),
+	KUNIT_CASE(clk_test_single_parent_mux_set_range_disjoint_child_last),
+	KUNIT_CASE(clk_test_single_parent_mux_set_range_disjoint_parent_last),
+	KUNIT_CASE(clk_test_single_parent_mux_set_range_round_rate_child_smaller),
+	{}
+};
+
+/*
+ * Test suite for a basic mux clock with one parent, with
+ * CLK_SET_RATE_PARENT on the child.
+ *
+ * These tests exercise the consumer API and check that the state of the
+ * child and parent are sane and consistent.
+ */
+static struct kunit_suite
+clk_single_parent_mux_test_suite = {
+	.name = "clk-single-parent-mux-test",
+	.init = clk_single_parent_mux_test_init,
+	.exit = clk_single_parent_mux_test_exit,
+	.test_cases = clk_single_parent_mux_test_cases,
+};
+
 static int clk_orphan_transparent_single_parent_mux_test_init(struct kunit *test)
 {
 	struct clk_single_parent_ctx *ctx;
@@ -401,14 +584,6 @@ static int clk_orphan_transparent_single_parent_mux_test_init(struct kunit *test
 	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.
@@ -455,7 +630,7 @@ static struct kunit_case clk_orphan_transparent_single_parent_mux_test_cases[] =
 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,
+	.exit = clk_single_parent_mux_test_exit,
 	.test_cases = clk_orphan_transparent_single_parent_mux_test_cases,
 };
 
@@ -1168,6 +1343,7 @@ kunit_test_suites(
 	&clk_range_test_suite,
 	&clk_range_maximize_test_suite,
 	&clk_range_minimize_test_suite,
+	&clk_single_parent_mux_test_suite,
 	&clk_uncached_test_suite
 );
 MODULE_LICENSE("GPL v2");
-- 
2.37.1

[PATCH v9 10/25] clk: tests: Add tests for mux with multiple parents

[Maxime Ripard]

We'll need to test a few corner cases that occur when we have a mux
clock whose default parent is missing.

For now, let's create the context structure and the trivial ops, along
with a test suite that just tests trivial things for now, without
considering the orphan case.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 121 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 121 insertions(+)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 06c9220873bb..1ccafd4fabff 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -108,6 +108,39 @@ static const struct clk_ops clk_dummy_single_parent_ops = {
 	.get_parent = clk_dummy_single_get_parent,
 };
 
+struct clk_multiple_parent_ctx {
+	struct clk_dummy_context parents_ctx[2];
+	struct clk_hw hw;
+	u8 current_parent;
+};
+
+static int clk_multiple_parents_mux_set_parent(struct clk_hw *hw, u8 index)
+{
+	struct clk_multiple_parent_ctx *ctx =
+		container_of(hw, struct clk_multiple_parent_ctx, hw);
+
+	if (index >= clk_hw_get_num_parents(hw))
+		return -EINVAL;
+
+	ctx->current_parent = index;
+
+	return 0;
+}
+
+static u8 clk_multiple_parents_mux_get_parent(struct clk_hw *hw)
+{
+	struct clk_multiple_parent_ctx *ctx =
+		container_of(hw, struct clk_multiple_parent_ctx, hw);
+
+	return ctx->current_parent;
+}
+
+static const struct clk_ops clk_multiple_parents_mux_ops = {
+	.get_parent = clk_multiple_parents_mux_get_parent,
+	.set_parent = clk_multiple_parents_mux_set_parent,
+	.determine_rate = __clk_mux_determine_rate_closest,
+};
+
 static int clk_test_init_with_ops(struct kunit *test, const struct clk_ops *ops)
 {
 	struct clk_dummy_context *ctx;
@@ -360,6 +393,93 @@ static struct kunit_suite clk_uncached_test_suite = {
 	.test_cases = clk_uncached_test_cases,
 };
 
+static int
+clk_multiple_parents_mux_test_init(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx;
+	const char *parents[2] = { "parent-0", "parent-1"};
+	int ret;
+
+	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	test->priv = ctx;
+
+	ctx->parents_ctx[0].hw.init = CLK_HW_INIT_NO_PARENT("parent-0",
+							    &clk_dummy_rate_ops,
+							    0);
+	ctx->parents_ctx[0].rate = DUMMY_CLOCK_RATE_1;
+	ret = clk_hw_register(NULL, &ctx->parents_ctx[0].hw);
+	if (ret)
+		return ret;
+
+	ctx->parents_ctx[1].hw.init = CLK_HW_INIT_NO_PARENT("parent-1",
+							    &clk_dummy_rate_ops,
+							    0);
+	ctx->parents_ctx[1].rate = DUMMY_CLOCK_RATE_2;
+	ret = clk_hw_register(NULL, &ctx->parents_ctx[1].hw);
+	if (ret)
+		return ret;
+
+	ctx->current_parent = 0;
+	ctx->hw.init = CLK_HW_INIT_PARENTS("test-mux", parents,
+					   &clk_multiple_parents_mux_ops,
+					   CLK_SET_RATE_PARENT);
+	ret = clk_hw_register(NULL, &ctx->hw);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void
+clk_multiple_parents_mux_test_exit(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+
+	clk_hw_unregister(&ctx->hw);
+	clk_hw_unregister(&ctx->parents_ctx[0].hw);
+	clk_hw_unregister(&ctx->parents_ctx[1].hw);
+}
+
+/*
+ * Test that for a clock with multiple parents, clk_get_parent()
+ * actually returns the current one.
+ */
+static void
+clk_test_multiple_parents_mux_get_parent(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent = clk_hw_get_clk(&ctx->parents_ctx[0].hw, NULL);
+
+	KUNIT_EXPECT_TRUE(test, clk_is_match(clk_get_parent(clk), parent));
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
+static struct kunit_case clk_multiple_parents_mux_test_cases[] = {
+	KUNIT_CASE(clk_test_multiple_parents_mux_get_parent),
+	{}
+};
+
+/*
+ * Test suite for a basic mux clock with two parents, with
+ * CLK_SET_RATE_PARENT on the child.
+ *
+ * These tests exercise the consumer API and check that the state of the
+ * child and parents are sane and consistent.
+ */
+static struct kunit_suite
+clk_multiple_parents_mux_test_suite = {
+	.name = "clk-multiple-parents-mux-test",
+	.init = clk_multiple_parents_mux_test_init,
+	.exit = clk_multiple_parents_mux_test_exit,
+	.test_cases = clk_multiple_parents_mux_test_cases,
+};
+
 struct clk_single_parent_ctx {
 	struct clk_dummy_context parent_ctx;
 	struct clk_hw hw;
@@ -1339,6 +1459,7 @@ static struct kunit_suite clk_range_minimize_test_suite = {
 
 kunit_test_suites(
 	&clk_test_suite,
+	&clk_multiple_parents_mux_test_suite,
 	&clk_orphan_transparent_single_parent_test_suite,
 	&clk_range_test_suite,
 	&clk_range_maximize_test_suite,
-- 
2.37.1

[PATCH v9 11/25] clk: tests: Add some tests for orphan with multiple parents

[Maxime Ripard]

Let's leverage the dummy mux with multiple parents we have to create a
mux whose default parent will never be registered, and thus will always
be orphan by default.

We can then create some tests to make sure that the clock API behaves
properly in such a case, and that the transition to a non-orphan clock
when we change the parent is done properly.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 237 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 237 insertions(+)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 1ccafd4fabff..ceed49c5a88b 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -480,6 +480,242 @@ clk_multiple_parents_mux_test_suite = {
 	.test_cases = clk_multiple_parents_mux_test_cases,
 };
 
+static int
+clk_orphan_transparent_multiple_parent_mux_test_init(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx;
+	const char *parents[2] = { "missing-parent", "proper-parent"};
+	int ret;
+
+	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	test->priv = ctx;
+
+	ctx->parents_ctx[1].hw.init = CLK_HW_INIT_NO_PARENT("proper-parent",
+							    &clk_dummy_rate_ops,
+							    0);
+	ctx->parents_ctx[1].rate = DUMMY_CLOCK_INIT_RATE;
+	ret = clk_hw_register(NULL, &ctx->parents_ctx[1].hw);
+	if (ret)
+		return ret;
+
+	ctx->hw.init = CLK_HW_INIT_PARENTS("test-orphan-mux", parents,
+					   &clk_multiple_parents_mux_ops,
+					   CLK_SET_RATE_PARENT);
+	ret = clk_hw_register(NULL, &ctx->hw);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void
+clk_orphan_transparent_multiple_parent_mux_test_exit(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+
+	clk_hw_unregister(&ctx->hw);
+	clk_hw_unregister(&ctx->parents_ctx[1].hw);
+}
+
+/*
+ * Test that, for a mux whose current parent hasn't been registered yet and is
+ * thus orphan, clk_get_parent() will return NULL.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_get_parent(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+
+	KUNIT_EXPECT_PTR_EQ(test, clk_get_parent(clk), NULL);
+
+	clk_put(clk);
+}
+
+/*
+ * Test that, for a mux whose current parent hasn't been registered yet,
+ * calling clk_set_parent() to a valid parent will properly update the
+ * mux parent and its orphan status.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_parent(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent, *new_parent;
+	int ret;
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+
+	ret = clk_set_parent(clk, parent);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	new_parent = clk_get_parent(clk);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+	KUNIT_EXPECT_TRUE(test, clk_is_match(parent, new_parent));
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
+/*
+ * Test that, for a mux that started orphan but got switched to a valid
+ * parent, the rate of the mux and its new parent are consistent.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_parent_get_rate(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long parent_rate, rate;
+	int ret;
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+
+	parent_rate = clk_get_rate(parent);
+	KUNIT_ASSERT_GT(test, parent_rate, 0);
+
+	ret = clk_set_parent(clk, parent);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, parent_rate, rate);
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
+/*
+ * Test that, for a mux that started orphan but got switched to a valid
+ * parent, calling clk_set_rate_range() will affect the parent state if
+ * its rate is out of range.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_modified(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long rate;
+	int ret;
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+
+	ret = clk_set_parent(clk, parent);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate_range(clk, DUMMY_CLOCK_RATE_1, DUMMY_CLOCK_RATE_2);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
+/*
+ * Test that, for a mux whose current parent hasn't been registered yet,
+ * calling clk_set_rate_range() will succeed, and will be taken into
+ * account when rounding a rate.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_range_round_rate(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	unsigned long rate;
+	int ret;
+
+	ret = clk_set_rate_range(clk, DUMMY_CLOCK_RATE_1, DUMMY_CLOCK_RATE_2);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_round_rate(clk, DUMMY_CLOCK_RATE_1 - 1000);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
+}
+
+/*
+ * Test that, for a mux that started orphan, was assigned and rate and
+ * then got switched to a valid parent, its rate is eventually within
+ * range.
+ *
+ * FIXME: Even though we update the rate as part of clk_set_parent(), we
+ * don't evaluate whether that new rate is within range and needs to be
+ * adjusted.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_range_set_parent_get_rate(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long rate;
+	int ret;
+
+	kunit_skip(test, "This needs to be fixed in the core.");
+
+	clk_hw_set_rate_range(hw, DUMMY_CLOCK_RATE_1, DUMMY_CLOCK_RATE_2);
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+
+	ret = clk_set_parent(clk, parent);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
+static struct kunit_case clk_orphan_transparent_multiple_parent_mux_test_cases[] = {
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_get_parent),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_get_rate),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_modified),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_range_round_rate),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_range_set_parent_get_rate),
+	{}
+};
+
+/*
+ * Test suite for a basic mux clock with two parents. The default parent
+ * isn't registered, only the second parent is. By default, the clock
+ * will thus be orphan.
+ *
+ * These tests exercise the behaviour of the consumer API when dealing
+ * with an orphan clock, and how we deal with the transition to a valid
+ * parent.
+ */
+static struct kunit_suite clk_orphan_transparent_multiple_parent_mux_test_suite = {
+	.name = "clk-orphan-transparent-multiple-parent-mux-test",
+	.init = clk_orphan_transparent_multiple_parent_mux_test_init,
+	.exit = clk_orphan_transparent_multiple_parent_mux_test_exit,
+	.test_cases = clk_orphan_transparent_multiple_parent_mux_test_cases,
+};
+
 struct clk_single_parent_ctx {
 	struct clk_dummy_context parent_ctx;
 	struct clk_hw hw;
@@ -1460,6 +1696,7 @@ static struct kunit_suite clk_range_minimize_test_suite = {
 kunit_test_suites(
 	&clk_test_suite,
 	&clk_multiple_parents_mux_test_suite,
+	&clk_orphan_transparent_multiple_parent_mux_test_suite,
 	&clk_orphan_transparent_single_parent_test_suite,
 	&clk_range_test_suite,
 	&clk_range_maximize_test_suite,
-- 
2.37.1

[PATCH v9 12/25] clk: Take into account uncached clocks in clk_set_rate_range()

[Maxime Ripard]

clk_set_rate_range() will use the last requested rate for the clock when
it calls into the driver set_rate hook.

However, if CLK_GET_RATE_NOCACHE is set on that clock, the last
requested rate might not be matching the current rate of the clock. In
such a case, let's read out the rate from the hardware and use that in
our set_rate instead.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c      |  6 +++++-
 drivers/clk/clk_test.c | 31 +++++++++++++++++++++++++++++++
 2 files changed, 36 insertions(+), 1 deletion(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index caa2eb640441..53b28e63deae 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -2373,6 +2373,10 @@ static int clk_set_rate_range_nolock(struct clk *clk,
 		goto out;
 	}
 
+	rate = clk->core->req_rate;
+	if (clk->core->flags & CLK_GET_RATE_NOCACHE)
+		rate = clk_core_get_rate_recalc(clk->core);
+
 	/*
 	 * Since the boundaries have been changed, let's give the
 	 * opportunity to the provider to adjust the clock rate based on
@@ -2390,7 +2394,7 @@ 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);
+	rate = clamp(rate, min, max);
 	ret = clk_core_set_rate_nolock(clk->core, rate);
 	if (ret) {
 		/* rollback the changes */
diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index ceed49c5a88b..d3e121f21ae2 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -375,9 +375,40 @@ static void clk_test_uncached_set_range(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * Test that for an uncached clock, clk_set_rate_range() will work
+ * properly if the rate has changed in hardware.
+ *
+ * In this case, it means that if the rate wasn't initially in the range
+ * we're trying to set, but got changed at some point into the range
+ * without the kernel knowing about it, its rate shouldn't be affected.
+ */
+static void clk_test_uncached_updated_rate_set_range(struct kunit *test)
+{
+	struct clk_dummy_context *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	unsigned long rate;
+
+	/* We change the rate behind the clock framework's back */
+	ctx->rate = DUMMY_CLOCK_RATE_1 + 1000;
+	KUNIT_ASSERT_EQ(test,
+			clk_set_rate_range(clk,
+					   DUMMY_CLOCK_RATE_1,
+					   DUMMY_CLOCK_RATE_2),
+			0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_RATE_1 + 1000);
+
+	clk_put(clk);
+}
+
 static struct kunit_case clk_uncached_test_cases[] = {
 	KUNIT_CASE(clk_test_uncached_get_rate),
 	KUNIT_CASE(clk_test_uncached_set_range),
+	KUNIT_CASE(clk_test_uncached_updated_rate_set_range),
 	{}
 };
 
-- 
2.37.1

[PATCH v9 13/25] clk: Set req_rate on reparenting

[Maxime Ripard]

If a non-rate clock started by default with a parent that never
registered, core->req_rate will be 0. The expectation is that whenever
the parent will be registered, req_rate will be updated with the new
value that has just been computed.

However, if that clock is a mux, clk_set_parent() can also make that
clock no longer orphan. In this case however, we never update req_rate.

The natural solution to this would be to update core->rate and
core->req_rate in clk_reparent() by calling clk_recalc().

However, this doesn't work in all cases. Indeed, clk_recalc() is called
by __clk_set_parent_before(), __clk_set_parent() and
clk_core_reparent(). Both __clk_set_parent_before() and __clk_set_parent
will call clk_recalc() with the enable_lock taken through a call to
clk_enable_lock(), the underlying locking primitive being a spinlock.

clk_recalc() calls the backing driver .recalc_rate hook, and that
implementation might sleep if the underlying device uses a bus with
accesses that might sleep, such as i2c.

In such a situation, we would end up sleeping while holding a spinlock,
and thus in an atomic section.

In order to work around this, we can move the core->rate and
core->req_rate update to the clk_recalc() calling sites, after the
enable_lock has been released if it was taken.

The only situation that could still be problematic is the
clk_core_reparent() -> clk_reparent() case that doesn't have any
locking. clk_core_reparent() is itself called by clk_hw_reparent(),
which is then called by 4 drivers:

  * clk-stm32mp1.c, stm32/clk-stm32-core.c and tegra/clk-tegra210-emc.c
    use it in their set_parent implementation. The set_parent hook is
    only called by __clk_set_parent() and clk_change_rate(), both of
    them calling it without the enable_lock taken.

  * clk/tegra/clk-tegra124-emc.c calls it as part of its set_rate
    implementation. set_rate is only called by clk_change_rate(), again
    without the enable_lock taken.

In both cases we can't end up in a situation where the clk_hw_reparent()
caller would hold a spinlock, so it seems like this is a good
workaround.

Let's also add some unit tests to make sure we cover the original bug.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c      |  22 ++++
 drivers/clk/clk_test.c | 239 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 261 insertions(+)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 53b28e63deae..91bb1ea0e147 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1765,6 +1765,23 @@ static void clk_core_update_orphan_status(struct clk_core *core, bool is_orphan)
 		clk_core_update_orphan_status(child, is_orphan);
 }
 
+/*
+ * Update the orphan rate and req_rate of @core and all its children.
+ */
+static void clk_core_update_orphan_child_rates(struct clk_core *core)
+{
+	struct clk_core *child;
+	unsigned long parent_rate = 0;
+
+	if (core->parent)
+		parent_rate = core->parent->rate;
+
+	core->rate = core->req_rate = clk_recalc(core, parent_rate);
+
+	hlist_for_each_entry(child, &core->children, child_node)
+		clk_core_update_orphan_child_rates(child);
+}
+
 static void clk_reparent(struct clk_core *core, struct clk_core *new_parent)
 {
 	bool was_orphan = core->orphan;
@@ -1834,6 +1851,8 @@ static struct clk_core *__clk_set_parent_before(struct clk_core *core,
 	clk_reparent(core, parent);
 	clk_enable_unlock(flags);
 
+	clk_core_update_orphan_child_rates(core);
+
 	return old_parent;
 }
 
@@ -1878,6 +1897,8 @@ static int __clk_set_parent(struct clk_core *core, struct clk_core *parent,
 		flags = clk_enable_lock();
 		clk_reparent(core, old_parent);
 		clk_enable_unlock(flags);
+
+		clk_core_update_orphan_child_rates(core);
 		__clk_set_parent_after(core, old_parent, parent);
 
 		return ret;
@@ -2506,6 +2527,7 @@ static void clk_core_reparent(struct clk_core *core,
 				  struct clk_core *new_parent)
 {
 	clk_reparent(core, new_parent);
+	clk_core_update_orphan_child_rates(core);
 	__clk_recalc_accuracies(core);
 	__clk_recalc_rates(core, POST_RATE_CHANGE);
 }
diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index d3e121f21ae2..d1b1372f7aaa 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -594,6 +594,41 @@ clk_test_orphan_transparent_multiple_parent_mux_set_parent(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * Test that, for a mux that started orphan but got switched to a valid
+ * parent, calling clk_drop_range() on the mux won't affect the parent
+ * rate.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_parent_drop_range(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long parent_rate, new_parent_rate;
+	int ret;
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+
+	parent_rate = clk_get_rate(parent);
+	KUNIT_ASSERT_GT(test, parent_rate, 0);
+
+	ret = clk_set_parent(clk, parent);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_drop_range(clk);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	new_parent_rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, new_parent_rate, 0);
+	KUNIT_EXPECT_EQ(test, parent_rate, new_parent_rate);
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
 /*
  * Test that, for a mux that started orphan but got switched to a valid
  * parent, the rate of the mux and its new parent are consistent.
@@ -625,6 +660,39 @@ clk_test_orphan_transparent_multiple_parent_mux_set_parent_get_rate(struct kunit
 	clk_put(clk);
 }
 
+/*
+ * Test that, for a mux that started orphan but got switched to a valid
+ * parent, calling clk_put() on the mux won't affect the parent rate.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_parent_put(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk *clk, *parent;
+	unsigned long parent_rate, new_parent_rate;
+	int ret;
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+
+	clk = clk_hw_get_clk(&ctx->hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, clk);
+
+	parent_rate = clk_get_rate(parent);
+	KUNIT_ASSERT_GT(test, parent_rate, 0);
+
+	ret = clk_set_parent(clk, parent);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	clk_put(clk);
+
+	new_parent_rate = clk_get_rate(parent);
+	KUNIT_ASSERT_GT(test, new_parent_rate, 0);
+	KUNIT_EXPECT_EQ(test, parent_rate, new_parent_rate);
+
+	clk_put(parent);
+}
+
 /*
  * Test that, for a mux that started orphan but got switched to a valid
  * parent, calling clk_set_rate_range() will affect the parent state if
@@ -658,6 +726,43 @@ clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_modified(st
 	clk_put(clk);
 }
 
+/*
+ * Test that, for a mux that started orphan but got switched to a valid
+ * parent, calling clk_set_rate_range() won't affect the parent state if
+ * its rate is within range.
+ */
+static void
+clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_untouched(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long parent_rate, new_parent_rate;
+	int ret;
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
+
+	parent_rate = clk_get_rate(parent);
+	KUNIT_ASSERT_GT(test, parent_rate, 0);
+
+	ret = clk_set_parent(clk, parent);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate_range(clk,
+				 DUMMY_CLOCK_INIT_RATE - 1000,
+				 DUMMY_CLOCK_INIT_RATE + 1000);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	new_parent_rate = clk_get_rate(parent);
+	KUNIT_ASSERT_GT(test, new_parent_rate, 0);
+	KUNIT_EXPECT_EQ(test, parent_rate, new_parent_rate);
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
 /*
  * Test that, for a mux whose current parent hasn't been registered yet,
  * calling clk_set_rate_range() will succeed, and will be taken into
@@ -724,8 +829,11 @@ clk_test_orphan_transparent_multiple_parent_mux_set_range_set_parent_get_rate(st
 static struct kunit_case clk_orphan_transparent_multiple_parent_mux_test_cases[] = {
 	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_get_parent),
 	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_drop_range),
 	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_get_rate),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_put),
 	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_modified),
+	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_untouched),
 	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_range_round_rate),
 	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_range_set_parent_get_rate),
 	{}
@@ -1021,6 +1129,136 @@ static struct kunit_suite clk_orphan_transparent_single_parent_test_suite = {
 	.test_cases = clk_orphan_transparent_single_parent_mux_test_cases,
 };
 
+struct clk_single_parent_two_lvl_ctx {
+	struct clk_dummy_context parent_parent_ctx;
+	struct clk_dummy_context parent_ctx;
+	struct clk_hw hw;
+};
+
+static int
+clk_orphan_two_level_root_last_test_init(struct kunit *test)
+{
+	struct clk_single_parent_two_lvl_ctx *ctx;
+	int ret;
+
+	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	test->priv = ctx;
+
+	ctx->parent_ctx.hw.init =
+		CLK_HW_INIT("intermediate-parent",
+			    "root-parent",
+			    &clk_dummy_single_parent_ops,
+			    CLK_SET_RATE_PARENT);
+	ret = clk_hw_register(NULL, &ctx->parent_ctx.hw);
+	if (ret)
+		return ret;
+
+	ctx->hw.init =
+		CLK_HW_INIT("test-clk", "intermediate-parent",
+			    &clk_dummy_single_parent_ops,
+			    CLK_SET_RATE_PARENT);
+	ret = clk_hw_register(NULL, &ctx->hw);
+	if (ret)
+		return ret;
+
+	ctx->parent_parent_ctx.rate = DUMMY_CLOCK_INIT_RATE;
+	ctx->parent_parent_ctx.hw.init =
+		CLK_HW_INIT_NO_PARENT("root-parent",
+				      &clk_dummy_rate_ops,
+				      0);
+	ret = clk_hw_register(NULL, &ctx->parent_parent_ctx.hw);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void
+clk_orphan_two_level_root_last_test_exit(struct kunit *test)
+{
+	struct clk_single_parent_two_lvl_ctx *ctx = test->priv;
+
+	clk_hw_unregister(&ctx->hw);
+	clk_hw_unregister(&ctx->parent_ctx.hw);
+	clk_hw_unregister(&ctx->parent_parent_ctx.hw);
+}
+
+/*
+ * Test that, for a clock whose parent used to be orphan, clk_get_rate()
+ * will return the proper rate.
+ */
+static void
+clk_orphan_two_level_root_last_test_get_rate(struct kunit *test)
+{
+	struct clk_single_parent_two_lvl_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	unsigned long rate;
+
+	rate = clk_get_rate(clk);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_INIT_RATE);
+
+	clk_put(clk);
+}
+
+/*
+ * Test that, for a clock whose parent used to be orphan,
+ * clk_set_rate_range() won't affect its rate if it is already within
+ * range.
+ *
+ * See (for Exynos 4210):
+ * https://lore.kernel.org/linux-clk/366a0232-bb4a-c357-6aa8-636e398e05eb@samsung.com/
+ */
+static void
+clk_orphan_two_level_root_last_test_set_range(struct kunit *test)
+{
+	struct clk_single_parent_two_lvl_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	unsigned long rate;
+	int ret;
+
+	ret = clk_set_rate_range(clk,
+				 DUMMY_CLOCK_INIT_RATE - 1000,
+				 DUMMY_CLOCK_INIT_RATE + 1000);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_INIT_RATE);
+
+	clk_put(clk);
+}
+
+static struct kunit_case
+clk_orphan_two_level_root_last_test_cases[] = {
+	KUNIT_CASE(clk_orphan_two_level_root_last_test_get_rate),
+	KUNIT_CASE(clk_orphan_two_level_root_last_test_set_range),
+	{}
+};
+
+/*
+ * Test suite for a basic, transparent, clock with a parent that is also
+ * such a clock. The parent's parent is registered last, while the
+ * parent and its child are registered in that order. The intermediate
+ * and leaf clocks will thus be orphan when registered, but the leaf
+ * clock itself will always have its parent and will never be
+ * reparented. Indeed, it's only orphan because its parent is.
+ *
+ * These tests exercise the behaviour of the consumer API when dealing
+ * with an orphan clock, and how we deal with the transition to a valid
+ * parent.
+ */
+static struct kunit_suite
+clk_orphan_two_level_root_last_test_suite = {
+	.name = "clk-orphan-two-level-root-last-test",
+	.init = clk_orphan_two_level_root_last_test_init,
+	.exit = clk_orphan_two_level_root_last_test_exit,
+	.test_cases = clk_orphan_two_level_root_last_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
@@ -1729,6 +1967,7 @@ kunit_test_suites(
 	&clk_multiple_parents_mux_test_suite,
 	&clk_orphan_transparent_multiple_parent_mux_test_suite,
 	&clk_orphan_transparent_single_parent_test_suite,
+	&clk_orphan_two_level_root_last_test_suite,
 	&clk_range_test_suite,
 	&clk_range_maximize_test_suite,
 	&clk_range_minimize_test_suite,
-- 
2.37.1

[PATCH v9 14/25] clk: Change clk_core_init_rate_req prototype

[Maxime Ripard]

The expectation is that a clk_rate_request structure is supposed to be
initialized using clk_core_init_rate_req(), yet the rate we want to
request still needs to be set by hand.

Let's just pass the rate as a function argument so that callers don't
have any extra work to do.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 91bb1ea0e147..75cfde9f917f 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1380,13 +1380,16 @@ static int clk_core_determine_round_nolock(struct clk_core *core,
 }
 
 static void clk_core_init_rate_req(struct clk_core * const core,
-				   struct clk_rate_request *req)
+				   struct clk_rate_request *req,
+				   unsigned long rate)
 {
 	struct clk_core *parent;
 
 	if (WARN_ON(!core || !req))
 		return;
 
+	req->rate = rate;
+
 	parent = core->parent;
 	if (parent) {
 		req->best_parent_hw = parent->hw;
@@ -1412,7 +1415,7 @@ static int clk_core_round_rate_nolock(struct clk_core *core,
 		return 0;
 	}
 
-	clk_core_init_rate_req(core, req);
+	clk_core_init_rate_req(core, req, req->rate);
 
 	if (clk_core_can_round(core))
 		return clk_core_determine_round_nolock(core, req);
@@ -2004,11 +2007,10 @@ static struct clk_core *clk_calc_new_rates(struct clk_core *core,
 	if (clk_core_can_round(core)) {
 		struct clk_rate_request req;
 
-		req.rate = rate;
 		req.min_rate = min_rate;
 		req.max_rate = max_rate;
 
-		clk_core_init_rate_req(core, &req);
+		clk_core_init_rate_req(core, &req, rate);
 
 		ret = clk_core_determine_round_nolock(core, &req);
 		if (ret < 0)
-- 
2.37.1

[PATCH v9 15/25] clk: Move clk_core_init_rate_req() from clk_core_round_rate_nolock() to its caller

[Maxime Ripard]

The clk_rate_request structure is used internally as an argument for
the clk_core_determine_round_nolock() and clk_core_round_rate_nolock().

In both cases, the clk_core_init_rate_req() function is used to
initialize the clk_rate_request structure.

However, the expectation on who gets to call that function is
inconsistent between those two functions. Indeed,
clk_core_determine_round_nolock() will assume the structure is properly
initialized and will just use it.

On the other hand, clk_core_round_rate_nolock() will call
clk_core_init_rate_req() itself, expecting the caller to have filled
only a minimal set of parameters (rate, min_rate and max_rate).

If we ignore the calling convention inconsistency, this leads to a
second inconsistency for drivers:

   * If they get called by the framework through
     clk_core_round_rate_nolock(), the rate, min_rate and max_rate
     fields will be filled by the caller, and the best_parent_rate and
     best_parent_hw fields will get filled by clk_core_init_rate_req().

   * If they get called by a driver through __clk_determine_rate (and
     thus clk_core_round_rate_nolock), only best_parent_rate and
     best_parent_hw are being explicitly set by the framework. Even
     though we can reasonably expect rate to be set, only one of the 6
     in-tree users explicitly set min_rate and max_rate.

   * If they get called by the framework through
     clk_core_determine_round_nolock(), then we have two callpaths.
     Either it will be called by clk_core_round_rate_nolock() itself, or
     it will be called by clk_calc_new_rates(), which will properly
     initialize rate, min_rate, max_rate itself, and best_parent_rate
     and best_parent_hw through clk_core_init_rate_req().

Even though the first and third case seems equivalent, they aren't when
the clock has CLK_SET_RATE_PARENT. Indeed, in such a case
clk_core_round_rate_nolock() will call itself on the current parent
clock with the same clk_rate_request structure.

The clk_core_init_rate_req() function will then be called on the parent
clock, with the child clk_rate_request pointer and will fill the
best_parent_rate and best_parent_hw fields with the parent context.

When the whole recursion stops and the call returns, the initial caller
will end up with a clk_rate_request structure with some informations of
the child clock (rate, min_rate, max_rate) and some others of the last
clock up the tree whose child had CLK_SET_RATE_PARENT (best_parent_hw,
best_parent_rate).

In the most common case, best_parent_rate is going to be equal on all
the parent clocks so it's not a big deal. However, best_parent_hw is
going to point to a clock that never has been a valid parent for that
clock which is definitely confusing.

In order to fix the calling inconsistency, let's move the
clk_core_init_rate_req() calls to the callers, which will also help a
bit with the clk_core_round_rate_nolock() recursion.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 75cfde9f917f..553e1e9eb300 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1415,8 +1415,6 @@ static int clk_core_round_rate_nolock(struct clk_core *core,
 		return 0;
 	}
 
-	clk_core_init_rate_req(core, req, req->rate);
-
 	if (clk_core_can_round(core))
 		return clk_core_determine_round_nolock(core, req);
 	else if (core->flags & CLK_SET_RATE_PARENT)
@@ -1464,8 +1462,8 @@ unsigned long clk_hw_round_rate(struct clk_hw *hw, unsigned long rate)
 	int ret;
 	struct clk_rate_request req;
 
+	clk_core_init_rate_req(hw->core, &req, rate);
 	clk_core_get_boundaries(hw->core, &req.min_rate, &req.max_rate);
-	req.rate = rate;
 
 	ret = clk_core_round_rate_nolock(hw->core, &req);
 	if (ret)
@@ -1497,8 +1495,8 @@ long clk_round_rate(struct clk *clk, unsigned long rate)
 	if (clk->exclusive_count)
 		clk_core_rate_unprotect(clk->core);
 
+	clk_core_init_rate_req(clk->core, &req, rate);
 	clk_core_get_boundaries(clk->core, &req.min_rate, &req.max_rate);
-	req.rate = rate;
 
 	ret = clk_core_round_rate_nolock(clk->core, &req);
 
@@ -2209,8 +2207,8 @@ static unsigned long clk_core_req_round_rate_nolock(struct clk_core *core,
 	if (cnt < 0)
 		return cnt;
 
+	clk_core_init_rate_req(core, &req, req_rate);
 	clk_core_get_boundaries(core, &req.min_rate, &req.max_rate);
-	req.rate = req_rate;
 
 	ret = clk_core_round_rate_nolock(core, &req);
 
-- 
2.37.1

[PATCH v9 16/25] clk: Introduce clk_hw_init_rate_request()

[Maxime Ripard]

clk-divider instantiates clk_rate_request internally for its round_rate
implementations to share the code with its determine_rate
implementations.

However, it's missing a few fields (min_rate, max_rate) that would be
initialized properly if it was using clk_core_init_rate_req().

Let's create the clk_hw_init_rate_request() function for clock providers
to be able to share the code to instation clk_rate_requests with the
framework. This will also be useful for some tests introduced in later
patches.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk-divider.c    | 20 ++++++++++----------
 drivers/clk/clk.c            | 20 ++++++++++++++++++++
 include/linux/clk-provider.h |  6 ++++++
 3 files changed, 36 insertions(+), 10 deletions(-)

diff --git a/drivers/clk/clk-divider.c b/drivers/clk/clk-divider.c
index f6b2bf558486..a2c2b5203b0a 100644
--- a/drivers/clk/clk-divider.c
+++ b/drivers/clk/clk-divider.c
@@ -386,13 +386,13 @@ long divider_round_rate_parent(struct clk_hw *hw, struct clk_hw *parent,
 			       const struct clk_div_table *table,
 			       u8 width, unsigned long flags)
 {
-	struct clk_rate_request req = {
-		.rate = rate,
-		.best_parent_rate = *prate,
-		.best_parent_hw = parent,
-	};
+	struct clk_rate_request req;
 	int ret;
 
+	clk_hw_init_rate_request(hw, &req, rate);
+	req.best_parent_rate = *prate;
+	req.best_parent_hw = parent;
+
 	ret = divider_determine_rate(hw, &req, table, width, flags);
 	if (ret)
 		return ret;
@@ -408,13 +408,13 @@ long divider_ro_round_rate_parent(struct clk_hw *hw, struct clk_hw *parent,
 				  const struct clk_div_table *table, u8 width,
 				  unsigned long flags, unsigned int val)
 {
-	struct clk_rate_request req = {
-		.rate = rate,
-		.best_parent_rate = *prate,
-		.best_parent_hw = parent,
-	};
+	struct clk_rate_request req;
 	int ret;
 
+	clk_hw_init_rate_request(hw, &req, rate);
+	req.best_parent_rate = *prate;
+	req.best_parent_hw = parent;
+
 	ret = divider_ro_determine_rate(hw, &req, table, width, flags, val);
 	if (ret)
 		return ret;
diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 553e1e9eb300..96b372ff23c2 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1400,6 +1400,26 @@ static void clk_core_init_rate_req(struct clk_core * const core,
 	}
 }
 
+/**
+ * clk_hw_init_rate_request - Initializes a clk_rate_request
+ * @hw: the clk for which we want to submit a rate request
+ * @req: the clk_rate_request structure we want to initialise
+ * @rate: the rate which is to be requested
+ *
+ * Initializes a clk_rate_request structure to submit to
+ * __clk_determine_rate() or similar functions.
+ */
+void clk_hw_init_rate_request(const struct clk_hw *hw,
+			      struct clk_rate_request *req,
+			      unsigned long rate)
+{
+	if (WARN_ON(!hw || !req))
+		return;
+
+	clk_core_init_rate_req(hw->core, req, rate);
+}
+EXPORT_SYMBOL_GPL(clk_hw_init_rate_request);
+
 static bool clk_core_can_round(struct clk_core * const core)
 {
 	return core->ops->determine_rate || core->ops->round_rate;
diff --git a/include/linux/clk-provider.h b/include/linux/clk-provider.h
index 9a14cfa0d201..d857717aa386 100644
--- a/include/linux/clk-provider.h
+++ b/include/linux/clk-provider.h
@@ -42,6 +42,8 @@ struct dentry;
  * struct clk_rate_request - Structure encoding the clk constraints that
  * a clock user might require.
  *
+ * Should be initialized by calling clk_hw_init_rate_request().
+ *
  * @rate:		Requested clock rate. This field will be adjusted by
  *			clock drivers according to hardware capabilities.
  * @min_rate:		Minimum rate imposed by clk users.
@@ -60,6 +62,10 @@ struct clk_rate_request {
 	struct clk_hw *best_parent_hw;
 };
 
+void clk_hw_init_rate_request(const struct clk_hw *hw,
+			      struct clk_rate_request *req,
+			      unsigned long rate);
+
 /**
  * struct clk_duty - Struture encoding the duty cycle ratio of a clock
  *
-- 
2.37.1

[PATCH v9 17/25] clk: Add our request boundaries in clk_core_init_rate_req

[Maxime Ripard]

The expectation is that a new clk_rate_request is initialized through a
call to clk_core_init_rate_req().

However, at the moment it only fills the parent rate and clk_hw pointer,
but omits the other fields such as the clock rate boundaries.

Some users of that function will update them after calling it, but most
don't.

As we are passed the clk_core pointer, we have access to those
boundaries in clk_core_init_rate_req() however, so let's just fill it
there and remove it from the few callers that do it right.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 96b372ff23c2..2794bd3bef4b 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1389,6 +1389,7 @@ static void clk_core_init_rate_req(struct clk_core * const core,
 		return;
 
 	req->rate = rate;
+	clk_core_get_boundaries(core, &req->min_rate, &req->max_rate);
 
 	parent = core->parent;
 	if (parent) {
@@ -1483,7 +1484,6 @@ unsigned long clk_hw_round_rate(struct clk_hw *hw, unsigned long rate)
 	struct clk_rate_request req;
 
 	clk_core_init_rate_req(hw->core, &req, rate);
-	clk_core_get_boundaries(hw->core, &req.min_rate, &req.max_rate);
 
 	ret = clk_core_round_rate_nolock(hw->core, &req);
 	if (ret)
@@ -1516,7 +1516,6 @@ long clk_round_rate(struct clk *clk, unsigned long rate)
 		clk_core_rate_unprotect(clk->core);
 
 	clk_core_init_rate_req(clk->core, &req, rate);
-	clk_core_get_boundaries(clk->core, &req.min_rate, &req.max_rate);
 
 	ret = clk_core_round_rate_nolock(clk->core, &req);
 
@@ -2025,9 +2024,6 @@ static struct clk_core *clk_calc_new_rates(struct clk_core *core,
 	if (clk_core_can_round(core)) {
 		struct clk_rate_request req;
 
-		req.min_rate = min_rate;
-		req.max_rate = max_rate;
-
 		clk_core_init_rate_req(core, &req, rate);
 
 		ret = clk_core_determine_round_nolock(core, &req);
@@ -2228,7 +2224,6 @@ static unsigned long clk_core_req_round_rate_nolock(struct clk_core *core,
 		return cnt;
 
 	clk_core_init_rate_req(core, &req, req_rate);
-	clk_core_get_boundaries(core, &req.min_rate, &req.max_rate);
 
 	ret = clk_core_round_rate_nolock(core, &req);
 
-- 
2.37.1

[PATCH v9 18/25] clk: Switch from __clk_determine_rate to clk_core_round_rate_nolock

[Maxime Ripard]

clk_mux_determine_rate_flags() will call into __clk_determine_rate()
with a clk_hw pointer, while it has access to the clk_core pointer
already.

This leads to back and forth between clk_hw and clk_core, while
__clk_determine_rate will only call clk_core_round_rate_nolock() with
the clk_core pointer it retrieved from the clk_hw.

Let's simplify things a bit by calling into clk_core_round_rate_nolock
directly.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 2794bd3bef4b..f8a8bdd552d6 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -536,6 +536,9 @@ static bool mux_is_better_rate(unsigned long rate, unsigned long now,
 	return now <= rate && now > best;
 }
 
+static int clk_core_round_rate_nolock(struct clk_core *core,
+				      struct clk_rate_request *req);
+
 int clk_mux_determine_rate_flags(struct clk_hw *hw,
 				 struct clk_rate_request *req,
 				 unsigned long flags)
@@ -549,8 +552,12 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	if (core->flags & CLK_SET_RATE_NO_REPARENT) {
 		parent = core->parent;
 		if (core->flags & CLK_SET_RATE_PARENT) {
-			ret = __clk_determine_rate(parent ? parent->hw : NULL,
-						   &parent_req);
+			if (!parent) {
+				req->rate = 0;
+				return 0;
+			}
+
+			ret = clk_core_round_rate_nolock(parent, &parent_req);
 			if (ret)
 				return ret;
 
@@ -573,7 +580,7 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 
 		if (core->flags & CLK_SET_RATE_PARENT) {
 			parent_req = *req;
-			ret = __clk_determine_rate(parent->hw, &parent_req);
+			ret = clk_core_round_rate_nolock(parent, &parent_req);
 			if (ret)
 				continue;
 		} else {
-- 
2.37.1

[PATCH v9 19/25] clk: Introduce clk_core_has_parent()

[Maxime Ripard]

We will need to know if a clk_core pointer has a given parent in other
functions, so let's create a clk_core_has_parent() function that
clk_has_parent() will call into.

For good measure, let's add some unit tests as well to make sure it
works properly.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c      | 36 ++++++++++++++++++++---------------
 drivers/clk/clk_test.c | 43 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 64 insertions(+), 15 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index f8a8bdd552d6..c1bb64827060 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -539,6 +539,26 @@ static bool mux_is_better_rate(unsigned long rate, unsigned long now,
 static int clk_core_round_rate_nolock(struct clk_core *core,
 				      struct clk_rate_request *req);
 
+static bool clk_core_has_parent(struct clk_core *core, const struct clk_core *parent)
+{
+	unsigned int i;
+
+	/* Optimize for the case where the parent is already the parent. */
+	if (core == parent)
+		return true;
+
+	for (i = 0; i < core->num_parents; i++) {
+		struct clk_core *tmp = clk_core_get_parent_by_index(core, i);
+		if (!tmp)
+			continue;
+
+		if (tmp == parent)
+			return true;
+	}
+
+	return false;
+}
+
 int clk_mux_determine_rate_flags(struct clk_hw *hw,
 				 struct clk_rate_request *req,
 				 unsigned long flags)
@@ -2574,25 +2594,11 @@ void clk_hw_reparent(struct clk_hw *hw, struct clk_hw *new_parent)
  */
 bool clk_has_parent(struct clk *clk, struct clk *parent)
 {
-	struct clk_core *core, *parent_core;
-	int i;
-
 	/* NULL clocks should be nops, so return success if either is NULL. */
 	if (!clk || !parent)
 		return true;
 
-	core = clk->core;
-	parent_core = parent->core;
-
-	/* Optimize for the case where the parent is already the parent. */
-	if (core->parent == parent_core)
-		return true;
-
-	for (i = 0; i < core->num_parents; i++)
-		if (!strcmp(core->parents[i].name, parent_core->name))
-			return true;
-
-	return false;
+	return clk_core_has_parent(clk->core, parent->core);
 }
 EXPORT_SYMBOL_GPL(clk_has_parent);
 
diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index d1b1372f7aaa..7068517428e2 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -491,8 +491,32 @@ clk_test_multiple_parents_mux_get_parent(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * Test that for a clock with a multiple parents, clk_has_parent()
+ * actually reports all of them as parents.
+ */
+static void
+clk_test_multiple_parents_mux_has_parent(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[0].hw, NULL);
+	KUNIT_EXPECT_TRUE(test, clk_has_parent(clk, parent));
+	clk_put(parent);
+
+	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_EXPECT_TRUE(test, clk_has_parent(clk, parent));
+	clk_put(parent);
+
+	clk_put(clk);
+}
+
 static struct kunit_case clk_multiple_parents_mux_test_cases[] = {
 	KUNIT_CASE(clk_test_multiple_parents_mux_get_parent),
+	KUNIT_CASE(clk_test_multiple_parents_mux_has_parent),
 	{}
 };
 
@@ -918,6 +942,24 @@ clk_test_single_parent_mux_get_parent(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * Test that for a clock with a single parent, clk_has_parent() actually
+ * reports it as a parent.
+ */
+static void
+clk_test_single_parent_mux_has_parent(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent = clk_hw_get_clk(&ctx->parent_ctx.hw, NULL);
+
+	KUNIT_EXPECT_TRUE(test, clk_has_parent(clk, parent));
+
+	clk_put(parent);
+	clk_put(clk);
+}
+
 /*
  * Test that for a clock that can't modify its rate and with a single
  * parent, if we set disjoints range on the parent and then the child,
@@ -1022,6 +1064,7 @@ clk_test_single_parent_mux_set_range_round_rate_child_smaller(struct kunit *test
 
 static struct kunit_case clk_single_parent_mux_test_cases[] = {
 	KUNIT_CASE(clk_test_single_parent_mux_get_parent),
+	KUNIT_CASE(clk_test_single_parent_mux_has_parent),
 	KUNIT_CASE(clk_test_single_parent_mux_set_range_disjoint_child_last),
 	KUNIT_CASE(clk_test_single_parent_mux_set_range_disjoint_parent_last),
 	KUNIT_CASE(clk_test_single_parent_mux_set_range_round_rate_child_smaller),
-- 
2.37.1

[PATCH v9 20/25] clk: Constify clk_has_parent()

[Maxime Ripard]

clk_has_parent() doesn't modify the clocks being passed, so let's make
it const.

Suggested-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c   | 2 +-
 include/linux/clk.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index c1bb64827060..8b9ff78e7b53 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -2592,7 +2592,7 @@ void clk_hw_reparent(struct clk_hw *hw, struct clk_hw *new_parent)
  *
  * Returns true if @parent is a possible parent for @clk, false otherwise.
  */
-bool clk_has_parent(struct clk *clk, struct clk *parent)
+bool clk_has_parent(const struct clk *clk, const struct clk *parent)
 {
 	/* NULL clocks should be nops, so return success if either is NULL. */
 	if (!clk || !parent)
diff --git a/include/linux/clk.h b/include/linux/clk.h
index c13061cabdfc..1ef013324237 100644
--- a/include/linux/clk.h
+++ b/include/linux/clk.h
@@ -799,7 +799,7 @@ int clk_set_rate_exclusive(struct clk *clk, unsigned long rate);
  *
  * Returns true if @parent is a possible parent for @clk, false otherwise.
  */
-bool clk_has_parent(struct clk *clk, struct clk *parent);
+bool clk_has_parent(const struct clk *clk, const struct clk *parent);
 
 /**
  * clk_set_rate_range - set a rate range for a clock source
-- 
2.37.1

[PATCH v9 21/25] clk: Stop forwarding clk_rate_requests to the parent

[Maxime Ripard]

If the clock cannot modify its rate and has CLK_SET_RATE_PARENT,
clk_mux_determine_rate_flags(), clk_core_round_rate_nolock() and a
number of drivers will forward the clk_rate_request to the parent clock.

clk_core_round_rate_nolock() will pass the pointer directly, which means
that we pass a clk_rate_request to the parent that has the rate,
min_rate and max_rate of the child, and the best_parent_rate and
best_parent_hw fields will be relative to the child as well, so will
point to our current clock and its rate. The most common case for
CLK_SET_RATE_PARENT is that the child and parent clock rates will be
equal, so the rate field isn't a worry, but the other fields are.

Similarly, if the parent clock driver ever modifies the best_parent_rate
or best_parent_hw, this will be applied to the child once the call to
clk_core_round_rate_nolock() is done. best_parent_hw is probably not
going to be a valid parent, and best_parent_rate might lead to a parent
rate change different to the one that was initially computed.

clk_mux_determine_rate_flags() and the affected drivers will copy the
request before forwarding it to the parents, so they won't be affected
by the latter issue, but the former is still going to be there and will
lead to erroneous data and context being passed to the various clock
drivers in the same sub-tree.

Let's create two new functions, clk_core_forward_rate_req() and
clk_hw_forward_rate_request() for the framework and the clock providers
that will copy a request from a child clock and update the context to
match the parent's. We also update the relevant call sites in the
framework and drivers to use that new function.

Let's also add a test to make sure we avoid regressions there.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/at91/clk-generated.c  |   5 +-
 drivers/clk/at91/clk-master.c     |   9 +-
 drivers/clk/at91/clk-peripheral.c |   4 +-
 drivers/clk/clk-composite.c       |   6 +-
 drivers/clk/clk.c                 |  84 ++++++++++++--
 drivers/clk/clk_test.c            | 182 ++++++++++++++++++++++++++++++
 include/linux/clk-provider.h      |   5 +
 7 files changed, 279 insertions(+), 16 deletions(-)

diff --git a/drivers/clk/at91/clk-generated.c b/drivers/clk/at91/clk-generated.c
index d429ba52a719..943ea67bf135 100644
--- a/drivers/clk/at91/clk-generated.c
+++ b/drivers/clk/at91/clk-generated.c
@@ -136,7 +136,6 @@ static int clk_generated_determine_rate(struct clk_hw *hw,
 {
 	struct clk_generated *gck = to_clk_generated(hw);
 	struct clk_hw *parent = NULL;
-	struct clk_rate_request req_parent = *req;
 	long best_rate = -EINVAL;
 	unsigned long min_rate, parent_rate;
 	int best_diff = -1;
@@ -192,7 +191,9 @@ static int clk_generated_determine_rate(struct clk_hw *hw,
 		goto end;
 
 	for (div = 1; div < GENERATED_MAX_DIV + 2; div++) {
-		req_parent.rate = req->rate * div;
+		struct clk_rate_request req_parent;
+
+		clk_hw_forward_rate_request(hw, req, parent, &req_parent, req->rate * div);
 		if (__clk_determine_rate(parent, &req_parent))
 			continue;
 		clk_generated_best_diff(req, parent, req_parent.rate, div,
diff --git a/drivers/clk/at91/clk-master.c b/drivers/clk/at91/clk-master.c
index 164e2959c7cf..b7cd1924de52 100644
--- a/drivers/clk/at91/clk-master.c
+++ b/drivers/clk/at91/clk-master.c
@@ -581,7 +581,6 @@ static int clk_sama7g5_master_determine_rate(struct clk_hw *hw,
 					     struct clk_rate_request *req)
 {
 	struct clk_master *master = to_clk_master(hw);
-	struct clk_rate_request req_parent = *req;
 	struct clk_hw *parent;
 	long best_rate = LONG_MIN, best_diff = LONG_MIN;
 	unsigned long parent_rate;
@@ -618,11 +617,15 @@ static int clk_sama7g5_master_determine_rate(struct clk_hw *hw,
 		goto end;
 
 	for (div = 0; div < MASTER_PRES_MAX + 1; div++) {
+		struct clk_rate_request req_parent;
+		unsigned long req_rate;
+
 		if (div == MASTER_PRES_MAX)
-			req_parent.rate = req->rate * 3;
+			req_rate = req->rate * 3;
 		else
-			req_parent.rate = req->rate << div;
+			req_rate = req->rate << div;
 
+		clk_hw_forward_rate_request(hw, req, parent, &req_parent, req_rate);
 		if (__clk_determine_rate(parent, &req_parent))
 			continue;
 
diff --git a/drivers/clk/at91/clk-peripheral.c b/drivers/clk/at91/clk-peripheral.c
index e14fa5ac734c..5104d4025484 100644
--- a/drivers/clk/at91/clk-peripheral.c
+++ b/drivers/clk/at91/clk-peripheral.c
@@ -269,7 +269,6 @@ static int clk_sam9x5_peripheral_determine_rate(struct clk_hw *hw,
 {
 	struct clk_sam9x5_peripheral *periph = to_clk_sam9x5_peripheral(hw);
 	struct clk_hw *parent = clk_hw_get_parent(hw);
-	struct clk_rate_request req_parent = *req;
 	unsigned long parent_rate = clk_hw_get_rate(parent);
 	unsigned long tmp_rate;
 	long best_rate = LONG_MIN;
@@ -302,8 +301,9 @@ static int clk_sam9x5_peripheral_determine_rate(struct clk_hw *hw,
 		goto end;
 
 	for (shift = 0; shift <= PERIPHERAL_MAX_SHIFT; shift++) {
-		req_parent.rate = req->rate << shift;
+		struct clk_rate_request req_parent;
 
+		clk_hw_forward_rate_request(hw, req, parent, &req_parent, req->rate << shift);
 		if (__clk_determine_rate(parent, &req_parent))
 			continue;
 
diff --git a/drivers/clk/clk-composite.c b/drivers/clk/clk-composite.c
index b9c5f904f535..edfa94641bbf 100644
--- a/drivers/clk/clk-composite.c
+++ b/drivers/clk/clk-composite.c
@@ -85,10 +85,11 @@ static int clk_composite_determine_rate(struct clk_hw *hw,
 		req->best_parent_hw = NULL;
 
 		if (clk_hw_get_flags(hw) & CLK_SET_RATE_NO_REPARENT) {
-			struct clk_rate_request tmp_req = *req;
+			struct clk_rate_request tmp_req;
 
 			parent = clk_hw_get_parent(mux_hw);
 
+			clk_hw_forward_rate_request(hw, req, parent, &tmp_req, req->rate);
 			ret = clk_composite_determine_rate_for_parent(rate_hw,
 								      &tmp_req,
 								      parent,
@@ -104,12 +105,13 @@ static int clk_composite_determine_rate(struct clk_hw *hw,
 		}
 
 		for (i = 0; i < clk_hw_get_num_parents(mux_hw); i++) {
-			struct clk_rate_request tmp_req = *req;
+			struct clk_rate_request tmp_req;
 
 			parent = clk_hw_get_parent_by_index(mux_hw, i);
 			if (!parent)
 				continue;
 
+			clk_hw_forward_rate_request(hw, req, parent, &tmp_req, req->rate);
 			ret = clk_composite_determine_rate_for_parent(rate_hw,
 								      &tmp_req,
 								      parent,
diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 8b9ff78e7b53..a99451628f13 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -536,6 +536,10 @@ static bool mux_is_better_rate(unsigned long rate, unsigned long now,
 	return now <= rate && now > best;
 }
 
+static void clk_core_init_rate_req(struct clk_core * const core,
+				   struct clk_rate_request *req,
+				   unsigned long rate);
+
 static int clk_core_round_rate_nolock(struct clk_core *core,
 				      struct clk_rate_request *req);
 
@@ -559,6 +563,25 @@ static bool clk_core_has_parent(struct clk_core *core, const struct clk_core *pa
 	return false;
 }
 
+static void
+clk_core_forward_rate_req(struct clk_core *core,
+			  const struct clk_rate_request *old_req,
+			  struct clk_core *parent,
+			  struct clk_rate_request *req,
+			  unsigned long parent_rate)
+{
+	if (WARN_ON(!clk_core_has_parent(core, parent)))
+		return;
+
+	clk_core_init_rate_req(parent, req, parent_rate);
+
+	if (req->min_rate < old_req->min_rate)
+		req->min_rate = old_req->min_rate;
+
+	if (req->max_rate > old_req->max_rate)
+		req->max_rate = old_req->max_rate;
+}
+
 int clk_mux_determine_rate_flags(struct clk_hw *hw,
 				 struct clk_rate_request *req,
 				 unsigned long flags)
@@ -566,17 +589,19 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	struct clk_core *core = hw->core, *parent, *best_parent = NULL;
 	int i, num_parents, ret;
 	unsigned long best = 0;
-	struct clk_rate_request parent_req = *req;
 
 	/* if NO_REPARENT flag set, pass through to current parent */
 	if (core->flags & CLK_SET_RATE_NO_REPARENT) {
 		parent = core->parent;
 		if (core->flags & CLK_SET_RATE_PARENT) {
+			struct clk_rate_request parent_req;
+
 			if (!parent) {
 				req->rate = 0;
 				return 0;
 			}
 
+			clk_core_forward_rate_req(core, req, parent, &parent_req, req->rate);
 			ret = clk_core_round_rate_nolock(parent, &parent_req);
 			if (ret)
 				return ret;
@@ -594,23 +619,29 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 	/* find the parent that can provide the fastest rate <= rate */
 	num_parents = core->num_parents;
 	for (i = 0; i < num_parents; i++) {
+		unsigned long parent_rate;
+
 		parent = clk_core_get_parent_by_index(core, i);
 		if (!parent)
 			continue;
 
 		if (core->flags & CLK_SET_RATE_PARENT) {
-			parent_req = *req;
+			struct clk_rate_request parent_req;
+
+			clk_core_forward_rate_req(core, req, parent, &parent_req, req->rate);
 			ret = clk_core_round_rate_nolock(parent, &parent_req);
 			if (ret)
 				continue;
+
+			parent_rate = parent_req.rate;
 		} else {
-			parent_req.rate = clk_core_get_rate_nolock(parent);
+			parent_rate = clk_core_get_rate_nolock(parent);
 		}
 
-		if (mux_is_better_rate(req->rate, parent_req.rate,
+		if (mux_is_better_rate(req->rate, parent_rate,
 				       best, flags)) {
 			best_parent = parent;
-			best = parent_req.rate;
+			best = parent_rate;
 		}
 	}
 
@@ -1448,6 +1479,31 @@ void clk_hw_init_rate_request(const struct clk_hw *hw,
 }
 EXPORT_SYMBOL_GPL(clk_hw_init_rate_request);
 
+/**
+ * clk_hw_forward_rate_request - Forwards a clk_rate_request to a clock's parent
+ * @hw: the original clock that got the rate request
+ * @old_req: the original clk_rate_request structure we want to forward
+ * @parent: the clk we want to forward @old_req to
+ * @req: the clk_rate_request structure we want to initialise
+ * @parent_rate: The rate which is to be requested to @parent
+ *
+ * Initializes a clk_rate_request structure to submit to a clock parent
+ * in __clk_determine_rate() or similar functions.
+ */
+void clk_hw_forward_rate_request(const struct clk_hw *hw,
+				 const struct clk_rate_request *old_req,
+				 const struct clk_hw *parent,
+				 struct clk_rate_request *req,
+				 unsigned long parent_rate)
+{
+	if (WARN_ON(!hw || !old_req || !parent || !req))
+		return;
+
+	clk_core_forward_rate_req(hw->core, old_req,
+				  parent->core, req,
+				  parent_rate);
+}
+
 static bool clk_core_can_round(struct clk_core * const core)
 {
 	return core->ops->determine_rate || core->ops->round_rate;
@@ -1456,6 +1512,8 @@ static bool clk_core_can_round(struct clk_core * const core)
 static int clk_core_round_rate_nolock(struct clk_core *core,
 				      struct clk_rate_request *req)
 {
+	int ret;
+
 	lockdep_assert_held(&prepare_lock);
 
 	if (!core) {
@@ -1465,8 +1523,20 @@ static int clk_core_round_rate_nolock(struct clk_core *core,
 
 	if (clk_core_can_round(core))
 		return clk_core_determine_round_nolock(core, req);
-	else if (core->flags & CLK_SET_RATE_PARENT)
-		return clk_core_round_rate_nolock(core->parent, req);
+
+	if (core->flags & CLK_SET_RATE_PARENT) {
+		struct clk_rate_request parent_req;
+
+		clk_core_forward_rate_req(core, req, core->parent, &parent_req, req->rate);
+		ret = clk_core_round_rate_nolock(core->parent, &parent_req);
+		if (ret)
+			return ret;
+
+		req->best_parent_rate = parent_req.rate;
+		req->rate = parent_req.rate;
+
+		return 0;
+	}
 
 	req->rate = core->rate;
 	return 0;
diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 7068517428e2..5a5b7a8baba2 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -1024,6 +1024,36 @@ clk_test_single_parent_mux_set_range_disjoint_parent_last(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * Test that for a clock that can't modify its rate and with a single
+ * parent, if we set a range on the parent and then call
+ * clk_round_rate(), the boundaries of the parent are taken into
+ * account.
+ */
+static void
+clk_test_single_parent_mux_set_range_round_rate_parent_only(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long rate;
+	int ret;
+
+	parent = clk_get_parent(clk);
+	KUNIT_ASSERT_PTR_NE(test, parent, NULL);
+
+	ret = clk_set_rate_range(parent, DUMMY_CLOCK_RATE_1, DUMMY_CLOCK_RATE_2);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_round_rate(clk, DUMMY_CLOCK_RATE_1 - 1000);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2);
+
+	clk_put(clk);
+}
+
 /*
  * Test that for a clock that can't modify its rate and with a single
  * parent, if we set a range on the parent and a more restrictive one on
@@ -1062,12 +1092,52 @@ clk_test_single_parent_mux_set_range_round_rate_child_smaller(struct kunit *test
 	clk_put(clk);
 }
 
+/*
+ * Test that for a clock that can't modify its rate and with a single
+ * parent, if we set a range on the child and a more restrictive one on
+ * the parent, and then call clk_round_rate(), the boundaries of the
+ * two clocks are taken into account.
+ */
+static void
+clk_test_single_parent_mux_set_range_round_rate_parent_smaller(struct kunit *test)
+{
+	struct clk_single_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent;
+	unsigned long rate;
+	int ret;
+
+	parent = clk_get_parent(clk);
+	KUNIT_ASSERT_PTR_NE(test, parent, NULL);
+
+	ret = clk_set_rate_range(parent, DUMMY_CLOCK_RATE_1 + 1000, DUMMY_CLOCK_RATE_2 - 1000);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate_range(clk, DUMMY_CLOCK_RATE_1, DUMMY_CLOCK_RATE_2);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_round_rate(clk, DUMMY_CLOCK_RATE_1 - 1000);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1 + 1000);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2 - 1000);
+
+	rate = clk_round_rate(clk, DUMMY_CLOCK_RATE_2 + 1000);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1 + 1000);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_2 - 1000);
+
+	clk_put(clk);
+}
+
 static struct kunit_case clk_single_parent_mux_test_cases[] = {
 	KUNIT_CASE(clk_test_single_parent_mux_get_parent),
 	KUNIT_CASE(clk_test_single_parent_mux_has_parent),
 	KUNIT_CASE(clk_test_single_parent_mux_set_range_disjoint_child_last),
 	KUNIT_CASE(clk_test_single_parent_mux_set_range_disjoint_parent_last),
 	KUNIT_CASE(clk_test_single_parent_mux_set_range_round_rate_child_smaller),
+	KUNIT_CASE(clk_test_single_parent_mux_set_range_round_rate_parent_only),
+	KUNIT_CASE(clk_test_single_parent_mux_set_range_round_rate_parent_smaller),
 	{}
 };
 
@@ -2005,7 +2075,119 @@ static struct kunit_suite clk_range_minimize_test_suite = {
 	.test_cases = clk_range_minimize_test_cases,
 };
 
+struct clk_leaf_mux_ctx {
+	struct clk_multiple_parent_ctx mux_ctx;
+	struct clk_hw hw;
+};
+
+static int
+clk_leaf_mux_set_rate_parent_test_init(struct kunit *test)
+{
+	struct clk_leaf_mux_ctx *ctx;
+	const char *top_parents[2] = { "parent-0", "parent-1" };
+	int ret;
+
+	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+	test->priv = ctx;
+
+	ctx->mux_ctx.parents_ctx[0].hw.init = CLK_HW_INIT_NO_PARENT("parent-0",
+								    &clk_dummy_rate_ops,
+								    0);
+	ctx->mux_ctx.parents_ctx[0].rate = DUMMY_CLOCK_RATE_1;
+	ret = clk_hw_register(NULL, &ctx->mux_ctx.parents_ctx[0].hw);
+	if (ret)
+		return ret;
+
+	ctx->mux_ctx.parents_ctx[1].hw.init = CLK_HW_INIT_NO_PARENT("parent-1",
+								    &clk_dummy_rate_ops,
+								    0);
+	ctx->mux_ctx.parents_ctx[1].rate = DUMMY_CLOCK_RATE_2;
+	ret = clk_hw_register(NULL, &ctx->mux_ctx.parents_ctx[1].hw);
+	if (ret)
+		return ret;
+
+	ctx->mux_ctx.current_parent = 0;
+	ctx->mux_ctx.hw.init = CLK_HW_INIT_PARENTS("test-mux", top_parents,
+						   &clk_multiple_parents_mux_ops,
+						   0);
+	ret = clk_hw_register(NULL, &ctx->mux_ctx.hw);
+	if (ret)
+		return ret;
+
+	ctx->hw.init = CLK_HW_INIT_HW("test-clock", &ctx->mux_ctx.hw,
+				      &clk_dummy_single_parent_ops,
+				      CLK_SET_RATE_PARENT);
+	ret = clk_hw_register(NULL, &ctx->hw);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void clk_leaf_mux_set_rate_parent_test_exit(struct kunit *test)
+{
+	struct clk_leaf_mux_ctx *ctx = test->priv;
+
+	clk_hw_unregister(&ctx->hw);
+	clk_hw_unregister(&ctx->mux_ctx.hw);
+	clk_hw_unregister(&ctx->mux_ctx.parents_ctx[0].hw);
+	clk_hw_unregister(&ctx->mux_ctx.parents_ctx[1].hw);
+}
+
+/*
+ * Test that, for a clock that will forward any rate request to its
+ * parent, the rate request structure returned by __clk_determine_rate
+ * is sane and will be what we expect.
+ */
+static void clk_leaf_mux_set_rate_parent_determine_rate(struct kunit *test)
+{
+	struct clk_leaf_mux_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk_rate_request req;
+	unsigned long rate;
+	int ret;
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_EQ(test, rate, DUMMY_CLOCK_RATE_1);
+
+	clk_hw_init_rate_request(hw, &req, DUMMY_CLOCK_RATE_2);
+
+	ret = __clk_determine_rate(hw, &req);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	KUNIT_EXPECT_EQ(test, req.rate, DUMMY_CLOCK_RATE_2);
+	KUNIT_EXPECT_EQ(test, req.best_parent_rate, DUMMY_CLOCK_RATE_2);
+	KUNIT_EXPECT_PTR_EQ(test, req.best_parent_hw, &ctx->mux_ctx.hw);
+
+	clk_put(clk);
+}
+
+static struct kunit_case clk_leaf_mux_set_rate_parent_test_cases[] = {
+	KUNIT_CASE(clk_leaf_mux_set_rate_parent_determine_rate),
+	{}
+};
+
+/*
+ * Test suite for a clock whose parent is a mux with multiple parents.
+ * The leaf clock has CLK_SET_RATE_PARENT, and will forward rate
+ * requests to the mux, which will then select which parent is the best
+ * fit for a given rate.
+ *
+ * These tests exercise the behaviour of muxes, and the proper selection
+ * of parents.
+  */
+static struct kunit_suite clk_leaf_mux_set_rate_parent_test_suite = {
+	.name = "clk-leaf-mux-set-rate-parent",
+	.init = clk_leaf_mux_set_rate_parent_test_init,
+	.exit = clk_leaf_mux_set_rate_parent_test_exit,
+	.test_cases = clk_leaf_mux_set_rate_parent_test_cases,
+};
+
 kunit_test_suites(
+	&clk_leaf_mux_set_rate_parent_test_suite,
 	&clk_test_suite,
 	&clk_multiple_parents_mux_test_suite,
 	&clk_orphan_transparent_multiple_parent_mux_test_suite,
diff --git a/include/linux/clk-provider.h b/include/linux/clk-provider.h
index d857717aa386..8bce6c524f29 100644
--- a/include/linux/clk-provider.h
+++ b/include/linux/clk-provider.h
@@ -65,6 +65,11 @@ struct clk_rate_request {
 void clk_hw_init_rate_request(const struct clk_hw *hw,
 			      struct clk_rate_request *req,
 			      unsigned long rate);
+void clk_hw_forward_rate_request(const struct clk_hw *core,
+				 const struct clk_rate_request *old_req,
+				 const struct clk_hw *parent,
+				 struct clk_rate_request *req,
+				 unsigned long parent_rate);
 
 /**
  * struct clk_duty - Struture encoding the duty cycle ratio of a clock
-- 
2.37.1

[PATCH v9 22/25] clk: Zero the clk_rate_request structure

[Maxime Ripard]

In order to make sure we don't carry anything over from an already
existing clk_rate_request pointer we would pass to
clk_core_init_rate_req(), let's zero the entire structure before
initializing it.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index a99451628f13..fc145ea40f4e 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1446,6 +1446,8 @@ static void clk_core_init_rate_req(struct clk_core * const core,
 	if (WARN_ON(!core || !req))
 		return;
 
+	memset(req, 0, sizeof(*req));
+
 	req->rate = rate;
 	clk_core_get_boundaries(core, &req->min_rate, &req->max_rate);
 
-- 
2.37.1

[PATCH v9 23/25] clk: Introduce the clk_hw_get_rate_range function

[Maxime Ripard]

Some clock providers are hand-crafting their clk_rate_request, and need
to figure out the current boundaries of their clk_hw to fill it
properly.

Let's create such a function for clock providers.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk.c            | 16 ++++++++++++++++
 include/linux/clk-provider.h |  2 ++
 2 files changed, 18 insertions(+)

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index fc145ea40f4e..453e2ff10961 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -683,6 +683,22 @@ static void clk_core_get_boundaries(struct clk_core *core,
 		*max_rate = min(*max_rate, clk_user->max_rate);
 }
 
+/*
+ * clk_hw_get_rate_range() - returns the clock rate range for a hw clk
+ * @hw: the hw clk we want to get the range from
+ * @min_rate: pointer to the variable that will hold the minimum
+ * @max_rate: pointer to the variable that will hold the maximum
+ *
+ * Fills the @min_rate and @max_rate variables with the minimum and
+ * maximum that clock can reach.
+ */
+void clk_hw_get_rate_range(struct clk_hw *hw, unsigned long *min_rate,
+			   unsigned long *max_rate)
+{
+	clk_core_get_boundaries(hw->core, min_rate, max_rate);
+}
+EXPORT_SYMBOL_GPL(clk_hw_get_rate_range);
+
 static bool clk_core_check_boundaries(struct clk_core *core,
 				      unsigned long min_rate,
 				      unsigned long max_rate)
diff --git a/include/linux/clk-provider.h b/include/linux/clk-provider.h
index 8bce6c524f29..8724a3547a79 100644
--- a/include/linux/clk-provider.h
+++ b/include/linux/clk-provider.h
@@ -1267,6 +1267,8 @@ int clk_mux_determine_rate_flags(struct clk_hw *hw,
 				 struct clk_rate_request *req,
 				 unsigned long flags);
 void clk_hw_reparent(struct clk_hw *hw, struct clk_hw *new_parent);
+void clk_hw_get_rate_range(struct clk_hw *hw, unsigned long *min_rate,
+			   unsigned long *max_rate);
 void clk_hw_set_rate_range(struct clk_hw *hw, unsigned long min_rate,
 			   unsigned long max_rate);
 
-- 
2.37.1

[PATCH v9 24/25] clk: qcom: clk-rcg2: Take clock boundaries into consideration for gfx3d

[Maxime Ripard]

The gfx3d clock is hand-crafting its own clk_rate_request in
clk_gfx3d_determine_rate to pass to the parent of that clock.

However, since the clk_rate_request is zero'd at creation, it will have
a max_rate of 0 which will break any code depending on the clock
boundaries.

That includes the recent commit 948fb0969eae ("clk: Always clamp the
rounded rate") which will clamp the rate given to clk_round_rate() to
the current clock boundaries.

For the gfx3d clock, it means that since both the min_rate and max_rate
fields are set at zero, clk_round_rate() now always return 0.

Let's initialize the min_rate and max_rate fields properly for that
clock.

Fixes: 948fb0969eae ("clk: Always clamp the rounded rate")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/qcom/clk-rcg2.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
index 28019edd2a50..ee536b457952 100644
--- a/drivers/clk/qcom/clk-rcg2.c
+++ b/drivers/clk/qcom/clk-rcg2.c
@@ -908,6 +908,15 @@ static int clk_gfx3d_determine_rate(struct clk_hw *hw,
 		req->best_parent_hw = p2;
 	}
 
+	clk_hw_get_rate_range(req->best_parent_hw,
+			      &parent_req.min_rate, &parent_req.max_rate);
+
+	if (req->min_rate > parent_req.min_rate)
+		parent_req.min_rate = req->min_rate;
+
+	if (req->max_rate < parent_req.max_rate)
+		parent_req.max_rate = req->max_rate;
+
 	ret = __clk_determine_rate(req->best_parent_hw, &parent_req);
 	if (ret)
 		return ret;
-- 
2.37.1

[PATCH v9 25/25] clk: tests: Add missing test case for ranges

[Maxime Ripard]

Let's add a test on the rate range after a reparenting. This fails for
now, but it's worth having it to document the corner cases we don't
support yet.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
---
 drivers/clk/clk_test.c | 53 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)

diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
index 5a5b7a8baba2..8e0c76ab1a6d 100644
--- a/drivers/clk/clk_test.c
+++ b/drivers/clk/clk_test.c
@@ -514,9 +514,62 @@ clk_test_multiple_parents_mux_has_parent(struct kunit *test)
 	clk_put(clk);
 }
 
+/*
+ * Test that for a clock with a multiple parents, if we set a range on
+ * that clock and the parent is changed, its rate after the reparenting
+ * is still within the range we asked for.
+ *
+ * FIXME: clk_set_parent() only does the reparenting but doesn't
+ * reevaluate whether the new clock rate is within its boundaries or
+ * not.
+ */
+static void
+clk_test_multiple_parents_mux_set_range_set_parent_get_rate(struct kunit *test)
+{
+	struct clk_multiple_parent_ctx *ctx = test->priv;
+	struct clk_hw *hw = &ctx->hw;
+	struct clk *clk = clk_hw_get_clk(hw, NULL);
+	struct clk *parent1, *parent2;
+	unsigned long rate;
+	int ret;
+
+	kunit_skip(test, "This needs to be fixed in the core.");
+
+	parent1 = clk_hw_get_clk(&ctx->parents_ctx[0].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent1);
+	KUNIT_ASSERT_TRUE(test, clk_is_match(clk_get_parent(clk), parent1));
+
+	parent2 = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent2);
+
+	ret = clk_set_rate(parent1, DUMMY_CLOCK_RATE_1);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate(parent2, DUMMY_CLOCK_RATE_2);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_rate_range(clk,
+				 DUMMY_CLOCK_RATE_1 - 1000,
+				 DUMMY_CLOCK_RATE_1 + 1000);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = clk_set_parent(clk, parent2);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	rate = clk_get_rate(clk);
+	KUNIT_ASSERT_GT(test, rate, 0);
+	KUNIT_EXPECT_GE(test, rate, DUMMY_CLOCK_RATE_1 - 1000);
+	KUNIT_EXPECT_LE(test, rate, DUMMY_CLOCK_RATE_1 + 1000);
+
+	clk_put(parent2);
+	clk_put(parent1);
+	clk_put(clk);
+}
+
 static struct kunit_case clk_multiple_parents_mux_test_cases[] = {
 	KUNIT_CASE(clk_test_multiple_parents_mux_get_parent),
 	KUNIT_CASE(clk_test_multiple_parents_mux_has_parent),
+	KUNIT_CASE(clk_test_multiple_parents_mux_set_range_set_parent_get_rate),
 	{}
 };
 
-- 
2.37.1

Re: [PATCH v9 00/25] clk: More clock rate fixes and tests

[Alexander Stein]

Hello Maxime,

Am Dienstag, 16. August 2022, 13:25:05 CEST schrieb Maxime Ripard:
> Thanks to the feedback I got on the previous series, I found and fixed a
> number of bugs in the clock framework and how it deals with rates,
> especially when it comes to orphan clocks.
> 
> In order to make sure this doesn't pop up again as a regression, I've
> extended the number of tests.
> 
> The first patch reintroduces the clk_set_rate_range call on clk_put, but
> this time will only do so if there was a range set on that clock to
> begin with. It should be less intrusive, and reduce the number of
> potential side effects considerably.
> 
> We then have a fix for the qcom rcg2 issue that has been reported
> recently.
> 
> All the other patches should be probably be flagged as fixes, but
> they've never seem to have shown any real-world issues until now, and
> they aren't all really trivial to backport either, so I'm not sure it's
> worth it.
> 
> There's also some documentation improvements for recalc_rate and
> clk_get_rate to hopefully make the documentation less ambiguous and
> acknowledge that recalc_rate() returning 0 on error is fine.
> 
> Let me know what you think,
> Maxime
> 
> Changes from v8:
>   - Fixed a regression when probing a clock driver backed by a device
> accessed through a bus that might sleep

With this v9 series my TQMa8MPxL platform boots without issue, also with the 
i2c attached audio codec driver successfully probed. Thanks!

Best regards,
Alexander

> Changes from v7:
>   - Dropped the RPi fixes
>   - Rebased on 6.0-rc1
> 
> Changes from v6:
>   - Fixed a kernel build bot warning
> 
> Changes from v5:
>   - Rebased on current next (next-20220711)
>   - Dropped clk_get_rate_range, and used a custom function instead
>   - Switched all tests to use clk_hw_get_clk() instead of struct clk_hw->clk
> - Removed some intermediate variables
>   - Added some comments
>   - Dropped clk_get_parent() changes
>   - Dropped test on clk_hw pointer non-NULL in clk_hw_get_name
>   - Made clk_has_parent more const
> 
> Changes from v4:
>   - Fix build breakage on SAM9x60
> 
> Changes from v3:
>   - constness warning fix in clk_core_forward_rate_req
> 
> Changes from v2:
>   - Rebased on top of current next
>   - Fixed locking issue in clk_get_rate_range
> 
> Changes from v1:
>   - Rebased on top of next-20220428
>   - Dropped the patch to prevent non-orphan clocks from registering if
>     their recalc_rate hook returns 0
>   - Added some patches to clarify the clk_get_rate and recalc_rate
>     documentation
>   - Dropped the patch to skip the range setup on an orphan clock that
>     was introducing a regression on RaspberryPi3 when a monitor wasn't
>     connected at boot
>   - Added a patch to skip the rate clamping in clk_round_rate() when
>     min_rate == max_rate == 0
>   - Added a new set of functions to query the clk boundaries and fix a
>     regression with the RaspberryPi4
>   - Fixed all the drivers hand-crafting their clk_rate_request
>   - Reworded the test suite descriptions
>   - Reordered a few patches to ease the review
>   - Reworded some commit logs to better explain the issues they address
>   - Collected the Tested-by of Alexander and Marek
>   - More tests
> 
> Maxime Ripard (25):
>   clk: test: Switch to clk_hw_get_clk
>   clk: Drop the rate range on clk_put()
>   clk: Skip clamping when rounding if there's no boundaries
>   clk: Mention that .recalc_rate can return 0 on error
>   clk: Clarify clk_get_rate() expectations
>   clk: tests: Add test suites description
>   clk: tests: Add reference to the orphan mux bug report
>   clk: tests: Add tests for uncached clock
>   clk: tests: Add tests for single parent mux
>   clk: tests: Add tests for mux with multiple parents
>   clk: tests: Add some tests for orphan with multiple parents
>   clk: Take into account uncached clocks in clk_set_rate_range()
>   clk: Set req_rate on reparenting
>   clk: Change clk_core_init_rate_req prototype
>   clk: Move clk_core_init_rate_req() from clk_core_round_rate_nolock()
>     to its caller
>   clk: Introduce clk_hw_init_rate_request()
>   clk: Add our request boundaries in clk_core_init_rate_req
>   clk: Switch from __clk_determine_rate to clk_core_round_rate_nolock
>   clk: Introduce clk_core_has_parent()
>   clk: Constify clk_has_parent()
>   clk: Stop forwarding clk_rate_requests to the parent
>   clk: Zero the clk_rate_request structure
>   clk: Introduce the clk_hw_get_rate_range function
>   clk: qcom: clk-rcg2: Take clock boundaries into consideration for
>     gfx3d
>   clk: tests: Add missing test case for ranges
> 
>  drivers/clk/at91/clk-generated.c  |    5 +-
>  drivers/clk/at91/clk-master.c     |    9 +-
>  drivers/clk/at91/clk-peripheral.c |    4 +-
>  drivers/clk/clk-composite.c       |    6 +-
>  drivers/clk/clk-divider.c         |   20 +-
>  drivers/clk/clk.c                 |  288 ++++--
>  drivers/clk/clk_test.c            | 1413 ++++++++++++++++++++++++++++-
>  drivers/clk/qcom/clk-rcg2.c       |    9 +
>  include/linux/clk-provider.h      |   18 +-
>  include/linux/clk.h               |    2 +-
>  10 files changed, 1665 insertions(+), 109 deletions(-)

Re: [PATCH v9 00/25] clk: More clock rate fixes and tests

[Naresh Kamboju]

On Tue, 16 Aug 2022 at 16:55, Maxime Ripard <maxime@cerno.tech> wrote:
>
> Hi,
>
> Thanks to the feedback I got on the previous series, I found and fixed a
> number of bugs in the clock framework and how it deals with rates,
> especially when it comes to orphan clocks.
>
> In order to make sure this doesn't pop up again as a regression, I've
> extended the number of tests.
>
> The first patch reintroduces the clk_set_rate_range call on clk_put, but
> this time will only do so if there was a range set on that clock to
> begin with. It should be less intrusive, and reduce the number of
> potential side effects considerably.
>
> We then have a fix for the qcom rcg2 issue that has been reported
> recently.
>
> All the other patches should be probably be flagged as fixes, but
> they've never seem to have shown any real-world issues until now, and
> they aren't all really trivial to backport either, so I'm not sure it's
> worth it.
>
> There's also some documentation improvements for recalc_rate and
> clk_get_rate to hopefully make the documentation less ambiguous and
> acknowledge that recalc_rate() returning 0 on error is fine.
>
> Let me know what you think,
> Maxime
>
> Changes from v8:
>   - Fixed a regression when probing a clock driver backed by a device accessed
>     through a bus that might sleep
>
> Changes from v7:
>   - Dropped the RPi fixes
>   - Rebased on 6.0-rc1
>
> Changes from v6:
>   - Fixed a kernel build bot warning
>
> Changes from v5:
>   - Rebased on current next (next-20220711)
>   - Dropped clk_get_rate_range, and used a custom function instead
>   - Switched all tests to use clk_hw_get_clk() instead of struct clk_hw->clk
>   - Removed some intermediate variables
>   - Added some comments
>   - Dropped clk_get_parent() changes
>   - Dropped test on clk_hw pointer non-NULL in clk_hw_get_name
>   - Made clk_has_parent more const
>
> Changes from v4:
>   - Fix build breakage on SAM9x60
>
> Changes from v3:
>   - constness warning fix in clk_core_forward_rate_req
>
> Changes from v2:
>   - Rebased on top of current next
>   - Fixed locking issue in clk_get_rate_range
>
> Changes from v1:
>   - Rebased on top of next-20220428
>   - Dropped the patch to prevent non-orphan clocks from registering if
>     their recalc_rate hook returns 0
>   - Added some patches to clarify the clk_get_rate and recalc_rate
>     documentation
>   - Dropped the patch to skip the range setup on an orphan clock that
>     was introducing a regression on RaspberryPi3 when a monitor wasn't
>     connected at boot
>   - Added a patch to skip the rate clamping in clk_round_rate() when
>     min_rate == max_rate == 0
>   - Added a new set of functions to query the clk boundaries and fix a
>     regression with the RaspberryPi4
>   - Fixed all the drivers hand-crafting their clk_rate_request
>   - Reworded the test suite descriptions
>   - Reordered a few patches to ease the review
>   - Reworded some commit logs to better explain the issues they address
>   - Collected the Tested-by of Alexander and Marek
>   - More tests
>
> Maxime Ripard (25):
>   clk: test: Switch to clk_hw_get_clk
>   clk: Drop the rate range on clk_put()
>   clk: Skip clamping when rounding if there's no boundaries
>   clk: Mention that .recalc_rate can return 0 on error
>   clk: Clarify clk_get_rate() expectations
>   clk: tests: Add test suites description
>   clk: tests: Add reference to the orphan mux bug report
>   clk: tests: Add tests for uncached clock
>   clk: tests: Add tests for single parent mux
>   clk: tests: Add tests for mux with multiple parents
>   clk: tests: Add some tests for orphan with multiple parents
>   clk: Take into account uncached clocks in clk_set_rate_range()
>   clk: Set req_rate on reparenting
>   clk: Change clk_core_init_rate_req prototype
>   clk: Move clk_core_init_rate_req() from clk_core_round_rate_nolock()
>     to its caller
>   clk: Introduce clk_hw_init_rate_request()
>   clk: Add our request boundaries in clk_core_init_rate_req
>   clk: Switch from __clk_determine_rate to clk_core_round_rate_nolock
>   clk: Introduce clk_core_has_parent()
>   clk: Constify clk_has_parent()
>   clk: Stop forwarding clk_rate_requests to the parent
>   clk: Zero the clk_rate_request structure
>   clk: Introduce the clk_hw_get_rate_range function
>   clk: qcom: clk-rcg2: Take clock boundaries into consideration for
>     gfx3d
>   clk: tests: Add missing test case for ranges
>
>  drivers/clk/at91/clk-generated.c  |    5 +-
>  drivers/clk/at91/clk-master.c     |    9 +-
>  drivers/clk/at91/clk-peripheral.c |    4 +-
>  drivers/clk/clk-composite.c       |    6 +-
>  drivers/clk/clk-divider.c         |   20 +-
>  drivers/clk/clk.c                 |  288 ++++--
>  drivers/clk/clk_test.c            | 1413 ++++++++++++++++++++++++++++-
>  drivers/clk/qcom/clk-rcg2.c       |    9 +
>  include/linux/clk-provider.h      |   18 +-
>  include/linux/clk.h               |    2 +-
>  10 files changed, 1665 insertions(+), 109 deletions(-)

I have done build testing, boot testing and LTP syscalls testing on
arm64 devices Rpi4 and db410c all test pass.

Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Naresh Kamboju <naresh.kamboju@linaro.org>

Build link:
https://builds.tuxbuild.com/2DUDJVgnEeotDGSNrcOuFswbTRm/

Test runs links:
https://lkft.validation.linaro.org/scheduler/job/5402030
https://lkft.validation.linaro.org/scheduler/job/5402031

- Naresh

Re: [PATCH v9 00/25] clk: More clock rate fixes and tests

[Maxime Ripard]

[-- Attachment #1: Type: text/plain, Size: 1885 bytes --]

Stephen, Mike,

On Tue, Aug 16, 2022 at 01:25:05PM +0200, Maxime Ripard wrote:
> Thanks to the feedback I got on the previous series, I found and fixed a
> number of bugs in the clock framework and how it deals with rates,
> especially when it comes to orphan clocks.
> 
> In order to make sure this doesn't pop up again as a regression, I've
> extended the number of tests.
> 
> The first patch reintroduces the clk_set_rate_range call on clk_put, but
> this time will only do so if there was a range set on that clock to
> begin with. It should be less intrusive, and reduce the number of
> potential side effects considerably.
> 
> We then have a fix for the qcom rcg2 issue that has been reported
> recently.
> 
> All the other patches should be probably be flagged as fixes, but
> they've never seem to have shown any real-world issues until now, and
> they aren't all really trivial to backport either, so I'm not sure it's
> worth it.
> 
> There's also some documentation improvements for recalc_rate and
> clk_get_rate to hopefully make the documentation less ambiguous and
> acknowledge that recalc_rate() returning 0 on error is fine.

I'm not sure what to do at that point.

Back in July, you felt uncomfortable merging that series so close to the
merge window. I've sent a new series as soon as -rc1 was out, got a
kernelci run to test it and on a number of other platforms. And now
we're close to rc4, which means that you're going to bring the "we're
too late now for 6.1, let's target 6.2".

I believe this series fixes a number of real bugs in the CCF, in
addition to extending quite a lot the unit test coverage of the
framework. But if you just don't care, please just say so. I really
don't want to waste any more time rebasing and sending that series, and
pinging you on a regular basis if it's not going anywhere.

Maxime

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH v9 00/25] clk: More clock rate fixes and tests

[Stephen Boyd]

Quoting Maxime Ripard (2022-09-02 07:53:05)
> 
> I believe this series fixes a number of real bugs in the CCF, in
> addition to extending quite a lot the unit test coverage of the
> framework. But if you just don't care, please just say so. I really
> don't want to waste any more time rebasing and sending that series, and
> pinging you on a regular basis if it's not going anywhere.
> 

I've merged the patch series to clk-next. If things go sideways in the
next week we can figure out what to do. Thanks for sticking with it!

Re: [PATCH v9 00/25] clk: More clock rate fixes and tests

[Maxime Ripard]

[-- Attachment #1: Type: text/plain, Size: 755 bytes --]

Hi Stephen,

On Sat, Sep 17, 2022 at 01:31:35AM -0700, Stephen Boyd wrote:
> Quoting Maxime Ripard (2022-09-02 07:53:05)
> > 
> > I believe this series fixes a number of real bugs in the CCF, in
> > addition to extending quite a lot the unit test coverage of the
> > framework. But if you just don't care, please just say so. I really
> > don't want to waste any more time rebasing and sending that series, and
> > pinging you on a regular basis if it's not going anywhere.
> > 
> 
> I've merged the patch series to clk-next. If things go sideways in the
> next week we can figure out what to do. Thanks for sticking with it!

It looks like you picked it up in a separate branch, but didnt't
merge/push the merge into clk-next?

Maxime

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH v9 13/25] clk: Set req_rate on reparenting

[Marek Szyprowski]

Hi Maxime,

On 16.08.2022 13:25, Maxime Ripard wrote:
> If a non-rate clock started by default with a parent that never
> registered, core->req_rate will be 0. The expectation is that whenever
> the parent will be registered, req_rate will be updated with the new
> value that has just been computed.
>
> However, if that clock is a mux, clk_set_parent() can also make that
> clock no longer orphan. In this case however, we never update req_rate.
>
> The natural solution to this would be to update core->rate and
> core->req_rate in clk_reparent() by calling clk_recalc().
>
> However, this doesn't work in all cases. Indeed, clk_recalc() is called
> by __clk_set_parent_before(), __clk_set_parent() and
> clk_core_reparent(). Both __clk_set_parent_before() and __clk_set_parent
> will call clk_recalc() with the enable_lock taken through a call to
> clk_enable_lock(), the underlying locking primitive being a spinlock.
>
> clk_recalc() calls the backing driver .recalc_rate hook, and that
> implementation might sleep if the underlying device uses a bus with
> accesses that might sleep, such as i2c.
>
> In such a situation, we would end up sleeping while holding a spinlock,
> and thus in an atomic section.
>
> In order to work around this, we can move the core->rate and
> core->req_rate update to the clk_recalc() calling sites, after the
> enable_lock has been released if it was taken.
>
> The only situation that could still be problematic is the
> clk_core_reparent() -> clk_reparent() case that doesn't have any
> locking. clk_core_reparent() is itself called by clk_hw_reparent(),
> which is then called by 4 drivers:
>
>    * clk-stm32mp1.c, stm32/clk-stm32-core.c and tegra/clk-tegra210-emc.c
>      use it in their set_parent implementation. The set_parent hook is
>      only called by __clk_set_parent() and clk_change_rate(), both of
>      them calling it without the enable_lock taken.
>
>    * clk/tegra/clk-tegra124-emc.c calls it as part of its set_rate
>      implementation. set_rate is only called by clk_change_rate(), again
>      without the enable_lock taken.
>
> In both cases we can't end up in a situation where the clk_hw_reparent()
> caller would hold a spinlock, so it seems like this is a good
> workaround.
>
> Let's also add some unit tests to make sure we cover the original bug.
>
> Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b

Well, I don't have good news. This patch has my 'tested-by', but this 
version again doesn't work properly on Meson G12B.

Linux next-20220929, which contains this patch as commit cb1b1dd96241 
("clk: Set req_rate on reparenting") fails to boot on Odroid-N2. I only 
see a freeze once the modules (I see some messages from meson drm and 
cpu_freq) are loaded. Could you remind me how to help debugging this 
issue? I will try to identify which clock causes the issue. Reverting 
$subject on top of linux-next fixes/hides the problem.


> Signed-off-by: Maxime Ripard <maxime@cerno.tech>
> ---
>   drivers/clk/clk.c      |  22 ++++
>   drivers/clk/clk_test.c | 239 +++++++++++++++++++++++++++++++++++++++++
>   2 files changed, 261 insertions(+)
>
> diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> index 53b28e63deae..91bb1ea0e147 100644
> --- a/drivers/clk/clk.c
> +++ b/drivers/clk/clk.c
> @@ -1765,6 +1765,23 @@ static void clk_core_update_orphan_status(struct clk_core *core, bool is_orphan)
>   		clk_core_update_orphan_status(child, is_orphan);
>   }
>   
> +/*
> + * Update the orphan rate and req_rate of @core and all its children.
> + */
> +static void clk_core_update_orphan_child_rates(struct clk_core *core)
> +{
> +	struct clk_core *child;
> +	unsigned long parent_rate = 0;
> +
> +	if (core->parent)
> +		parent_rate = core->parent->rate;
> +
> +	core->rate = core->req_rate = clk_recalc(core, parent_rate);
> +
> +	hlist_for_each_entry(child, &core->children, child_node)
> +		clk_core_update_orphan_child_rates(child);
> +}
> +
>   static void clk_reparent(struct clk_core *core, struct clk_core *new_parent)
>   {
>   	bool was_orphan = core->orphan;
> @@ -1834,6 +1851,8 @@ static struct clk_core *__clk_set_parent_before(struct clk_core *core,
>   	clk_reparent(core, parent);
>   	clk_enable_unlock(flags);
>   
> +	clk_core_update_orphan_child_rates(core);
> +
>   	return old_parent;
>   }
>   
> @@ -1878,6 +1897,8 @@ static int __clk_set_parent(struct clk_core *core, struct clk_core *parent,
>   		flags = clk_enable_lock();
>   		clk_reparent(core, old_parent);
>   		clk_enable_unlock(flags);
> +
> +		clk_core_update_orphan_child_rates(core);
>   		__clk_set_parent_after(core, old_parent, parent);
>   
>   		return ret;
> @@ -2506,6 +2527,7 @@ static void clk_core_reparent(struct clk_core *core,
>   				  struct clk_core *new_parent)
>   {
>   	clk_reparent(core, new_parent);
> +	clk_core_update_orphan_child_rates(core);
>   	__clk_recalc_accuracies(core);
>   	__clk_recalc_rates(core, POST_RATE_CHANGE);
>   }
> diff --git a/drivers/clk/clk_test.c b/drivers/clk/clk_test.c
> index d3e121f21ae2..d1b1372f7aaa 100644
> --- a/drivers/clk/clk_test.c
> +++ b/drivers/clk/clk_test.c
> @@ -594,6 +594,41 @@ clk_test_orphan_transparent_multiple_parent_mux_set_parent(struct kunit *test)
>   	clk_put(clk);
>   }
>   
> +/*
> + * Test that, for a mux that started orphan but got switched to a valid
> + * parent, calling clk_drop_range() on the mux won't affect the parent
> + * rate.
> + */
> +static void
> +clk_test_orphan_transparent_multiple_parent_mux_set_parent_drop_range(struct kunit *test)
> +{
> +	struct clk_multiple_parent_ctx *ctx = test->priv;
> +	struct clk_hw *hw = &ctx->hw;
> +	struct clk *clk = clk_hw_get_clk(hw, NULL);
> +	struct clk *parent;
> +	unsigned long parent_rate, new_parent_rate;
> +	int ret;
> +
> +	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
> +	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
> +
> +	parent_rate = clk_get_rate(parent);
> +	KUNIT_ASSERT_GT(test, parent_rate, 0);
> +
> +	ret = clk_set_parent(clk, parent);
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +
> +	ret = clk_drop_range(clk);
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +
> +	new_parent_rate = clk_get_rate(clk);
> +	KUNIT_ASSERT_GT(test, new_parent_rate, 0);
> +	KUNIT_EXPECT_EQ(test, parent_rate, new_parent_rate);
> +
> +	clk_put(parent);
> +	clk_put(clk);
> +}
> +
>   /*
>    * Test that, for a mux that started orphan but got switched to a valid
>    * parent, the rate of the mux and its new parent are consistent.
> @@ -625,6 +660,39 @@ clk_test_orphan_transparent_multiple_parent_mux_set_parent_get_rate(struct kunit
>   	clk_put(clk);
>   }
>   
> +/*
> + * Test that, for a mux that started orphan but got switched to a valid
> + * parent, calling clk_put() on the mux won't affect the parent rate.
> + */
> +static void
> +clk_test_orphan_transparent_multiple_parent_mux_set_parent_put(struct kunit *test)
> +{
> +	struct clk_multiple_parent_ctx *ctx = test->priv;
> +	struct clk *clk, *parent;
> +	unsigned long parent_rate, new_parent_rate;
> +	int ret;
> +
> +	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
> +	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
> +
> +	clk = clk_hw_get_clk(&ctx->hw, NULL);
> +	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, clk);
> +
> +	parent_rate = clk_get_rate(parent);
> +	KUNIT_ASSERT_GT(test, parent_rate, 0);
> +
> +	ret = clk_set_parent(clk, parent);
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +
> +	clk_put(clk);
> +
> +	new_parent_rate = clk_get_rate(parent);
> +	KUNIT_ASSERT_GT(test, new_parent_rate, 0);
> +	KUNIT_EXPECT_EQ(test, parent_rate, new_parent_rate);
> +
> +	clk_put(parent);
> +}
> +
>   /*
>    * Test that, for a mux that started orphan but got switched to a valid
>    * parent, calling clk_set_rate_range() will affect the parent state if
> @@ -658,6 +726,43 @@ clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_modified(st
>   	clk_put(clk);
>   }
>   
> +/*
> + * Test that, for a mux that started orphan but got switched to a valid
> + * parent, calling clk_set_rate_range() won't affect the parent state if
> + * its rate is within range.
> + */
> +static void
> +clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_untouched(struct kunit *test)
> +{
> +	struct clk_multiple_parent_ctx *ctx = test->priv;
> +	struct clk_hw *hw = &ctx->hw;
> +	struct clk *clk = clk_hw_get_clk(hw, NULL);
> +	struct clk *parent;
> +	unsigned long parent_rate, new_parent_rate;
> +	int ret;
> +
> +	parent = clk_hw_get_clk(&ctx->parents_ctx[1].hw, NULL);
> +	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, parent);
> +
> +	parent_rate = clk_get_rate(parent);
> +	KUNIT_ASSERT_GT(test, parent_rate, 0);
> +
> +	ret = clk_set_parent(clk, parent);
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +
> +	ret = clk_set_rate_range(clk,
> +				 DUMMY_CLOCK_INIT_RATE - 1000,
> +				 DUMMY_CLOCK_INIT_RATE + 1000);
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +
> +	new_parent_rate = clk_get_rate(parent);
> +	KUNIT_ASSERT_GT(test, new_parent_rate, 0);
> +	KUNIT_EXPECT_EQ(test, parent_rate, new_parent_rate);
> +
> +	clk_put(parent);
> +	clk_put(clk);
> +}
> +
>   /*
>    * Test that, for a mux whose current parent hasn't been registered yet,
>    * calling clk_set_rate_range() will succeed, and will be taken into
> @@ -724,8 +829,11 @@ clk_test_orphan_transparent_multiple_parent_mux_set_range_set_parent_get_rate(st
>   static struct kunit_case clk_orphan_transparent_multiple_parent_mux_test_cases[] = {
>   	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_get_parent),
>   	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent),
> +	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_drop_range),
>   	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_get_rate),
> +	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_put),
>   	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_modified),
> +	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_parent_set_range_untouched),
>   	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_range_round_rate),
>   	KUNIT_CASE(clk_test_orphan_transparent_multiple_parent_mux_set_range_set_parent_get_rate),
>   	{}
> @@ -1021,6 +1129,136 @@ static struct kunit_suite clk_orphan_transparent_single_parent_test_suite = {
>   	.test_cases = clk_orphan_transparent_single_parent_mux_test_cases,
>   };
>   
> +struct clk_single_parent_two_lvl_ctx {
> +	struct clk_dummy_context parent_parent_ctx;
> +	struct clk_dummy_context parent_ctx;
> +	struct clk_hw hw;
> +};
> +
> +static int
> +clk_orphan_two_level_root_last_test_init(struct kunit *test)
> +{
> +	struct clk_single_parent_two_lvl_ctx *ctx;
> +	int ret;
> +
> +	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
> +	if (!ctx)
> +		return -ENOMEM;
> +	test->priv = ctx;
> +
> +	ctx->parent_ctx.hw.init =
> +		CLK_HW_INIT("intermediate-parent",
> +			    "root-parent",
> +			    &clk_dummy_single_parent_ops,
> +			    CLK_SET_RATE_PARENT);
> +	ret = clk_hw_register(NULL, &ctx->parent_ctx.hw);
> +	if (ret)
> +		return ret;
> +
> +	ctx->hw.init =
> +		CLK_HW_INIT("test-clk", "intermediate-parent",
> +			    &clk_dummy_single_parent_ops,
> +			    CLK_SET_RATE_PARENT);
> +	ret = clk_hw_register(NULL, &ctx->hw);
> +	if (ret)
> +		return ret;
> +
> +	ctx->parent_parent_ctx.rate = DUMMY_CLOCK_INIT_RATE;
> +	ctx->parent_parent_ctx.hw.init =
> +		CLK_HW_INIT_NO_PARENT("root-parent",
> +				      &clk_dummy_rate_ops,
> +				      0);
> +	ret = clk_hw_register(NULL, &ctx->parent_parent_ctx.hw);
> +	if (ret)
> +		return ret;
> +
> +	return 0;
> +}
> +
> +static void
> +clk_orphan_two_level_root_last_test_exit(struct kunit *test)
> +{
> +	struct clk_single_parent_two_lvl_ctx *ctx = test->priv;
> +
> +	clk_hw_unregister(&ctx->hw);
> +	clk_hw_unregister(&ctx->parent_ctx.hw);
> +	clk_hw_unregister(&ctx->parent_parent_ctx.hw);
> +}
> +
> +/*
> + * Test that, for a clock whose parent used to be orphan, clk_get_rate()
> + * will return the proper rate.
> + */
> +static void
> +clk_orphan_two_level_root_last_test_get_rate(struct kunit *test)
> +{
> +	struct clk_single_parent_two_lvl_ctx *ctx = test->priv;
> +	struct clk_hw *hw = &ctx->hw;
> +	struct clk *clk = clk_hw_get_clk(hw, NULL);
> +	unsigned long rate;
> +
> +	rate = clk_get_rate(clk);
> +	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_INIT_RATE);
> +
> +	clk_put(clk);
> +}
> +
> +/*
> + * Test that, for a clock whose parent used to be orphan,
> + * clk_set_rate_range() won't affect its rate if it is already within
> + * range.
> + *
> + * See (for Exynos 4210):
> + * https://lore.kernel.org/linux-clk/366a0232-bb4a-c357-6aa8-636e398e05eb@samsung.com/
> + */
> +static void
> +clk_orphan_two_level_root_last_test_set_range(struct kunit *test)
> +{
> +	struct clk_single_parent_two_lvl_ctx *ctx = test->priv;
> +	struct clk_hw *hw = &ctx->hw;
> +	struct clk *clk = clk_hw_get_clk(hw, NULL);
> +	unsigned long rate;
> +	int ret;
> +
> +	ret = clk_set_rate_range(clk,
> +				 DUMMY_CLOCK_INIT_RATE - 1000,
> +				 DUMMY_CLOCK_INIT_RATE + 1000);
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +
> +	rate = clk_get_rate(clk);
> +	KUNIT_ASSERT_GT(test, rate, 0);
> +	KUNIT_EXPECT_EQ(test, rate, DUMMY_CLOCK_INIT_RATE);
> +
> +	clk_put(clk);
> +}
> +
> +static struct kunit_case
> +clk_orphan_two_level_root_last_test_cases[] = {
> +	KUNIT_CASE(clk_orphan_two_level_root_last_test_get_rate),
> +	KUNIT_CASE(clk_orphan_two_level_root_last_test_set_range),
> +	{}
> +};
> +
> +/*
> + * Test suite for a basic, transparent, clock with a parent that is also
> + * such a clock. The parent's parent is registered last, while the
> + * parent and its child are registered in that order. The intermediate
> + * and leaf clocks will thus be orphan when registered, but the leaf
> + * clock itself will always have its parent and will never be
> + * reparented. Indeed, it's only orphan because its parent is.
> + *
> + * These tests exercise the behaviour of the consumer API when dealing
> + * with an orphan clock, and how we deal with the transition to a valid
> + * parent.
> + */
> +static struct kunit_suite
> +clk_orphan_two_level_root_last_test_suite = {
> +	.name = "clk-orphan-two-level-root-last-test",
> +	.init = clk_orphan_two_level_root_last_test_init,
> +	.exit = clk_orphan_two_level_root_last_test_exit,
> +	.test_cases = clk_orphan_two_level_root_last_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
> @@ -1729,6 +1967,7 @@ kunit_test_suites(
>   	&clk_multiple_parents_mux_test_suite,
>   	&clk_orphan_transparent_multiple_parent_mux_test_suite,
>   	&clk_orphan_transparent_single_parent_test_suite,
> +	&clk_orphan_two_level_root_last_test_suite,
>   	&clk_range_test_suite,
>   	&clk_range_maximize_test_suite,
>   	&clk_range_minimize_test_suite,

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland

Re: [PATCH v9 13/25] clk: Set req_rate on reparenting

[Stephen Boyd]

Quoting Marek Szyprowski (2022-10-03 02:59:07)
> On 16.08.2022 13:25, Maxime Ripard wrote:
> > If a non-rate clock started by default with a parent that never
> > registered, core->req_rate will be 0. The expectation is that whenever
> > the parent will be registered, req_rate will be updated with the new
> > value that has just been computed.
> >
> > However, if that clock is a mux, clk_set_parent() can also make that
> > clock no longer orphan. In this case however, we never update req_rate.
> >
> > The natural solution to this would be to update core->rate and
> > core->req_rate in clk_reparent() by calling clk_recalc().
> >
> > However, this doesn't work in all cases. Indeed, clk_recalc() is called
> > by __clk_set_parent_before(), __clk_set_parent() and
> > clk_core_reparent(). Both __clk_set_parent_before() and __clk_set_parent
> > will call clk_recalc() with the enable_lock taken through a call to
> > clk_enable_lock(), the underlying locking primitive being a spinlock.
> >
> > clk_recalc() calls the backing driver .recalc_rate hook, and that
> > implementation might sleep if the underlying device uses a bus with
> > accesses that might sleep, such as i2c.
> >
> > In such a situation, we would end up sleeping while holding a spinlock,
> > and thus in an atomic section.
> >
> > In order to work around this, we can move the core->rate and
> > core->req_rate update to the clk_recalc() calling sites, after the
> > enable_lock has been released if it was taken.
> >
> > The only situation that could still be problematic is the
> > clk_core_reparent() -> clk_reparent() case that doesn't have any
> > locking. clk_core_reparent() is itself called by clk_hw_reparent(),
> > which is then called by 4 drivers:
> >
> >    * clk-stm32mp1.c, stm32/clk-stm32-core.c and tegra/clk-tegra210-emc.c
> >      use it in their set_parent implementation. The set_parent hook is
> >      only called by __clk_set_parent() and clk_change_rate(), both of
> >      them calling it without the enable_lock taken.
> >
> >    * clk/tegra/clk-tegra124-emc.c calls it as part of its set_rate
> >      implementation. set_rate is only called by clk_change_rate(), again
> >      without the enable_lock taken.
> >
> > In both cases we can't end up in a situation where the clk_hw_reparent()
> > caller would hold a spinlock, so it seems like this is a good
> > workaround.
> >
> > Let's also add some unit tests to make sure we cover the original bug.
> >
> > Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
> > Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
> 
> Well, I don't have good news. This patch has my 'tested-by', but this 
> version again doesn't work properly on Meson G12B.
> 
> Linux next-20220929, which contains this patch as commit cb1b1dd96241 
> ("clk: Set req_rate on reparenting") fails to boot on Odroid-N2. I only 
> see a freeze once the modules (I see some messages from meson drm and 
> cpu_freq) are loaded. Could you remind me how to help debugging this 
> issue? I will try to identify which clock causes the issue. Reverting 
> $subject on top of linux-next fixes/hides the problem.
> 
> 
> > Signed-off-by: Maxime Ripard <maxime@cerno.tech>
> > ---
> >   drivers/clk/clk.c      |  22 ++++
> >   drivers/clk/clk_test.c | 239 +++++++++++++++++++++++++++++++++++++++++
> >   2 files changed, 261 insertions(+)
> >
> > diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> > index 53b28e63deae..91bb1ea0e147 100644
> > --- a/drivers/clk/clk.c
> > +++ b/drivers/clk/clk.c
> > @@ -1765,6 +1765,23 @@ static void clk_core_update_orphan_status(struct clk_core *core, bool is_orphan)
> >               clk_core_update_orphan_status(child, is_orphan);
> >   }
> >   
> > +/*
> > + * Update the orphan rate and req_rate of @core and all its children.
> > + */
> > +static void clk_core_update_orphan_child_rates(struct clk_core *core)
> > +{
> > +     struct clk_core *child;
> > +     unsigned long parent_rate = 0;
> > +
> > +     if (core->parent)
> > +             parent_rate = core->parent->rate;
> > +
> > +     core->rate = core->req_rate = clk_recalc(core, parent_rate);
> > +
> > +     hlist_for_each_entry(child, &core->children, child_node)
> > +             clk_core_update_orphan_child_rates(child);
> > +}
> > +
> >   static void clk_reparent(struct clk_core *core, struct clk_core *new_parent)
> >   {
> >       bool was_orphan = core->orphan;
> > @@ -2506,6 +2527,7 @@ static void clk_core_reparent(struct clk_core *core,
> >                                 struct clk_core *new_parent)
> >   {
> >       clk_reparent(core, new_parent);
> > +     clk_core_update_orphan_child_rates(core);
> >       __clk_recalc_accuracies(core);
> >       __clk_recalc_rates(core, POST_RATE_CHANGE);

I see a problem. __clk_recalc_rates() uses 'core->rate' as "old rate"
but we'll have already destroyed that by calling
clk_core_update_orphan_child_rates() and assigning 'core->rate' to the
recalc_rate. Are clk notifiers being used? If so, it will probably be
confused because the notifier will see the same rate as what was set
instead of the old rate. cpufreq is probably the biggest user of clk
notifiers.

We should add a test for that so when a clk is reparented the old rate
is still what we expected it to be when the notifier is called.

Also, clk_core_update_orphan_child_rates() is poorly named. It doesn't
care at all that the clk is an orphan. It seems like another
__clk_recalc_rates() without the notifier. I have no idea why we need
another recalc rates. Possibly setting the req_rate in
__clk_recalc_rates() is sufficient. Or maybe we should bail out if the
clk doesn't have the orphan bit set.

Re: [PATCH v9 13/25] clk: Set req_rate on reparenting

[Maxime Ripard]

On Tue, Oct 04, 2022 at 01:59:50PM -0700, Stephen Boyd wrote:
> Quoting Marek Szyprowski (2022-10-03 02:59:07)
> > On 16.08.2022 13:25, Maxime Ripard wrote:
> > > If a non-rate clock started by default with a parent that never
> > > registered, core->req_rate will be 0. The expectation is that whenever
> > > the parent will be registered, req_rate will be updated with the new
> > > value that has just been computed.
> > >
> > > However, if that clock is a mux, clk_set_parent() can also make that
> > > clock no longer orphan. In this case however, we never update req_rate.
> > >
> > > The natural solution to this would be to update core->rate and
> > > core->req_rate in clk_reparent() by calling clk_recalc().
> > >
> > > However, this doesn't work in all cases. Indeed, clk_recalc() is called
> > > by __clk_set_parent_before(), __clk_set_parent() and
> > > clk_core_reparent(). Both __clk_set_parent_before() and __clk_set_parent
> > > will call clk_recalc() with the enable_lock taken through a call to
> > > clk_enable_lock(), the underlying locking primitive being a spinlock.
> > >
> > > clk_recalc() calls the backing driver .recalc_rate hook, and that
> > > implementation might sleep if the underlying device uses a bus with
> > > accesses that might sleep, such as i2c.
> > >
> > > In such a situation, we would end up sleeping while holding a spinlock,
> > > and thus in an atomic section.
> > >
> > > In order to work around this, we can move the core->rate and
> > > core->req_rate update to the clk_recalc() calling sites, after the
> > > enable_lock has been released if it was taken.
> > >
> > > The only situation that could still be problematic is the
> > > clk_core_reparent() -> clk_reparent() case that doesn't have any
> > > locking. clk_core_reparent() is itself called by clk_hw_reparent(),
> > > which is then called by 4 drivers:
> > >
> > >    * clk-stm32mp1.c, stm32/clk-stm32-core.c and tegra/clk-tegra210-emc.c
> > >      use it in their set_parent implementation. The set_parent hook is
> > >      only called by __clk_set_parent() and clk_change_rate(), both of
> > >      them calling it without the enable_lock taken.
> > >
> > >    * clk/tegra/clk-tegra124-emc.c calls it as part of its set_rate
> > >      implementation. set_rate is only called by clk_change_rate(), again
> > >      without the enable_lock taken.
> > >
> > > In both cases we can't end up in a situation where the clk_hw_reparent()
> > > caller would hold a spinlock, so it seems like this is a good
> > > workaround.
> > >
> > > Let's also add some unit tests to make sure we cover the original bug.
> > >
> > > Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
> > > Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
> > 
> > Well, I don't have good news. This patch has my 'tested-by', but this 
> > version again doesn't work properly on Meson G12B.
> > 
> > Linux next-20220929, which contains this patch as commit cb1b1dd96241 
> > ("clk: Set req_rate on reparenting") fails to boot on Odroid-N2. I only 
> > see a freeze once the modules (I see some messages from meson drm and 
> > cpu_freq) are loaded. Could you remind me how to help debugging this 
> > issue? I will try to identify which clock causes the issue. Reverting 
> > $subject on top of linux-next fixes/hides the problem.
> > 
> > 
> > > Signed-off-by: Maxime Ripard <maxime@cerno.tech>
> > > ---
> > >   drivers/clk/clk.c      |  22 ++++
> > >   drivers/clk/clk_test.c | 239 +++++++++++++++++++++++++++++++++++++++++
> > >   2 files changed, 261 insertions(+)
> > >
> > > diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> > > index 53b28e63deae..91bb1ea0e147 100644
> > > --- a/drivers/clk/clk.c
> > > +++ b/drivers/clk/clk.c
> > > @@ -1765,6 +1765,23 @@ static void clk_core_update_orphan_status(struct clk_core *core, bool is_orphan)
> > >               clk_core_update_orphan_status(child, is_orphan);
> > >   }
> > >   
> > > +/*
> > > + * Update the orphan rate and req_rate of @core and all its children.
> > > + */
> > > +static void clk_core_update_orphan_child_rates(struct clk_core *core)
> > > +{
> > > +     struct clk_core *child;
> > > +     unsigned long parent_rate = 0;
> > > +
> > > +     if (core->parent)
> > > +             parent_rate = core->parent->rate;
> > > +
> > > +     core->rate = core->req_rate = clk_recalc(core, parent_rate);
> > > +
> > > +     hlist_for_each_entry(child, &core->children, child_node)
> > > +             clk_core_update_orphan_child_rates(child);
> > > +}
> > > +
> > >   static void clk_reparent(struct clk_core *core, struct clk_core *new_parent)
> > >   {
> > >       bool was_orphan = core->orphan;
> > > @@ -2506,6 +2527,7 @@ static void clk_core_reparent(struct clk_core *core,
> > >                                 struct clk_core *new_parent)
> > >   {
> > >       clk_reparent(core, new_parent);
> > > +     clk_core_update_orphan_child_rates(core);
> > >       __clk_recalc_accuracies(core);
> > >       __clk_recalc_rates(core, POST_RATE_CHANGE);
> 
> I see a problem. __clk_recalc_rates() uses 'core->rate' as "old rate"
> but we'll have already destroyed that by calling
> clk_core_update_orphan_child_rates() and assigning 'core->rate' to the
> recalc_rate. Are clk notifiers being used? If so, it will probably be
> confused because the notifier will see the same rate as what was set
> instead of the old rate. cpufreq is probably the biggest user of clk
> notifiers.

That's a very good point... Which raises another one. Would it be ok to
notify users on a reparenting? It would make sense to me, since the rate
could be affected, but it's not been done so far so I'm not sure what
the implications might be

> We should add a test for that so when a clk is reparented the old rate
> is still what we expected it to be when the notifier is called.

I can do it, but I'm not sure what you want to test exactly. Let's
assume we have a mux with a given rate, we change the parent of that
mux, the rate is likely to be changed as well and we should put in the
notifier that the old_rate is the first parent's, and the new rate the
one of the new parent?

> Also, clk_core_update_orphan_child_rates() is poorly named. It doesn't
> care at all that the clk is an orphan. It seems like another
> __clk_recalc_rates() without the notifier. I have no idea why we need
> another recalc rates.

You're right, the only difference between the two (aside from the
notifiers) is that req_rate is also updated in
clk_core_update_orphan_child_rates().

> Possibly setting the req_rate in __clk_recalc_rates() is sufficient.
> Or maybe we should bail out if the clk doesn't have the orphan bit
> set.

Either way makes sense to me, the latter is probably less intrusive, but
the former allows to consolidate __clk_recalc_rates() and
clk_core_update_orphan_child_rates(). Which one would you prefer?

Maxime

Re: [PATCH v9 13/25] clk: Set req_rate on reparenting

[Maxime Ripard]

[-- Attachment #1: Type: text/plain, Size: 3654 bytes --]

Replying to my own questions after some time working on this.

On Mon, Oct 10, 2022 at 11:56:08AM +0200, Maxime Ripard wrote:
> > > > +/*
> > > > + * Update the orphan rate and req_rate of @core and all its children.
> > > > + */
> > > > +static void clk_core_update_orphan_child_rates(struct clk_core *core)
> > > > +{
> > > > +     struct clk_core *child;
> > > > +     unsigned long parent_rate = 0;
> > > > +
> > > > +     if (core->parent)
> > > > +             parent_rate = core->parent->rate;
> > > > +
> > > > +     core->rate = core->req_rate = clk_recalc(core, parent_rate);
> > > > +
> > > > +     hlist_for_each_entry(child, &core->children, child_node)
> > > > +             clk_core_update_orphan_child_rates(child);
> > > > +}
> > > > +
> > > >   static void clk_reparent(struct clk_core *core, struct clk_core *new_parent)
> > > >   {
> > > >       bool was_orphan = core->orphan;
> > > > @@ -2506,6 +2527,7 @@ static void clk_core_reparent(struct clk_core *core,
> > > >                                 struct clk_core *new_parent)
> > > >   {
> > > >       clk_reparent(core, new_parent);
> > > > +     clk_core_update_orphan_child_rates(core);
> > > >       __clk_recalc_accuracies(core);
> > > >       __clk_recalc_rates(core, POST_RATE_CHANGE);
> > 
> > I see a problem. __clk_recalc_rates() uses 'core->rate' as "old rate"
> > but we'll have already destroyed that by calling
> > clk_core_update_orphan_child_rates() and assigning 'core->rate' to the
> > recalc_rate. Are clk notifiers being used? If so, it will probably be
> > confused because the notifier will see the same rate as what was set
> > instead of the old rate. cpufreq is probably the biggest user of clk
> > notifiers.
> 
> That's a very good point... Which raises another one. Would it be ok to
> notify users on a reparenting? It would make sense to me, since the rate
> could be affected, but it's not been done so far so I'm not sure what
> the implications might be

Turns out it's already done, and the rates were indeed off like you
pointed out.

> > We should add a test for that so when a clk is reparented the old rate
> > is still what we expected it to be when the notifier is called.
> 
> I can do it, but I'm not sure what you want to test exactly. Let's
> assume we have a mux with a given rate, we change the parent of that
> mux, the rate is likely to be changed as well and we should put in the
> notifier that the old_rate is the first parent's, and the new rate the
> one of the new parent?

I implemented this, and this catches the issue you pointed out, so it
looks like a decent test :)

> > Also, clk_core_update_orphan_child_rates() is poorly named. It doesn't
> > care at all that the clk is an orphan. It seems like another
> > __clk_recalc_rates() without the notifier. I have no idea why we need
> > another recalc rates.
> 
> You're right, the only difference between the two (aside from the
> notifiers) is that req_rate is also updated in
> clk_core_update_orphan_child_rates().
> 
> > Possibly setting the req_rate in __clk_recalc_rates() is sufficient.
> > Or maybe we should bail out if the clk doesn't have the orphan bit
> > set.
> 
> Either way makes sense to me, the latter is probably less intrusive, but
> the former allows to consolidate __clk_recalc_rates() and
> clk_core_update_orphan_child_rates(). Which one would you prefer?

I ended up removing clk_core_update_orphan_child_rates() entirely,
adding the test, and submitting it here:
https://lore.kernel.org/linux-clk/20221010-rpi-clk-fixes-again-v1-0-d87ba82ac404@cerno.tech/

Maxime

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH v9 13/25] clk: Set req_rate on reparenting

[Stephen Boyd]

Quoting Maxime Ripard (2022-10-10 07:52:56)
> > > I see a problem. __clk_recalc_rates() uses 'core->rate' as "old rate"
> > > but we'll have already destroyed that by calling
> > > clk_core_update_orphan_child_rates() and assigning 'core->rate' to the
> > > recalc_rate. Are clk notifiers being used? If so, it will probably be
> > > confused because the notifier will see the same rate as what was set
> > > instead of the old rate. cpufreq is probably the biggest user of clk
> > > notifiers.
> > 
> > That's a very good point... Which raises another one. Would it be ok to
> > notify users on a reparenting? It would make sense to me, since the rate
> > could be affected, but it's not been done so far so I'm not sure what
> > the implications might be
> 
> Turns out it's already done, and the rates were indeed off like you
> pointed out.

Yes reparenting does notifiers.

> 
> > > We should add a test for that so when a clk is reparented the old rate
> > > is still what we expected it to be when the notifier is called.
> > 
> > I can do it, but I'm not sure what you want to test exactly. Let's
> > assume we have a mux with a given rate, we change the parent of that
> > mux, the rate is likely to be changed as well and we should put in the
> > notifier that the old_rate is the first parent's, and the new rate the
> > one of the new parent?
> 
> I implemented this, and this catches the issue you pointed out, so it
> looks like a decent test :)

Awesome!

> 
> > > Also, clk_core_update_orphan_child_rates() is poorly named. It doesn't
> > > care at all that the clk is an orphan. It seems like another
> > > __clk_recalc_rates() without the notifier. I have no idea why we need
> > > another recalc rates.
> > 
> > You're right, the only difference between the two (aside from the
> > notifiers) is that req_rate is also updated in
> > clk_core_update_orphan_child_rates().
> > 
> > > Possibly setting the req_rate in __clk_recalc_rates() is sufficient.
> > > Or maybe we should bail out if the clk doesn't have the orphan bit
> > > set.
> > 
> > Either way makes sense to me, the latter is probably less intrusive, but
> > the former allows to consolidate __clk_recalc_rates() and
> > clk_core_update_orphan_child_rates(). Which one would you prefer?
> 
> I ended up removing clk_core_update_orphan_child_rates() entirely,
> adding the test, and submitting it here:
> https://lore.kernel.org/linux-clk/20221010-rpi-clk-fixes-again-v1-0-d87ba82ac404@cerno.tech/
> 

Cool, thanks. Let me go stack that on top. I'd prefer to send that
branch to Linus in the next couple days.