clk: fractional-divider: Improve approximation when zero based and export

The fractional divider approximation does not utilize the full available range for clocks that are flagged CLK_FRAC_DIVIDER_ZERO_BASED. This patchset aims to fix that.

It also adds test cases for the edge cases of fractional divider clocks with and without the CLK_FRAC_DIVIDER_ZERO_BASED flag to highlight the changes.

Finally, it also exports clk_fractional_divider_general_approximation so that the test cases (but also other users like rockchip clk driver) can be compiled as a module.

Unfortunately, I have no boards to test this patch. So all we have are the unit tests. It seems the only user of this flag in mainline is drivers/clk/imx/clk-composite-7ulp.c, therefore I'm cc-ing get_maintainers.pl --git-blame -f drivers/clk/imx/clk-composite-7ulp.c in the hopes of a wider audience.

Thank you for considering this contribution,
Frank

P.S.: V4 was short-lived, because it triggered a compiler warning on clang. I'm sorry for the noise. -- Frank

V4: https://lore.kernel.org/all/[email protected]/
V3: https://lore.kernel.org/all/[email protected]/
V2: https://lore.kernel.org/lkml/[email protected]/
V1: https://lore.kernel.org/lkml/[email protected]/

Changes in V5 (since V4):
- Fix compiler warnings on clang by removing superfluous variable max_m in the test cases for the maximum denominator (n) and max_n in the test cases for max numerator (m).
Thank you, Intel, for your kernel test robot!

Changes in V4 (since V3):
- Export clk_fractional_divider_general_approximation so that users (e.g., the testcases) can be compiled as modules.
- Change test cases so that they test clk_fractional_divider_general_approximation again (like in V2) instead of clk_fd_round_rate (like in V3), but keeping the structure of V3 with separate file and individual test cases for each edge case.

Changes in V3 (since V2):
- Completely reworked the test cases
- Moved tests to separate file as per Stephen's request
- Move each edge case into their individual test case as per Stephen's request
- Test clk_fd_round_rate instead of clk_fractional_divider_general_approximation as testing the latter broke builds

Changes in V2 (since V1):
- Added test case as requested by Stephen Boyd
- Fixed commit message as the Cc: was missing a closing bracket, so that the original mail unfortunately did not go out to A. s. Dong.

Frank Oltmanns (2):
clk: fractional-divider: Improve approximation when zero based and
export
clk: fractional-divider: tests: Add test suite for edge cases

 drivers/clk/.kunitconfig                  |   1 +
 drivers/clk/Kconfig                       |   7 +
 drivers/clk/Makefile                      |   1 +
 drivers/clk/clk-fractional-divider.c      |  27 +++-
 drivers/clk/clk-fractional-divider_test.c | 157 ++++++++++++++++++++++
 5 files changed, 186 insertions(+), 7 deletions(-)
 create mode 100644 drivers/clk/clk-fractional-divider_test.c

2.41.0

Consider the CLK_FRAC_DIVIDER_ZERO_BASED flag when finding the best approximation for m and n. By doing so, increase the range of valid values for the numerator and denominator by 1.

Furthermore, export the approximation function so that users of this function can be compiled as modules.

---
 drivers/clk/clk-fractional-divider.c | 27 ++++++++++++++++++++-------
 1 file changed, 20 insertions(+), 7 deletions(-)

index 479297763e70..5067e067e906 100644
@@ -123,6 +123,7 @@ void clk_fractional_divider_general_approximation(struct clk_hw *hw,
 						  unsigned long *m, unsigned long *n)
 {
 	struct clk_fractional_divider *fd = to_clk_fd(hw);
+	unsigned long max_m, max_n;
 
 	/*
 	 * Get rate closer to *parent_rate to guarantee there is no overflow

@@ -138,10 +139,17 @@ void clk_fractional_divider_general_approximation(struct clk_hw *hw,
 			rate <<= scale - fd->nwidth;
 	}
 
-	rational_best_approximation(rate, *parent_rate,
-			GENMASK(fd->mwidth - 1, 0), GENMASK(fd->nwidth - 1, 0),
-			m, n);
+	if (fd->flags & CLK_FRAC_DIVIDER_ZERO_BASED) {
+		max_m = 1 << fd->mwidth;
+		max_n = 1 << fd->nwidth;
+	} else {
+		max_m = GENMASK(fd->mwidth - 1, 0);
+		max_n = GENMASK(fd->nwidth - 1, 0);
+	}
+
+	rational_best_approximation(rate, *parent_rate, max_m, max_n, m, n);
 }
+EXPORT_SYMBOL_GPL(clk_fractional_divider_general_approximation);
 
 static long clk_fd_round_rate(struct clk_hw *hw, unsigned long rate,
 			      unsigned long *parent_rate)

@@ -169,13 +177,18 @@ static int clk_fd_set_rate(struct clk_hw *hw, unsigned long rate,
 {
 	struct clk_fractional_divider *fd = to_clk_fd(hw);
 	unsigned long flags = 0;
-	unsigned long m, n;
+	unsigned long m, n, max_m, max_n;
 	u32 mmask, nmask;
 	u32 val;
 
-	rational_best_approximation(rate, parent_rate,
-			GENMASK(fd->mwidth - 1, 0), GENMASK(fd->nwidth - 1, 0),
-			&m, &n);
+	if (fd->flags & CLK_FRAC_DIVIDER_ZERO_BASED) {
+		max_m = 1 << fd->mwidth;
+		max_n = 1 << fd->nwidth;
+	} else {
+		max_m = GENMASK(fd->mwidth - 1, 0);
+		max_n = GENMASK(fd->nwidth - 1, 0);
+	}
+	rational_best_approximation(rate, parent_rate, max_m, max_n, &m, &n);
 
 	if (fd->flags & CLK_FRAC_DIVIDER_ZERO_BASED) {
 		m--;

2.41.0

Quoting Frank Oltmanns (2023-06-17 06:10:40)

Applied to clk-next

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

In light of the recent discovery that the fractional divisor approximation does not utilize the full available range for clocks that are flagged CLK_FRAC_DIVIDER_ZERO_BASED [1], implement tests for the edge cases of this clock type.

---
Please note: I get two checkpatch warnings for this patch:
- Concerning the help text in Kconfig.
- Regarding the file being added, asking if MAINTAINERS needs updating.

Both the help text as well as the MAINTAINERS file seem fine to me.

As expected, when the tests are run *without* PATCH 1, the two "zero based" test cases fail:

================= clk-fd-test (4 subtests) =================
[PASSED] clk_fd_test_approximation_max_denominator
[PASSED] clk_fd_test_approximation_max_numerator
# clk_fd_test_approximation_max_denominator_zero_based: EXPECTATION FAILED at drivers/clk/clk-fractional-divider_test.c:104 Expected n == max_n, but

    n == 7 (0x7)
    max_n == 8 (0x8)
[FAILED] clk_fd_test_approximation_max_denominator_zero_based
# clk_fd_test_approximation_max_numerator_zero_based: EXPECTATION FAILED at drivers/clk/clk-fractional-divider_test.c:134
Expected m == max_m, but
    m == 7 (0x7)
    max_m == 8 (0x8)

[FAILED] clk_fd_test_approximation_max_numerator_zero_based # clk-fd-test: pass:2 fail:2 skip:0 total:4
# Totals: pass:2 fail:2 skip:0 total:4 =================== [FAILED] clk-fd-test ===================

Best regards,
Frank

 drivers/clk/.kunitconfig                  |   1 +
 drivers/clk/Kconfig                       |   7 +
 drivers/clk/Makefile                      |   1 +
 drivers/clk/clk-fractional-divider_test.c | 157 ++++++++++++++++++++++
 4 files changed, 166 insertions(+)
 create mode 100644 drivers/clk/clk-fractional-divider_test.c

index 2fbeb71316f8..efa12ac2b3f2 100644
@@ -2,4 +2,5 @@ CONFIG_KUNIT=y
 CONFIG_COMMON_CLK=y
 CONFIG_CLK_KUNIT_TEST=y
 CONFIG_CLK_GATE_KUNIT_TEST=y
+CONFIG_CLK_FD_KUNIT_TEST=y
 CONFIG_UML_PCI_OVER_VIRTIO=n

index 016814e15536..3fbb40cb5551 100644
@@ -513,4 +513,11 @@ config CLK_GATE_KUNIT_TEST
 	help
 	  Kunit test for the basic clk gate type.
 
+config CLK_FD_KUNIT_TEST
+	tristate "Basic fractional divider type Kunit test" if !KUNIT_ALL_TESTS
+	depends on KUNIT
+	default KUNIT_ALL_TESTS
+	help
+	  Kunit test for the clk-fractional-divider type.
+
 endif

index 0aebef17edc6..9d2337c12dd1 100644
@@ -12,6 +12,7 @@ obj-$(CONFIG_COMMON_CLK)	+= clk-multiplier.o
 obj-$(CONFIG_COMMON_CLK)	+= clk-mux.o
 obj-$(CONFIG_COMMON_CLK)	+= clk-composite.o
 obj-$(CONFIG_COMMON_CLK)	+= clk-fractional-divider.o
+obj-$(CONFIG_CLK_FD_KUNIT_TEST) += clk-fractional-divider_test.o
 obj-$(CONFIG_COMMON_CLK)	+= clk-gpio.o
 ifeq ($(CONFIG_OF), y)
 obj-$(CONFIG_COMMON_CLK)	+= clk-conf.o

2.41.0

Quoting Frank Oltmanns (2023-06-17 06:10:41)

Applied to clk-next

I made some changes but they're mostly cosmetic.

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

Hi,

Are there remarks or questions on this patchset? Anything that needs to be improved?

Thanks,
Frank

Quoting Frank Oltmanns (2023-07-17 01:24:49)

I haven't taken a look yet. You sent this almost exactly when I started a weeks long trip with minimal network access.

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