[PATCH libpwm] pwmtestperf: Implement -I for inversed polarity requests

[PATCH libpwm] pwmtestperf: Implement -I for inversed polarity requests

[Uwe Kleine-König]

The driver handling for .duty_offset_ns > 0 isn't always trivial to get
right. So make pwmtestperf capable to do some tests with a non-zero duty
offset.

Fix a typo in the manpage en passant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@baylibre.com>
---
Hello,

triggered by a driver submission by Andrea della Porta[1], this increases
test coverage to (hopefully) catch issues around driver handling of
duty offset.

Tesing that on stm32 indeed found a bug, will send a patch for that
later.

Best regards
Uwe

[1] https://lore.kernel.org/linux-pwm/adkrHkANCzxO8KUP@monoceros/

 pwmtestperf.1 | 13 +++++++++++--
 pwmtestperf.c | 17 ++++++++++++++++-
 2 files changed, 27 insertions(+), 3 deletions(-)

diff --git a/pwmtestperf.1 b/pwmtestperf.1
index 74dc8e30f32a..944564c14a0b 100644
--- a/pwmtestperf.1
+++ b/pwmtestperf.1
@@ -13,6 +13,7 @@ pwmtestperf \- Applies a simple test pattern to a PWM output
 .IB periodlength ]
 .RB [ \-S\~\c
 .IB step ]
+.RB [ \-I ]
 .RB [ \-v ]
 .YS
 .SH Description
@@ -27,7 +28,7 @@ and
 
 The fixed period is specified by
 .I periodlength
-and the duty length is increased in each steps by
+and the duty length is increased in each step by
 .I step 
 up from 0 to
 .IR periodlength .
@@ -35,7 +36,11 @@ up from 0 to
 .I step 
 is negative, it decreases from
 .I periodlength
-to 0.
+to 0. With
+.B \-I
+given the duty offset in each request is chosen to describe an inversed
+polarity waveform, that is a falling edge at the end of each period. Otherwise
+the duty offset is zero.
 .
 .SH Options
 .
@@ -68,6 +73,10 @@ in nanoseconds defines the length of the periods of the waveform to be set.
 in nanoseconds defines the step the duty length is increased in each step. (Or decreased if negative.)
 .
 .TP
+.B \-I
+Use requests with inverted polarity.
+.
+.TP
 .B \-v
 Emit some information during runtime.
 .
diff --git a/pwmtestperf.c b/pwmtestperf.c
index 72abf3987459..95be2b6fe64e 100644
--- a/pwmtestperf.c
+++ b/pwmtestperf.c
@@ -22,16 +22,21 @@ int main(int argc, char *const argv[])
 	int ret;
 	int opt;
 	long long step = 1;
+	bool inverted = false;
 
 	unsigned int chipno = 0;
 	unsigned int pwmno = 0;
 
-	while ((opt = getopt(argc, argv, "c:p:P:S:")) != -1) {
+	while ((opt = getopt(argc, argv, "c:Ip:P:S:")) != -1) {
 		switch (opt) {
 		case 'c':
 			chipno = atoi(optarg);
 			break;
 
+		case 'I':
+			inverted = true;
+			break;
+
 		case 'p':
 			pwmno = atoi(optarg);
 			break;
@@ -67,6 +72,16 @@ int main(int argc, char *const argv[])
 	}
 
 	for (wf.duty_length_ns = (step > 0 ? 0 : wf.period_length_ns); wf.duty_length_ns <= wf.period_length_ns; wf.duty_length_ns += step) {
+
+		/*
+		 * .duty_length_ns must be < .period_length_ns, so don't invert
+		 * for .duty_length_ns == 0.
+		 */
+		if (inverted && wf.duty_length_ns)
+			wf.duty_offset_ns = wf.period_length_ns - wf.duty_length_ns;
+		else
+			wf.duty_offset_ns = 0;
+
 		ret = pwm_set_waveform(pwm, &wf);
 		if (ret < 0) {
 			perror("Failed to configure PWM");
-- 
2.47.3

Re: [PATCH libpwm] pwmtestperf: Implement -I for inversed polarity requests

[Uwe Kleine-König]

[-- Attachment #1: Type: text/plain, Size: 476 bytes --]

Hello,

On Wed, Apr 15, 2026 at 04:37:25PM +0200, Uwe Kleine-König wrote:
> The driver handling for .duty_offset_ns > 0 isn't always trivial to get
> right. So make pwmtestperf capable to do some tests with a non-zero duty
> offset.
> 
> Fix a typo in the manpage en passant.
> 
> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@baylibre.com>

I applied this patch to
https://git.kernel.org/pub/scm/linux/kernel/git/ukleinek/libpwm.git main

Best regards
Uwe

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]