Its value is calculated in sun4i_pwm_apply() and is used only there.
Cc: stable@vger.kernel.org Signed-off-by: Max Kellermann max.kellermann@gmail.com --- drivers/pwm/pwm-sun4i.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-)
diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c index 91ca67651abd..c7c564a6bf36 100644 --- a/drivers/pwm/pwm-sun4i.c +++ b/drivers/pwm/pwm-sun4i.c @@ -89,7 +89,6 @@ struct sun4i_pwm_chip { void __iomem *base; spinlock_t ctrl_lock; const struct sun4i_pwm_data *data; - unsigned long next_period[2]; };
static inline struct sun4i_pwm_chip *to_sun4i_pwm_chip(struct pwm_chip *chip) @@ -237,6 +236,7 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm, int ret; unsigned int delay_us, prescaler = 0; unsigned long now; + unsigned long next_period; bool bypass;
pwm_get_state(pwm, &cstate); @@ -284,7 +284,7 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
val = (duty & PWM_DTY_MASK) | PWM_PRD(period); sun4i_pwm_writel(sun4i_pwm, val, PWM_CH_PRD(pwm->hwpwm)); - sun4i_pwm->next_period[pwm->hwpwm] = jiffies + + next_period = jiffies + nsecs_to_jiffies(cstate.period + 1000);
if (state->polarity != PWM_POLARITY_NORMAL) @@ -306,9 +306,8 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
/* We need a full period to elapse before disabling the channel. */ now = jiffies; - if (time_before(now, sun4i_pwm->next_period[pwm->hwpwm])) { - delay_us = jiffies_to_usecs(sun4i_pwm->next_period[pwm->hwpwm] - - now); + if (time_before(now, next_period)) { + delay_us = jiffies_to_usecs(next_period - now); if ((delay_us / 500) > MAX_UDELAY_MS) msleep(delay_us / 1000 + 1); else
Basically this code did "jiffies + period - jiffies", and we can simply eliminate the "jiffies" time stamp here.
Cc: stable@vger.kernel.org Signed-off-by: Max Kellermann max.kellermann@gmail.com --- drivers/pwm/pwm-sun4i.c | 19 +++++++------------ 1 file changed, 7 insertions(+), 12 deletions(-)
diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c index c7c564a6bf36..b44deececb8b 100644 --- a/drivers/pwm/pwm-sun4i.c +++ b/drivers/pwm/pwm-sun4i.c @@ -234,9 +234,8 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm, struct pwm_state cstate; u32 ctrl, duty = 0, period = 0, val; int ret; + unsigned int delay_jiffies; unsigned int delay_us, prescaler = 0; - unsigned long now; - unsigned long next_period; bool bypass;
pwm_get_state(pwm, &cstate); @@ -284,8 +283,6 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
val = (duty & PWM_DTY_MASK) | PWM_PRD(period); sun4i_pwm_writel(sun4i_pwm, val, PWM_CH_PRD(pwm->hwpwm)); - next_period = jiffies + - nsecs_to_jiffies(cstate.period + 1000);
if (state->polarity != PWM_POLARITY_NORMAL) ctrl &= ~BIT_CH(PWM_ACT_STATE, pwm->hwpwm); @@ -305,14 +302,12 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm, return 0;
/* We need a full period to elapse before disabling the channel. */ - now = jiffies; - if (time_before(now, next_period)) { - delay_us = jiffies_to_usecs(next_period - now); - if ((delay_us / 500) > MAX_UDELAY_MS) - msleep(delay_us / 1000 + 1); - else - usleep_range(delay_us, delay_us * 2); - } + delay_jiffies = nsecs_to_jiffies(cstate.period + 1000); + delay_us = jiffies_to_usecs(delay_jiffies); + if ((delay_us / 500) > MAX_UDELAY_MS) + msleep(delay_us / 1000 + 1); + else + usleep_range(delay_us, delay_us * 2);
spin_lock(&sun4i_pwm->ctrl_lock); ctrl = sun4i_pwm_readl(sun4i_pwm, PWM_CTRL_REG);
On Tue, Jan 25, 2022 at 01:34:28PM +0100, Max Kellermann wrote:
Basically this code did "jiffies + period - jiffies", and we can simply eliminate the "jiffies" time stamp here.
Cc: stable@vger.kernel.org
I don't see how this is relevant for stable, but the change looks fine:
Acked-by: Uwe Kleine-König u.kleine-koenig@pengutronix.de
Thanks Uwe
This fixes a problem that was supposed to be addressed by commit 6eefb79d6f5bc ("pwm: sun4i: Remove erroneous else branch") - backlight could not be switched off on some Allwinner A20. The commit was correct, but was not a reliable fix for the problem, which was timing related.
The real problem for the backlight switching problem was that sleeping for a full period did not work, because delay_us is always zero.
It is zero because the period (plus 1 microsecond) is rounded down to the next "jiffies", but the period is less than one jiffy.
On my Cubieboard 2, the period is 5ms, and 1 jiffy (at the default HZ=100) is 10ms, so nsecs_to_jiffies(10ms+1us)=0.
The roundtrip from nanoseconds to jiffies and back to microseconds is an unnecessary loss of precision; always rounding down (via nsecs_to_jiffies()) then causes the breakage.
This patch eliminates this roundtrip, and directly converts from nanoseconds to microseconds (for usleep_range()), using DIV_ROUND_UP_ULL() to force rounding up. This way, the sleep time is never zero, and after the sleep, we are guaranteed to be in a different period, and the device is ready for another control command for sure.
Cc: stable@vger.kernel.org Signed-off-by: Max Kellermann max.kellermann@gmail.com --- drivers/pwm/pwm-sun4i.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c index b44deececb8b..8527a655034c 100644 --- a/drivers/pwm/pwm-sun4i.c +++ b/drivers/pwm/pwm-sun4i.c @@ -234,7 +234,6 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm, struct pwm_state cstate; u32 ctrl, duty = 0, period = 0, val; int ret; - unsigned int delay_jiffies; unsigned int delay_us, prescaler = 0; bool bypass;
@@ -302,8 +301,7 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm, return 0;
/* We need a full period to elapse before disabling the channel. */ - delay_jiffies = nsecs_to_jiffies(cstate.period + 1000); - delay_us = jiffies_to_usecs(delay_jiffies); + delay_us = DIV_ROUND_UP_ULL(cstate.period, NSEC_PER_USEC); if ((delay_us / 500) > MAX_UDELAY_MS) msleep(delay_us / 1000 + 1); else
On Tue, Jan 25, 2022 at 01:34:29PM +0100, Max Kellermann wrote:
This fixes a problem that was supposed to be addressed by commit 6eefb79d6f5bc ("pwm: sun4i: Remove erroneous else branch") - backlight could not be switched off on some Allwinner A20. The commit was correct, but was not a reliable fix for the problem, which was timing related.
The real problem for the backlight switching problem was that sleeping for a full period did not work, because delay_us is always zero.
It is zero because the period (plus 1 microsecond) is rounded down to the next "jiffies", but the period is less than one jiffy.
On my Cubieboard 2, the period is 5ms, and 1 jiffy (at the default HZ=100) is 10ms, so nsecs_to_jiffies(10ms+1us)=0.
The roundtrip from nanoseconds to jiffies and back to microseconds is an unnecessary loss of precision; always rounding down (via nsecs_to_jiffies()) then causes the breakage.
This patch eliminates this roundtrip, and directly converts from nanoseconds to microseconds (for usleep_range()), using DIV_ROUND_UP_ULL() to force rounding up. This way, the sleep time is never zero, and after the sleep, we are guaranteed to be in a different period, and the device is ready for another control command for sure.
Cc: stable@vger.kernel.org Signed-off-by: Max Kellermann max.kellermann@gmail.com
Sounds reasonable
Acked-by; Uwe Kleine-König u.kleine-koenig@pengutronix.de
Best regards Uwe
On Thu, Feb 24, 2022 at 05:51:35PM +0100, Uwe Kleine-König wrote:
On Tue, Jan 25, 2022 at 01:34:29PM +0100, Max Kellermann wrote:
This fixes a problem that was supposed to be addressed by commit 6eefb79d6f5bc ("pwm: sun4i: Remove erroneous else branch") - backlight could not be switched off on some Allwinner A20. The commit was correct, but was not a reliable fix for the problem, which was timing related.
The real problem for the backlight switching problem was that sleeping for a full period did not work, because delay_us is always zero.
It is zero because the period (plus 1 microsecond) is rounded down to the next "jiffies", but the period is less than one jiffy.
On my Cubieboard 2, the period is 5ms, and 1 jiffy (at the default HZ=100) is 10ms, so nsecs_to_jiffies(10ms+1us)=0.
The roundtrip from nanoseconds to jiffies and back to microseconds is an unnecessary loss of precision; always rounding down (via nsecs_to_jiffies()) then causes the breakage.
This patch eliminates this roundtrip, and directly converts from nanoseconds to microseconds (for usleep_range()), using DIV_ROUND_UP_ULL() to force rounding up. This way, the sleep time is never zero, and after the sleep, we are guaranteed to be in a different period, and the device is ready for another control command for sure.
Cc: stable@vger.kernel.org Signed-off-by: Max Kellermann max.kellermann@gmail.com
Sounds reasonable
Acked-by; Uwe Kleine-König u.kleine-koenig@pengutronix.de
/me had problems with is keyboard, the Shift-key didn't work and so I typed a ; instead a :. To make patchwork pick up my tag, I'll repeat it here:
Acked-by: Uwe Kleine-König u.kleine-koenig@pengutronix.de
Thanks Uwe
Hello,
On Tue, Jan 25, 2022 at 01:34:27PM +0100, Max Kellermann wrote:
Its value is calculated in sun4i_pwm_apply() and is used only there.
Cc: stable@vger.kernel.org
I think I'd drop this. This isn't a fix worth on it's own to be backported and if this is needed for one of the next patches, the stable maintainers will notice themselves (and it might be worth to shuffle this series to make the fixes come first).
Signed-off-by: Max Kellermann max.kellermann@gmail.com
Other than that, LGTM:
Reviewed-by: Uwe Kleine-König u.kleine-koenig@pengutronix.de
Thanks Uwe
On 2022/01/25 15:31, Uwe Kleine-König u.kleine-koenig@pengutronix.de wrote:
I think I'd drop this. This isn't a fix worth on it's own to be backported and if this is needed for one of the next patches, the stable maintainers will notice themselves (and it might be worth to shuffle this series to make the fixes come first).
The first two patches are preparation for the third patch, which fixes the actual bug.
Of course, I could have done everything in one patch, but I thought splitting the first two out makes review easier. This way, every step is almost trivial.
If you want me to fold the three patches into one, I can do that. But I can't reorder them (or backport only the bug fix to stable).
Max
On Tue, Jan 25, 2022 at 03:39:16PM +0100, Max Kellermann wrote:
On 2022/01/25 15:31, Uwe Kleine-König u.kleine-koenig@pengutronix.de wrote:
I think I'd drop this. This isn't a fix worth on it's own to be backported and if this is needed for one of the next patches, the stable maintainers will notice themselves (and it might be worth to shuffle this series to make the fixes come first).
The first two patches are preparation for the third patch, which fixes the actual bug.
Of course, I could have done everything in one patch, but I thought splitting the first two out makes review easier. This way, every step is almost trivial.
If you want me to fold the three patches into one, I can do that. But I can't reorder them (or backport only the bug fix to stable).
That sounds fine. Note my statement "I'd drop this" only refers to the Cc: stable line.
Best regards Uwe
On Tue, Jan 25, 2022 at 03:31:58PM +0100, Uwe Kleine-König wrote:
Hello,
On Tue, Jan 25, 2022 at 01:34:27PM +0100, Max Kellermann wrote:
Its value is calculated in sun4i_pwm_apply() and is used only there.
Cc: stable@vger.kernel.org
I think I'd drop this. This isn't a fix worth on it's own to be backported and if this is needed for one of the next patches, the stable maintainers will notice themselves (and it might be worth to shuffle this series to make the fixes come first).
Signed-off-by: Max Kellermann max.kellermann@gmail.com
Other than that, LGTM:
Reviewed-by: Uwe Kleine-König u.kleine-koenig@pengutronix.de
Does that apply to patches 2 & 3 as well?
Thierry
On Thu, Feb 24, 2022 at 02:04:23PM +0100, Thierry Reding wrote:
On Tue, Jan 25, 2022 at 03:31:58PM +0100, Uwe Kleine-König wrote:
Hello,
On Tue, Jan 25, 2022 at 01:34:27PM +0100, Max Kellermann wrote:
Its value is calculated in sun4i_pwm_apply() and is used only there.
Cc: stable@vger.kernel.org
I think I'd drop this. This isn't a fix worth on it's own to be backported and if this is needed for one of the next patches, the stable maintainers will notice themselves (and it might be worth to shuffle this series to make the fixes come first).
Signed-off-by: Max Kellermann max.kellermann@gmail.com
Other than that, LGTM:
Reviewed-by: Uwe Kleine-König u.kleine-koenig@pengutronix.de
Does that apply to patches 2 & 3 as well?
No, at that time I only looked at patch 1.
I just looked at 2 and 3 and will reply to them individually.
Best regards Uwe
On Tue, Jan 25, 2022 at 01:34:27PM +0100, Max Kellermann wrote:
Its value is calculated in sun4i_pwm_apply() and is used only there.
Cc: stable@vger.kernel.org Signed-off-by: Max Kellermann max.kellermann@gmail.com
drivers/pwm/pwm-sun4i.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-)
Patches applied, though I dropped the Cc: stable on patches 1 and 2.
Thanks, Thierry
linux-stable-mirror@lists.linaro.org
-
Max Kellermann -
Max Kellermann -
Thierry Reding -
Uwe Kleine-König