Message ID | 1447345684-108039-1-git-send-email-qipeng.zha@intel.com |
---|---|
State | Superseded |
Headers | show |
On Fri, Nov 13, 2015 at 12:28:04AM +0800, Qipeng Zha wrote: > For Broxton PWM controller, base unit is defined as 8bit integer > and 14bit fraction, so need to update base unit setting to output > wave with right frequency. > a) add scaler for each board setting; > b) remove validity check of base unit for special board, let pwm > user to handle this; > > Signed-off-by: Qipeng Zha <qipeng.zha@intel.com> > --- > drivers/pwm/pwm-lpss.c | 20 +++++++++++--------- > drivers/pwm/pwm-lpss.h | 1 + > 2 files changed, 12 insertions(+), 9 deletions(-) > > diff --git a/drivers/pwm/pwm-lpss.c b/drivers/pwm/pwm-lpss.c > index 2504410..5a907db 100644 > --- a/drivers/pwm/pwm-lpss.c > +++ b/drivers/pwm/pwm-lpss.c > @@ -14,6 +14,7 @@ > */ > > #include <linux/io.h> > +#include <linux/time.h> > #include <linux/kernel.h> > #include <linux/module.h> > #include <linux/pm_runtime.h> > @@ -24,11 +25,9 @@ > #define PWM_ENABLE BIT(31) > #define PWM_SW_UPDATE BIT(30) > #define PWM_BASE_UNIT_SHIFT 8 > -#define PWM_BASE_UNIT_MASK 0x00ffff00 > +#define PWM_BASE_UNIT_MASK 0x3fffff00 Isn't this going to potentially write reserved bits on non-Broxton platforms? Previously the upper 8 bits were masked out, but now only the upper 2 bits are masked out. What about the other 6? Perhaps it'd be better to parameterize the mask in a way similar to the scaler value? Also it's unclear to me how critical this is. Initial patches for Broxton support were merged into Linus' tree yesterday. Presumably they had received some testing before, but nobody can't have noticed or this bug wouldn't exist in the patches that were merged. Does this break any existing setups and hence should go into v4.4 along with the initial Broxton support? Thierry
On Thu, Nov 12, 2015 at 01:52:36PM +0100, Thierry Reding wrote: > On Fri, Nov 13, 2015 at 12:28:04AM +0800, Qipeng Zha wrote: > > For Broxton PWM controller, base unit is defined as 8bit integer > > and 14bit fraction, so need to update base unit setting to output > > wave with right frequency. > > a) add scaler for each board setting; > > b) remove validity check of base unit for special board, let pwm > > user to handle this; > > > > Signed-off-by: Qipeng Zha <qipeng.zha@intel.com> > > --- > > drivers/pwm/pwm-lpss.c | 20 +++++++++++--------- > > drivers/pwm/pwm-lpss.h | 1 + > > 2 files changed, 12 insertions(+), 9 deletions(-) > > > > diff --git a/drivers/pwm/pwm-lpss.c b/drivers/pwm/pwm-lpss.c > > index 2504410..5a907db 100644 > > --- a/drivers/pwm/pwm-lpss.c > > +++ b/drivers/pwm/pwm-lpss.c > > @@ -14,6 +14,7 @@ > > */ > > > > #include <linux/io.h> > > +#include <linux/time.h> > > #include <linux/kernel.h> > > #include <linux/module.h> > > #include <linux/pm_runtime.h> > > @@ -24,11 +25,9 @@ > > #define PWM_ENABLE BIT(31) > > #define PWM_SW_UPDATE BIT(30) > > #define PWM_BASE_UNIT_SHIFT 8 > > -#define PWM_BASE_UNIT_MASK 0x00ffff00 > > +#define PWM_BASE_UNIT_MASK 0x3fffff00 > > Isn't this going to potentially write reserved bits on non-Broxton > platforms? Previously the upper 8 bits were masked out, but now only the > upper 2 bits are masked out. What about the other 6? > > Perhaps it'd be better to parameterize the mask in a way similar to the > scaler value? Or call the field "base_unit_bits" which holds 16 for BSW/BYT and 22 for BXT, and calculate both scaler and mask from that in pwm_lpss_config()? Actually I think we should make the driver private structure look like this: struct pwm_lpss_chip { struct pwm_chip chip; void __iomem *regs; const struct pwm_lpss_boardinfo *info; }; and then set ->info to point to the platform data in probe(). pwm_lpss_config() can then refer lpwm->info->base_unit_bits and lpwm->info->clk_rate. > Also it's unclear to me how critical this is. Initial patches for > Broxton support were merged into Linus' tree yesterday. Presumably they > had received some testing before, but nobody can't have noticed or this > bug wouldn't exist in the patches that were merged. Does this break any > existing setups and hence should go into v4.4 along with the initial > Broxton support? Nobody outside Intel should have Broxtons yet so it should not break any existing users. -- To unsubscribe from this list: send the line "unsubscribe linux-pwm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Nov 12, 2015 at 04:50:25PM +0200, Mika Westerberg wrote: > On Thu, Nov 12, 2015 at 01:52:36PM +0100, Thierry Reding wrote: > > On Fri, Nov 13, 2015 at 12:28:04AM +0800, Qipeng Zha wrote: > > > For Broxton PWM controller, base unit is defined as 8bit integer > > > and 14bit fraction, so need to update base unit setting to output > > > wave with right frequency. > > > a) add scaler for each board setting; > > > b) remove validity check of base unit for special board, let pwm > > > user to handle this; > > > > > > Signed-off-by: Qipeng Zha <qipeng.zha@intel.com> > > > --- > > > drivers/pwm/pwm-lpss.c | 20 +++++++++++--------- > > > drivers/pwm/pwm-lpss.h | 1 + > > > 2 files changed, 12 insertions(+), 9 deletions(-) > > > > > > diff --git a/drivers/pwm/pwm-lpss.c b/drivers/pwm/pwm-lpss.c > > > index 2504410..5a907db 100644 > > > --- a/drivers/pwm/pwm-lpss.c > > > +++ b/drivers/pwm/pwm-lpss.c > > > @@ -14,6 +14,7 @@ > > > */ > > > > > > #include <linux/io.h> > > > +#include <linux/time.h> > > > #include <linux/kernel.h> > > > #include <linux/module.h> > > > #include <linux/pm_runtime.h> > > > @@ -24,11 +25,9 @@ > > > #define PWM_ENABLE BIT(31) > > > #define PWM_SW_UPDATE BIT(30) > > > #define PWM_BASE_UNIT_SHIFT 8 > > > -#define PWM_BASE_UNIT_MASK 0x00ffff00 > > > +#define PWM_BASE_UNIT_MASK 0x3fffff00 > > > > Isn't this going to potentially write reserved bits on non-Broxton > > platforms? Previously the upper 8 bits were masked out, but now only the > > upper 2 bits are masked out. What about the other 6? > > > > Perhaps it'd be better to parameterize the mask in a way similar to the > > scaler value? > > Or call the field "base_unit_bits" which holds 16 for BSW/BYT and 22 for > BXT, and calculate both scaler and mask from that in pwm_lpss_config()? Yeah, that should work as well. > Actually I think we should make the driver private structure look like > this: > > struct pwm_lpss_chip { > struct pwm_chip chip; > void __iomem *regs; > const struct pwm_lpss_boardinfo *info; > }; > > and then set ->info to point to the platform data in probe(). > > pwm_lpss_config() can then refer lpwm->info->base_unit_bits and > lpwm->info->clk_rate. Yeah, that thought had crossed my mind as well. I would've probably waited until a third field was copied from info to chip before requesting that change, but since you already brought it up, feel free to send a patch. > > Also it's unclear to me how critical this is. Initial patches for > > Broxton support were merged into Linus' tree yesterday. Presumably they > > had received some testing before, but nobody can't have noticed or this > > bug wouldn't exist in the patches that were merged. Does this break any > > existing setups and hence should go into v4.4 along with the initial > > Broxton support? > > Nobody outside Intel should have Broxtons yet so it should not break any > existing users. Okay. Thierry
diff --git a/drivers/pwm/pwm-lpss.c b/drivers/pwm/pwm-lpss.c index 2504410..5a907db 100644 --- a/drivers/pwm/pwm-lpss.c +++ b/drivers/pwm/pwm-lpss.c @@ -14,6 +14,7 @@ */ #include <linux/io.h> +#include <linux/time.h> #include <linux/kernel.h> #include <linux/module.h> #include <linux/pm_runtime.h> @@ -24,11 +25,9 @@ #define PWM_ENABLE BIT(31) #define PWM_SW_UPDATE BIT(30) #define PWM_BASE_UNIT_SHIFT 8 -#define PWM_BASE_UNIT_MASK 0x00ffff00 +#define PWM_BASE_UNIT_MASK 0x3fffff00 #define PWM_ON_TIME_DIV_MASK 0x000000ff #define PWM_DIVISION_CORRECTION 0x2 -#define PWM_LIMIT (0x8000 + PWM_DIVISION_CORRECTION) -#define NSECS_PER_SEC 1000000000UL /* Size of each PWM register space if multiple */ #define PWM_SIZE 0x400 @@ -37,12 +36,14 @@ struct pwm_lpss_chip { struct pwm_chip chip; void __iomem *regs; unsigned long clk_rate; + unsigned long scaler; }; /* BayTrail */ const struct pwm_lpss_boardinfo pwm_lpss_byt_info = { .clk_rate = 25000000, .npwm = 1, + .scaler = 65536 }; EXPORT_SYMBOL_GPL(pwm_lpss_byt_info); @@ -50,6 +51,7 @@ EXPORT_SYMBOL_GPL(pwm_lpss_byt_info); const struct pwm_lpss_boardinfo pwm_lpss_bsw_info = { .clk_rate = 19200000, .npwm = 1, + .scaler = 65536 }; EXPORT_SYMBOL_GPL(pwm_lpss_bsw_info); @@ -57,6 +59,7 @@ EXPORT_SYMBOL_GPL(pwm_lpss_bsw_info); const struct pwm_lpss_boardinfo pwm_lpss_bxt_info = { .clk_rate = 19200000, .npwm = 4, + .scaler = 4194304 }; EXPORT_SYMBOL_GPL(pwm_lpss_bxt_info); @@ -85,13 +88,13 @@ static int pwm_lpss_config(struct pwm_chip *chip, struct pwm_device *pwm, struct pwm_lpss_chip *lpwm = to_lpwm(chip); u8 on_time_div; unsigned long c; - unsigned long long base_unit, freq = NSECS_PER_SEC; + unsigned long long base_unit, freq = NSEC_PER_SEC; u32 ctrl; do_div(freq, period_ns); - /* The equation is: base_unit = ((freq / c) * 65536) + correction */ - base_unit = freq * 65536; + /* The equation is: base_unit = ((freq / c) * scaler) + correction */ + base_unit = freq * lpwm->scaler; c = lpwm->clk_rate; if (!c) @@ -99,8 +102,6 @@ static int pwm_lpss_config(struct pwm_chip *chip, struct pwm_device *pwm, do_div(base_unit, c); base_unit += PWM_DIVISION_CORRECTION; - if (base_unit > PWM_LIMIT) - return -EINVAL; if (duty_ns <= 0) duty_ns = 1; @@ -110,7 +111,7 @@ static int pwm_lpss_config(struct pwm_chip *chip, struct pwm_device *pwm, ctrl = pwm_lpss_read(pwm); ctrl &= ~(PWM_BASE_UNIT_MASK | PWM_ON_TIME_DIV_MASK); - ctrl |= (u16) base_unit << PWM_BASE_UNIT_SHIFT; + ctrl |= (u32) base_unit << PWM_BASE_UNIT_SHIFT; ctrl |= on_time_div; /* request PWM to update on next cycle */ ctrl |= PWM_SW_UPDATE; @@ -157,6 +158,7 @@ struct pwm_lpss_chip *pwm_lpss_probe(struct device *dev, struct resource *r, return ERR_CAST(lpwm->regs); lpwm->clk_rate = info->clk_rate; + lpwm->scaler = info->scaler; lpwm->chip.dev = dev; lpwm->chip.ops = &pwm_lpss_ops; lpwm->chip.base = -1; diff --git a/drivers/pwm/pwm-lpss.h b/drivers/pwm/pwm-lpss.h index e8cf337..834be4c 100644 --- a/drivers/pwm/pwm-lpss.h +++ b/drivers/pwm/pwm-lpss.h @@ -21,6 +21,7 @@ struct pwm_lpss_chip; struct pwm_lpss_boardinfo { unsigned long clk_rate; unsigned int npwm; + unsigned long scaler; }; extern const struct pwm_lpss_boardinfo pwm_lpss_byt_info;
For Broxton PWM controller, base unit is defined as 8bit integer and 14bit fraction, so need to update base unit setting to output wave with right frequency. a) add scaler for each board setting; b) remove validity check of base unit for special board, let pwm user to handle this; Signed-off-by: Qipeng Zha <qipeng.zha@intel.com> --- drivers/pwm/pwm-lpss.c | 20 +++++++++++--------- drivers/pwm/pwm-lpss.h | 1 + 2 files changed, 12 insertions(+), 9 deletions(-)