| Message ID | 20230529-pm8941-pwrkey-debounce-v1-2-c043a6d5c814@linaro.org (mailing list archive) |
|---|---|
| State | Not Applicable |
| Headers | show |
| Series | Fix pm8941-pwrkey debounce programming | expand |
On 29.05.2023 21:55, Caleb Connolly wrote: > Since PM8998/PM660, the power key debounce register was redefined to > support shorter debounce times. On PM8941 the shortest debounce time > (represented by register value 0) was 15625us, on PM8998 the shortest > debounce time is 62us, with the default being 2ms. > > Adjust the bit shift to correctly program debounce on PM8998 and newer. > > Fixes: 68c581d5e7d8 ("Input: add Qualcomm PM8941 power key driver") > Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org> > --- > This patch shouldn't be backported earlier then 5.4, as that is the > first kernel with support for PM8998. > --- > drivers/input/misc/pm8941-pwrkey.c | 19 +++++++++++++++---- > 1 file changed, 15 insertions(+), 4 deletions(-) > > diff --git a/drivers/input/misc/pm8941-pwrkey.c b/drivers/input/misc/pm8941-pwrkey.c > index b6a27ebae977..74d77d8aaeff 100644 > --- a/drivers/input/misc/pm8941-pwrkey.c > +++ b/drivers/input/misc/pm8941-pwrkey.c > @@ -50,7 +50,10 @@ > #define PON_RESIN_PULL_UP BIT(0) > > #define PON_DBC_CTL 0x71 > -#define PON_DBC_DELAY_MASK 0x7 > +#define PON_DBC_DELAY_MASK_GEN1 0x7 > +#define PON_DBC_DELAY_MASK_GEN2 0xf > +#define PON_DBC_SHIFT_GEN1 6 > +#define PON_DBC_SHIFT_GEN2 14 mask+shift -> field_prep/get? Nice find! Konrad > > struct pm8941_data { > unsigned int pull_up_bit; > @@ -247,7 +250,7 @@ static int pm8941_pwrkey_probe(struct platform_device *pdev) > struct device *parent; > struct device_node *regmap_node; > const __be32 *addr; > - u32 req_delay; > + u32 req_delay, mask, delay_shift; > int error; > > if (of_property_read_u32(pdev->dev.of_node, "debounce", &req_delay)) > @@ -336,12 +339,20 @@ static int pm8941_pwrkey_probe(struct platform_device *pdev) > pwrkey->input->phys = pwrkey->data->phys; > > if (pwrkey->data->supports_debounce_config) { > - req_delay = (req_delay << 6) / USEC_PER_SEC; > + if (pwrkey->subtype >= PON_SUBTYPE_GEN2_PRIMARY) { > + mask = PON_DBC_DELAY_MASK_GEN2; > + delay_shift = PON_DBC_SHIFT_GEN2; > + } else { > + mask = PON_DBC_DELAY_MASK_GEN1; > + delay_shift = PON_DBC_SHIFT_GEN1; > + } > + > + req_delay = (req_delay << delay_shift) / USEC_PER_SEC; > req_delay = ilog2(req_delay); > > error = regmap_update_bits(pwrkey->regmap, > pwrkey->baseaddr + PON_DBC_CTL, > - PON_DBC_DELAY_MASK, > + mask, > req_delay); > if (error) { > dev_err(&pdev->dev, "failed to set debounce: %d\n", >
On 30/05/2023 10:36, Konrad Dybcio wrote: > > > On 29.05.2023 21:55, Caleb Connolly wrote: >> Since PM8998/PM660, the power key debounce register was redefined to >> support shorter debounce times. On PM8941 the shortest debounce time >> (represented by register value 0) was 15625us, on PM8998 the shortest >> debounce time is 62us, with the default being 2ms. >> >> Adjust the bit shift to correctly program debounce on PM8998 and newer. >> >> Fixes: 68c581d5e7d8 ("Input: add Qualcomm PM8941 power key driver") >> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org> >> --- >> This patch shouldn't be backported earlier then 5.4, as that is the >> first kernel with support for PM8998. >> --- >> drivers/input/misc/pm8941-pwrkey.c | 19 +++++++++++++++---- >> 1 file changed, 15 insertions(+), 4 deletions(-) >> >> diff --git a/drivers/input/misc/pm8941-pwrkey.c b/drivers/input/misc/pm8941-pwrkey.c >> index b6a27ebae977..74d77d8aaeff 100644 >> --- a/drivers/input/misc/pm8941-pwrkey.c >> +++ b/drivers/input/misc/pm8941-pwrkey.c >> @@ -50,7 +50,10 @@ >> #define PON_RESIN_PULL_UP BIT(0) >> >> #define PON_DBC_CTL 0x71 >> -#define PON_DBC_DELAY_MASK 0x7 >> +#define PON_DBC_DELAY_MASK_GEN1 0x7 >> +#define PON_DBC_DELAY_MASK_GEN2 0xf >> +#define PON_DBC_SHIFT_GEN1 6 >> +#define PON_DBC_SHIFT_GEN2 14 > mask+shift -> field_prep/get? I figured it was better to keep it consistent and try to minimise the diff so that backporting is easier. Migrating over to bitfield helpers probably makes sense to do as a separate change - maybe if a new platform comes along and requires even more additional complexity? > > Nice find! > > Konrad >> >> struct pm8941_data { >> unsigned int pull_up_bit; >> @@ -247,7 +250,7 @@ static int pm8941_pwrkey_probe(struct platform_device *pdev) >> struct device *parent; >> struct device_node *regmap_node; >> const __be32 *addr; >> - u32 req_delay; >> + u32 req_delay, mask, delay_shift; >> int error; >> >> if (of_property_read_u32(pdev->dev.of_node, "debounce", &req_delay)) >> @@ -336,12 +339,20 @@ static int pm8941_pwrkey_probe(struct platform_device *pdev) >> pwrkey->input->phys = pwrkey->data->phys; >> >> if (pwrkey->data->supports_debounce_config) { >> - req_delay = (req_delay << 6) / USEC_PER_SEC; >> + if (pwrkey->subtype >= PON_SUBTYPE_GEN2_PRIMARY) { >> + mask = PON_DBC_DELAY_MASK_GEN2; >> + delay_shift = PON_DBC_SHIFT_GEN2; >> + } else { >> + mask = PON_DBC_DELAY_MASK_GEN1; >> + delay_shift = PON_DBC_SHIFT_GEN1; >> + } >> + >> + req_delay = (req_delay << delay_shift) / USEC_PER_SEC; >> req_delay = ilog2(req_delay); >> >> error = regmap_update_bits(pwrkey->regmap, >> pwrkey->baseaddr + PON_DBC_CTL, >> - PON_DBC_DELAY_MASK, >> + mask, >> req_delay); >> if (error) { >> dev_err(&pdev->dev, "failed to set debounce: %d\n", >>
On Mon, May 29, 2023 at 08:55:07PM +0100, Caleb Connolly wrote: > Since PM8998/PM660, the power key debounce register was redefined to > support shorter debounce times. On PM8941 the shortest debounce time > (represented by register value 0) was 15625us, on PM8998 the shortest > debounce time is 62us, with the default being 2ms. > > Adjust the bit shift to correctly program debounce on PM8998 and newer. > > Fixes: 68c581d5e7d8 ("Input: add Qualcomm PM8941 power key driver") > Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org> Applied, thank you.
diff --git a/drivers/input/misc/pm8941-pwrkey.c b/drivers/input/misc/pm8941-pwrkey.c index b6a27ebae977..74d77d8aaeff 100644 --- a/drivers/input/misc/pm8941-pwrkey.c +++ b/drivers/input/misc/pm8941-pwrkey.c @@ -50,7 +50,10 @@ #define PON_RESIN_PULL_UP BIT(0) #define PON_DBC_CTL 0x71 -#define PON_DBC_DELAY_MASK 0x7 +#define PON_DBC_DELAY_MASK_GEN1 0x7 +#define PON_DBC_DELAY_MASK_GEN2 0xf +#define PON_DBC_SHIFT_GEN1 6 +#define PON_DBC_SHIFT_GEN2 14 struct pm8941_data { unsigned int pull_up_bit; @@ -247,7 +250,7 @@ static int pm8941_pwrkey_probe(struct platform_device *pdev) struct device *parent; struct device_node *regmap_node; const __be32 *addr; - u32 req_delay; + u32 req_delay, mask, delay_shift; int error; if (of_property_read_u32(pdev->dev.of_node, "debounce", &req_delay)) @@ -336,12 +339,20 @@ static int pm8941_pwrkey_probe(struct platform_device *pdev) pwrkey->input->phys = pwrkey->data->phys; if (pwrkey->data->supports_debounce_config) { - req_delay = (req_delay << 6) / USEC_PER_SEC; + if (pwrkey->subtype >= PON_SUBTYPE_GEN2_PRIMARY) { + mask = PON_DBC_DELAY_MASK_GEN2; + delay_shift = PON_DBC_SHIFT_GEN2; + } else { + mask = PON_DBC_DELAY_MASK_GEN1; + delay_shift = PON_DBC_SHIFT_GEN1; + } + + req_delay = (req_delay << delay_shift) / USEC_PER_SEC; req_delay = ilog2(req_delay); error = regmap_update_bits(pwrkey->regmap, pwrkey->baseaddr + PON_DBC_CTL, - PON_DBC_DELAY_MASK, + mask, req_delay); if (error) { dev_err(&pdev->dev, "failed to set debounce: %d\n",
Since PM8998/PM660, the power key debounce register was redefined to support shorter debounce times. On PM8941 the shortest debounce time (represented by register value 0) was 15625us, on PM8998 the shortest debounce time is 62us, with the default being 2ms. Adjust the bit shift to correctly program debounce on PM8998 and newer. Fixes: 68c581d5e7d8 ("Input: add Qualcomm PM8941 power key driver") Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org> --- This patch shouldn't be backported earlier then 5.4, as that is the first kernel with support for PM8998. --- drivers/input/misc/pm8941-pwrkey.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-)