| Message ID | 20231006094033.3082468-1-andriy.shevchenko@linux.intel.com |
|---|---|
| State | Accepted |
| Commit | d3386552155c6e9128e49fff86d7acfe4b13f949 |
| Headers | show |
| Series | [v1,1/2] pinctrl: cherryview: Avoid duplicated I/O | expand |
On Fri, Oct 06, 2023 at 12:40:33PM +0300, Andy Shevchenko wrote: > @@ -1119,28 +1099,26 @@ static struct pinctrl_desc chv_pinctrl_desc = { > static int chv_gpio_get(struct gpio_chip *chip, unsigned int offset) > { > struct intel_pinctrl *pctrl = gpiochip_get_data(chip); > - unsigned long flags; > u32 ctrl0, cfg; > > - raw_spin_lock_irqsave(&chv_lock, flags); > - ctrl0 = chv_readl(pctrl, offset, CHV_PADCTRL0); > - raw_spin_unlock_irqrestore(&chv_lock, flags); > + scoped_guard(raw_spinlock_irqsave, &chv_lock) > + ctrl0 = chv_readl(pctrl, offset, CHV_PADCTRL0); > > cfg = ctrl0 & CHV_PADCTRL0_GPIOCFG_MASK; > cfg >>= CHV_PADCTRL0_GPIOCFG_SHIFT; > > if (cfg == CHV_PADCTRL0_GPIOCFG_GPO) > return !!(ctrl0 & CHV_PADCTRL0_GPIOTXSTATE); > + Unrelated whitespace change. Otherwise looks good, Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
On Fri, Oct 06, 2023 at 01:26:41PM +0300, Mika Westerberg wrote: > On Fri, Oct 06, 2023 at 12:40:33PM +0300, Andy Shevchenko wrote: ... > > if (cfg == CHV_PADCTRL0_GPIOCFG_GPO) > > return !!(ctrl0 & CHV_PADCTRL0_GPIOTXSTATE); > > + > > Unrelated whitespace change. Will drop it. > Otherwise looks good, > Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Thank you!
On Fri, Oct 06, 2023 at 01:24:12PM +0300, Mika Westerberg wrote: > On Fri, Oct 06, 2023 at 12:40:32PM +0300, Andy Shevchenko wrote: ... > Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Pushed to my review and testing queue, thanks!
diff --git a/drivers/pinctrl/intel/pinctrl-cherryview.c b/drivers/pinctrl/intel/pinctrl-cherryview.c index f047f7bf4afb..9b9f18f50c1d 100644 --- a/drivers/pinctrl/intel/pinctrl-cherryview.c +++ b/drivers/pinctrl/intel/pinctrl-cherryview.c @@ -612,9 +612,14 @@ static void chv_writel(struct intel_pinctrl *pctrl, unsigned int pin, unsigned i } /* When Pad Cfg is locked, driver can only change GPIOTXState or GPIORXState */ +static bool chv_pad_is_locked(u32 ctrl1) +{ + return ctrl1 & CHV_PADCTRL1_CFGLOCK; +} + static bool chv_pad_locked(struct intel_pinctrl *pctrl, unsigned int offset) { - return chv_readl(pctrl, offset, CHV_PADCTRL1) & CHV_PADCTRL1_CFGLOCK; + return chv_pad_is_locked(chv_readl(pctrl, offset, CHV_PADCTRL1)); } static void chv_pin_dbg_show(struct pinctrl_dev *pctldev, struct seq_file *s, @@ -623,13 +628,11 @@ static void chv_pin_dbg_show(struct pinctrl_dev *pctldev, struct seq_file *s, struct intel_pinctrl *pctrl = pinctrl_dev_get_drvdata(pctldev); unsigned long flags; u32 ctrl0, ctrl1; - bool locked; raw_spin_lock_irqsave(&chv_lock, flags); ctrl0 = chv_readl(pctrl, offset, CHV_PADCTRL0); ctrl1 = chv_readl(pctrl, offset, CHV_PADCTRL1); - locked = chv_pad_locked(pctrl, offset); raw_spin_unlock_irqrestore(&chv_lock, flags); @@ -646,7 +649,7 @@ static void chv_pin_dbg_show(struct pinctrl_dev *pctldev, struct seq_file *s, seq_printf(s, "0x%08x 0x%08x", ctrl0, ctrl1); - if (locked) + if (chv_pad_is_locked(ctrl1)) seq_puts(s, " [LOCKED]"); }
In some cases we already read the value from the register followed by a reading of it again for other purposes, but the both reads are under the lock and bits we are insterested in are not going to change (they are not volatile from HW perspective). Hence, no need to read the same registeer twice. Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> --- drivers/pinctrl/intel/pinctrl-cherryview.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)