On Fri, 2023-10-20 at 11:32 -0500, Miles Glenn wrote:
> On Fri, 2023-10-20 at 11:42 +0200, Philippe Mathieu-Daudé wrote:
> > On 20/10/23 04:51, Andrew Jeffery wrote:
> > > On Thu, 2023-10-19 at 15:40 -0500, Glenn Miles wrote:
> > > > > The pca9552 INPUT0 and INPUT1 registers are supposed to
> > > > > hold the logical values of the LED pins. A logical 0
> > > > > should be seen in the INPUT0/1 registers for a pin when
> > > > > its corresponding LSn bits are set to 0, which is also
> > > > > the state needed for turning on an LED in a typical
> > > > > usage scenario. Existing code was doing the opposite
> > > > > and setting INPUT0/1 bit to a 1 when the LSn bit was
> > > > > set to 0, so this commit fixes that.
> > > > >
> > > > > Signed-off-by: Glenn Miles <mil...@linux.vnet.ibm.com>
> > > > > ---
> > > > >
> > > > > Changes from prior version:
> > > > > - Added comment regarding pca953X
> > > > > - Added cover letter
> > > > >
> > > > > hw/misc/pca9552.c | 18 +++++++++++++-----
> > > > > tests/qtest/pca9552-test.c | 6 +++---
> > > > > 2 files changed, 16 insertions(+), 8 deletions(-)
> > > > >
> > > > > diff --git a/hw/misc/pca9552.c b/hw/misc/pca9552.c
> > > > > index fff19e369a..445f56a9e8 100644
> > > > > --- a/hw/misc/pca9552.c
> > > > > +++ b/hw/misc/pca9552.c
> > > > > @@ -36,7 +36,10 @@ typedef struct PCA955xClass PCA955xClass;
> > > > >
> > > > > DECLARE_CLASS_CHECKERS(PCA955xClass, PCA955X,
> > > > > TYPE_PCA955X)
> > > > > -
> > > > > +/*
> > > > > + * Note: The LED_ON and LED_OFF configuration values for the
> > > > > PCA955X
> > > > > + * chips are the reverse of the PCA953X family of
> > > > > chips.
> > > > > + */
> > > > > #define PCA9552_LED_ON 0x0
> > > > > #define PCA9552_LED_OFF 0x1
> > > > > #define PCA9552_LED_PWM0 0x2
> > > > > @@ -112,13 +115,18 @@ static void
> > > > > pca955x_update_pin_input(PCA955xState *s)
> > > > >
> > > > > switch (config) {
> > > > > case PCA9552_LED_ON:
> > > > > - qemu_set_irq(s->gpio[i], 1);
> > > > > - s->regs[input_reg] |= 1 << input_shift;
> > > > > - break;
> > > > > - case PCA9552_LED_OFF:
> > > > > + /* Pin is set to 0V to turn on LED */
> > > > > qemu_set_irq(s->gpio[i], 0);
> > > > > s->regs[input_reg] &= ~(1 << input_shift);
> > > > > break;
> > > > > + case PCA9552_LED_OFF:
> > > > > + /*
> > > > > + * Pin is set to Hi-Z to turn off LED and
> > > > > + * pullup sets it to a logical 1.
> > > > > + */
> > > > > + qemu_set_irq(s->gpio[i], 1);
> > > > > + s->regs[input_reg] |= 1 << input_shift;
> > > > > + break;
> > >
> > > So the witherspoon-bmc machine was a user of the PCA9552 outputs as
> > > LEDs. I guess its LEDs were in the wrong state the whole time? That
> > > looks like the only user though, and shouldn't be negatively
> > > affected.
> >
> > Usually GPIO polarity is a machine/board property, not a device
> > one.
> >
> > We could use the LED API (hw/misc/led.h) which initialize each
> > output with GpioPolarity.
> >
>
> Thanks for your comments! This piqued my curiosity so I decided to
> run a test with the witherspoon-bmc machine. Without my changes, I ran
> the following command to turn off LED13 on the pca9552 which I had
> previously set to "on":
>
> qom-set /machine/unattached/device[18] led13 off
>
> I had GDB connected at the time with a breakpoint set on
> led_set_state() so that I could see what was happening. Due to the
> inversion bug, I expected the pca9552 code to set the pin low and also
> set the irq low, which it did. The connected LED's on this pca9552
> were all configured as GPIO_POLARITY_ACTIVE_LOW, so I expected that
> setting the irq state low would actually turn on the LED. Instead it
> turned off the LED.
>
> Reviewing the LED code, I believe the problem lies here:
>
> static void led_set_state_gpio_handler(void *opaque, int line, int
> new_state)
> {
> LEDState *s = LED(opaque);
>
> assert(line == 0);
> led_set_state(s, !!new_state != s->gpio_active_high);
> }
>
>
> In my test, new_state was 0 and gpio_active_high was 0, resulting in
> the boolean expression of ( 0 != 0 ) which is false and results in
> turning off the LED. So, this looks like a bug to me.
>
> For another simpler example, if the LED polarity was set to
> GPIO_POLARITY_ACTIVE_HIGH, then s->gpio_active_high would be 1. Then,
> if we set the irq line to 1, wouldn't we expect the LED to turn on?
> However, as the code stands, it would actually turn the LED off. So, I
> think we can remove one of the "!"'s from in front of new_state. Then,
> if the LED is active high and the irq line is set high, it would turn
> on the LED. Correct?
Seems reasonable to me. Looks like Philippe added the LED behaviour in
Fddb67f6402b8 ("hw/misc/led: Allow connecting from GPIO output"), so
his input would be helpful. Perhaps send a fix for review?
Andrew
