On Thu, Oct 04, 2012 at 03:02:26PM +0530, Viresh Kumar wrote: > On 4 October 2012 14:50, Uwe Kleine-König > <u.kleine-koenig@xxxxxxxxxxxxxx> wrote: > > >> So, we actually need to do "Low-High" 9 times instead of "High-Low". > >> So, initializing val to 0 should fix it? > > I don't think this is enough. If you cut off the last half clock of the > > first sequence above doing 9 times low-high isn't enough. So you have to > > do high + 9x low-high to assert 9 full cycles. > > I am not cutting the last half clock. val is the variable which keeps > track of value to be > set on the line. I am asking to start from zero. I meant the sequence that created the stall, not the one intending to clear it. If you remove the last rising edge from that the SCL line is initially low. > You mean, we should do a high first and then 9 low-high? But this line was high > initially.. what difference will it make by making it high again? > > Sorry, i am not a I2C expert, so need some help :) :( > >> >> + * @scl_gpio_flags: flag for gpio_request_one of scl_gpio. 0 implies > >> >> + * GPIOF_OUT_INIT_LOW. > >> >> + * @sda_gpio_flags: flag for gpio_request_one of sda_gpio. 0 implies > >> >> + * GPIOF_OUT_INIT_LOW. > >> > >> > Didn't you want to change this to GPIOF_OUT_INIT_HIGH | > >> > GPIOF_OPEN_DRAIN? Hmm, I just wonder how to distinguish > >> > GPIOF_OUT_INIT_LOW from unset. Hmm, maybe assume unset because _LOW is > >> > wrong? > >> > >> Hmmm.... Hmmmm... Hmmm... :) > >> Two things: > >> - Why should we default it to GPIOF_OPEN_DRAIN? Open drain would mean, > >> gpio_set_value() will not write one for it, but would put it in input mode. > >> I don't think that would be correct, as an generic case. > > As the i2c-bus is open drain and multi-master capable it's wrong to pull > > it to 1 because this can result in a short circuit. Even in a > > single-master scenario (where you can pull SCL in both directions) you > > must not drive SDA to 1. > > Hmm... Hopefully i understood it well this time. > - We actually don't need these flags then. > > Always pass: > - scl_flags: GPIOF_OPEN_DRAIN | GPIOF_OUT_INIT_HIGH > - sda_flags: GPIOF_IN > > Why would anybody else require something different for any SoC? IMHO this should work, yes. (At least conceptually, not necessary for all implementations.) Best regards Uwe -- Pengutronix e.K. | Uwe Kleine-König | Industrial Linux Solutions | http://www.pengutronix.de/ | -- To unsubscribe from this list: send the line "unsubscribe linux-i2c" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
