On Thu, Sep 3, 2020 at 3:53 AM Dmitry Osipenko <digetx@xxxxxxxxx> wrote:
>
> The tegra_i2c_flush_fifos() shouldn't sleep in atomic transfer and jiffies
> are not updating if interrupts are disabled. Hence let's use proper delay
> functions and use ktime API in order not to hang atomic transfer. Note
> that this patch doesn't fix any known problem because normally FIFO is
> flushed at the time of starting a new transfer.
> + /*
> + * ktime_get() may take up to couple milliseconds in a worst case
> + * and normally FIFOs are flushed, hence let's check the state before
> + * proceeding to polling.
> + */
Everything, including above can be done by using macros from iopoll.h. Why not?
> + if (!(i2c_readl(i2c_dev, offset) & mask))
> + return 0;
> +
> + ktime = ktime_get();
> + ktimeout = ktime_add_ms(ktime, 1000);
> +
> while (i2c_readl(i2c_dev, offset) & mask) {
> - if (time_after(jiffies, timeout)) {
> - dev_warn(i2c_dev->dev, "timeout waiting for fifo flush\n");
> - return -ETIMEDOUT;
> - }
> - usleep_range(1000, 2000);
> + if (ktime_after(ktime, ktimeout))
> + goto err_timeout;
> +
> + if (i2c_dev->is_curr_atomic_xfer)
> + mdelay(1);
> + else
> + fsleep(1000);
> +
> + ktime = ktime_get();
> }
> return 0;
> +
> +err_timeout:
> + dev_err(i2c_dev->dev, "FIFO flushing timed out\n");
> +
> + return -ETIMEDOUT;
> }
--
With Best Regards,
Andy Shevchenko
