On Sun, Aug 09, 2020 at 09:25:16PM +0800, Kent Gibson wrote:
> Add support for requesting lines using the GPIO_V2_GET_LINE_IOCTL, and
> returning their current values using GPIO_V2_LINE_GET_VALUES_IOCTL.
>
> Signed-off-by: Kent Gibson <warthog...@gmail.com>
> ---
>
[snip]
> +static long line_get_values(struct line *line, void __user *ip)
> +{
> + struct gpio_v2_line_values lv;
> + DECLARE_BITMAP(vals, GPIO_V2_LINES_MAX);
> + DECLARE_BITMAP(mask, GPIO_V2_LINES_MAX);
> + struct gpio_desc **descs;
> + int ret, i, didx, num_get = 0;
> +
> + /* NOTE: It's ok to read values of output lines. */
> + if (copy_from_user(&lv, ip, sizeof(lv)))
> + return -EFAULT;
> +
> + if (lv.mask == 0)
> + return -EINVAL;
> +
> + bitmap_from_u64(mask, lv.mask);
> + num_get = bitmap_weight(mask, line->num_descs);
Nuts, that doesn't do what I intend, and is part of a failed experiment
that I thought I had reverted.
Assume that in the next version the mask handling will be similar to
line_set_values_locked in patch 09.
i.e.
for (i = 0; i < line->num_descs; i++)
if (lv.mask & BIT_ULL(i))
num_get++;
and the subsequent mask tests will be against lv.mask instead of the
mask bitmap declared above (which will be gone).
I won't do a v4 just yet though - there is sure to be other things
to go in there.
Cheers,
Kent.
