[PATCH v4 02/11] drm/bridge: it6505: fix aux operation for edid read

[AngeloGioacchino Del Regno]

Il 26/09/24 09:47, Hermes Wu ha scritto:
> From: Hermes Wu <Hermes.wu at ite.com.tw>
> 
> The EDID read operation can reach the maximum size of the AUX FIFO buffer.
> 
> Fixes: b5c84a9edcd4 ("drm/bridge: add it6505 driver")
> Signed-off-by: Hermes Wu <Hermes.wu at ite.com.tw>
> ---
>   drivers/gpu/drm/bridge/ite-it6505.c | 8 ++++++--
>   1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/bridge/ite-it6505.c b/drivers/gpu/drm/bridge/ite-it6505.c
> index 28a8043229d3..b451d3c2ac1d 100644
> --- a/drivers/gpu/drm/bridge/ite-it6505.c
> +++ b/drivers/gpu/drm/bridge/ite-it6505.c
> @@ -1078,8 +1078,11 @@ static ssize_t it6505_aux_do_transfer(struct it6505 *it6505,
>   	int i, ret_size, ret = 0, request_size;

int fifo_max_sz = (cmd == CMD_AUX_I2C_EDID_READ) ?
		  AUX_FIFO_MAX_SIZE : 4;

which later becomes

int fifo_max_sz = (cmd == CMD_AUX_I2C_EDID_READ || cmd == CMD_AUX_GET_KSV_LIST) ?
		  AUX_FIFO_MAX_SIZE : 4;

otherwise you can do it "the long way"...

if (cmd == .....)
	fifo_max_sz = AUX_FIFO_MAX_SIZE;
else
	fifo_max_sz = 4;

The point is, cmd won't change during iterations, so it's useless to put that if
branch inside of that loop below.

>   
>   	mutex_lock(&it6505->aux_lock);
> -	for (i = 0; i < size; i += 4) {
> -		request_size = min((int)size - i, 4);

		request_size = min_t(int, (int)size - i, fifo_max_sz);

P.S.: Also, I'd consider changing this to a do-while instead...

Cheers,
Angelo

> +	for (i = 0; i < size; ) {
> +		if (cmd == CMD_AUX_I2C_EDID_READ)
> +			request_size = min_t(int, (int)size - i, AUX_FIFO_MAX_SIZE);
> +		else
> +			request_size = min_t(int, (int)size - i, 4);
>   		ret_size = it6505_aux_operation(it6505, cmd, address + i,
>   						buffer + i, request_size,
>   						reply);
> @@ -1088,6 +1091,7 @@ static ssize_t it6505_aux_do_transfer(struct it6505 *it6505,
>   			goto aux_op_err;
>   		}
>   
> +		i += request_size;
>   		ret += ret_size;
>   	}
>