Re: [alsa-devel] [PATCH v4 08/15] soundwire: Add Slave status handling helpers

[Pierre-Louis Bossart]

On 12/3/17 11:08 AM, Vinod Koul wrote:
> On Fri, Dec 01, 2017 at 05:36:47PM -0600, Pierre-Louis Bossart wrote:
>
> > > +/* called with bus_lock held */
> > > +static int sdw_get_device_num(struct sdw_slave *slave)
> > > +{
> > > +	int bit;
> > > +
> > > +	bit = find_first_zero_bit(slave->bus->assigned, SDW_MAX_DEVICES);
> > > +	if (bit == SDW_MAX_DEVICES) {
> > > +		bit = -ENODEV;
> > > +		goto err;
> >
> > My brain is starting to fry but is this correct? Bit11 seems like a valid
> > value. Should it be bit > 15 (assuming bit 12,13,14 are set to avoid using
> > groups and master)?
>
> this is correct. You are confusing SDW concept and API return types!
> That should be hint for you to start weekend if you didn't do so :D
>
> This API returns max value it was provided (last arg) if it doesn't
> find free bit. That's an indication to caller that we ran out of devices
> hence ENODEV error!

Can you just make sure bit11 is included?

> > > +static int sdw_program_device_num(struct sdw_bus *bus)
> > > +{
> > > +	u8 buf[SDW_NUM_DEV_ID_REGISTERS] = {0};
> > > +	struct sdw_slave *slave, *_s;
> > > +	struct sdw_slave_id id;
> > > +	struct sdw_msg msg;
> > > +	bool found = false;
> > > +	int count = 0, ret;
> > > +	u64 addr;
> > > +
> > > +	/* No Slave, so use raw xfer api */
> > > +	ret = sdw_fill_msg(&msg, NULL, SDW_SCP_DEVID_0,
> > > +			SDW_NUM_DEV_ID_REGISTERS, 0, SDW_MSG_FLAG_READ, buf);
> > > +	if (ret < 0)
> > > +		return ret;
> > > +
> > > +	do {
> > > +		ret = sdw_transfer(bus, NULL, &msg);
> > > +		if (ret == -ENODATA) { /* end of device id reads */
> > > +			ret = 0;
> > > +			break;
> > > +		}
> > > +		if (ret < 0) {
> > > +			dev_err(bus->dev, "DEVID read fail:%d\n", ret);
> > > +			break;
> > > +		}
> > > +
> > > +		/*
> > > +		 * Construct the addr and extract. Cast the higher shift
> > > +		 * bits to avoid truncation due to size limit.
> > > +		 */
> > > +		addr = buf[5] | (buf[4] << 8) | (buf[3] << 16) |
> > > +			(buf[2] << 24) | ((unsigned long long)buf[1] << 32) |
> > > +			((unsigned long long)buf[0] << 40);
> > > +
> > > +		sdw_extract_slave_id(bus, addr, &id);
> > > +
> > > +		/* Now compare with entries */
> > > +		list_for_each_entry_safe(slave, _s, &bus->slaves, node) {
> > > +			if (sdw_compare_devid(slave, id) == 0) {
> > > +				found = true;
> > > +
> > > +				/*
> > > +				 * Assign a new dev_num to this Slave and
> > > +				 * not mark it present. It will be marked
> > > +				 * present after it reports ATTACHED on new
> > > +				 * dev_num
> > > +				 */
> > > +				ret = sdw_assign_device_num(slave);
> > > +				if (ret) {
> > > +					dev_err(slave->bus->dev,
> > > +						"Assign dev_num failed:%d",
> > > +						ret);
> > > +					return ret;
> > > +				}
> > > +
> > > +				break;
> > > +			}
> > > +		}
> > > +
> > > +		if (found == false) {
> > > +			/* TODO: Park this device in Group 13 */
> > > +			dev_err(bus->dev, "Slave Entry not found");
> > > +		}
> > > +
> > > +		count++;
> > > +
> > > +	} while (ret == 0 && count < (SDW_MAX_DEVICES * 2));
> >
> > explain that the last condition is intentional - this is not a bug -, some
> > devices can drop off during enumeration and rejoin so might be counted
> > twice.
>
> ok will add