Re: [PATCH v4 2/8] i2c: add I2C Address Translator (ATR) support

[Tomi Valkeinen]

Hi Andy,

On 01/11/2022 16:30, Andy Shevchenko wrote:
> On Tue, Nov 01, 2022 at 03:20:26PM +0200, Tomi Valkeinen wrote:
> > From: Luca Ceresoli <luca@xxxxxxxxxxxxxxxx>
> >
> > An ATR is a device that looks similar to an i2c-mux: it has an I2C
> > slave "upstream" port and N master "downstream" ports, and forwards
> > transactions from upstream to the appropriate downstream port. But is
> > is different in that the forwarded transaction has a different slave
> > address. The address used on the upstream bus is called the "alias"
> > and is (potentially) different from the physical slave address of the
> > downstream chip.
> >
> > Add a helper file (just like i2c-mux.c for a mux or switch) to allow
> > implementing ATR features in a device driver. The helper takes care or
> > adapter creation/destruction and translates addresses at each transaction.
>
> ...
>
> >      i2c-topology
> >      muxes/i2c-mux-gpio
> >      i2c-sysfs
> > +   muxes/i2c-atr
>
> Doesn't make sense to group muxes/*, that they are following each other?

Ok.

> ...
>
> > +I2C ADDRESS TRANSLATOR (ATR)
> > +M:	Luca Ceresoli <luca@xxxxxxxxxxxxxxxx>
>
> Hmm... Are you going to maintain this? Or Review? Why not?

We haven't discussed with Luca if he wants to maintain this (this is 
mostly his code). But, indeed, I should add my name there.

> > +L:	linux-i2c@xxxxxxxxxxxxxxx
> > +S:	Maintained
> > +F:	drivers/i2c/i2c-atr.c
> > +F:	include/linux/i2c-atr.h
>
> ...
>
> > +		void *new_buf = kmalloc_array(num, sizeof(chan->orig_addrs[0]),
> > +					      GFP_KERNEL);
> > +		if (new_buf == NULL)
> > +			return -ENOMEM;
>
> Isn't it better to write this as
>
> 		void *new_buf;
>
> 		new_buf = kmalloc_array(num, sizeof(chan->orig_addrs[0]), GFP_KERNEL);
> 		if (!new_buf)
> 			return -ENOMEM;

Ok.

> Remarks:
> - note the style of the conditional
> - why is it void?

No idea. I'll change it.

> Also, does it make sense to use krealloc_array() or is it complete replacement
> of the data?

The whole array will be rewritten, so we don't need to preserve the 
current data.

The buffer allocated here (i.e. orig_addrs) is only used for the 
duration of the i2c_atr_master_xfer(). So, we could allocate a new 
buffer for every xfer call, but to avoid that, we retain the old buffer. 
Any old data in the buffer can be discarded.

> > +		kfree(chan->orig_addrs);
> > +		chan->orig_addrs = new_buf;
> > +		chan->orig_addrs_size = num;
>
> ...
>
> > +static void i2c_atr_unmap_msgs(struct i2c_atr_chan *chan, struct i2c_msg msgs[],
> > +			       int num)
>
> [] in the function parameter is longer than * and actually doesn't make
> difference in C. Ditto for the rest of similar cases.

Ok. I missed a few, it seems.

> ...
>
> > +static int i2c_atr_smbus_xfer(struct i2c_adapter *adap, u16 addr,
> > +			      unsigned short flags, char read_write, u8 command,
> > +			      int size, union i2c_smbus_data *data)
>
> Can flags be fixed size (yes I understand that in our case short would probably
> never be different to u16, but for the sake of clearness)?

The parameters and their types come from the ops in struct i2c_algorithm.

> ...
>
> > +static int i2c_atr_attach_client(struct i2c_adapter *adapter,
> > +				 const struct i2c_board_info *info,
> > +				 const struct i2c_client *client)
> > +{
> > +	struct i2c_atr_chan *chan = adapter->algo_data;
> > +	struct i2c_atr *atr = chan->atr;
> > +	struct i2c_atr_cli2alias_pair *c2a;
>
> > +	u16 alias_id = 0;
>
> Can we split assignment from the definition and locate it closer to the first
> use?

Actually, I don't think we need to initialize it at all. If 
attach_client() fails, we don't care about alias_id. If attach_client() 
succeeds, it _must_ return alias_id.

> > +	int ret = 0;
>
> Useless assignment.

Yep.

> > +
> > +	c2a = kzalloc(sizeof(*c2a), GFP_KERNEL);
> > +	if (!c2a)
> > +		return -ENOMEM;
> > +
> > +	ret = atr->ops->attach_client(atr, chan->chan_id, info, client,
> > +				      &alias_id);
>
> On one line looks better.

I agree, but it doesn't fit into 80 characters. I personally think 
that's a too narrow a limit, but some maintainers absolutely require max 
80 chars, so I try to limit the lines to 80 unless it looks really ugly.

> > +	if (ret)
> > +		goto err_free;
> > +	if (alias_id == 0) {
> > +		ret = -EINVAL;
> > +		goto err_free;
> > +	}
> > +
> > +	c2a->client = client;
> > +	c2a->alias = alias_id;
> > +	list_add(&c2a->node, &chan->alias_list);
> > +
> > +	return 0;
> > +
> > +err_free:
> > +	kfree(c2a);
> > +	return ret;
> > +}
>
> ...
>
> > +int i2c_atr_add_adapter(struct i2c_atr *atr, u32 chan_id,
> > +			struct fwnode_handle *bus_handle)
> > +{
> > +	struct i2c_adapter *parent = atr->parent;
> > +	struct device *dev = atr->dev;
> > +	struct i2c_atr_chan *chan;
> > +	char *symlink_name;
> > +	int ret;
> > +
> > +	if (chan_id >= atr->max_adapters) {
> > +		dev_err(dev, "No room for more i2c-atr adapters\n");
> > +		return -EINVAL;
> > +	}
> > +
> > +	if (atr->adapter[chan_id]) {
> > +		dev_err(dev, "Adapter %d already present\n", chan_id);
> > +		return -EEXIST;
> > +	}
> > +
> > +	chan = kzalloc(sizeof(*chan), GFP_KERNEL);
> > +	if (!chan)
> > +		return -ENOMEM;
> > +
> > +	chan->atr = atr;
> > +	chan->chan_id = chan_id;
> > +	INIT_LIST_HEAD(&chan->alias_list);
> > +	mutex_init(&chan->orig_addrs_lock);
> > +
> > +	snprintf(chan->adap.name, sizeof(chan->adap.name), "i2c-%d-atr-%d",
> > +		 i2c_adapter_id(parent), chan_id);
> > +	chan->adap.owner = THIS_MODULE;
> > +	chan->adap.algo = &atr->algo;
> > +	chan->adap.algo_data = chan;
> > +	chan->adap.dev.parent = dev;
> > +	chan->adap.retries = parent->retries;
> > +	chan->adap.timeout = parent->timeout;
> > +	chan->adap.quirks = parent->quirks;
> > +	chan->adap.lock_ops = &i2c_atr_lock_ops;
> > +	chan->adap.attach_ops = &i2c_atr_attach_ops;
> > +
> > +	if (bus_handle) {
> > +		device_set_node(&chan->adap.dev, fwnode_handle_get(bus_handle));
> > +	} else {
> > +		struct fwnode_handle *atr_node;
> > +		struct fwnode_handle *child;
> > +		u32 reg;
> > +
> > +		atr_node = device_get_named_child_node(dev, "i2c-atr");
> > +
> > +		fwnode_for_each_child_node(atr_node, child) {
> > +			ret = fwnode_property_read_u32(child, "reg", &reg);
> > +			if (ret)
> > +				continue;
> > +			if (chan_id == reg)
> > +				break;
> > +		}
> > +
> > +		device_set_node(&chan->adap.dev, child);
> > +		fwnode_handle_put(atr_node);
> > +	}
>
> It seems you have OF independent code, but by some reason you included of.h
> instead of property.h. Am I right?

Just an leftover from the conversion from of to fwnode.

> > +	ret = i2c_add_adapter(&chan->adap);
> > +	if (ret) {
> > +		dev_err(dev, "failed to add atr-adapter %u (error=%d)\n",
> > +			chan_id, ret);
> > +		goto err_add_adapter;
> > +	}
> > +
> > +	symlink_name = kasprintf(GFP_KERNEL, "channel-%u", chan_id);
>
> No NULL check?

Right, missed that.

> > +	WARN(sysfs_create_link(&chan->adap.dev.kobj, &dev->kobj, "atr_device"),
> > +	     "can't create symlink to atr device\n");
> > +	WARN(sysfs_create_link(&dev->kobj, &chan->adap.dev.kobj, symlink_name),
> > +	     "can't create symlink for channel %u\n", chan_id);
>
> Why WARNs? sysfs has already some in their implementation.

True, and I can drop these if required. But afaics, sysfs_create_link 
only warns if there's a duplicate entry, not for other errors.

> > +
> > +	kfree(symlink_name);
> > +
> > +	dev_dbg(dev, "Added ATR child bus %d\n", i2c_adapter_id(&chan->adap));
> > +
> > +	atr->adapter[chan_id] = &chan->adap;
> > +	return 0;
> > +
> > +err_add_adapter:
> > +	mutex_destroy(&chan->orig_addrs_lock);
> > +	kfree(chan);
> > +	return ret;
> > +}
>
> ...
>
> > +	struct fwnode_handle *fwnode = adap->dev.fwnode;
>
> Please don't dereference fwnode like this, we have dev_fwnode() for that.

Ok.

> ...
>
> > +	if (atr->adapter[chan_id] == NULL) {
>
> !

Yep.

> > +		dev_err(dev, "Adapter %d does not exist\n", chan_id);
> > +		return;
> > +	}
>
> ...
>
> > +	snprintf(symlink_name, sizeof(symlink_name),
> > +		 "channel-%u", chan->chan_id);
>
> Once line?

80 char limit here too. But I see that this is (kind of) broken. In the 
i2c_atr_add_adapter() I used dynamic alloc for the symlink_name, but 
here we still have the fixed size buffer.

> ...
>
> > +	atr_size = struct_size(atr, adapter, max_adapters);
>
> > +	if (atr_size == SIZE_MAX)
> > +		return ERR_PTR(-EOVERFLOW);
>
> Dunno if you really need this to be separated from devm_kzalloc(), either way
> you will get an error, but in embedded case it will be -ENOMEM.

Yep. Well... I kind of like it to be explicit. Calling alloc(SIZE_MAX) 
doesn't feel nice.

> > +	atr = devm_kzalloc(dev, atr_size, GFP_KERNEL);
> > +	if (!atr)
> > +		return ERR_PTR(-ENOMEM);
>
> ...
>
> > +EXPORT_SYMBOL_GPL(i2c_atr_delete);
>
> I would put these to their own namespace from day 1.

What would be the namespace? Isn't this something that should be 
subsystem-wide decision? I have to admit I have never used symbol 
namespaces, and don't know much about them.

> ...
>
> > +/**
> > + * Helper to add I2C ATR features to a device driver.
> > + */
>
> ??? Copy'n'paste typo?

No idea where that is from... I'll fix it.

> > +struct i2c_atr {
> > +	/* private: internal use only */
> > +
> > +	struct i2c_adapter *parent;
> > +	struct device *dev;
> > +	const struct i2c_atr_ops *ops;
> > +
> > +	void *priv;
> > +
> > +	struct i2c_algorithm algo;
> > +	struct mutex lock;
> > +	int max_adapters;
> > +
> > +	struct i2c_adapter *adapter[0];
>
> No VLAs.

Ok.

I'm not arguing against any of the comments you've made, I think they 
are all valid, but I want to point out that many of them are in a code 
copied from i2c-mux.

Whether there's any value in keeping i2c-mux and i2c-atr similar in 
design/style... Maybe not.

> > +};
>
> ...
>
> > +int i2c_atr_add_adapter(struct i2c_atr *atr, u32 chan_id,
> > +			struct fwnode_handle *bus_np);
>
> Missing
>
> struct fwnode_handle;
>
> at the top of the file?

Ok.

 Tomi