Re: [PATCH v6 1/3] ubi: Introduce block devices for UBI volumes

[Ezequiel Garcia <ezequiel.garcia@free-electrons.com>]

On Mon, Feb 17, 2014 at 05:22:31PM +0200, Artem Bityutskiy wrote:
> On Sun, 2014-02-16 at 17:03 -0300, Ezequiel Garcia wrote:
> > +static int ubiblock_read(struct ubiblock *dev, char *buffer, int pos, int len)
> > +{
> > +	int leb, offset, ret;
> > +	int bytes_left = len;
> > +	int to_read = len;
> 
> "pos" sounds like byte offset, which cannot be true because 'int' would
> be too short type for it. 
> 

[..]
> > +
> > +	pos = blk_rq_pos(req) << 9;
> 
> So 'pos' is actually the 512-byte sector number? Would you please better
> name then, something self-documenting like 'sec' or 'sector' ?
> 

No, 'pos' is the byte offset. See the << 9, which is the custom
translation from 512-byte sector into byte offset.

I'm completely sure which type to use, size_t ? loff_t ? off_t ?

> > +
> > +	/*
> > +	 * Let's prevent the device from being removed while we're doing I/O
> > +	 * work. Notice that this means we serialize all the I/O operations,
> > +	 * but it's probably of no impact given the NAND core serializes
> > +	 * flash acceses anywafy.
> > +	 */
> > +	mutex_lock(&dev->vol_mutex);
> > +	ret = ubiblock_read(dev, req->buffer, pos, len);
> > +	mutex_unlock(&dev->vol_mutex);
> 
> Would you please a better name for 'vol_mutex'. Just makes me confused
> because this is what we use in UBI to lock the entire volume. And here
> it is different mutex. Let's express the code in clearer terms and use
> something like just 'device_lock' or something which would suggest that
> this is locks the entire ubiblock device.
> 

Sure, no problem.

Thanks for the review!
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com