Re: [PATCH 2/4] thermal: imx: Add support for reading OCOTP through nvmem

[Leonard Crestez]

On Fri, 2017-07-14 at 09:48 +0100, Srinivas Kandagatla wrote:
> On 12/07/17 07:36, Shawn Guo wrote:

> > > +static int nvmem_cell_read_u32(struct device* dev, const char *cell_id, u32 *val)
> > > +{
> > > +	struct nvmem_cell *cell;
> > > +	void *buf;
> > > +	size_t len;
> > > +
> > > +	cell = nvmem_cell_get(dev, cell_id);
> > > +	if (IS_ERR(cell))
> > > +		return PTR_ERR(cell);
> > > +
> > > +	buf = nvmem_cell_read(cell, &len);
> > > +	if (IS_ERR(buf)) {
> > > +		nvmem_cell_put(cell);
> > > +		return PTR_ERR(buf);
> > > +	}
> > > +	if (len != sizeof(*val)) {
> > > +		kfree(buf);
> > > +		nvmem_cell_put(cell);
> > > +		return -EINVAL;
> > > +	}
> > > +	memcpy(val, buf, sizeof(*val));

> This can overflow the memory allocated to val, we should be careful hereÂ
> not to do so.
> limit this to sizeof(u32) should be good. Also add some sanity checks toÂ
> make sure that len is atleast 4 bytes.

I'm not sure what you mean, isn't this already done? There is an
explicit check above that the read len is exactly as expected. It's
just that the limit is written as sizeof(*val) rather than sizeof(u32).

> > > +
> > > +	kfree(buf);
> > > +	nvmem_cell_put(cell);
> > > +	return 0;
> > > +}
> > The function looks nothing IMX specific, and could be a nvmem core
> > function?
> > 
> > @Srinivas, thoughts?
> Yep, this function looks generic, can be moved to nvmem layer.

Ok, next version of the series will have this function in nvmem core as
a separate commit.