Re: mtd: remove driver-core BUS_ID_SIZE

[Atsushi Nemoto <anemo@mba.ocn.ne.jp>]

On Fri, 29 May 2009 14:33:41 +0100, David Woodhouse <dwmw2@infradead.org> wrote:
> > > This might work better. Nemoto-san, can you test and confirm?

Sure, I will test it on next week.

> > > diff --git a/drivers/mtd/nand/txx9ndfmc.c b/drivers/mtd/nand/txx9ndfmc.c
> > 
> > >                if (plat->ch_mask != 1) {
> > > +                       char *devname = dev_name(&dev->dev);
> > >                        txx9_priv->cs = i;
> > > +                       txx9_priv->mtdname = kmalloc(strlen(devname) + 3,
> > > +                                                    GFP_KERNEL);
> > 
> > kasprintf() might be nicer here?
> 
> True; thanks. (Er, how have I managed not to know that asprintf()
> exists, for the last 20-odd years?)
> 
> diff --git a/drivers/mtd/nand/txx9ndfmc.c b/drivers/mtd/nand/txx9ndfmc.c
...
>  		} else {
>  			txx9_priv->cs = -1;
> -			strcpy(txx9_priv->mtdname, dev_name(&dev->dev));
> +			txx9_priv->mtdname = dev_name(&dev->dev);

Use kstrdup() here, otherwise nand_scan() failure will cause memory
leak.

Or you can add conditional kfree into failure path of nand_scan(), but
I think saving few bytes is not worth to complicate an error (and
removal) path.

>  		}
>  		if (plat->wide_mask & (1 << i))
>  			chip->options |= NAND_BUSWIDTH_16;
> @@ -385,6 +391,8 @@ static int __exit txx9ndfmc_remove(struct platform_device *dev)
>  		kfree(drvdata->parts[i]);
>  #endif
>  		del_mtd_device(mtd);
> +		if (txx9_priv->mtdname != dev_name(&dev->dev))
> +			kfree(txx9_priv->mtdname);

If you did kstrdup() as above, you can just call kfree unconditionally.

---
Atsushi Nemoto