[PATCH] OneNAND MTD support

Thomas Gleixner tglx at linutronix.de
Mon Jul 11 03:47:21 EDT 2005

On Mon, 2005-07-11 at 10:53 +0900, Kyungmin Park wrote:
> It's derived from omap-nand-flash.c, But it's not sended to MTD.
> OK. I done

I removed the redundant dependencies in drivers/mtd/nand and
drivers/mtd/maps too.

> > 
> > +EXPORT_SYMBOL(onenand_scan);
> > +EXPORT_SYMBOL(onenand_release);
> > 
> > Would you mind to change those exports to EXPORT_SYMBOL_GPL ?
> In fact I not sure and also don't know license issue. 
> If we develop the some commercial mtd user software then can we use mtd
> device code exported with GPL? Is there no problem?

User space software is not concerned of kernel licences.

If you talk about a proprietary closed source kernel module which needs
those exports, you are stepping into a legal grey area regardless of the
exports style. I don't see a point where it makes sense to have a closed
source board driver, as it just initializes a data structure and calls
generic handling code. 

> > Are device_id and version_id necessary for anything else than 
> > scan ? If
> > no, please remove.
> device_id necessary to control DDP chip. But I'm not yet tested with DDP
> chip.


Some nitpicks though

+	/* The 10 msec is enough */
+	timeout = jiffies + msecs_to_jiffies(10);
+	while (time_before(jiffies, timeout)) {

On CPU's which have HZ=100 (1 jiffy == 10ms) this might be critical, as
you might set the timeout immidiately before the timer interrupt
happens. Using a 20ms timeout will keep you on the safe side in any case

Please clean up your includes to avoid redundant includes of header


More information about the linux-mtd mailing list