Re: [PATCH] mtd: nand: atmel: fix of_irq_get() error check

[Boris Brezillon <boris.brezillon@free-electrons.com>]

Le Thu, 10 Aug 2017 13:12:09 +0300,
Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> a écrit :

> On 8/10/2017 12:59 PM, Boris Brezillon wrote:
> 
> >> of_irq_get() may return 0 as well as negative error number on failure,
> >> while the driver only checks for the negative values.  The driver would
> >> then call devm_request_irq() for IRQ0  in its probe method and never get
> >> a valid interrupt.
> >>
> >> Check for 'nc->irq <= 0' instead and return -ENXIO from the driver's probe
> >> iff of_irq_get() returned 0.
> >>
> >> Fixes: f88fc122cc34 ("mtd: nand: Cleanup/rework the atmel_nand driver")
> >> Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
> >>
> >> ---
> >> The patch is against the 'master' branch of the 'linux-mtd.git' repo --
> >> the 'nand/fixes'  branch seems too old...
> >>
> >>   drivers/mtd/nand/atmel/nand-controller.c |   12 ++++++------
> >>   1 file changed, 6 insertions(+), 6 deletions(-)
> >>
> >> Index: linux-mtd/drivers/mtd/nand/atmel/nand-controller.c
> >> ===================================================================
> >> --- linux-mtd.orig/drivers/mtd/nand/atmel/nand-controller.c
> >> +++ linux-mtd/drivers/mtd/nand/atmel/nand-controller.c
> >> @@ -2078,11 +2078,11 @@ atmel_hsmc_nand_controller_legacy_init(s
> >>   	}
> >>   
> >>   	nc->irq = of_irq_get(nand_np, 0);
> >> -	if (nc->irq < 0) {
> >> -		ret = nc->irq;
> >> -		if (ret != -EPROBE_DEFER)
> >> +	if (nc->irq <= 0) {
> >> +		if (nc->irq != -EPROBE_DEFER)
> >>   			dev_err(dev, "Failed to get IRQ number (err = %d)\n",
> >> -				ret);
> >> +				nc->irq);  
> > 
> > nc->irq can be 0 here which doesn't make any sense for an error return
> > code, so I'd suggest to still use ret and move the 'ret = nc->irq ?:
> > -ENXIO' line before doing the != -EPROBE_DEFER test.  
> 
>     I tried to keep reporting the original of_irq_get() result. Can change it 
> if you prefer it another way...

Yep, I prefer to have an error core here. No need to send a new
version, I already fixed the patch when applying.

BTW, it's applied to nand/next not nand/fixes because this bug does not
really impact users as long as the 'interrupts' property is correctly
defined in the DT (which AFAICT is the case).

> 
> >> +		ret = nc->irq ?: -ENXIO;  
> > 
> > Why ENXIO and not EINVAL?  
> 
>     That's what platform_get_irq() (another of_irq_get() user) returns if it 
> doesn't find the IRQ resource.

Okay.

> 
> >>   		goto out;
> >>   	}
> >>     
> [...]
> MBR, Sergei