[ldv-project] [PATCH] mtd: rawnand: mtk: avoid underflow in mtk_nfc_nand_chip_init()

[Evgeny Novikov]

Hi Miquel,

17.09.2020, 19:30, "Miquel Raynal" <miquel.raynal at bootlin.com>:
> Hi Evgeny,
>
> Evgeny Novikov <novikov at ispras.ru> wrote on Wed, 16 Sep 2020 22:40:45
> +0300:
>
>>  If of_get_property() will set nsels to negative values the driver may
>
> Is this really a possible case?
>
> Looking at the OF code, I don't think it can ever happen...

I do not know whether this is possible now or it will be possible one day
in the future.

Our tool assumed that this can happen according to the type (int).

Also, I can not find any driver that may suffer from this. Two examples
where negative cases are forbidden intentionally or unintentionally:
- drivers/sbus/char/openprom.c
- drivers/video/fbdev/omap2/omapfb/dss/dsi.c

Best regards,
Evgeny

>>  allocate insufficient memory for chip. Moreover, there may be underflow
>>  for devm_kzalloc(). This can result in various bad consequences later.
>>  The patch causes mtk_nfc_nand_chip_init() to fail for negative values of
>>  nsels.
>>
>>  Found by Linux Driver Verification project (linuxtesting.org).
>>
>>  Signed-off-by: Evgeny Novikov <novikov at ispras.ru>
>>  ---
>>   drivers/mtd/nand/raw/mtk_nand.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>>  diff --git a/drivers/mtd/nand/raw/mtk_nand.c b/drivers/mtd/nand/raw/mtk_nand.c
>>  index ad1b55dab211..df98a2eec240 100644
>>  --- a/drivers/mtd/nand/raw/mtk_nand.c
>>  +++ b/drivers/mtd/nand/raw/mtk_nand.c
>>  @@ -1376,7 +1376,7 @@ static int mtk_nfc_nand_chip_init(struct device *dev, struct mtk_nfc *nfc,
>>                   return -ENODEV;
>>
>>           nsels /= sizeof(u32);
>>  - if (!nsels || nsels > MTK_NAND_MAX_NSELS) {
>>  + if (nsels <= 0 || nsels > MTK_NAND_MAX_NSELS) {
>>                   dev_err(dev, "invalid reg property size %d\n", nsels);
>>                   return -EINVAL;
>>           }
>
> Thanks,
> Miquèl