Re: [PATCH] media: isl6421: fix possible uninitialized-variable access in isl6421_set_voltage()

From: Mauro Carvalho Chehab
Date: Fri Nov 19 2021 - 11:28:39 EST


Em Sat, 31 Jul 2021 00:38:06 -0700
Tuo Li <islituo@xxxxxxxxx> escreveu:

> A memory block is allocated through kmalloc(), and its return value is
> assigned to the pointer isl6421. Then isl6421 is assigned to the
> varialbe fe->sec_priv. The function isl6421_set_voltage() is called with
> the argument fe. In this function, fe->sec_priv is assigned to isl6421.
> Thus the pointer isl6421 in the function isl6421_attach() and the function
> isl6421_set_voltage() point to the same memory. However, isl6421->is_off
> is not initialized but it is accessed at line 75:
> if (isl6421->is_off && !is_off)
>
> To fix this possible uninitialized-variable access, isl6421->is_off is
> initialized to false in the function isl6421_attach().
>
> Reported-by: TOTE Robot <oslab@xxxxxxxxxxxxxxx>
> Signed-off-by: Tuo Li <islituo@xxxxxxxxx>
> ---
> drivers/media/dvb-frontends/isl6421.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/media/dvb-frontends/isl6421.c b/drivers/media/dvb-frontends/isl6421.c
> index 43b0dfc6f453..ea101f66ea88 100644
> --- a/drivers/media/dvb-frontends/isl6421.c
> +++ b/drivers/media/dvb-frontends/isl6421.c
> @@ -185,6 +185,7 @@ struct dvb_frontend *isl6421_attach(struct dvb_frontend *fe, struct i2c_adapter
> isl6421->config = ISL6421_ISEL1;
> isl6421->i2c = i2c;
> isl6421->i2c_addr = i2c_addr;
> + isl6421->is_off = false;
> fe->sec_priv = isl6421;

No need. The driver always initialize it:

/* default configuration */
isl6421->config = ISL6421_ISEL1;
isl6421->i2c = i2c;
isl6421->i2c_addr = i2c_addr;
fe->sec_priv = isl6421;

/* bits which should be forced to '1' */
isl6421->override_or = override_set;

/* bits which should be forced to '0' */
isl6421->override_and = ~override_clear;

/* detect if it is present or not */
if (isl6421_set_voltage(fe, SEC_VOLTAGE_OFF)) {
kfree(isl6421);
fe->sec_priv = NULL;
return NULL;
}

isl6421->is_off = true;
^^^^^^^^^^^^^^^^^^^^^^^

(except if errors, in which case the struct is freed)

It sounds that your bot is misleading you to write an useless patch

;-)

Regards,
Mauro