Re: [PATCH 2/2] powercap: arm_scmi: Fix a NULL vs IS_ERR() bug

Dan Carpenter <dan.carpenter@xxxxxxxxxx> · Wed, 6 Jul 2022 18:32:33 +0300

On Wed, Jul 06, 2022 at 04:20:41PM +0100, Cristian Marussi wrote:
> May I ask which static checker you use ? Sparse/smatch and W=1 did not
> spot any of these issues (including other in the series) in my workflow ...
> 

These are Smatch warnings:

$ kchecker drivers/powercap/arm_scmi_powercap.c

Using test/ version of smatch

  CALL    scripts/checksyscalls.sh
  CALL    scripts/atomic/check-atomics.sh
  DESCEND objtool
  CC [M]  drivers/powercap/arm_scmi_powercap.o
  CHECK   drivers/powercap/arm_scmi_powercap.c
drivers/powercap/arm_scmi_powercap.c:429 scmi_powercap_probe() warn: unsigned 'pr->num_zones' is never less than zero.
drivers/powercap/arm_scmi_powercap.c:494 scmi_powercap_probe() error: uninitialized symbol 'ret'.
drivers/powercap/arm_scmi_powercap.c:521 scmi_powercap_init() warn: 'scmi_top_pcntrl' is an error pointer or valid
$

The problem is that the "is an error pointer or valid" requires the
cross function DB to work and that takes forever (over night on my
system).

regards,
dan carpenter