Re: Thermal: int340x_thermal: expose acpi thermal relationship tables

[Jacob Pan <jacob.jun.pan@linux.intel.com>]

On Mon, 13 Oct 2014 20:41:11 +0300
Dan Carpenter <dan.carpenter@oracle.com> wrote:

> Hello Jacob Pan,
> 
> The patch 52b1c69d7e3c: "Thermal: int340x_thermal: expose acpi
> thermal relationship tables" from Sep 3, 2014, leads to the following
> static checker warning:
> 
> 	drivers/thermal/int340x_thermal/acpi_thermal_rel.c:327
> acpi_thermal_rel_ioctl() warn: passing casted pointer '&count' to
> 'acpi_parse_trt()' 64 vs 32.
> 
> drivers/thermal/int340x_thermal/acpi_thermal_rel.c
>    315  static long acpi_thermal_rel_ioctl(struct file *f, unsigned
> int cmd, 316                                     unsigned long __arg)
>    317  {
>    318          int ret = 0;
>    319          unsigned long length = 0;
>    320          unsigned long count = 0;
>    321          char __user *arg = (void __user *)__arg;
>    322          struct trt *trts;
>    323          struct art *arts;
>    324  
>    325          switch (cmd) {
>    326          case ACPI_THERMAL_GET_TRT_COUNT:
>    327                  ret = acpi_parse_trt(acpi_thermal_rel_handle,
> (int *)&count, ^^^^^^^^^^^^
> The reason for the static checker warning is that this will not work
> on big endian 64 bit systems.  It's most likely not a concern but it's
> still pretty ugly.
> 
> You could just make "count" an int and remove the casting.  The copy
> to user will still work if it's an int because the user pointer has
> been casted to unsigned long __user pointer.
you are right, I will make the change. Thanks for pointing this out and
sorry for the late response, I was traveling.

> 
>    328                                  &trts, false);
>    329                  kfree(trts);
>    330                  if (!ret)
>    331                          return put_user(count, (unsigned long
> __user *)__arg); 332                  return ret;
>    333          case ACPI_THERMAL_GET_TRT_LEN:
> 
> regards,
> dan carpenter

[Jacob Pan]