Re: [bug report] firewire: ohci: use guard macro to serialize accesses to phy registers

[Takashi S. <o-t...@sa...>]

Hi,

On Thu, Aug 15, 2024 at 10:08:27AM +0300, Dan Carpenter wrote:
> Hello Takashi Sakamoto,
> 
> Commit 404957c1e207 ("firewire: ohci: use guard macro to serialize
> accesses to phy registers") from Aug 5, 2024 (linux-next), leads to
> the following Smatch static checker warning:
> 
> 	drivers/firewire/ohci.c:1927 initiated_reset()
> 	warn: potential negative cast to bool 'reg'
> 
> drivers/firewire/ohci.c
>     1923 static bool initiated_reset(struct fw_ohci *ohci)
>     1924 {
>     1925         int reg;
>     1926 
> --> 1927         guard(mutex)(&ohci->phy_reg_mutex);
> 
> I'm not sure why Smatch is blaming this line...
> 
>     1928 
>     1929         // Select page 7
>     1930         reg = write_phy_reg(ohci, 7, 0xe0);
>     1931         if (reg < 0)
>     1932                 return reg;
> 
> This is a bool function.  I think these should be return false, instead of
> return negative/true.
> 
>     1933 
>     1934         reg = read_phy_reg(ohci, 8);
>     1935         if (reg < 0)
>     1936                 return reg;
>                          ^^^^^^^^^^
> 
>     1937 
>     1938         // set PMODE bit
>     1939         reg |= 0x40;
>     1940         reg = write_phy_reg(ohci, 8, reg);
>     1941         if (reg < 0)
>     1942                 return reg;
>                          ^^^^^^^^^^
> 
>     1943 
>     1944         // read register 12
>     1945         reg = read_phy_reg(ohci, 12);
>     1946         if (reg < 0)
>     1947                 return reg;
>                          ^^^^^^^^^^
> 
>     1948 
>     1949         // bit 3 indicates "initiated reset"
>     1950         return !!((reg & 0x08) == 0x08);
>     1951 }

Indeed. It is my mistake, and I'll fix it soon.


Thanks

Takashi Sakamoto