Re: [PATCH v3 3/4] include/asm-generic/io.h: remove performing pointer arithmetic on a null pointer

["Arnd Bergmann" <arnd@arndb.de>]

On Mon, Dec 5, 2022, at 09:30, Song Chen wrote:
> kernel test robot reports below warnings:
>
>    In file included from kernel/trace/trace_events_synth.c:18:
>    In file included from include/linux/trace_events.h:9:
>    In file included from include/linux/hardirq.h:11:
>    In file included from ./arch/hexagon/include/generated/asm/hardirq.h:1:
>    In file included from include/asm-generic/hardirq.h:17:
>    In file included from include/linux/irq.h:20:
>    In file included from include/linux/io.h:13:
>    In file included from arch/hexagon/include/asm/io.h:334:
>    include/asm-generic/io.h:547:31: warning: performing pointer arithmetic
> 	on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
>            val = __raw_readb(PCI_IOBASE + addr);
>                              ~~~~~~~~~~ ^
>    include/asm-generic/io.h:560:61: warning: performing pointer arithmetic
> 	on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
>            val = __le16_to_cpu((__le16 __force)__raw_readw(PCI_IOBASE + addr));
>                                                            ~~~~~~~~~~ ^
>    include/uapi/linux/byteorder/little_endian.h:37:51: note:
> 		expanded from macro '__le16_to_cpu'
>    #define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
>
> The reason could be constant literal zero converted to any pointer type decays
> into the null pointer constant.
>
> I'm not sure why those warnings are only triggered when building hexagon instead
> of x86 or arm, but anyway, i found a work around:
>
> 	void *pci_iobase = PCI_IOBASE;
> 	val = __raw_readb(pci_iobase + addr);
>
> The pointer is not evaluated at compile time, so the warnings are removed.
>
> Signed-off-by: Song Chen <chensong_2000@189.cn>
> Reported-by: kernel test robot <lkp@intel.com>

The code is still wrong, you just hide the warning, so no, this is
not a correct fix. When PCI_IOBASE is NULL, any call to
inb() etc is a NULL pointer dereference that immediately crashes
the kernel, so the correct solution is to not allow building code
that uses port I/O on kernels that are configured not to
support port I/O.

We have discussed this bit multiple times, and Niklas Schnelle
last posted his series to fix this as an RFC in [1].

      Arnd

[1] https://lore.kernel.org/lkml/20220429135108.2781579-1-schnelle@linux.ibm.com/