Re: Out-of-bounds access in vread_hpet()

[Andy Lutomirski]

On Mon, Jun 9, 2014 at 11:36 PM, Richard Weinberger <richard@xxxxxx> wrote:
> FYI:
> ________________________________________________________________________________________________________
> *** CID 1222081:  Out-of-bounds access  (ARRAY_VS_SINGLETON)
> /arch/x86/vdso/vclock_gettime.c: 38 in vread_hpet()
> 32     #ifdef CONFIG_HPET_TIMER
> 33     extern u8 hpet_page
> 34      __attribute__((visibility("hidden")));
> 35
> 36     static notrace cycle_t vread_hpet(void)
> 37     {
>>>>     CID 1222081:  Out-of-bounds access  (ARRAY_VS_SINGLETON)
>>>>     Using "&hpet_page" as an array.  This might corrupt or misinterpret adjacent memory locations.
> 38      return *(const volatile u32 *)(&hpet_page + HPET_COUNTER);

That's a false positive.  HPET_COUNTER is a compile-time constant
index of a register, and hpet_page is the base address where the thing
is mapped.

I can change the code if there's something non-ugly that makes it
clearer, but ISTM that this particular warning might not make so much
sense when the offset is a real constant like this.

Hmm.  Maybe hpet_page could be redefined as const char hpet_page[PAGE_SIZE].

--Andy
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/