On 23 July 2015 at 12:04, Andrew Jones <drjo...@redhat.com> wrote:
> One more observation; to make this work for aarch64 you'll need to modify
> your instruction modification. The add immediate can't be changed with
> +-1 like it can for arm. Please make sure the code works for both, even
> if you need to do
>
> #ifdef __arm__
> asm volatile(arm asm)
> #else
> asm volatile(aarch64 asm)
> #endif
Yes I already did something like this on my original repo (ptr points
to seg0), will update it:
#ifdef ARCH_ARM
(int *)seg0++;
} else {
(int *)seg0--;
}
#elif ARCH_AARCH64
*ptr |= 1 << 11;
*ptr &= ~(1 << 10);
} else {
*ptr |= 1 << 10;
*ptr &= ~(1 << 11);
}
#endif
> Also, if you need a flip-flopping flag, why not use xor instead of
> a counter mod 2?
You are right xor is more straightforward.
Best regards.
