On Thu, Aug 19, 2021 at 04:06:45PM +0200, Peter Zijlstra wrote:

quoted

  * We can implement (1) by checking if we hit zero (ZF=1)
  * We can implement (2) by checking if the new value is < 0 (SF=1).
    We then need to catch the case where the old value was < 0 but the
    new value is 0. I think this is (SF=0 && OF=1).

So maybe the second check is actually SF != OF? I could benefit from some
x86 expertise here, but hopefully you get the idea.

Right, so the first condition is ZF=1, we hit zero.
The second condition is SF=1, the result is negative.

I'm not sure we need OF, if we hit that condition we've already lost.
But it's easy enough to add I suppose.

If we can skip the OF... we can do something like this:

static inline bool refcount_dec_and_test(refcount_t *r)
{
	asm_volatile_goto (LOCK_PREFIX "decl %[var]\n\t"
			   "jz %l[cc_zero]\n\t"
			   "jns 1f\n\t"
			   "ud1 %[var], %%ebx\n\t"
			   "1:"
			   : : [var] "m" (r->refs.counter)
			   : "memory" : cc_zero);

	return false;

cc_zero:
	smp_acquire__after_ctrl_dep();
	return true;
}

where we encode the whole refcount_warn_saturate() thing into UD1. The
first argument is @r and the second argument the REFCOUNT_* thing
encoded in register space.

It would mean adding something 'clever' to the #UD handler that decodes
the trapping instruction and extracts these arguments, but this is the
smallest I could get it.