Re: [PATCH] a new lock algorithm with lock value and CS counts in the same atomic int

[Jens Gustedt <jens.gustedt@inria.fr>]

[-- Attachment #1: Type: text/plain, Size: 1368 bytes --]

Hello Alexander,

On Sun, 18 Jun 2017 13:01:18 +0300 (MSK) Alexander Monakov
<amonakov@ispras.ru> wrote:

> On Fri, 16 Jun 2017, Jens Gustedt wrote:
> >  void __lock(volatile int *l)
> >  {
> > -	if (libc.threads_minus_1)
> > -		while (a_swap(l, 1)) __wait(l, l+1, 1, 1);
> > +	/* This test is mostly useless, now. Leaving it to the
> > first CAS is
> > +	   probably just as efficient. */
> > +	if (libc.threads_minus_1) {  
> [...]
> >  void __unlock(volatile int *l)
> >  {
> > -	if (l[0]) {
> > -		a_store(l, 0);
> > -		if (l[1]) __wake(l, 1, 1);
> > +	if (a_fetch_add(l, INT_MAX) != -INT_MAX) {
> > +		__syscall(SYS_futex, l, FUTEX_WAKE|FUTEX_PRIVATE,
> > 1); }
> >  }  
> 
> This looks wrong in single-threaded case, __lock doesn't touch the
> lock, but __unlock modifies it unconditionally.

Right, probably there should be the same test as for the lock case. Or
we should drop that test all along. I don't think that it still serves
much purpose here. This is just trading one memory access against
another.


Thanks
Jens

-- 
:: INRIA Nancy Grand Est ::: Camus ::::::: ICube/ICPS :::
:: ::::::::::::::: office Strasbourg : +33 368854536   ::
:: :::::::::::::::::::::: gsm France : +33 651400183   ::
:: ::::::::::::::: gsm international : +49 15737185122 ::
:: http://icube-icps.unistra.fr/index.php/Jens_Gustedt ::

[-- Attachment #2: Digitale Signatur von OpenPGP --]
[-- Type: application/pgp-signature, Size: 195 bytes --]