Re: [PATCH] locking/pvqspinlock/x86: Use _Q_LOCKED_VAL in PV_UNLOCK_ASM macro

[Uros Bizjak]

On Mon, Apr 22, 2024 at 9:45 PM Boqun Feng <boqun.feng@xxxxxxxxx> wrote:

> > diff --git a/arch/x86/include/asm/qspinlock_paravirt.h b/arch/x86/include/asm/qspinlock_paravirt.h
> > index 466af57b8ed6..0a985784be9b 100644
> > --- a/arch/x86/include/asm/qspinlock_paravirt.h
> > +++ b/arch/x86/include/asm/qspinlock_paravirt.h
> > @@ -40,7 +40,7 @@ __PV_CALLEE_SAVE_REGS_THUNK(__pv_queued_spin_unlock_slowpath, ".spinlock.text");
> >  #define PV_UNLOCK_ASM                                                        \
> >       FRAME_BEGIN                                                     \
> >       "push  %rdx\n\t"                                                \
> > -     "mov   $0x1,%eax\n\t"                                           \
> > +     "mov   $" __stringify(_Q_LOCKED_VAL) ",%eax\n\t"                \
> >       "xor   %edx,%edx\n\t"                                           \
> >       LOCK_PREFIX "cmpxchg %dl,(%rdi)\n\t"                            \
> >       "jne   .slowpath\n\t"                                           \
>
> There is also a "cmp $0x1,%al" afterwards.

Not anymore. Please note that the patch is against -tip, where we have:

https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/commit/?id=91095666125a666c8f20c2323b742c53165c0325

Thanks,
Uros.