Re: [PATCH] locking/pvqspinlock: Correct the type of "old" variable in pv_kick_node()

[maobibo]

On 2024/7/22 下午11:13, Waiman Long wrote:
> On 7/22/24 02:54, maobibo wrote:
> > Uros,
> >
> > Sorry for late reply because of weekend time. This modification works 
> > well for me.
> >
> > In later I want to define pv_node::state as int type on LoongArch 
> > because there is only int32/int64 cmpxchg is better supported on the 
> > system, however that is another issue.
>
> I thought about changing pv_node::state to int, but that requires 
> changing the pv_wait() call to use int too. That doesn't work because 
> pv_wait() is also used to monitor the state of the lock byte of the 
> qspinlock. In essence, we can't use pvqspinlock if 8-bit cmpxchg isn't 
> supported.
Hi Longman,

yes, your are right. pv_wait also monitors the state of the qspinlock 
lock byte, so the parameter type about pv_wait must be u8.

LoongArch supports byte cmpxchg, only that word cmpxchg has higher 
efficiency. By the draft kernel compiling test, pvqspinlock can improve 
performance greatly when pCPU is shared by multiple vCPUs.

And thanks for your remaindering.

Regards
Bibo Mao
> Cheers,
> Longman
>
> > Tested-by:Bibo Mao <maobibo@xxxxxxxxxxx>
> >
> > On 2024/7/22 上午12:45, Uros Bizjak wrote:
> > > "enum vcpu_state" is not compatible with "u8" type for all targets,
> > > resulting in:
> > >
> > > error: initialization of 'u8 *' {aka 'unsigned char *'} from 
> > > incompatible pointer type 'enum vcpu_state *'
> > >
> > > for LoongArch. Correct the type of "old" variable to "u8".
> > >
> > > Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > > Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> > > Cc: Will Deacon <will@xxxxxxxxxx>
> > > Cc: Waiman Long <longman@xxxxxxxxxx>
> > > Cc: Boqun Feng <boqun.feng@xxxxxxxxx>
> > > Fixes: fea0e1820b51 ("locking/pvqspinlock: Use try_cmpxchg() in 
> > > qspinlock_paravirt.h")
> > > Reported-by: Bibo Mao <maobibo@xxxxxxxxxxx>
> > > Signed-off-by: Uros Bizjak <ubizjak@xxxxxxxxx>
> > > Closes: 
> > > https://lore.kernel.org/lkml/20240719024010.3296488-1-maobibo@xxxxxxxxxxx/ 
> > >
> > > ---
> > >   kernel/locking/qspinlock_paravirt.h | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/kernel/locking/qspinlock_paravirt.h 
> > > b/kernel/locking/qspinlock_paravirt.h
> > > index f5a36e67b593..ac2e22502741 100644
> > > --- a/kernel/locking/qspinlock_paravirt.h
> > > +++ b/kernel/locking/qspinlock_paravirt.h
> > > @@ -357,7 +357,7 @@ static void pv_wait_node(struct mcs_spinlock 
> > > *node, struct mcs_spinlock *prev)
> > >   static void pv_kick_node(struct qspinlock *lock, struct 
> > > mcs_spinlock *node)
> > >   {
> > >       struct pv_node *pn = (struct pv_node *)node;
> > > -    enum vcpu_state old = vcpu_halted;
> > > +    u8 old = vcpu_halted;
> > >       /*
> > >        * If the vCPU is indeed halted, advance its state to match 
> > > that of
> > >        * pv_wait_node(). If OTOH this fails, the vCPU was running and 
> > > will