Re: [RFC PATCH] locking/atomic: arch/mips: Fix atomic{_64,}_sub_if_positive

[Thomas Bogendoerfer]

On Thu, Jul 29, 2021 at 08:31:11PM +0800, hev wrote:
> Hi, Thomas,
> 
> On Thu, Jul 29, 2021 at 5:53 PM Thomas Bogendoerfer
> <tsbogend@xxxxxxxxxxxxxxxx> wrote:
> >
> > On Thu, Jul 29, 2021 at 04:25:49PM +0800, Rui Wang wrote:
> > > This looks like a typo and that caused atomic64 test failed.
> > >
> > > Signed-off-by: Rui Wang <wangrui@xxxxxxxxxxx>
> > > Signed-off-by: hev <r@xxxxxx>
> > > ---
> > >  arch/mips/include/asm/atomic.h | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/arch/mips/include/asm/atomic.h b/arch/mips/include/asm/atomic.h
> > > index 95e1f7f3597f..a0b9e7c1e4fc 100644
> > > --- a/arch/mips/include/asm/atomic.h
> > > +++ b/arch/mips/include/asm/atomic.h
> > > @@ -206,7 +206,7 @@ ATOMIC_OPS(atomic64, xor, s64, ^=, xor, lld, scd)
> > >   * The function returns the old value of @v minus @i.
> > >   */
> > >  #define ATOMIC_SIP_OP(pfx, type, op, ll, sc)                         \
> > > -static __inline__ int arch_##pfx##_sub_if_positive(type i, pfx##_t * v)      \
> > > +static __inline__ type arch_##pfx##_sub_if_positive(type i, pfx##_t * v)     \
> > >  {                                                                    \
> > >       type temp, result;                                              \
> > >                                                                       \
> >
> > sub_if_postive looks unused to me. Could you send a patch removing it
> > instead ? riscv also has a sub_if_positive implementation, which looks
> > unused.
> I found atomic{_64,}_dec_if_postive is based on sub_if_postive, and
> used in many places:
> 
> kernel/kmod.c:    if (atomic_dec_if_positive(&kmod_concurrent_max) < 0) {
> kernel/kmod.c:
> atomic_dec_if_positive(&kmod_concurrent_max) >= 0,
> kernel/module.c:        ret = atomic_dec_if_positive(&module->refcnt);
> ...
> drivers/net/ethernet/mellanox/mlx5/core/eswitch.c:
> atomic64_dec_if_positive(&esw->user_count);
> drivers/net/netdevsim/fib.c:        atomic64_dec_if_positive(&entry->num);
> drivers/net/netdevsim/fib.c:        atomic64_dec_if_positive(&entry->num);
> 
> Are you sure to remove it?

in that case, let's keep it. There is a fallback for atomic_dec_if_positive,
that's why I didn't notice where it is used.

Thomas.

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]