On Wed, Apr 14, 2021 at 02:45:43PM +0200, Peter Zijlstra wrote:
> On Wed, Apr 14, 2021 at 12:16:38PM +0200, Peter Zijlstra wrote:
> > On Wed, Apr 14, 2021 at 11:05:24AM +0200, Peter Zijlstra wrote:
> >
> > > That made me look at the qspinlock code, and queued_spin_*lock() uses
> > > atomic_try_cmpxchg_acquire(), which means any arch that uses qspinlock
> > > and has RCpc atomics will give us massive pain.
> > >
> > > Current archs using qspinlock are: x86, arm64, power, sparc64, mips and
> > > openrisc (WTF?!).
> > >
> > > Of those, x86 and sparc are TSO archs with SC atomics, arm64 has RCsc
> > > atomics, power has RCtso atomics (and is the arch we all hate for having
> > > RCtso locks).
> > >
> > > Now MIPS has all sorts of ill specified barriers, but last time looked
> > > at it it didn't actually use any of that and stuck to using smp_mb(), so
> > > it will have RCsc atomics.
> > >
> > > /me goes look at wth openrisc is.. doesn't even appear to have
> > > asm/barrier.h :-/ Looking at wikipedia it also doesn't appear to
> > > actually have hardware ...
> >
> > FWIW this is broken, anything SMP *MUST* define mb(), at the very least.
>
> As near as I can tell this should do. The arch spec only lists this one
> instruction and the text makes it sound like a completion barrier.
Yes, I will submit this patch.
The l.msync instruction completes all load/store operations.
The l.lwa/l.swa pair (used for xchg/cmpxchg) also implies l.msync.
> ---
> arch/openrisc/include/asm/barrier.h | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
> diff --git a/arch/openrisc/include/asm/barrier.h
> b/arch/openrisc/include/asm/barrier.h
> new file mode 100644
> index 000000000000..7538294721be
> --- /dev/null
> +++ b/arch/openrisc/include/asm/barrier.h
> @@ -0,0 +1,9 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +#ifndef __ASM_BARRIER_H
> +#define __ASM_BARRIER_H
> +
> +#define mb() asm volatile ("l.msync" ::: "memory")
> +
> +#include <asm-generic/barrier.h>
> +
> +#endif /* __ASM_BARRIER_H */
