RE: [PATCH 1/2]AArch64 Fix 128-bit sequential consistency atomic operations.

[Tamar Christina <Tamar.Christina@arm.com>]

> -----Original Message-----
> From: Kyrylo Tkachov <Kyrylo.Tkachov@arm.com>
> Sent: Tuesday, July 12, 2022 2:46 PM
> To: Tamar Christina <Tamar.Christina@arm.com>; gcc-patches@gcc.gnu.org
> Cc: nd <nd@arm.com>; Richard Earnshaw <Richard.Earnshaw@arm.com>;
> Marcus Shawcroft <Marcus.Shawcroft@arm.com>; Richard Sandiford
> <Richard.Sandiford@arm.com>
> Subject: RE: [PATCH 1/2]AArch64 Fix 128-bit sequential consistency atomic
> operations.
> 
> Hi Tamar,
> 
> Let me be the latest to offer my apologies for the slow review.
> 
> > -----Original Message-----
> > From: Tamar Christina <Tamar.Christina@arm.com>
> > Sent: Wednesday, June 8, 2022 3:49 PM
> > To: gcc-patches@gcc.gnu.org
> > Cc: nd <nd@arm.com>; Richard Earnshaw <Richard.Earnshaw@arm.com>;
> > Marcus Shawcroft <Marcus.Shawcroft@arm.com>; Kyrylo Tkachov
> > <Kyrylo.Tkachov@arm.com>; Richard Sandiford
> > <Richard.Sandiford@arm.com>
> > Subject: [PATCH 1/2]AArch64 Fix 128-bit sequential consistency atomic
> > operations.
> >
> > Hi All,
> >
> > The AArch64 implementation of 128-bit atomics is broken.
> >
> > For 128-bit atomics we rely on pthread barriers to correct guard the
> > address in the pointer to get correct memory ordering.  However for
> > 128-bit atomics the address under the lock is different from the
> > original pointer.
> >
> > This means that one of the values under the atomic operation is not
> > protected properly and so we fail during when the user has requested
> > sequential consistency as there's no barrier to enforce this
> > requirement.
> >
> > As such users have resorted to adding an
> >
> > #ifdef GCC
> > <emit barrier>
> > #endif
> >
> > around the use of these atomics.
> >
> > This corrects the issue by issuing a barrier only when
> > __ATOMIC_SEQ_CST was requested.  To remedy this performance hit I
> > think we should revisit using a similar approach to out-line-atomics
> > for the 128-bit atomics.
> >
> > Note that I believe I need the empty file due to the include_next
> > chain but I am not entirely sure.  I have hand verified that the
> > barriers are inserted for atomic seq cst.
> >
> > Bootstrapped Regtested on aarch64-none-linux-gnu and no issues.
> >
> > Ok for master? and for backporting to GCC 12, 11 and 10?
> 
> I'll admit I'm not too familiar with the mechanics of libatomic but...
> 
> >
> > Thanks,
> > Tamar
> >
> > libatomic/ChangeLog:
> >
> > 	PR target/102218
> > 	* config/aarch64/aarch64-config.h: New file.
> > 	* config/aarch64/host-config.h: New file.
> >
> > --- inline copy of patch --
> > diff --git a/libatomic/config/aarch64/aarch64-config.h
> > b/libatomic/config/aarch64/aarch64-config.h
> > new file mode 100644
> > index
> > 0000000000000000000000000000000000000000..d3474fa8ff80cb0c3ddbf8c4
> > 8acd931d2339d33d
> > --- /dev/null
> > +++ b/libatomic/config/aarch64/aarch64-config.h
> > @@ -0,0 +1,23 @@
> > +/* Copyright (C) 2022 Free Software Foundation, Inc.
> > +
> > +   This file is part of the GNU Atomic Library (libatomic).
> > +
> > +   Libatomic is free software; you can redistribute it and/or modify it
> > +   under the terms of the GNU General Public License as published by
> > +   the Free Software Foundation; either version 3 of the License, or
> > +   (at your option) any later version.
> > +
> > +   Libatomic is distributed in the hope that it will be useful, but
> > + WITHOUT
> > ANY
> > +   WARRANTY; without even the implied warranty of MERCHANTABILITY or
> > FITNESS
> > +   FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
> > +   more details.
> > +
> > +   Under Section 7 of GPL version 3, you are granted additional
> > +   permissions described in the GCC Runtime Library Exception, version
> > +   3.1, as published by the Free Software Foundation.
> > +
> > +   You should have received a copy of the GNU General Public License and
> > +   a copy of the GCC Runtime Library Exception along with this program;
> > +   see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
> > +   <http://www.gnu.org/licenses/>.  */
> > +
> > diff --git a/libatomic/config/aarch64/host-config.h
> > b/libatomic/config/aarch64/host-config.h
> > new file mode 100644
> > index
> > 0000000000000000000000000000000000000000..f445a47d25ef5cc51cd21670
> > 69500245d07bf1bc
> > --- /dev/null
> > +++ b/libatomic/config/aarch64/host-config.h
> > @@ -0,0 +1,46 @@
> > +/* Copyright (C) 2022 Free Software Foundation, Inc.
> > +
> > +   This file is part of the GNU Atomic Library (libatomic).
> > +
> > +   Libatomic is free software; you can redistribute it and/or modify it
> > +   under the terms of the GNU General Public License as published by
> > +   the Free Software Foundation; either version 3 of the License, or
> > +   (at your option) any later version.
> > +
> > +   Libatomic is distributed in the hope that it will be useful, but
> > + WITHOUT
> > ANY
> > +   WARRANTY; without even the implied warranty of MERCHANTABILITY or
> > FITNESS
> > +   FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
> > +   more details.
> > +
> > +   Under Section 7 of GPL version 3, you are granted additional
> > +   permissions described in the GCC Runtime Library Exception, version
> > +   3.1, as published by the Free Software Foundation.
> > +
> > +   You should have received a copy of the GNU General Public License and
> > +   a copy of the GCC Runtime Library Exception along with this program;
> > +   see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
> > +   <http://www.gnu.org/licenses/>.  */
> > +
> > +/* Avoiding the DMB (or kernel helper) can be a good thing.  */
> > +#define WANT_SPECIALCASE_RELAXED
> > +
> > +/* Glibc, at least, uses acq_rel in its pthread mutex
> > +   implementation.  If the user is asking for seq_cst,
> > +   this is insufficient.  */
> > +
> > +static inline void __attribute__((always_inline, artificial))
> > +pre_seq_barrier(int model) {
> > +  if (model == __ATOMIC_SEQ_CST)
> > +    __atomic_thread_fence (__ATOMIC_SEQ_CST); }
> > +
> > +static inline void __attribute__((always_inline, artificial))
> > +post_seq_barrier(int model) {
> > +  pre_seq_barrier(model);
> > +}
> > +
> > +#define pre_post_seq_barrier 1
> > +
> > +#include_next <host-config.h>
> 
> ... This does looks sensible and similar to what's done on powerpc, which is
> similar to the aarch64 target in this regard.
> However, there is already a host-config.h in config/linux/aarch64/host-
> config.h . Does this file end up including the one in config/linux?
> If so, does this mean that this works correctly (i.e. was tested) for aarch64-
> none-elf as well as Linux?
>

Hi,

We don't build libatomic on any elf platforms.  It has a default unsupported
flag which we don't override.  Indeed we don't produce libatomic.a for elf
and a simple example fails to link as well..

Regards,
Tamar

> Thanks,
> Kyrill
> 
> >
> >
> >
> >
> > --