Re: [PATCH v4 06/13] riscv: Improve zacas fully-ordered cmpxchg()

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

To: Andrew Jones <ajones@xxxxxxxxxxxxxxxx>
Subject: Re: [PATCH v4 06/13] riscv: Improve zacas fully-ordered cmpxchg()
From: Andrea Parri <parri.andrea@xxxxxxxxx>
Date: Wed, 31 Jul 2024 12:33:54 +0200
Cc: Alexandre Ghiti <alexghiti@xxxxxxxxxxxx>, Jonathan Corbet <corbet@xxxxxxx>, Paul Walmsley <paul.walmsley@xxxxxxxxxx>, Palmer Dabbelt <palmer@xxxxxxxxxxx>, Albert Ou <aou@xxxxxxxxxxxxxxxxx>, Conor Dooley <conor@xxxxxxxxxx>, Rob Herring <robh@xxxxxxxxxx>, Krzysztof Kozlowski <krzk+dt@xxxxxxxxxx>, Nathan Chancellor <nathan@xxxxxxxxxx>, Peter Zijlstra <peterz@xxxxxxxxxxxxx>, Ingo Molnar <mingo@xxxxxxxxxx>, Will Deacon <will@xxxxxxxxxx>, Waiman Long <longman@xxxxxxxxxx>, Boqun Feng <boqun.feng@xxxxxxxxx>, Arnd Bergmann <arnd@xxxxxxxx>, Leonardo Bras <leobras@xxxxxxxxxx>, Guo Ren <guoren@xxxxxxxxxx>, linux-doc@xxxxxxxxxxxxxxx, devicetree@xxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, linux-riscv@xxxxxxxxxxxxxxxxxxx, linux-arch@xxxxxxxxxxxxxxx, Andrea Parri <andrea@xxxxxxxxxxxx>
In-reply-to: <20240731-260cce60e1a6cd06670d1b24@orel>
References: <20240731072405.197046-1-alexghiti@rivosinc.com> <20240731072405.197046-7-alexghiti@rivosinc.com> <20240731-260cce60e1a6cd06670d1b24@orel>

> >   amocas.X.aqrl   a5,a4,(s1)
> 
> We won't get release semantics if the exchange fails. Does that matter?

Conditional (atomic) RMW operations are unordered/relaxed on failure, so
we should be okay here; cf. e.g. Documentation/atomic_t.txt.

  Andrea

References:
- [PATCH v4 00/13] Zacas/Zabha support and qspinlocks
  - From: Alexandre Ghiti
- [PATCH v4 06/13] riscv: Improve zacas fully-ordered cmpxchg()
  - From: Alexandre Ghiti
- Re: [PATCH v4 06/13] riscv: Improve zacas fully-ordered cmpxchg()
  - From: Andrew Jones

Prev by Date: Re: [PATCH] ARM: support PREEMPT_DYNAMIC
Next by Date: Re: [PATCH 2/2] mm: tlb: add tlb swap entries batch async release
Previous by thread: Re: [PATCH v4 06/13] riscv: Improve zacas fully-ordered cmpxchg()
Next by thread: [PATCH v4 07/13] riscv: Implement arch_cmpxchg128() using Zacas
Index(es):
- Date
- Thread

[Index of Archives] [Linux Kernel] [Kernel Newbies] [x86 Platform Driver] [Netdev] [Linux Wireless] [Netfilter] [Bugtraq] [Linux Filesystems] [Yosemite Discussion] [MIPS Linux] [ARM Linux] [Linux Security] [Linux RAID] [Samba] [Device Mapper]