Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 12 Feb 2020 16:47:56 +0100
From:   Peter Zijlstra <peterz@infradead.org>
To:     Michal Simek <michal.simek@xilinx.com>
Cc:     linux-kernel@vger.kernel.org, monstr@monstr.eu, git@xilinx.com,
        arnd@arndb.de, Stefan Asserhall <stefan.asserhall@xilinx.com>,
        Ingo Molnar <mingo@redhat.com>, Will Deacon <will@kernel.org>
Subject: Re: [PATCH 6/7] microblaze: Implement architecture spinlock
Message-ID: <20200212154756.GY14897@hirez.programming.kicks-ass.net>
References: <cover.1581522136.git.michal.simek@xilinx.com>
 <ed53474e9ca6736353afd10ebe7ea98e4c6c459e.1581522136.git.michal.simek@xilinx.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ed53474e9ca6736353afd10ebe7ea98e4c6c459e.1581522136.git.michal.simek@xilinx.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Feb 12, 2020 at 04:42:28PM +0100, Michal Simek wrote:
> From: Stefan Asserhall <stefan.asserhall@xilinx.com>
> 
> Using exclusive loads/stores to implement spinlocks which can be used on
> SMP systems.
> 
> Signed-off-by: Stefan Asserhall <stefan.asserhall@xilinx.com>
> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
> ---
> 
>  arch/microblaze/include/asm/spinlock.h       | 240 +++++++++++++++++++
>  arch/microblaze/include/asm/spinlock_types.h |  25 ++
>  2 files changed, 265 insertions(+)
>  create mode 100644 arch/microblaze/include/asm/spinlock.h
>  create mode 100644 arch/microblaze/include/asm/spinlock_types.h
> 
> diff --git a/arch/microblaze/include/asm/spinlock.h b/arch/microblaze/include/asm/spinlock.h
> new file mode 100644
> index 000000000000..0199ea9f7f0f
> --- /dev/null
> +++ b/arch/microblaze/include/asm/spinlock.h
> @@ -0,0 +1,240 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright (C) 2013-2020 Xilinx, Inc.
> + */
> +
> +#ifndef _ASM_MICROBLAZE_SPINLOCK_H
> +#define _ASM_MICROBLAZE_SPINLOCK_H
> +
> +/*
> + * Unlocked value: 0
> + * Locked value: 1
> + */
> +#define arch_spin_is_locked(x)	(READ_ONCE((x)->lock) != 0)
> +
> +static inline void arch_spin_lock(arch_spinlock_t *lock)
> +{
> +	unsigned long tmp;
> +
> +	__asm__ __volatile__ (
> +		/* load conditional address in %1 to %0 */
> +		"1:	lwx	 %0, %1, r0;\n"
> +		/* not zero? try again */
> +		"	bnei	%0, 1b;\n"
> +		/* increment lock by 1 */
> +		"	addi	%0, r0, 1;\n"
> +		/* attempt store */
> +		"	swx	%0, %1, r0;\n"
> +		/* checking msr carry flag */
> +		"	addic	%0, r0, 0;\n"
> +		/* store failed (MSR[C] set)? try again */
> +		"	bnei	%0, 1b;\n"
> +		/* Outputs: temp variable for load result */
> +		: "=&r" (tmp)
> +		/* Inputs: lock address */
> +		: "r" (&lock->lock)
> +		: "cc", "memory"
> +	);
> +}

That's a test-and-set spinlock if I read it correctly. Why? that's the
worst possible spinlock implementation possible.