Re: [PATCH v3 08/12] KVM: arm64: selftests: Add light-weight spinlock support

From: Andrew Jones <drjones@redhat.com>
To: Raghavendra Rao Ananta <rananta@google.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
	Marc Zyngier <maz@kernel.org>, James Morse <james.morse@arm.com>,
	Alexandru Elisei <alexandru.elisei@arm.com>,
	Suzuki K Poulose <suzuki.poulose@arm.com>,
	kvm@vger.kernel.org, Catalin Marinas <catalin.marinas@arm.com>,
	Peter Shier <pshier@google.com>,
	linux-kernel@vger.kernel.org, Will Deacon <will@kernel.org>,
	kvmarm@lists.cs.columbia.edu,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v3 08/12] KVM: arm64: selftests: Add light-weight spinlock support
Date: Fri, 3 Sep 2021 10:25:34 +0200	[thread overview]
Message-ID: <20210903082534.jz3r2defqnrt2ee6@gator.home> (raw)
In-Reply-To: <20210901211412.4171835-9-rananta@google.com>

On Wed, Sep 01, 2021 at 09:14:08PM +0000, Raghavendra Rao Ananta wrote:
> Add a simpler version of spinlock support for ARM64 for
> the guests to use.
> 
> The implementation is loosely based on the spinlock
> implementation in kvm-unit-tests.
> 
> Signed-off-by: Raghavendra Rao Ananta <rananta@google.com>
> ---
>  tools/testing/selftests/kvm/Makefile          |  2 +-
>  .../selftests/kvm/include/aarch64/spinlock.h  | 13 +++++++++
>  .../selftests/kvm/lib/aarch64/spinlock.c      | 27 +++++++++++++++++++
>  3 files changed, 41 insertions(+), 1 deletion(-)
>  create mode 100644 tools/testing/selftests/kvm/include/aarch64/spinlock.h
>  create mode 100644 tools/testing/selftests/kvm/lib/aarch64/spinlock.c
> 
> diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
> index 5d05801ab816..61f0d376af99 100644
> --- a/tools/testing/selftests/kvm/Makefile
> +++ b/tools/testing/selftests/kvm/Makefile
> @@ -35,7 +35,7 @@ endif
>  
>  LIBKVM = lib/assert.c lib/elf.c lib/io.c lib/kvm_util.c lib/rbtree.c lib/sparsebit.c lib/test_util.c lib/guest_modes.c lib/perf_test_util.c
>  LIBKVM_x86_64 = lib/x86_64/apic.c lib/x86_64/processor.c lib/x86_64/vmx.c lib/x86_64/svm.c lib/x86_64/ucall.c lib/x86_64/handlers.S
> -LIBKVM_aarch64 = lib/aarch64/processor.c lib/aarch64/ucall.c lib/aarch64/handlers.S
> +LIBKVM_aarch64 = lib/aarch64/processor.c lib/aarch64/ucall.c lib/aarch64/handlers.S lib/aarch64/spinlock.c
>  LIBKVM_s390x = lib/s390x/processor.c lib/s390x/ucall.c lib/s390x/diag318_test_handler.c
>  
>  TEST_GEN_PROGS_x86_64 = x86_64/cr4_cpuid_sync_test
> diff --git a/tools/testing/selftests/kvm/include/aarch64/spinlock.h b/tools/testing/selftests/kvm/include/aarch64/spinlock.h
> new file mode 100644
> index 000000000000..cf0984106d14
> --- /dev/null
> +++ b/tools/testing/selftests/kvm/include/aarch64/spinlock.h
> @@ -0,0 +1,13 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +
> +#ifndef SELFTEST_KVM_ARM64_SPINLOCK_H
> +#define SELFTEST_KVM_ARM64_SPINLOCK_H
> +
> +struct spinlock {
> +	int v;
> +};
> +
> +extern void spin_lock(struct spinlock *lock);
> +extern void spin_unlock(struct spinlock *lock);
> +
> +#endif /* SELFTEST_KVM_ARM64_SPINLOCK_H */
> diff --git a/tools/testing/selftests/kvm/lib/aarch64/spinlock.c b/tools/testing/selftests/kvm/lib/aarch64/spinlock.c
> new file mode 100644
> index 000000000000..6d66a3dac237
> --- /dev/null
> +++ b/tools/testing/selftests/kvm/lib/aarch64/spinlock.c
> @@ -0,0 +1,27 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * ARM64 Spinlock support
> + */
> +#include <stdint.h>
> +
> +#include "spinlock.h"
> +
> +void spin_lock(struct spinlock *lock)
> +{
> +	uint32_t val, res;
> +
> +	asm volatile(
> +	"1:	ldaxr	%w0, [%2]\n"
> +	"	cbnz	%w0, 1b\n"
> +	"	mov	%w0, #1\n"
> +	"	stxr	%w1, %w0, [%2]\n"
> +	"	cbnz	%w1, 1b\n"
> +	: "=&r" (val), "=&r" (res)
> +	: "r" (&lock->v)
> +	: "memory");
> +}
> +
> +void spin_unlock(struct spinlock *lock)
> +{
> +	asm volatile("stlr wzr, [%0]\n"	: : "r" (&lock->v) : "memory");
> +}
> -- 

Reviewed-by: Andrew Jones <drjones@redhat.com>

It makes sense that the explicit barriers in kvm-unit-tests weren't also
inherited, because we already have the implicit barriers with these ld/st
instruction variants. (I suppose we could improve the kvm-unit-tests
implementation at some point.)

Thanks,
drew