Re: [RFC] Add implicit barriers to irqsave/restore class of functions

From: Vineet Gupta
Date: Wed Apr 03 2013 - 09:29:54 EST


Hi Christian,

On 04/03/2013 06:40 PM, Christian Ruppert wrote:
> This patch adds implicit memory barriers to irqsave/restore functions of
> the ARC architecture port in line with what is done in other architectures.
>
> It seems to fix several seemingly unrelated issues in our platform but for
> the moment it is insufficiently tested (and might even be incomplete).
> Please comment.

I think this is not needed. Semantically we need barrier for spinlocks, not irq
save/restore - although spin locks are one of the primary users of irq
save/restore API.

So repeating what Thomas already said, the barrier already exists for
PREEMPT_COUNT, we need to make sure they are present for !PREEMPT_COUNT.

-Vineet

> Signed-off-by: Christian Ruppert <christian.ruppert@xxxxxxxxxx>
> ---
> arch/arc/include/asm/irqflags.h | 8 ++++----
> 1 files changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/arch/arc/include/asm/irqflags.h b/arch/arc/include/asm/irqflags.h
> index ccd8480..c8147d1 100644
> --- a/arch/arc/include/asm/irqflags.h
> +++ b/arch/arc/include/asm/irqflags.h
> @@ -39,7 +39,7 @@ static inline long arch_local_irq_save(void)
> " flag.nz %0 \n"
> : "=r"(temp), "=r"(flags)
> : "n"((STATUS_E1_MASK | STATUS_E2_MASK))
> - : "cc");
> + : "memory", "cc");
>
> return flags;
> }
> @@ -53,7 +53,7 @@ static inline void arch_local_irq_restore(unsigned long flags)
> __asm__ __volatile__(
> " flag %0 \n"
> :
> - : "r"(flags));
> + : "r"(flags) : "memory");
> }
>
> /*
> @@ -73,7 +73,7 @@ static inline void arch_local_irq_disable(void)
> " and %0, %0, %1 \n"
> " flag %0 \n"
> : "=&r"(temp)
> - : "n"(~(STATUS_E1_MASK | STATUS_E2_MASK)));
> + : "n"(~(STATUS_E1_MASK | STATUS_E2_MASK)) : "memory");
> }
>
> /*
> @@ -85,7 +85,7 @@ static inline long arch_local_save_flags(void)
>
> __asm__ __volatile__(
> " lr %0, [status32] \n"
> - : "=&r"(temp));
> + : "=&r"(temp) : : "memory");
>
> return temp;
> }

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/