This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [RFC] [PATCH] powerpc: Fix missing barriers in atomic_exchange_and_add_{acq,rel}
- From: Adhemerval Zanella <azanella at linux dot vnet dot ibm dot com>
- To: libc-alpha at sourceware dot org
- Date: Tue, 25 Nov 2014 13:07:41 -0200
- Subject: Re: [RFC] [PATCH] powerpc: Fix missing barriers in atomic_exchange_and_add_{acq,rel}
- Authentication-results: sourceware.org; auth=none
- References: <1413921274 dot 8483 dot 65 dot camel at triegel dot csb>
Hi Torvald,
On 21-10-2014 17:54, Torvald Riegel wrote:
> \
> })
> +#define atomic_exchange_and_add_acq(mem, value) \
> + ({ \
> + __typeof (*(mem)) __result2; \
> + __result2 = atomic_exchange_and_add (mem, value); \
> + atomic_read_barrier (); \
> + __result2;
Although it is not wrong by using a 'atomic_read_barrier' (lwsync), it adds a more
expensive synchronization than required (isync). I would prefer if we use the
already defined __arch_compare_and_exchange_val_[32|64]_[acq|rel] operations on powerpc.
> \
> + })
> +#define atomic_exchange_and_add_rel(mem, value) \
> + ({ \
> + atomic_write_barrier (); \
> + atomic_exchange_and_add (mem, value); \
> + })
>
> #define atomic_increment_val(mem) \
> ({ \